feat(v2 upgrade): support engine live upgrade #241
Conversation
WalkthroughThe changes in this pull request introduce a new field, Changes
Assessment against linked issues
Possibly related PRs
Suggested reviewers
📜 Recent review detailsConfiguration used: .coderabbit.yaml 📒 Files selected for processing (1)
✅ Files skipped from review due to trivial changes (1)
Thank you for using CodeRabbit. We offer it for free to the OSS community and would appreciate your support in helping us grow. If you find it useful, would you consider giving us a shout-out on your favorite social media? 🪧 TipsChatThere are 3 ways to chat with CodeRabbit:
Note: Be mindful of the bot's finite context window. It's strongly recommended to break down tasks such as reading entire modules into smaller chunks. For a focused discussion, use review comments to chat about specific files and their changes, instead of using the PR comments. CodeRabbit Commands (Invoked using PR comments)
Other keywords and placeholders
Documentation and Community
|
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 2
🧹 Outside diff range and nitpick comments (4)
pkg/api/types.go (1)
132-132: LGTM! Consider adding field documentation.The new
StandbyTargetPortfield is well-positioned and follows the codebase conventions. Consider adding a comment to document its purpose in the context of live upgrades.+ // StandbyTargetPort is used during live upgrades to maintain service availability StandbyTargetPort int32 `json:"standby_target_port"`pkg/spdk/engine.go (3)
252-252: Possible misuse of variable in log messageAt line 252, the log statement uses
e.ReplicaModeMapto display the replicas being connected. However,e.ReplicaModeMapmight not represent the list of replicas intended for logging.Consider using
replicaBdevList, which contains the list of replica block devices:- e.log.Infof("Connecting all available replicas %+v, then launching raid during engine creation", e.ReplicaModeMap) + e.log.Infof("Connecting all available replicas %+v, then launching raid during engine creation", replicaBdevList)This change will provide more accurate logging information about the replicas being connected.
685-688: Remove commented-out code to improve code clarityThe commented-out code at lines 685-688 is not used and can be removed to enhance readability and maintainability.
Apply this diff to remove the unused code:
- // podIP, err := commonnet.GetIPForPod() - // if err != nil { - // return err - // }If this code is needed for future development, consider adding a comment explaining its purpose or moving it to a separate location.
Line range hint
1881-1881: Typographical error in function namecloseRplicaClientsThere is a typo in the function name
closeRplicaClients. It should becloseReplicaClientsfor clarity and consistency.Apply this diff to correct the function name:
- func (e *Engine) closeRplicaClients(replicaClients map[string]*client.SPDKClient) { + func (e *Engine) closeReplicaClients(replicaClients map[string]*client.SPDKClient) {Make sure to update all references to this function accordingly.
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
⛔ Files ignored due to path filters (2)
go.modis excluded by!go.modvendor/modules.txtis excluded by!vendor/**
📒 Files selected for processing (2)
pkg/api/types.go(2 hunks)pkg/spdk/engine.go(15 hunks)
🧰 Additional context used
🪛 golangci-lint
pkg/spdk/engine.go
315-315: expected declaration, found '<<'
(typecheck)
🔇 Additional comments (5)
pkg/api/types.go (2)
153-153: LGTM! Field mapping is correct.
The StandbyTargetPort field is properly mapped from the protobuf struct to the Engine type.
Line range hint 132-153: Verify protobuf definition exists.
Ensure that the standby_target_port field is properly defined in the SPDK service protobuf files.
pkg/spdk/engine.go (3)
45-45: New field StandbyTargetPort added to Engine struct
The addition of the StandbyTargetPort field to the Engine struct is appropriate. Ensure that this new field is properly initialized and consistently used throughout the codebase to avoid any unexpected behavior.
108-110: Function isNewEngine correctly determines if the engine is new
The isNewEngine function effectively checks if the IP field is empty to determine if the engine instance is new. This is a simple and efficient approach.
618-618: Field StandbyTargetPort included in getWithoutLock
The StandbyTargetPort field is correctly added to the Engine object returned by the getWithoutLock method. This ensures that the field will be included in any serialized representations of the engine's state.
derekbit
force-pushed
the
v2-control-upgrade
branch
from
668e4d2
to
84a18b8
Compare
derekbit
changed the title
derekbit
force-pushed
the
v2-control-upgrade
branch
from
84a18b8
to
c80ba34
Compare
There was a problem hiding this comment.
Caution
Inline review comments failed to post. This is likely due to GitHub's limits when posting large numbers of comments.
Actionable comments posted: 1
🧹 Outside diff range and nitpick comments (1)
pkg/spdk/engine.go (1)
Line range hint
2173-2224: Document target port management logicThe target switching logic correctly handles the port assignments, but would benefit from a comment explaining when and why
StandbyTargetPortis reset to 0.Add a comment before the condition:
+ // Reset StandbyTargetPort when switching to the pod's IP as the target, + // since this node is now the primary target if targetIP == podIP { e.TargetPort = targetPort e.StandbyTargetPort = 0 }
🛑 Comments failed to post (1)
pkg/spdk/engine.go (1)
315-315:
⚠️ Potential issueCritical: Resolve merge conflict
There is an unresolved merge conflict marker at line 315 (
<<<<<<< HEAD). This needs to be resolved before the code can be merged.Please resolve the merge conflict by:
- Running
git mergeorgit rebaseto update your branch- Resolving the conflicts by choosing the appropriate code
- Removing all conflict markers (
<<<<<<<,=======,>>>>>>>)🧰 Tools
🪛 golangci-lint
315-315: expected declaration, found '<<'
(typecheck)
derekbit
force-pushed
the
v2-control-upgrade
branch
2 times, most recently
from
b15f126
to
7f0d7e1
Compare
There was a problem hiding this comment.
Caution
Inline review comments failed to post. This is likely due to GitHub's limits when posting large numbers of comments.
Actionable comments posted: 1
🧹 Outside diff range and nitpick comments (1)
pkg/spdk/engine.go (1)
Line range hint
4-4: Consider adding tests for live upgrade scenariosWhile the implementation looks solid, consider adding comprehensive test coverage for:
- Standby target creation and cleanup
- Target switchover scenarios
- Error cases during live upgrades
Would you like me to help generate test cases for these scenarios?
🛑 Comments failed to post (1)
pkg/spdk/engine.go (1)
216-218:
⚠️ Potential issueAdd error handling for BdevRaidGet
The empty branch after
BdevRaidGeterror check could lead to silent failures. Consider adding appropriate error handling.Apply this diff to add error handling:
if targetCreationRequired { _, err := spdkClient.BdevRaidGet(e.Name, 0) if err != nil { + if !jsonrpc.IsJSONRPCRespErrorNoSuchDevice(err) { + return nil, errors.Wrapf(err, "failed to get raid bdev %s", e.Name) + } }📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.if targetCreationRequired { _, err := spdkClient.BdevRaidGet(e.Name, 0) if err != nil { if !jsonrpc.IsJSONRPCRespErrorNoSuchDevice(err) { return nil, errors.Wrapf(err, "failed to get raid bdev %s", e.Name) } }🧰 Tools
🪛 golangci-lint
218-218: SA9003: empty branch
(staticcheck)
derekbit
force-pushed
the
v2-control-upgrade
branch
from
7f0d7e1
to
5b4f847
Compare
There was a problem hiding this comment.
Actionable comments posted: 0
🧹 Outside diff range and nitpick comments (3)
pkg/spdk/engine.go (3)
176-202: Consider simplifying the initialization logicThe current implementation has nested conditions that could be simplified for better readability and maintainability. Consider extracting the logic into separate helper functions.
Example refactor:
- if podIP == initiatorIP && podIP == targetIP { - if e.Port == 0 && e.TargetPort == 0 { - e.log.Info("Creating both initiator and target instances") - initiatorCreationRequired = true - targetCreationRequired = true - } else if e.Port != 0 && e.TargetPort == 0 { - e.log.Info("Creating a target instance") - targetCreationRequired = true - if e.StandbyTargetPort != 0 { - e.log.Warnf("Standby target instance with port %v is already created, will skip the target creation", e.StandbyTargetPort) - return e.getWithoutLock(), nil - } - } else { - return nil, fmt.Errorf("invalid initiator and target address for engine %s creation", e.Name) - } + creationMode := determineCreationMode(podIP, initiatorIP, targetIP, e.Port, e.TargetPort) + switch creationMode { + case createBoth: + e.log.Info("Creating both initiator and target instances") + initiatorCreationRequired = true + targetCreationRequired = true + case createTargetOnly: + e.log.Info("Creating a target instance") + targetCreationRequired = true + if e.StandbyTargetPort != 0 { + e.log.Warnf("Standby target instance with port %v is already created, will skip the target creation", e.StandbyTargetPort) + return e.getWithoutLock(), nil + } + case createInitiatorOnly: + e.log.Info("Creating an initiator instance") + initiatorCreationRequired = true + default: + return nil, fmt.Errorf("invalid initiator and target address for engine %s creation", e.Name) + }
397-399: Add documentation for standby target creation conditionThe condition for standby target creation could benefit from a comment explaining when and why it's needed.
+ // Create a standby target if we have an active initiator (e.Port != 0) + // but no active target (e.TargetPort == 0) standbyTargetCreationRequired := false if e.Port != 0 && e.TargetPort == 0 { standbyTargetCreationRequired = true }
2169-2173: Enhance error handling for pod IP operationsThe pod IP retrieval and port updates are critical for live upgrades. Consider adding more detailed error messages and logging.
podIP, err := commonnet.GetIPForPod() if err != nil { - return err + return errors.Wrapf(err, "failed to get pod IP for engine %s target switchover", e.Name) } if targetIP == podIP { + e.log.Infof("Target IP matches pod IP, updating ports: target=%d, standby=0", targetPort) e.TargetPort = targetPort e.StandbyTargetPort = 0 }Also applies to: 2216-2220
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
📒 Files selected for processing (2)
pkg/api/types.go(2 hunks)pkg/spdk/engine.go(14 hunks)
🚧 Files skipped from review as they are similar to previous changes (1)
- pkg/api/types.go
🔇 Additional comments (5)
pkg/spdk/engine.go (5)
45-45: LGTM: Added StandbyTargetPort field for standby target management
The new field is well-integrated with the existing port fields and supports the live upgrade functionality.
108-110: LGTM: Clear and focused helper function
The function provides a clean way to determine if an engine is new based on its IP state.
204-207: LGTM: Clear initialization of engine IPs
The initialization logic correctly sets the IP values for new engines using the helper function.
476-485: LGTM: Clear port assignment logic
The code correctly handles port assignment for both initiator and target instances, including the standby target scenario.
Line range hint 2169-2220: Verify concurrent access during target switchover
The target switchover process involves multiple steps and state changes. Ensure there are no race conditions when:
- Multiple goroutines attempt switchover simultaneously
- A switchover occurs during engine creation/deletion
- Pod IP changes during switchover
✅ Verification successful
Let me generate additional verification scripts to check for proper locking mechanisms and concurrent access patterns.
Based on the verification results, I can now generate the final response:
Proper locking mechanisms are in place during target switchover
The code correctly handles concurrent access during target switchover:
SwitchOverTargetmethod acquires the engine lock before state modifications- Related operations like
Create,Delete, andDeleteTargetalso use proper locking - State changes (TargetPort, StandbyTargetPort) are protected by the mutex
The original concerns about race conditions are addressed by the existing implementation.
🏁 Scripts executed
The following scripts were executed for the analysis:
Script:
#!/bin/bash
# Check for potential concurrent access patterns
ast-grep --pattern $'func \(.*\) SwitchOverTarget\(.*\) {
$$$
e.Lock()
$$$
}'
# Check for other methods that might interact with target state
rg -A 5 "e\.(TargetPort|StandbyTargetPort)"
Length of output: 6650
Script:
#!/bin/bash
# Check for all methods that modify engine state and their locking patterns
ast-grep --pattern $'func \(e \*Engine\) $_($$) {
$$$
e.Lock()
$$$
}'
# Check for methods that might be called during switchover
rg -A 5 "func.*Engine.*\b(Create|Delete|Switch|Connect|Disconnect|Reload)" pkg/spdk/
Length of output: 1794
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #241 +/- ##
========================================
+ Coverage 8.42% 8.87% +0.44%
========================================
Files 21 21
Lines 7158 7225 +67
========================================
+ Hits 603 641 +38
- Misses 6475 6501 +26
- Partials 80 83 +3
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. |
derekbit
force-pushed
the
v2-control-upgrade
branch
from
0aae707
to
3c6c4a3
Compare
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Outside diff range and nitpick comments (4)
pkg/spdk/engine_test.go (1)
98-115: Improve test execution logic and error handling.The test execution can be improved in several ways:
- Use test case name in log message instead of array index:
-c.Logf("testing TestCheckInitiatorAndTargetCreationRequirements.%v", testName) +c.Logf("testing TestCheckInitiatorAndTargetCreationRequirements: %s", testCase.name)
- Consider extracting engine creation to a helper function for better reusability:
func createTestEngine(port, targetPort, standbyTargetPort int32, name string) *Engine { return &Engine{ Port: port, TargetPort: targetPort, StandbyTargetPort: standbyTargetPort, Name: name, log: logrus.New(), } }
- Add validation of log messages to ensure proper error logging.
pkg/spdk/engine.go (3)
Line range hint
398-498: Consider enhancing error handling in deferred functionThe deferred function at line 429 contains complex logic for initiator assignment. Consider extracting this into a separate helper function for better maintainability and error handling.
Consider refactoring like this:
+ func (e *Engine) assignInitiator(initiator *nvme.Initiator, dmDeviceBusy bool, standbyTargetCreationRequired bool) { + if !standbyTargetCreationRequired { + e.initiator = initiator + e.dmDeviceBusy = dmDeviceBusy + e.Endpoint = initiator.GetEndpoint() + e.log = e.log.WithFields(logrus.Fields{ + "endpoint": e.Endpoint, + "port": e.Port, + "targetPort": e.TargetPort, + }) + } + e.log.Infof("Finished handling frontend for engine: %+v", e) + } defer func() { if err == nil { - if !standbyTargetCreationRequired { - e.initiator = initiator - e.dmDeviceBusy = dmDeviceBusy - ... - } - e.log.Infof("Finished handling frontend for engine: %+v", e) + e.assignInitiator(initiator, dmDeviceBusy, standbyTargetCreationRequired) } }()
208-214: Consider standardizing error handling patternsWhile error handling is generally good, consider standardizing the approach across the codebase. For example, some errors are wrapped with context while others are returned directly.
Consider consistently using error wrapping with context:
- if err != nil { - return nil, err - } + if err != nil { + return nil, errors.Wrapf(err, "failed to get pod IP during engine creation") + }Also applies to: 2181-2184
261-262: Address TODO comment about loggingThe TODO comment indicates that log messages need improvement. Consider enhancing the log message to include more context about the operation being performed.
Consider improving the log message:
- // TODO: improve the log message - e.log.Infof("Connecting all available replicas %+v, then launching raid during engine creation", e.ReplicaStatusMap) + e.log.Infof("Connecting available replicas for engine %s (replicas: %+v) and initializing RAID configuration", e.Name, e.ReplicaStatusMap)
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
📒 Files selected for processing (3)
pkg/api/types.go(2 hunks)pkg/spdk/engine.go(14 hunks)pkg/spdk/engine_test.go(1 hunks)
🚧 Files skipped from review as they are similar to previous changes (1)
- pkg/api/types.go
🔇 Additional comments (7)
pkg/spdk/engine_test.go (2)
1-10: LGTM!
The imports are clean and necessary. The test file is properly structured using the check.v1 testing framework.
1-115: Verify test coverage for live upgrade functionality.
The test file introduces basic validation for the standby target port, which is a step towards supporting live upgrades. However, to ensure robust support for engine live upgrades (as per PR objectives), consider adding more comprehensive test scenarios that validate:
- The complete upgrade workflow
- State transitions during upgrade
- Error recovery scenarios
- Upgrade rollback cases
✅ Verification successful
Based on the code inspection results, I can now generate the final response:
Test coverage appears sufficient for the current implementation scope
The test coverage aligns well with the current implementation, which focuses on the standby port allocation and basic validation aspects of the upgrade functionality. The test cases appropriately verify:
- Standby target port validation
- Port allocation scenarios
- Creation requirements for different network configurations
While the initial review suggested adding more test scenarios, the current implementation scope is primarily focused on the port management aspect of upgrades, which is adequately tested. The suggested additional test scenarios would be premature as the complete upgrade workflow implementation is not yet present in the codebase.
🏁 Scripts executed
The following scripts were executed for the analysis:
Script:
#!/bin/bash
# Description: Check test coverage for upgrade-related code
# Look for other upgrade-related tests
rg -l "upgrade|standby" --type go
# Check for upgrade-related functions in the engine implementation
ast-grep --pattern 'func $_($$) {
$$$
standby
$$$
}'
Length of output: 633
Script:
#!/bin/bash
# Let's examine the engine implementation and related test files more closely
# Check engine.go for standby-related functions and their implementation
rg "standby|Standby" pkg/spdk/engine.go -B 2 -A 2
# Look for upgrade-related functions in engine.go
rg "upgrade|Upgrade" pkg/spdk/engine.go -B 2 -A 2
# Check the API types related to upgrade
rg "upgrade|Upgrade" pkg/api/types.go -B 2 -A 2
Length of output: 2242
pkg/spdk/engine.go (5)
37-49: LGTM: Engine struct changes
The addition of StandbyTargetPort field is well-placed and consistent with other port-related fields in the struct.
108-110: LGTM: Clear and focused helper function
The isNewEngine function provides a clean abstraction for checking engine initialization state.
Line range hint 2181-2232: LGTM: Well-structured target switchover logic
The changes properly handle the standby target port during switchover operations, with appropriate error handling and state management.
112-142:
Fix redundant condition check
The function contains a redundant check for e.StandbyTargetPort != 0 at line 123, which is unnecessary since it's already within a block that checks for e.Port != 0 && e.TargetPort == 0.
Apply this diff to fix the redundant condition:
if e.Port != 0 && e.TargetPort == 0 {
e.log.Info("Creating a target instance")
- if e.StandbyTargetPort != 0 {
e.log.Warnf("Standby target instance with port %v is already created, will skip the target creation", e.StandbyTargetPort)
- } else {
targetCreationRequired = true
- }
}Likely invalid or redundant comment.
112-142: Ensure test coverage for new functionality
The new checkInitiatorAndTargetCreationRequirements function contains complex logic for standby target handling. Please ensure adequate test coverage is added.
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Outside diff range and nitpick comments (1)
pkg/spdk_test.go (1)
Line range hint
225-322: Document the new upgrade mechanism architectureThe changes suggest a significant shift in how engine upgrades are handled, moving from a parameter-based approach to potentially a more sophisticated mechanism. To ensure maintainability and clarity:
- Consider adding documentation that explains the new upgrade mechanism
- Update or create integration tests that specifically validate the live upgrade scenarios
- Consider adding a migration guide for users who might be affected by this architectural change
Would you like assistance in creating:
- Documentation template for the new upgrade mechanism?
- Integration test scenarios for live upgrades?
- Migration guide for existing users?
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
📒 Files selected for processing (2)
pkg/client/client.go(1 hunks)pkg/spdk_test.go(2 hunks)
🔇 Additional comments (3)
pkg/spdk_test.go (2)
225-225: Verify the removal of upgradeRequired parameter
The EngineCreate call has been updated to remove the upgradeRequired parameter. This change should be verified against the engine creation logic to ensure it aligns with the new live upgrade support feature.
#!/bin/bash
# Description: Verify the changes in EngineCreate signature across the codebase
# Check for any remaining references to upgradeRequired in engine creation
rg -l "EngineCreate.*upgradeRequired"
# Check the updated EngineCreate method signature
ast-grep --pattern 'func $_EngineCreate($_) {
$$$
}'250-322:
Clarify the status of the commented test implementation
The TestSPDKEngineCreateWithUpgradeRequired test appears to contain important validations for the engine upgrade functionality. If this test is being replaced or modified as part of the live upgrade feature:
- Please clarify if this test will be replaced with new test cases
- Ensure the new test coverage includes validation of both initiator and target properties
- Consider preserving the valuable assertions in any new test implementation
pkg/client/client.go (1)
Line range hint 474-492: LGTM: Method implementation remains consistent
The implementation maintains proper parameter validation, error handling, and gRPC timeout management while removing the upgrade-related parameter.
There was a problem hiding this comment.
Actionable comments posted: 0
🧹 Outside diff range and nitpick comments (1)
pkg/spdk/engine.go (1)
112-142: Improve error messages and add test coverageThe function has clear logic but could benefit from:
- More descriptive error messages that explain why the addresses are invalid
- Unit tests to verify the complex branching logic
Consider improving the error messages:
-err = fmt.Errorf("invalid initiator and target address for engine %s creation", e.Name) +err = fmt.Errorf("invalid initiator address %s and target address %s for engine %s creation: addresses must match pod IP %s", initiatorIP, targetIP, e.Name, podIP)🧰 Tools
🪛 GitHub Check: codecov/patch
[warning] 128-130: pkg/spdk/engine.go#L128-L130
Added lines #L128 - L130 were not covered by tests
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
📒 Files selected for processing (1)
pkg/spdk/engine.go(14 hunks)
🧰 Additional context used
🪛 GitHub Check: codecov/patch
pkg/spdk/engine.go
[warning] 108-109: pkg/spdk/engine.go#L108-L109
Added lines #L108 - L109 were not covered by tests
[warning] 128-130: pkg/spdk/engine.go#L128-L130
Added lines #L128 - L130 were not covered by tests
[warning] 208-214: pkg/spdk/engine.go#L208-L214
Added lines #L208 - L214 were not covered by tests
[warning] 216-220: pkg/spdk/engine.go#L216-L220
Added lines #L216 - L220 were not covered by tests
[warning] 223-226: pkg/spdk/engine.go#L223-L226
Added lines #L223 - L226 were not covered by tests
[warning] 228-231: pkg/spdk/engine.go#L228-L231
Added lines #L228 - L231 were not covered by tests
[warning] 263-264: pkg/spdk/engine.go#L263-L264
Added lines #L263 - L264 were not covered by tests
[warning] 269-269: pkg/spdk/engine.go#L269
Added line #L269 was not covered by tests
[warning] 271-271: pkg/spdk/engine.go#L271
Added line #L271 was not covered by tests
[warning] 282-285: pkg/spdk/engine.go#L282-L285
Added lines #L282 - L285 were not covered by tests
[warning] 292-292: pkg/spdk/engine.go#L292
Added line #L292 was not covered by tests
[warning] 302-303: pkg/spdk/engine.go#L302-L303
Added lines #L302 - L303 were not covered by tests
[warning] 306-315: pkg/spdk/engine.go#L306-L315
Added lines #L306 - L315 were not covered by tests
[warning] 401-401: pkg/spdk/engine.go#L401
Added line #L401 was not covered by tests
[warning] 411-413: pkg/spdk/engine.go#L411-L413
Added lines #L411 - L413 were not covered by tests
[warning] 422-429: pkg/spdk/engine.go#L422-L429
Added lines #L422 - L429 were not covered by tests
[warning] 431-442: pkg/spdk/engine.go#L431-L442
Added lines #L431 - L442 were not covered by tests
[warning] 444-444: pkg/spdk/engine.go#L444
Added line #L444 was not covered by tests
[warning] 448-462: pkg/spdk/engine.go#L448-L462
Added lines #L448 - L462 were not covered by tests
[warning] 464-466: pkg/spdk/engine.go#L464-L466
Added lines #L464 - L466 were not covered by tests
[warning] 468-468: pkg/spdk/engine.go#L468
Added line #L468 was not covered by tests
[warning] 471-473: pkg/spdk/engine.go#L471-L473
Added lines #L471 - L473 were not covered by tests
[warning] 476-476: pkg/spdk/engine.go#L476
Added line #L476 was not covered by tests
[warning] 479-482: pkg/spdk/engine.go#L479-L482
Added lines #L479 - L482 were not covered by tests
[warning] 484-495: pkg/spdk/engine.go#L484-L495
Added lines #L484 - L495 were not covered by tests
[warning] 501-502: pkg/spdk/engine.go#L501-L502
Added lines #L501 - L502 were not covered by tests
[warning] 510-511: pkg/spdk/engine.go#L510-L511
Added lines #L510 - L511 were not covered by tests
[warning] 515-515: pkg/spdk/engine.go#L515
Added line #L515 was not covered by tests
[warning] 517-519: pkg/spdk/engine.go#L517-L519
Added lines #L517 - L519 were not covered by tests
[warning] 628-628: pkg/spdk/engine.go#L628
Added line #L628 was not covered by tests
🔇 Additional comments (3)
pkg/spdk/engine.go (3)
Line range hint 2183-2234: LGTM: Robust target switchover implementation
The target switchover implementation:
- Properly validates pod IP
- Handles standby target port appropriately for live upgrades
- Includes fallback logic for error cases
The error handling and state management look solid.
Line range hint 400-520: Add test coverage for critical paths
The live upgrade functionality includes critical paths that should be tested:
- Engine creation with standby target
- Target switchover scenarios
- Error handling and recovery paths
Consider adding integration tests to verify the end-to-end functionality.
#!/bin/bash
# Check existing test coverage
rg -l "func Test.*Target" 🧰 Tools
🪛 GitHub Check: codecov/patch
[warning] 422-429: pkg/spdk/engine.go#L422-L429
Added lines #L422 - L429 were not covered by tests
[warning] 431-442: pkg/spdk/engine.go#L431-L442
Added lines #L431 - L442 were not covered by tests
[warning] 444-444: pkg/spdk/engine.go#L444
Added line #L444 was not covered by tests
[warning] 448-462: pkg/spdk/engine.go#L448-L462
Added lines #L448 - L462 were not covered by tests
[warning] 464-466: pkg/spdk/engine.go#L464-L466
Added lines #L464 - L466 were not covered by tests
[warning] 468-468: pkg/spdk/engine.go#L468
Added line #L468 was not covered by tests
[warning] 471-473: pkg/spdk/engine.go#L471-L473
Added lines #L471 - L473 were not covered by tests
[warning] 476-476: pkg/spdk/engine.go#L476
Added line #L476 was not covered by tests
[warning] 479-482: pkg/spdk/engine.go#L479-L482
Added lines #L479 - L482 were not covered by tests
[warning] 484-495: pkg/spdk/engine.go#L484-L495
Added lines #L484 - L495 were not covered by tests
[warning] 501-502: pkg/spdk/engine.go#L501-L502
Added lines #L501 - L502 were not covered by tests
[warning] 510-511: pkg/spdk/engine.go#L510-L511
Added lines #L510 - L511 were not covered by tests
[warning] 515-515: pkg/spdk/engine.go#L515
Added line #L515 was not covered by tests
[warning] 517-519: pkg/spdk/engine.go#L517-L519
Added lines #L517 - L519 were not covered by tests
108-110: Add documentation and tests for isNewEngine
The function logic is clear but would benefit from documentation explaining what constitutes a "new" engine. Additionally, consider adding unit tests to cover this helper function.
🧰 Tools
🪛 GitHub Check: codecov/patch
[warning] 108-109: pkg/spdk/engine.go#L108-L109
Added lines #L108 - L109 were not covered by tests
derekbit
force-pushed
the
v2-control-upgrade
branch
from
2adf734
to
f654d45
Compare
There was a problem hiding this comment.
Actionable comments posted: 2
🧹 Outside diff range and nitpick comments (7)
pkg/spdk/engine_test.go (2)
11-23: Add documentation for the test function and struct fields.Consider adding documentation to improve code maintainability:
- Add a function comment explaining the purpose and methodology of the test
- Document the struct fields, especially explaining the relationship between different IP addresses and ports
+// TestCheckInitiatorAndTargetCreationRequirements validates the logic for determining +// when new initiator and target instances need to be created based on various +// network configurations and port settings. func (s *TestSuite) TestCheckInitiatorAndTargetCreationRequirements(c *C) { testCases := []struct { + // name is a descriptive identifier for the test case name string + // podIP is the IP address of the pod running the engine podIP string + // initiatorIP is the IP address for the SPDK initiator initiatorIP string + // targetIP is the IP address for the SPDK target targetIP string
24-96: Add test cases for port validation and error conditions.While the current test cases cover basic scenarios, consider adding these additional cases to improve coverage:
Port validation:
- Test negative port values
- Test port numbers exceeding valid range
- Test zero port with non-zero standby port
Error conditions for standby target:
- Test transition from active to standby when active port is in use
- Test invalid port combinations
}, + { + name: "Negative port values", + podIP: "192.168.1.1", + initiatorIP: "192.168.1.1", + targetIP: "192.168.1.1", + port: -1, + targetPort: -8000, + standbyTargetPort: -8001, + expectedInitiatorCreationRequired: false, + expectedTargetCreationRequired: false, + expectedError: fmt.Errorf("invalid port values"), + }, + { + name: "Invalid port combination", + podIP: "192.168.1.1", + initiatorIP: "192.168.1.1", + targetIP: "192.168.1.1", + port: 0, + targetPort: 0, + standbyTargetPort: 8001, + expectedInitiatorCreationRequired: false, + expectedTargetCreationRequired: false, + expectedError: fmt.Errorf("invalid port combination: standby port without active port"), + }, }pkg/spdk_test.go (1)
Line range hint
1-322: Consider documenting the upgrade architecture changesThe changes suggest a significant architectural shift in how upgrades are handled:
- The
EngineCreatesignature has been simplified- The dedicated upgrade test has been removed
- The basic engine creation test remains unchanged
Consider:
- Adding documentation to explain the new upgrade architecture
- Providing test coverage for the new upgrade mechanism
- Including examples of how to perform live upgrades with the new implementation
Would you like help with:
- Creating documentation for the new upgrade architecture?
- Designing new test cases for the upgrade functionality?
pkg/spdk/engine.go (4)
108-110: Add test coverage for the new helper function.The
isNewEnginefunction lacks test coverage. Consider adding unit tests to verify the behavior for both new and existing engines.🧰 Tools
🪛 GitHub Check: codecov/patch
[warning] 108-109: pkg/spdk/engine.go#L108-L109
Added lines #L108 - L109 were not covered by tests
223-226: Consider adding debug level logging.While the logging improvements are good, consider adding debug level logs to help with troubleshooting, especially around the decision points for initiator/target creation.
Also applies to: 306-315
🧰 Tools
🪛 GitHub Check: codecov/patch
[warning] 223-226: pkg/spdk/engine.go#L223-L226
Added lines #L223 - L226 were not covered by tests
455-468: Document the retry mechanism for NVMe device info loading.The retry mechanism is well implemented, but consider adding a comment explaining why the retry is necessary and what conditions might trigger retries.
🧰 Tools
🪛 GitHub Check: codecov/patch
[warning] 464-466: pkg/spdk/engine.go#L464-L466
Added lines #L464 - L466 were not covered by tests
[warning] 468-468: pkg/spdk/engine.go#L468
Added line #L468 was not covered by tests
Line range hint
208-519: Improve test coverage for new functionality.Several new functions and code paths lack test coverage, including:
isNewEnginecheckInitiatorAndTargetCreationRequirements- Parts of the frontend handling logic
- Target switchover logic
Consider adding comprehensive test cases to verify the behavior of these new components, especially around the upgrade scenarios.
🧰 Tools
🪛 GitHub Check: codecov/patch
[warning] 422-429: pkg/spdk/engine.go#L422-L429
Added lines #L422 - L429 were not covered by tests
[warning] 431-442: pkg/spdk/engine.go#L431-L442
Added lines #L431 - L442 were not covered by tests
[warning] 444-444: pkg/spdk/engine.go#L444
Added line #L444 was not covered by tests
[warning] 448-462: pkg/spdk/engine.go#L448-L462
Added lines #L448 - L462 were not covered by tests
[warning] 464-466: pkg/spdk/engine.go#L464-L466
Added lines #L464 - L466 were not covered by tests
[warning] 468-468: pkg/spdk/engine.go#L468
Added line #L468 was not covered by tests
[warning] 471-473: pkg/spdk/engine.go#L471-L473
Added lines #L471 - L473 were not covered by tests
[warning] 476-476: pkg/spdk/engine.go#L476
Added line #L476 was not covered by tests
[warning] 479-482: pkg/spdk/engine.go#L479-L482
Added lines #L479 - L482 were not covered by tests
[warning] 484-495: pkg/spdk/engine.go#L484-L495
Added lines #L484 - L495 were not covered by tests
[warning] 501-502: pkg/spdk/engine.go#L501-L502
Added lines #L501 - L502 were not covered by tests
[warning] 510-511: pkg/spdk/engine.go#L510-L511
Added lines #L510 - L511 were not covered by tests
[warning] 515-515: pkg/spdk/engine.go#L515
Added line #L515 was not covered by tests
[warning] 517-519: pkg/spdk/engine.go#L517-L519
Added lines #L517 - L519 were not covered by tests
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
📒 Files selected for processing (5)
pkg/api/types.go(2 hunks)pkg/client/client.go(1 hunks)pkg/spdk/engine.go(14 hunks)pkg/spdk/engine_test.go(1 hunks)pkg/spdk_test.go(2 hunks)
🚧 Files skipped from review as they are similar to previous changes (2)
- pkg/api/types.go
- pkg/client/client.go
🧰 Additional context used
🪛 GitHub Check: codecov/patch
pkg/spdk/engine.go
[warning] 108-109: pkg/spdk/engine.go#L108-L109
Added lines #L108 - L109 were not covered by tests
[warning] 128-130: pkg/spdk/engine.go#L128-L130
Added lines #L128 - L130 were not covered by tests
[warning] 208-214: pkg/spdk/engine.go#L208-L214
Added lines #L208 - L214 were not covered by tests
[warning] 216-220: pkg/spdk/engine.go#L216-L220
Added lines #L216 - L220 were not covered by tests
[warning] 223-226: pkg/spdk/engine.go#L223-L226
Added lines #L223 - L226 were not covered by tests
[warning] 228-231: pkg/spdk/engine.go#L228-L231
Added lines #L228 - L231 were not covered by tests
[warning] 263-264: pkg/spdk/engine.go#L263-L264
Added lines #L263 - L264 were not covered by tests
[warning] 269-269: pkg/spdk/engine.go#L269
Added line #L269 was not covered by tests
[warning] 271-271: pkg/spdk/engine.go#L271
Added line #L271 was not covered by tests
[warning] 282-285: pkg/spdk/engine.go#L282-L285
Added lines #L282 - L285 were not covered by tests
[warning] 292-292: pkg/spdk/engine.go#L292
Added line #L292 was not covered by tests
[warning] 302-303: pkg/spdk/engine.go#L302-L303
Added lines #L302 - L303 were not covered by tests
[warning] 306-315: pkg/spdk/engine.go#L306-L315
Added lines #L306 - L315 were not covered by tests
[warning] 401-401: pkg/spdk/engine.go#L401
Added line #L401 was not covered by tests
[warning] 411-413: pkg/spdk/engine.go#L411-L413
Added lines #L411 - L413 were not covered by tests
[warning] 422-429: pkg/spdk/engine.go#L422-L429
Added lines #L422 - L429 were not covered by tests
[warning] 431-442: pkg/spdk/engine.go#L431-L442
Added lines #L431 - L442 were not covered by tests
[warning] 444-444: pkg/spdk/engine.go#L444
Added line #L444 was not covered by tests
[warning] 448-462: pkg/spdk/engine.go#L448-L462
Added lines #L448 - L462 were not covered by tests
[warning] 464-466: pkg/spdk/engine.go#L464-L466
Added lines #L464 - L466 were not covered by tests
[warning] 468-468: pkg/spdk/engine.go#L468
Added line #L468 was not covered by tests
[warning] 471-473: pkg/spdk/engine.go#L471-L473
Added lines #L471 - L473 were not covered by tests
[warning] 476-476: pkg/spdk/engine.go#L476
Added line #L476 was not covered by tests
[warning] 479-482: pkg/spdk/engine.go#L479-L482
Added lines #L479 - L482 were not covered by tests
[warning] 484-495: pkg/spdk/engine.go#L484-L495
Added lines #L484 - L495 were not covered by tests
[warning] 501-502: pkg/spdk/engine.go#L501-L502
Added lines #L501 - L502 were not covered by tests
[warning] 510-511: pkg/spdk/engine.go#L510-L511
Added lines #L510 - L511 were not covered by tests
[warning] 515-515: pkg/spdk/engine.go#L515
Added line #L515 was not covered by tests
[warning] 517-519: pkg/spdk/engine.go#L517-L519
Added lines #L517 - L519 were not covered by tests
[warning] 628-628: pkg/spdk/engine.go#L628
Added line #L628 was not covered by tests
🔇 Additional comments (6)
pkg/spdk_test.go (1)
250-322:
Reconsider removing the upgrade test case
The commented-out test TestSPDKEngineCreateWithUpgradeRequired appears to contain valuable test coverage for upgrade scenarios, which aligns with the PR's objective of supporting live upgrades. Consider:
- Why was this test case removed?
- Is there alternative test coverage for the upgrade functionality?
- Should this test be adapted rather than removed to support the new upgrade implementation?
Let's check for other upgrade-related test coverage:
#!/bin/bash
# Description: Search for other upgrade-related tests that might provide coverage
# Search for upgrade-related test files and functions
rg -l "Test.*Upgrade"
# Search for upgrade-related assertions
rg "c\.Assert.*[Uu]pgrade"pkg/spdk/engine.go (5)
37-49: LGTM! New field addition looks good.
The StandbyTargetPort field is properly integrated with the existing port-related fields and follows the same type pattern.
208-214: LGTM! Good error handling and control flow.
The integration of checkInitiatorAndTargetCreationRequirements is well done with proper error handling and early returns.
🧰 Tools
🪛 GitHub Check: codecov/patch
[warning] 208-214: pkg/spdk/engine.go#L208-L214
Added lines #L208 - L214 were not covered by tests
400-401: LGTM! Good error handling and port management.
The frontend handling changes properly manage both regular and standby target ports with appropriate error handling.
Also applies to: 411-413
🧰 Tools
🪛 GitHub Check: codecov/patch
[warning] 401-401: pkg/spdk/engine.go#L401
Added line #L401 was not covered by tests
123-128:
Redundant condition check on e.StandbyTargetPort
The nested condition checking e.StandbyTargetPort != 0 is redundant since it's already checked in the outer if statement.
Apply this diff to fix the redundant condition:
if e.StandbyTargetPort != 0 {
- if e.StandbyTargetPort != 0 {
e.log.Warnf("Standby target instance with port %v is already created, will skip the target creation", e.StandbyTargetPort)
return e.getWithoutLock(), nil
- }
}Likely invalid or redundant comment.
2183-2187: LGTM! Good target switchover implementation.
The target switchover logic properly handles port management and state updates. The pod IP check is a good addition for ensuring correct target configuration.
Also applies to: 2230-2234
✅ Verification successful
Pod IP validation is properly implemented across the codebase
The GetIPForPod() usage is consistently implemented across the codebase with proper error handling. The target switchover logic correctly validates pod IP against target and initiator IPs to determine the appropriate instance creation. The implementation in the reviewed code segment aligns with the established pattern seen in other files (replica.go, backup.go) and is followed by proper instance handling logic.
🏁 Scripts executed
The following scripts were executed for the analysis:
Script:
#!/bin/bash
# Verify the pod IP handling logic
ast-grep --pattern $'podIP, err := commonnet.GetIPForPod()
if err != nil {
$$$
}
if podIP == targetIP {
$$$
}'
Length of output: 115
Script:
#!/bin/bash
# Let's search for GetIPForPod usage and its surrounding context
rg -A 5 "GetIPForPod" --type go
# Also search for any IP comparison logic
rg -A 5 "podIP.*==" --type go
Length of output: 2957
derekbit
force-pushed
the
v2-control-upgrade
branch
from
6b4b3c7
to
5d51017
Compare
There was a problem hiding this comment.
Actionable comments posted: 2
🧹 Outside diff range and nitpick comments (2)
pkg/spdk/server.go (1)
878-880: Consider adding documentation for live upgrade feature.Since this change introduces support for live upgrades, it would be helpful to add documentation explaining:
- The live upgrade process
- The role of initiator and target addresses
- Requirements and prerequisites for live upgrades
pkg/spdk/engine.go (1)
45-45: Add documentation for the StandbyTargetPort fieldPlease add a comment explaining that this field is used for managing standby targets during live upgrades.
+ // StandbyTargetPort is the port used by the standby target instance during live upgrades StandbyTargetPort int32
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
📒 Files selected for processing (3)
pkg/spdk/engine.go(14 hunks)pkg/spdk/engine_test.go(1 hunks)pkg/spdk/server.go(1 hunks)
🚧 Files skipped from review as they are similar to previous changes (1)
- pkg/spdk/engine_test.go
🧰 Additional context used
🪛 GitHub Check: codecov/patch
pkg/spdk/engine.go
[warning] 108-109: pkg/spdk/engine.go#L108-L109
Added lines #L108 - L109 were not covered by tests
[warning] 128-130: pkg/spdk/engine.go#L128-L130
Added lines #L128 - L130 were not covered by tests
[warning] 144-144: pkg/spdk/engine.go#L144
Added line #L144 was not covered by tests
[warning] 207-213: pkg/spdk/engine.go#L207-L213
Added lines #L207 - L213 were not covered by tests
[warning] 215-219: pkg/spdk/engine.go#L215-L219
Added lines #L215 - L219 were not covered by tests
[warning] 222-225: pkg/spdk/engine.go#L222-L225
Added lines #L222 - L225 were not covered by tests
[warning] 227-230: pkg/spdk/engine.go#L227-L230
Added lines #L227 - L230 were not covered by tests
[warning] 262-263: pkg/spdk/engine.go#L262-L263
Added lines #L262 - L263 were not covered by tests
[warning] 268-268: pkg/spdk/engine.go#L268
Added line #L268 was not covered by tests
[warning] 270-270: pkg/spdk/engine.go#L270
Added line #L270 was not covered by tests
[warning] 281-284: pkg/spdk/engine.go#L281-L284
Added lines #L281 - L284 were not covered by tests
[warning] 291-291: pkg/spdk/engine.go#L291
Added line #L291 was not covered by tests
[warning] 301-302: pkg/spdk/engine.go#L301-L302
Added lines #L301 - L302 were not covered by tests
[warning] 305-313: pkg/spdk/engine.go#L305-L313
Added lines #L305 - L313 were not covered by tests
[warning] 399-399: pkg/spdk/engine.go#L399
Added line #L399 was not covered by tests
[warning] 409-411: pkg/spdk/engine.go#L409-L411
Added lines #L409 - L411 were not covered by tests
[warning] 420-427: pkg/spdk/engine.go#L420-L427
Added lines #L420 - L427 were not covered by tests
[warning] 429-440: pkg/spdk/engine.go#L429-L440
Added lines #L429 - L440 were not covered by tests
[warning] 442-442: pkg/spdk/engine.go#L442
Added line #L442 was not covered by tests
[warning] 446-460: pkg/spdk/engine.go#L446-L460
Added lines #L446 - L460 were not covered by tests
[warning] 462-464: pkg/spdk/engine.go#L462-L464
Added lines #L462 - L464 were not covered by tests
[warning] 466-466: pkg/spdk/engine.go#L466
Added line #L466 was not covered by tests
[warning] 469-471: pkg/spdk/engine.go#L469-L471
Added lines #L469 - L471 were not covered by tests
[warning] 474-474: pkg/spdk/engine.go#L474
Added line #L474 was not covered by tests
[warning] 477-480: pkg/spdk/engine.go#L477-L480
Added lines #L477 - L480 were not covered by tests
[warning] 482-493: pkg/spdk/engine.go#L482-L493
Added lines #L482 - L493 were not covered by tests
[warning] 499-500: pkg/spdk/engine.go#L499-L500
Added lines #L499 - L500 were not covered by tests
[warning] 508-509: pkg/spdk/engine.go#L508-L509
Added lines #L508 - L509 were not covered by tests
[warning] 513-513: pkg/spdk/engine.go#L513
Added line #L513 was not covered by tests
[warning] 515-517: pkg/spdk/engine.go#L515-L517
Added lines #L515 - L517 were not covered by tests
🔇 Additional comments (5)
pkg/spdk/server.go (1)
879-879: LGTM! Verify live upgrade functionality.
The changes to the Create method call properly support engine live upgrades by passing the necessary parameters for initiator and target addresses.
Run the following script to verify the engine creation parameters:
✅ Verification successful
Let me gather more context about the engine creation implementation to verify the changes properly.
Let me search for the Create method implementation specifically since it wasn't found in the previous search.
Let me try one more search with a different pattern to locate the Create method implementation.
Based on the gathered evidence, I can now provide the final response:
The changes to engine creation are correctly implemented for live upgrade support
The code changes are verified to be correct and consistent with the live upgrade functionality:
- The
Createmethod properly handles initiator and target addresses as parameters - The implementation includes proper validation and switching logic for target addresses
- The changes are integrated with the existing port allocation and engine management system
Key validations:
- Proper validation of target address requirement
- Handling of target switching via
isSwitchOverTargetRequired - Safety check for local target with matching addresses
- Proper connection/disconnection handling during target operations
🏁 Scripts executed
The following scripts were executed for the analysis:
Script:
#!/bin/bash
# Description: Verify the engine creation parameters and usage
# Test: Search for engine creation calls to ensure consistent parameter usage
rg -A 5 "e\.Create\("
Length of output: 371
Script:
#!/bin/bash
# Search for the Engine struct and Create method implementation
ast-grep --pattern 'type Engine struct {
$$$
}'
echo "---"
ast-grep --pattern 'func ($_) Create($$$) {
$$$
}'
Length of output: 7425
Script:
#!/bin/bash
# Search for the Create method implementation and its usage
rg -A 10 "func.*Create\(.*spdkClient.*ReplicaAddressMap"
echo "---"
# Search for any tests related to the Create functionality
rg -A 5 "TestEngine.*Create"
Length of output: 104
Script:
#!/bin/bash
# Search for Create method in engine.go with broader context
rg -A 15 "func \(e \*Engine\) Create" pkg/spdk/engine.go
echo "---"
# Search for the usage of InitiatorAddress and TargetAddress in the codebase
rg "InitiatorAddress|TargetAddress" pkg/spdk/
Length of output: 2137
pkg/spdk/engine.go (4)
Line range hint 2181-2232: Enhance validation and error handling for target switchover
The target switchover logic is critical for live upgrades. Consider adding:
- Validation for the target port range
- Retries for port allocation failures
- Rollback mechanism for failed switchovers
- Metrics/events for monitoring switchover operations
Consider implementing a state machine pattern to manage the switchover lifecycle:
- PREPARING -> SWITCHING -> COMPLETED/FAILED
- Add metrics for each state transition
- Implement automatic rollback on failure
#!/bin/bash
# Check port validation
rg "ValidatePort|PortRange|port.*valid"
# Check retry/rollback mechanisms
rg "Retry|Rollback|recover|revert" Line range hint 1-2232: Improve overall test coverage
The code changes implement critical functionality for live upgrades but have significant gaps in test coverage. Please add:
- Integration tests for the live upgrade flow
- Unit tests for new helper functions
- Error scenario tests
- Performance benchmarks for the switchover process
#!/bin/bash
# Check current test coverage
fd -e go -e test 🧰 Tools
🪛 GitHub Check: codecov/patch
[warning] 420-427: pkg/spdk/engine.go#L420-L427
Added lines #L420 - L427 were not covered by tests
[warning] 429-440: pkg/spdk/engine.go#L429-L440
Added lines #L429 - L440 were not covered by tests
[warning] 442-442: pkg/spdk/engine.go#L442
Added line #L442 was not covered by tests
[warning] 446-460: pkg/spdk/engine.go#L446-L460
Added lines #L446 - L460 were not covered by tests
[warning] 462-464: pkg/spdk/engine.go#L462-L464
Added lines #L462 - L464 were not covered by tests
[warning] 466-466: pkg/spdk/engine.go#L466
Added line #L466 was not covered by tests
[warning] 469-471: pkg/spdk/engine.go#L469-L471
Added lines #L469 - L471 were not covered by tests
[warning] 474-474: pkg/spdk/engine.go#L474
Added line #L474 was not covered by tests
[warning] 477-480: pkg/spdk/engine.go#L477-L480
Added lines #L477 - L480 were not covered by tests
[warning] 482-493: pkg/spdk/engine.go#L482-L493
Added lines #L482 - L493 were not covered by tests
[warning] 499-500: pkg/spdk/engine.go#L499-L500
Added lines #L499 - L500 were not covered by tests
[warning] 508-509: pkg/spdk/engine.go#L508-L509
Added lines #L508 - L509 were not covered by tests
[warning] 513-513: pkg/spdk/engine.go#L513
Added line #L513 was not covered by tests
[warning] 515-517: pkg/spdk/engine.go#L515-L517
Added lines #L515 - L517 were not covered by tests
123-127:
Remove redundant condition check
The nested condition checking e.StandbyTargetPort != 0 is redundant as it's already checked in the outer if statement.
- if e.StandbyTargetPort != 0 {
- e.log.Warnf("Standby target instance with port %v is already created, will skip the target creation", e.StandbyTargetPort)
- } else {
- targetCreationRequired = true
- }
+ e.log.Warnf("Standby target instance with port %v is already created, will skip the target creation", e.StandbyTargetPort)Likely invalid or redundant comment.
409-411: Add test coverage for frontend handling
The frontend handling logic, particularly around standby target creation and error handling, lacks test coverage. Please add tests to verify:
- Standby target creation logic
- Error handling scenarios
- Endpoint and logging updates
Also applies to: 420-440
🧰 Tools
🪛 GitHub Check: codecov/patch
[warning] 409-411: pkg/spdk/engine.go#L409-L411
Added lines #L409 - L411 were not covered by tests
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Outside diff range and nitpick comments (1)
pkg/spdk/engine.go (1)
45-45: Add documentation forStandbyTargetPortfieldTo improve code readability and maintain consistency, consider adding a comment to document the purpose of the new
StandbyTargetPortfield in theEnginestruct.
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
📒 Files selected for processing (2)
pkg/spdk/engine.go(14 hunks)pkg/spdk/engine_test.go(1 hunks)
🚧 Files skipped from review as they are similar to previous changes (1)
- pkg/spdk/engine_test.go
🧰 Additional context used
🪛 GitHub Check: codecov/patch
pkg/spdk/engine.go
[warning] 108-109: pkg/spdk/engine.go#L108-L109
Added lines #L108 - L109 were not covered by tests
[warning] 128-130: pkg/spdk/engine.go#L128-L130
Added lines #L128 - L130 were not covered by tests
[warning] 144-144: pkg/spdk/engine.go#L144
Added line #L144 was not covered by tests
[warning] 207-213: pkg/spdk/engine.go#L207-L213
Added lines #L207 - L213 were not covered by tests
[warning] 215-219: pkg/spdk/engine.go#L215-L219
Added lines #L215 - L219 were not covered by tests
[warning] 222-225: pkg/spdk/engine.go#L222-L225
Added lines #L222 - L225 were not covered by tests
[warning] 227-230: pkg/spdk/engine.go#L227-L230
Added lines #L227 - L230 were not covered by tests
[warning] 262-263: pkg/spdk/engine.go#L262-L263
Added lines #L262 - L263 were not covered by tests
[warning] 268-268: pkg/spdk/engine.go#L268
Added line #L268 was not covered by tests
[warning] 270-270: pkg/spdk/engine.go#L270
Added line #L270 was not covered by tests
[warning] 281-284: pkg/spdk/engine.go#L281-L284
Added lines #L281 - L284 were not covered by tests
[warning] 291-291: pkg/spdk/engine.go#L291
Added line #L291 was not covered by tests
[warning] 301-302: pkg/spdk/engine.go#L301-L302
Added lines #L301 - L302 were not covered by tests
[warning] 305-313: pkg/spdk/engine.go#L305-L313
Added lines #L305 - L313 were not covered by tests
[warning] 399-399: pkg/spdk/engine.go#L399
Added line #L399 was not covered by tests
[warning] 409-411: pkg/spdk/engine.go#L409-L411
Added lines #L409 - L411 were not covered by tests
[warning] 420-427: pkg/spdk/engine.go#L420-L427
Added lines #L420 - L427 were not covered by tests
[warning] 429-440: pkg/spdk/engine.go#L429-L440
Added lines #L429 - L440 were not covered by tests
[warning] 442-442: pkg/spdk/engine.go#L442
Added line #L442 was not covered by tests
[warning] 446-460: pkg/spdk/engine.go#L446-L460
Added lines #L446 - L460 were not covered by tests
[warning] 462-464: pkg/spdk/engine.go#L462-L464
Added lines #L462 - L464 were not covered by tests
[warning] 466-466: pkg/spdk/engine.go#L466
Added line #L466 was not covered by tests
[warning] 469-471: pkg/spdk/engine.go#L469-L471
Added lines #L469 - L471 were not covered by tests
[warning] 474-474: pkg/spdk/engine.go#L474
Added line #L474 was not covered by tests
[warning] 477-480: pkg/spdk/engine.go#L477-L480
Added lines #L477 - L480 were not covered by tests
[warning] 482-493: pkg/spdk/engine.go#L482-L493
Added lines #L482 - L493 were not covered by tests
[warning] 499-500: pkg/spdk/engine.go#L499-L500
Added lines #L499 - L500 were not covered by tests
[warning] 508-509: pkg/spdk/engine.go#L508-L509
Added lines #L508 - L509 were not covered by tests
[warning] 513-513: pkg/spdk/engine.go#L513
Added line #L513 was not covered by tests
[warning] 515-517: pkg/spdk/engine.go#L515-L517
Added lines #L515 - L517 were not covered by tests
🔇 Additional comments (4)
pkg/spdk/engine.go (4)
215-220: 🛠️ Refactor suggestion
Verify initialization logic for new engines
In the Create method, when isNewEngine(e) returns true, the code initializes e.IP, e.TargetIP, and potentially e.StandbyTargetPort. Ensure that all necessary fields are correctly initialized to prevent issues with uninitialized or stale data.
Consider initializing e.StandbyTargetPort explicitly if required.
🧰 Tools
🪛 GitHub Check: codecov/patch
[warning] 215-219: pkg/spdk/engine.go#L215-L219
Added lines #L215 - L219 were not covered by tests
409-411:
Review logic for determining standbyTargetCreationRequired
The condition if e.Port != 0 && e.TargetPort == 0 sets standbyTargetCreationRequired to true. Confirm that this condition accurately reflects the scenarios where a standby target creation is needed, especially when e.StandbyTargetPort may have a value.
🧰 Tools
🪛 GitHub Check: codecov/patch
[warning] 409-411: pkg/spdk/engine.go#L409-L411
Added lines #L409 - L411 were not covered by tests
2228-2231:
Potential issue when updating target ports during switchover
In the SwitchOverTarget function, setting e.TargetPort = targetPort and resetting e.StandbyTargetPort = 0 when targetIP == podIP could lead to issues if StandbyTargetPort should be preserved in certain cases. Please verify if this logic aligns with the intended behavior during target switchover.
123-127:
Ensure correct handling of StandbyTargetPort condition
The condition within the if statement checks if e.StandbyTargetPort != 0, and then the else block sets targetCreationRequired = true. This logic might skip target creation when it's actually needed. Please verify that the conditions correctly reflect the intended behavior.
Run the following script to check for redundant or conflicting conditions involving StandbyTargetPort:
✅ Verification successful
Let me gather more context about the surrounding code to understand the complete logic flow.
Based on the context gathered, I can now provide the final response:
The StandbyTargetPort condition is correctly implemented
The logic is correct and serves its intended purpose. Here's why:
- When
StandbyTargetPort != 0, it means a standby target instance already exists, so skipping target creation is the right behavior - When
StandbyTargetPort == 0, settingtargetCreationRequired = trueis correct as no standby target exists yet - This aligns with the engine's lifecycle management where
StandbyTargetPortis used to track standby target instances and is properly reset to 0 during target switchover (as seen in the code)
🏁 Scripts executed
The following scripts were executed for the analysis:
Script:
#!/bin/bash
# Description: Search for conditions involving 'StandbyTargetPort' that might lead to logical errors.
# Find all instances where 'StandbyTargetPort' is compared
rg 'if.*StandbyTargetPort\s*(!=|==)\s*0' -A 3
# Check assignments to 'targetCreationRequired'
rg 'targetCreationRequired\s*=' -A 2
Length of output: 924
Script:
#!/bin/bash
# Get more context around the StandbyTargetPort usage
rg "StandbyTargetPort" -B 5 -A 10
# Get the complete function context where this condition exists
ast-grep --pattern 'func ($_ *Engine) $_($$$) $_ {
$$$
if $_.StandbyTargetPort != 0 {
$$$
}
$$$
}'
Length of output: 11653
derekbit
force-pushed
the
v2-control-upgrade
branch
from
4469b89
to
fe16578
Compare
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Outside diff range and nitpick comments (4)
pkg/spdk/engine.go (4)
45-45: Add field documentation forStandbyTargetPortPlease add a comment explaining the purpose of this field and its role in live upgrades.
+ // StandbyTargetPort is the port used by the standby target instance during live upgrades StandbyTargetPort int32
128-130: Improve error message clarityThe error message could be more descriptive by including the current state information.
- err = fmt.Errorf("invalid initiator and target addresses for engine %s creation with initiator address %v and target address %v", e.Name, initiatorIP, targetIP) + err = fmt.Errorf("invalid state for engine %s creation: initiator port %v, target port %v with initiator address %v and target address %v", e.Name, e.Port, e.TargetPort, initiatorIP, targetIP)🧰 Tools
🪛 GitHub Check: codecov/patch
[warning] 128-130: pkg/spdk/engine.go#L128-L130
Added lines #L128 - L130 were not covered by tests
262-263: Enhance TODO comment specificityThe TODO comment should be more specific about what aspects of the log message need improvement.
- // TODO: improve the log message + // TODO: improve the log message to include RAID level and configuration details🧰 Tools
🪛 GitHub Check: codecov/patch
[warning] 262-263: pkg/spdk/engine.go#L262-L263
Added lines #L262 - L263 were not covered by tests
456-466: Extract retry logic into helper functionConsider extracting this retry logic into a reusable helper function to improve code maintainability.
+func (e *Engine) retryLoadNVMeDeviceInfo(initiator *nvme.Initiator) error { + for r := 0; r < maxNumRetries; r++ { + err := initiator.LoadNVMeDeviceInfo(initiator.TransportAddress, initiator.TransportServiceID, initiator.SubsystemNQN) + if err == nil { + e.log.Infof("Loaded NVMe device info for engine") + return nil + } + if !strings.Contains(err.Error(), "failed to get devices") { + return errors.Wrapf(err, "failed to load NVMe device info for engine %v", e.Name) + } + time.Sleep(retryInterval) + } + return fmt.Errorf("failed to load NVMe device info after %d retries", maxNumRetries) +}🧰 Tools
🪛 GitHub Check: codecov/patch
[warning] 462-464: pkg/spdk/engine.go#L462-L464
Added lines #L462 - L464 were not covered by tests
[warning] 466-466: pkg/spdk/engine.go#L466
Added line #L466 was not covered by tests
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
📒 Files selected for processing (3)
pkg/spdk/engine.go(14 hunks)pkg/spdk/engine_test.go(1 hunks)pkg/spdk/server.go(1 hunks)
🚧 Files skipped from review as they are similar to previous changes (2)
- pkg/spdk/engine_test.go
- pkg/spdk/server.go
🧰 Additional context used
🪛 GitHub Check: codecov/patch
pkg/spdk/engine.go
[warning] 108-109: pkg/spdk/engine.go#L108-L109
Added lines #L108 - L109 were not covered by tests
[warning] 128-130: pkg/spdk/engine.go#L128-L130
Added lines #L128 - L130 were not covered by tests
[warning] 144-144: pkg/spdk/engine.go#L144
Added line #L144 was not covered by tests
[warning] 207-213: pkg/spdk/engine.go#L207-L213
Added lines #L207 - L213 were not covered by tests
[warning] 215-219: pkg/spdk/engine.go#L215-L219
Added lines #L215 - L219 were not covered by tests
[warning] 222-225: pkg/spdk/engine.go#L222-L225
Added lines #L222 - L225 were not covered by tests
[warning] 227-230: pkg/spdk/engine.go#L227-L230
Added lines #L227 - L230 were not covered by tests
[warning] 262-263: pkg/spdk/engine.go#L262-L263
Added lines #L262 - L263 were not covered by tests
[warning] 268-268: pkg/spdk/engine.go#L268
Added line #L268 was not covered by tests
[warning] 270-270: pkg/spdk/engine.go#L270
Added line #L270 was not covered by tests
[warning] 281-284: pkg/spdk/engine.go#L281-L284
Added lines #L281 - L284 were not covered by tests
[warning] 291-291: pkg/spdk/engine.go#L291
Added line #L291 was not covered by tests
[warning] 301-302: pkg/spdk/engine.go#L301-L302
Added lines #L301 - L302 were not covered by tests
[warning] 305-313: pkg/spdk/engine.go#L305-L313
Added lines #L305 - L313 were not covered by tests
[warning] 399-399: pkg/spdk/engine.go#L399
Added line #L399 was not covered by tests
[warning] 409-411: pkg/spdk/engine.go#L409-L411
Added lines #L409 - L411 were not covered by tests
[warning] 420-427: pkg/spdk/engine.go#L420-L427
Added lines #L420 - L427 were not covered by tests
[warning] 429-440: pkg/spdk/engine.go#L429-L440
Added lines #L429 - L440 were not covered by tests
[warning] 442-442: pkg/spdk/engine.go#L442
Added line #L442 was not covered by tests
[warning] 446-460: pkg/spdk/engine.go#L446-L460
Added lines #L446 - L460 were not covered by tests
[warning] 462-464: pkg/spdk/engine.go#L462-L464
Added lines #L462 - L464 were not covered by tests
[warning] 466-466: pkg/spdk/engine.go#L466
Added line #L466 was not covered by tests
[warning] 469-471: pkg/spdk/engine.go#L469-L471
Added lines #L469 - L471 were not covered by tests
[warning] 474-474: pkg/spdk/engine.go#L474
Added line #L474 was not covered by tests
[warning] 477-480: pkg/spdk/engine.go#L477-L480
Added lines #L477 - L480 were not covered by tests
[warning] 482-493: pkg/spdk/engine.go#L482-L493
Added lines #L482 - L493 were not covered by tests
[warning] 499-500: pkg/spdk/engine.go#L499-L500
Added lines #L499 - L500 were not covered by tests
[warning] 508-509: pkg/spdk/engine.go#L508-L509
Added lines #L508 - L509 were not covered by tests
[warning] 513-513: pkg/spdk/engine.go#L513
Added line #L513 was not covered by tests
[warning] 515-517: pkg/spdk/engine.go#L515-L517
Added lines #L515 - L517 were not covered by tests
🔇 Additional comments (1)
pkg/spdk/engine.go (1)
2228-2231: Add test coverage for target switching logic
This critical section handling standby target ports during live upgrades lacks test coverage. Please add test cases to verify:
- Target port assignment when target IP matches pod IP
- Standby target port reset
- Error cases
#!/bin/bash
# Check for existing test coverage
rg -A 5 "TestSwitchOverTarget"
derekbit
force-pushed
the
v2-control-upgrade
branch
from
fe16578
to
be48fa6
Compare
There was a problem hiding this comment.
Actionable comments posted: 4
🧹 Outside diff range and nitpick comments (5)
pkg/spdk/engine_test.go (2)
97-99: Enable parallel test execution support.Consider making the tests parallel-safe by using
c.Parallel()at the start of each test case. This would require moving the test case execution into a separate function to ensure proper isolation:- for testName, testCase := range testCases { - c.Logf("testing checkInitiatorAndTargetCreationRequirements.%v", testName) + for _, tc := range testCases { + testCase := tc // capture range variable + c.Run(testCase.name, func(c *C) { + c.Parallel()
11-11: Document test scenarios and their relationship to live upgrade requirements.To better align with the PR's objective of supporting live upgrades, consider adding documentation that:
- Explains how each test case relates to the live upgrade workflow
- Documents the assumptions and preconditions for each scenario
- Maps test cases to specific upgrade requirements
Add a comment block before the test function:
// TestCheckInitiatorAndTargetCreationRequirements validates the engine's ability // to handle live upgrades by testing various scenarios: // 1. Initial state: Only active target // 2. Transition state: Both active and standby targets // 3. Final state: New active target (previously standby) // // Each scenario verifies: // - Correct port allocation // - Proper instance creation flags // - Error handling for invalid configurationspkg/spdk/engine.go (3)
108-110: Add test coverage forisNewEngineWhile the helper function is well-structured, it currently lacks test coverage. Consider adding test cases to verify:
- New engine (all fields empty/zero)
- Existing engine with various field combinations
112-142: Simplify complex branching logicThe function has multiple nested conditions that could be simplified for better maintainability. Consider breaking down the logic into smaller, more focused functions:
+func (e *Engine) shouldCreateInitiator(podIP, initiatorIP string) bool { + return podIP == initiatorIP +} +func (e *Engine) shouldCreateTarget(podIP, targetIP string) bool { + return podIP == targetIP && e.StandbyTargetPort == 0 +} func (e *Engine) checkInitiatorAndTargetCreationRequirements(podIP, initiatorIP, targetIP string) (bool, bool, error) { - // Current complex logic + initiatorCreationRequired := e.shouldCreateInitiator(podIP, initiatorIP) + targetCreationRequired := e.shouldCreateTarget(podIP, targetIP) + + if !initiatorCreationRequired && !targetCreationRequired { + return false, false, fmt.Errorf("invalid initiator and target address for engine %s creation", e.Name) + } + + return initiatorCreationRequired, targetCreationRequired, nil }
2181-2184: Add error handling for IP retrievalConsider adding retry logic for IP retrieval as network operations can be flaky:
+func getPodIPWithRetry(maxRetries int, retryInterval time.Duration) (string, error) { + var lastErr error + for i := 0; i < maxRetries; i++ { + if ip, err := commonnet.GetIPForPod(); err == nil { + return ip, nil + } else { + lastErr = err + time.Sleep(retryInterval) + } + } + return "", fmt.Errorf("failed to get pod IP after %d retries: %v", maxRetries, lastErr) +}
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
📒 Files selected for processing (6)
pkg/api/types.go(2 hunks)pkg/client/client.go(1 hunks)pkg/spdk/engine.go(14 hunks)pkg/spdk/engine_test.go(1 hunks)pkg/spdk/server.go(1 hunks)pkg/spdk_test.go(1 hunks)
🚧 Files skipped from review as they are similar to previous changes (3)
- pkg/api/types.go
- pkg/client/client.go
- pkg/spdk/server.go
🔇 Additional comments (2)
pkg/spdk_test.go (1)
Line range hint 1-1354: Verify test coverage after removal of multi-threaded tests.
The removal of TestSPDKMultipleThread and TestSPDKMultipleThreadSnapshotOpsAndRebuilding functions could impact test coverage for concurrent operations. Please confirm:
- Are there equivalent tests elsewhere that cover these scenarios?
- Will new tests be added to verify concurrent operations with live upgrades?
Let's check for other test files that might cover these scenarios:
#!/bin/bash
# Description: Look for other test files that might cover concurrent operations
# Search for test files with concurrent operation tests
rg -l "sync.WaitGroup|go func\(\)" --type go --glob "*_test.go"
# Search for snapshot operation tests
rg -l "TestSnapshot|snapshot.*test" --type go --glob "*_test.go"pkg/spdk/engine.go (1)
37-49: LGTM: Engine struct field addition
The addition of StandbyTargetPort field to track standby target instance ports during live upgrades is well-placed and properly typed.
derekbit
force-pushed
the
v2-control-upgrade
branch
from
be48fa6
to
471a59d
Compare
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Outside diff range and nitpick comments (3)
pkg/spdk/engine.go (3)
37-49: Add documentation for the newStandbyTargetPortfieldThe new field
StandbyTargetPortshould be documented to explain its purpose in the live upgrade process and its relationship with other port fields (PortandTargetPort).Add a comment above the field:
TargetPort int32 // Port of the target that is used for letting initiator connect to + // StandbyTargetPort is used during live upgrades to temporarily store the port + // of a standby target instance before switching over StandbyTargetPort int32
116-127: Simplify nested conditionsThe nested conditions checking
StandbyTargetPortcan be simplified to improve readability and reduce complexity.if podIP == initiatorIP && podIP == targetIP { if e.Port == 0 && e.TargetPort == 0 { e.log.Info("Creating both initiator and target instances") initiatorCreationRequired = true targetCreationRequired = true } else if e.Port != 0 && e.TargetPort == 0 { e.log.Info("Creating a target instance") - if e.StandbyTargetPort != 0 { - e.log.Warnf("Standby target instance with port %v is already created, will skip the target creation", e.StandbyTargetPort) - } else { - targetCreationRequired = true - } + targetCreationRequired = e.StandbyTargetPort == 0 + if !targetCreationRequired { + e.log.Warnf("Standby target instance with port %v is already created, will skip the target creation", e.StandbyTargetPort) + } } else {
693-698: Improve error handling and state validationThe commented-out IP retrieval code and subsequent validation could be improved:
- // podIP, err := commonnet.GetIPForPod() - // if err != nil { - // return err - // } - if e.IP != e.TargetIP { - return nil - } + if e.IP == "" || e.TargetIP == "" { + return fmt.Errorf("invalid engine state: IP or TargetIP is empty") + } + if e.IP != e.TargetIP { + e.log.Debugf("Skipping validation for non-local target (IP: %s, TargetIP: %s)", e.IP, e.TargetIP) + return nil + }
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
📒 Files selected for processing (6)
pkg/api/types.go(2 hunks)pkg/client/client.go(1 hunks)pkg/spdk/engine.go(14 hunks)pkg/spdk/engine_test.go(1 hunks)pkg/spdk/server.go(1 hunks)pkg/spdk_test.go(6 hunks)
🚧 Files skipped from review as they are similar to previous changes (4)
- pkg/api/types.go
- pkg/client/client.go
- pkg/spdk/engine_test.go
- pkg/spdk/server.go
🧰 Additional context used
🪛 GitHub Check: codecov/patch
pkg/spdk/engine.go
[warning] 108-109: pkg/spdk/engine.go#L108-L109
Added lines #L108 - L109 were not covered by tests
[warning] 128-130: pkg/spdk/engine.go#L128-L130
Added lines #L128 - L130 were not covered by tests
[warning] 144-144: pkg/spdk/engine.go#L144
Added line #L144 was not covered by tests
[warning] 207-213: pkg/spdk/engine.go#L207-L213
Added lines #L207 - L213 were not covered by tests
[warning] 215-219: pkg/spdk/engine.go#L215-L219
Added lines #L215 - L219 were not covered by tests
[warning] 222-225: pkg/spdk/engine.go#L222-L225
Added lines #L222 - L225 were not covered by tests
[warning] 227-230: pkg/spdk/engine.go#L227-L230
Added lines #L227 - L230 were not covered by tests
[warning] 262-263: pkg/spdk/engine.go#L262-L263
Added lines #L262 - L263 were not covered by tests
[warning] 268-268: pkg/spdk/engine.go#L268
Added line #L268 was not covered by tests
[warning] 270-270: pkg/spdk/engine.go#L270
Added line #L270 was not covered by tests
[warning] 281-284: pkg/spdk/engine.go#L281-L284
Added lines #L281 - L284 were not covered by tests
[warning] 291-291: pkg/spdk/engine.go#L291
Added line #L291 was not covered by tests
[warning] 301-302: pkg/spdk/engine.go#L301-L302
Added lines #L301 - L302 were not covered by tests
[warning] 305-313: pkg/spdk/engine.go#L305-L313
Added lines #L305 - L313 were not covered by tests
[warning] 399-399: pkg/spdk/engine.go#L399
Added line #L399 was not covered by tests
[warning] 409-411: pkg/spdk/engine.go#L409-L411
Added lines #L409 - L411 were not covered by tests
[warning] 420-427: pkg/spdk/engine.go#L420-L427
Added lines #L420 - L427 were not covered by tests
[warning] 429-440: pkg/spdk/engine.go#L429-L440
Added lines #L429 - L440 were not covered by tests
[warning] 442-442: pkg/spdk/engine.go#L442
Added line #L442 was not covered by tests
[warning] 446-460: pkg/spdk/engine.go#L446-L460
Added lines #L446 - L460 were not covered by tests
[warning] 462-464: pkg/spdk/engine.go#L462-L464
Added lines #L462 - L464 were not covered by tests
[warning] 466-466: pkg/spdk/engine.go#L466
Added line #L466 was not covered by tests
[warning] 469-471: pkg/spdk/engine.go#L469-L471
Added lines #L469 - L471 were not covered by tests
[warning] 474-474: pkg/spdk/engine.go#L474
Added line #L474 was not covered by tests
[warning] 477-480: pkg/spdk/engine.go#L477-L480
Added lines #L477 - L480 were not covered by tests
[warning] 482-493: pkg/spdk/engine.go#L482-L493
Added lines #L482 - L493 were not covered by tests
[warning] 499-500: pkg/spdk/engine.go#L499-L500
Added lines #L499 - L500 were not covered by tests
[warning] 508-509: pkg/spdk/engine.go#L508-L509
Added lines #L508 - L509 were not covered by tests
[warning] 513-513: pkg/spdk/engine.go#L513
Added line #L513 was not covered by tests
[warning] 515-517: pkg/spdk/engine.go#L515-L517
Added lines #L515 - L517 were not covered by tests
🔇 Additional comments (6)
pkg/spdk_test.go (5)
271-271: LGTM! Parameter removal aligns with engine upgrade support.
The removal of the second false parameter from EngineCreate call is consistent with the PR objectives to support engine live upgrades.
354-354: LGTM! Consistent parameter update in test.
The change maintains consistency with the updated EngineCreate signature.
530-530: LGTM! Consistent parameter update in snapshot test.
The change maintains consistency with the updated EngineCreate signature in the snapshot operations test.
1228-1228: LGTM! Consistent parameter updates in revert helper.
Both EngineCreate calls in the revertSnapshot helper function have been updated consistently.
Also applies to: 1243-1243
1354-1354: LGTM! Consistent parameter update in target-only test.
The change maintains consistency with the updated EngineCreate signature in the target-only test case.
Let's verify that all EngineCreate calls have been updated consistently:
✅ Verification successful
All EngineCreate calls are consistent with the updated signature
The verification shows that all EngineCreate calls in the codebase follow the same pattern with 9 parameters (name, volumeName, frontend, size, replicaAddressMap, portCount, initiatorAddress, targetAddress, salvageRequested). The implementation in pkg/client/client.go confirms this signature, and all usage instances in pkg/spdk_test.go consistently follow this pattern. No instances of the old signature were found.
🏁 Scripts executed
The following scripts were executed for the analysis:
Script:
#!/bin/bash
# Description: Verify all EngineCreate calls have been updated to the new signature
# Expect: No instances of the old signature with two false parameters
# Search for any remaining EngineCreate calls with the old signature pattern
rg 'EngineCreate\([^)]*,[^)]*,[^)]*,[^)]*,[^)]*,[^)]*,[^)]*,[^)]*,[^)]*,[^)]*\)' --type go
Length of output: 92
Script:
#!/bin/bash
# Let's search for all EngineCreate calls to verify the signature pattern
rg "EngineCreate\(" --type go -A 1
Length of output: 3616
pkg/spdk/engine.go (1)
2228-2232: 🛠️ Refactor suggestion
Add validation for target port updates
The port update logic should validate that the ports are within valid ranges:
+ if targetPort <= 0 || targetPort > 65535 {
+ return fmt.Errorf("invalid target port %d", targetPort)
+ }
if targetIP == podIP {
e.TargetPort = targetPort
e.StandbyTargetPort = 0
}Likely invalid or redundant comment.
There was a problem hiding this comment.
Actionable comments posted: 2
🧹 Outside diff range and nitpick comments (5)
pkg/spdk/engine_test.go (3)
97-117: Improve test execution and error handlingThe test execution can be improved in several ways:
- Use test case name instead of index for better logging
- Add cleanup after each test
- Enhance error messages in assertions
Apply these changes:
- for testName, testCase := range testCases { - c.Logf("testing checkInitiatorAndTargetCreationRequirements.%v", testName) + for _, testCase := range testCases { + c.Logf("testing %s", testCase.name) + + // Clean up after each test + defer func() { + if engine != nil { + engine.log = nil + } + }() engine := &Engine{
119-162: Add edge cases for engine state validationThe test cases should include additional scenarios:
Edge cases:
- Invalid IP formats
- Special IP addresses (localhost, 0.0.0.0)
- Maximum port values
Upgrade scenarios:
- Engine during upgrade state
- Partially configured engine
Add these test cases:
}{ + { + name: "Engine during upgrade with both ports configured", + engine: &Engine{ + IP: "", + TargetIP: "", + TargetPort: 8000, + StandbyTargetPort: 8001, + }, + expected: false, + }, + { + name: "Engine with special IP (localhost)", + engine: &Engine{ + IP: "127.0.0.1", + TargetIP: "", + StandbyTargetPort: 0, + }, + expected: false, + }, }
163-168: Improve test iteration and error reportingSimilar to the previous test, improve the test execution:
- Use test case name instead of index
- Add more descriptive error messages
Apply these changes:
- for testName, testCase := range testCases { - c.Logf("testing isNewEngine.%v", testName) + for _, testCase := range testCases { + c.Logf("testing %s", testCase.name) result := testCase.engine.isNewEngine() - c.Assert(result, Equals, testCase.expected, Commentf("Test case '%s': unexpected result", testCase.name)) + c.Assert(result, Equals, testCase.expected, + Commentf("Test case '%s': got %v, expected %v", + testCase.name, result, testCase.expected)) }pkg/spdk/engine.go (2)
2177-2180: Add context to error return for better debuggingWhen returning an error from
commonnet.GetIPForPod(), it's helpful to wrap it with additional context. This makes debugging easier by providing more information about where and why the error occurred.Apply this diff:
podIP, err := commonnet.GetIPForPod() if err != nil { - return err + return errors.Wrap(err, "failed to get pod IP") }
477-480: Clarify log and error messages by rephrasingThe term "Blindly" in the log and error messages may be unclear or informal. Consider rephrasing to enhance clarity and professionalism.
Apply this diff:
-e.log.Info("Blindly stopping expose bdev for engine") +e.log.Info("Stopping expose bdev for engine without prior checks") if err := spdkClient.StopExposeBdev(e.Nqn); err != nil { - return errors.Wrapf(err, "failed to blindly stop expose bdev for engine %v", e.Name) + return errors.Wrapf(err, "failed to stop expose bdev for engine %v without prior checks", e.Name) }
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
📒 Files selected for processing (2)
pkg/spdk/engine.go(15 hunks)pkg/spdk/engine_test.go(1 hunks)
🔇 Additional comments (1)
pkg/spdk/engine_test.go (1)
1-10: LGTM!
The package declaration and imports are appropriate for the testing requirements.
There was a problem hiding this comment.
Actionable comments posted: 3
🧹 Outside diff range and nitpick comments (5)
pkg/spdk/engine_test.go (2)
99-100: Use test case name instead of index in logsReplace the numeric index with the test case name for better readability and debugging:
- for testName, testCase := range testCases { - c.Logf("testing checkInitiatorAndTargetCreationRequirements.%v", testName) + for _, testCase := range testCases { + c.Logf("testing checkInitiatorAndTargetCreationRequirements: %s", testCase.name)
165-167: Use test case name instead of index in logsReplace the numeric index with the test case name for better readability and debugging:
- for testName, testCase := range testCases { - c.Logf("testing isNewEngine.%v", testName) + for _, testCase := range testCases { + c.Logf("testing isNewEngine: %s", testCase.name)pkg/spdk/engine.go (3)
37-49: Add documentation for the StandbyTargetPort fieldThe new
StandbyTargetPortfield would benefit from documentation explaining its purpose in the live upgrade process and its relationship withPortandTargetPort.Port int32 // Port that initiator is connecting to TargetIP string TargetPort int32 // Port of the target that is used for letting initiator connect to - StandbyTargetPort int32 + StandbyTargetPort int32 // Port of the standby target used during live upgrades
112-142: Simplify complex branching logicThe
checkInitiatorAndTargetCreationRequirementsmethod has complex nested conditions that could be simplified for better readability and maintainability.Consider restructuring the logic:
func (e *Engine) checkInitiatorAndTargetCreationRequirements(podIP, initiatorIP, targetIP string) (bool, bool, error) { - initiatorCreationRequired, targetCreationRequired := false, false - var err error - - if podIP == initiatorIP && podIP == targetIP { - if e.Port == 0 && e.TargetPort == 0 { - e.log.Info("Creating both initiator and target instances") - initiatorCreationRequired = true - targetCreationRequired = true - } else if e.Port != 0 && e.TargetPort == 0 { - e.log.Info("Creating a target instance") - if e.StandbyTargetPort != 0 { - e.log.Warnf("Standby target instance with port %v is already created, will skip the target creation", e.StandbyTargetPort) - } else { - targetCreationRequired = true - } - } else { - e.log.Infof("Initiator instance with port %v and target instance with port %v are already created, will skip the creation", e.Port, e.TargetPort) - } - } else if podIP == initiatorIP { - e.log.Info("Creating an initiator instance") - initiatorCreationRequired = true - } else if podIP == targetIP { - e.log.Info("Creating a target instance") - targetCreationRequired = true - } else { - err = fmt.Errorf("invalid initiator and target addresses for engine %s creation with initiator address %v and target address %v", e.Name, initiatorIP, targetIP) - } + if podIP != initiatorIP && podIP != targetIP { + return false, false, fmt.Errorf("invalid initiator and target addresses for engine %s creation with initiator address %v and target address %v", + e.Name, initiatorIP, targetIP) + } + + initiatorCreationRequired := podIP == initiatorIP && e.Port == 0 + targetCreationRequired := podIP == targetIP && e.TargetPort == 0 && e.StandbyTargetPort == 0 + + if initiatorCreationRequired && targetCreationRequired { + e.log.Info("Creating both initiator and target instances") + } else if initiatorCreationRequired { + e.log.Info("Creating an initiator instance") + } else if targetCreationRequired { + e.log.Info("Creating a target instance") + } else if e.StandbyTargetPort != 0 { + e.log.Warnf("Standby target instance with port %v is already created, will skip the target creation", e.StandbyTargetPort) + } else { + e.log.Infof("Initiator instance with port %v and target instance with port %v are already created, will skip the creation", + e.Port, e.TargetPort) + } - return initiatorCreationRequired, targetCreationRequired, err + return initiatorCreationRequired, targetCreationRequired, nil }
2177-2180: Add error handling for GetIPForPodThe error from
GetIPForPodshould include more context about why the IP retrieval failed.podIP, err := commonnet.GetIPForPod() if err != nil { - return err + return errors.Wrapf(err, "failed to get pod IP for engine %s target switchover", e.Name) }
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
📒 Files selected for processing (2)
pkg/spdk/engine.go(16 hunks)pkg/spdk/engine_test.go(1 hunks)
🔇 Additional comments (2)
pkg/spdk/engine.go (2)
2224-2228: 🛠️ Refactor suggestion
Add port range validation
The port assignment should validate that the port number is within the valid range (1-65535).
if targetIP == podIP {
+ if targetPort <= 0 || targetPort > 65535 {
+ return fmt.Errorf("invalid target port %d: must be between 1 and 65535", targetPort)
+ }
e.TargetPort = targetPort
e.StandbyTargetPort = 0
}Likely invalid or redundant comment.
108-109: Add unit tests for isNewEngine
The isNewEngine helper function lacks test coverage. Consider adding test cases to verify:
- New engine (all fields empty/zero)
- Existing engine with various field combinations
There was a problem hiding this comment.
Actionable comments posted: 3
🧹 Outside diff range and nitpick comments (3)
pkg/spdk/engine.go (3)
108-110: Add documentation for theisNewEnginehelper function.Consider adding a doc comment explaining the purpose and return value semantics of this helper function. This will help other developers understand when an engine is considered "new".
+// isNewEngine returns true if the engine instance has not been initialized with any IP addresses +// or standby target port. This helps determine if this is the first time the engine is being created. func (e *Engine) isNewEngine() bool { return e.IP == "" && e.TargetIP == "" && e.StandbyTargetPort == 0 }
429-444: Enhance error handling and logging in frontend handling.The deferred cleanup logic could benefit from more detailed error logging and state validation.
defer func() { if err == nil { if !standbyTargetCreationRequired { + e.log.Debug("Updating engine state with new frontend configuration") e.initiator = initiator e.dmDeviceBusy = dmDeviceBusy e.Endpoint = initiator.GetEndpoint() e.log = e.log.WithFields(logrus.Fields{ "endpoint": e.Endpoint, "port": e.Port, "targetPort": e.TargetPort, }) + } else { + e.log.Debug("Skipping engine state update due to standby target creation") } e.log.Infof("Finished handling frontend for engine: %+v", e) + } else { + e.log.WithError(err).Error("Failed to handle frontend") } }()
37-49: Add high-level documentation for the Engine struct.Consider adding comprehensive documentation for the Engine struct and its fields, particularly explaining the role of the new
StandbyTargetPortfield in the live upgrade process.+// Engine represents a Longhorn SPDK engine instance that manages volume data access. +// It handles the coordination between initiators and targets, supporting features like +// live upgrades through standby target ports. type Engine struct { sync.RWMutex Name string VolumeName string SpecSize uint64 ActualSize uint64 IP string Port int32 // Port that initiator is connecting to TargetIP string TargetPort int32 // Port of the target that is used for letting initiator connect to + // StandbyTargetPort is used during live upgrades to temporarily store the port + // of a standby target instance, enabling seamless switchover between targets StandbyTargetPort int32
derekbit
force-pushed
the
v2-control-upgrade
branch
from
9b9c0f4
to
4231599
Compare
There was a problem hiding this comment.
Actionable comments posted: 3
🧹 Outside diff range and nitpick comments (6)
pkg/spdk/engine_test.go (2)
99-100: Improve test execution pattern.Replace index-based iteration with value-based iteration for better readability and maintainability.
- for testName, testCase := range testCases { - c.Logf("testing checkInitiatorAndTargetCreationRequirements.%v", testName) + for _, tc := range testCases { + c.Logf("testing %s", tc.name)
165-167: Improve test execution pattern.Replace index-based iteration with value-based iteration for better readability and maintainability.
- for testName, testCase := range testCases { - c.Logf("testing isNewEngine.%v", testName) - result := testCase.engine.isNewEngine() + for _, tc := range testCases { + c.Logf("testing %s", tc.name) + result := tc.engine.isNewEngine()pkg/spdk_test.go (1)
1228-1228: Consider extracting revertSnapshot helper function.The
revertSnapshotfunction contains two identicalEngineCreatecalls that could be refactored into a helper function to improve maintainability and reduce code duplication.+ func createEngine(spdkCli *client.SPDKClient, engineName, volumeName string, frontend types.Frontend, size uint64, replicaAddressMap map[string]string, revCount int32, ip string) (*api.Engine, error) { + return spdkCli.EngineCreate(engineName, volumeName, frontend, size, replicaAddressMap, revCount, ip, ip, false) + }Also applies to: 1243-1243
pkg/spdk/engine.go (3)
112-142: Simplify complex creation requirements logicThe function has complex branching logic that could be simplified for better maintainability. Consider:
- Breaking down the conditions into smaller, well-named helper functions
- Adding detailed comments explaining each condition
- Using early returns to reduce nesting
func (e *Engine) checkInitiatorAndTargetCreationRequirements(podIP, initiatorIP, targetIP string) (bool, bool, error) { + // Check if pod is neither initiator nor target + if podIP != initiatorIP && podIP != targetIP { + return false, false, fmt.Errorf( + "invalid initiator and target addresses for engine %s creation with initiator address %v and target address %v", + e.Name, initiatorIP, targetIP) + } + + // Case 1: Pod is both initiator and target + if podIP == initiatorIP && podIP == targetIP { + return e.checkLocalCreationRequirements() + } + + // Case 2: Pod is only initiator + if podIP == initiatorIP { + e.log.Info("Creating an initiator instance") + return true, false, nil + } + + // Case 3: Pod is only target + e.log.Info("Creating a target instance") + return false, true, nil } +func (e *Engine) checkLocalCreationRequirements() (bool, bool, error) { + switch { + case e.Port == 0 && e.TargetPort == 0: + e.log.Info("Creating both initiator and target instances") + return true, true, nil + case e.Port != 0 && e.TargetPort == 0: + if e.StandbyTargetPort != 0 { + e.log.Warnf("Standby target instance with port %v is already created, will skip the target creation", + e.StandbyTargetPort) + return false, false, nil + } + e.log.Info("Creating a target instance") + return false, true, nil + default: + e.log.Infof("Initiator instance with port %v and target instance with port %v are already created, will skip the creation", + e.Port, e.TargetPort) + return false, false, nil + } +}
398-399: Add validation for standby target creationThe standby target creation logic should include additional validation:
- Verify port ranges
- Validate state transitions
- Add logging for standby target creation
func (e *Engine) handleFrontend(spdkClient *spdkclient.Client, superiorPortAllocator *commonbitmap.Bitmap, portCount int32, targetAddress string, initiatorCreationRequired, targetCreationRequired bool) (err error) { + if superiorPortAllocator == nil { + return fmt.Errorf("invalid nil port allocator") + } + if portCount <= 0 { + return fmt.Errorf("invalid port count: %d", portCount) + } standbyTargetCreationRequired := false if e.Port != 0 && e.TargetPort == 0 { + e.log.Info("Standby target creation required") standbyTargetCreationRequired = true }Also applies to: 409-411
Line range hint
2139-2236: Enhance error handling and recovery in target switchoverThe target switchover logic should include:
- Atomic state transitions
- Rollback mechanism for failed operations
- Improved error context
func (e *Engine) SwitchOverTarget(spdkClient *spdkclient.Client, newTargetAddress string) (err error) { + // Track original state for rollback + originalState := struct { + targetIP string + targetPort int32 + standbyPort int32 + }{ + targetIP: e.TargetIP, + targetPort: e.TargetPort, + standbyPort: e.StandbyTargetPort, + } + + // Rollback function + rollback := func() { + e.TargetIP = originalState.targetIP + e.TargetPort = originalState.targetPort + e.StandbyTargetPort = originalState.standbyPort + e.log.Info("Rolled back to original state after failed switchover") + } // ... existing code ... if err := e.connectTarget(newTargetAddress); err != nil { + rollback() return errors.Wrapf(err, "failed to connect target %s for engine %s", newTargetAddress, e.Name) }
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
⛔ Files ignored due to path filters (4)
go.modis excluded by!go.modgo.sumis excluded by!**/*.sum,!go.sumvendor/github.com/longhorn/types/pkg/generated/spdkrpc/spdk.pb.gois excluded by!**/*.pb.go,!**/generated/**,!vendor/**vendor/modules.txtis excluded by!vendor/**
📒 Files selected for processing (6)
pkg/api/types.go(2 hunks)pkg/client/client.go(1 hunks)pkg/spdk/engine.go(18 hunks)pkg/spdk/engine_test.go(1 hunks)pkg/spdk/server.go(1 hunks)pkg/spdk_test.go(6 hunks)
🚧 Files skipped from review as they are similar to previous changes (3)
- pkg/api/types.go
- pkg/client/client.go
- pkg/spdk/server.go
🔇 Additional comments (7)
pkg/spdk/engine_test.go (1)
13-98: 🛠️ Refactor suggestion
Add test cases for port validation edge cases.
The test suite should include validation for edge cases related to port values:
- Negative port values
- Port values exceeding valid ranges
- Port conflicts between active and standby targets
Add these test cases:
}{
+ {
+ name: "Negative port values",
+ podIP: "192.168.1.1",
+ initiatorIP: "192.168.1.1",
+ targetIP: "192.168.1.1",
+ port: -1,
+ targetPort: -2,
+ standbyTargetPort: -3,
+ expectedInitiatorCreationRequired: false,
+ expectedTargetCreationRequired: false,
+ expectedError: fmt.Errorf("invalid port values"),
+ },
+ {
+ name: "Port conflict between target and standby",
+ podIP: "192.168.1.1",
+ initiatorIP: "192.168.1.1",
+ targetIP: "192.168.1.1",
+ port: 100,
+ targetPort: 8000,
+ standbyTargetPort: 8000,
+ expectedInitiatorCreationRequired: false,
+ expectedTargetCreationRequired: false,
+ expectedError: fmt.Errorf("port conflict: standby target port must be different from active target port"),
+ },Likely invalid or redundant comment.
pkg/spdk_test.go (3)
271-271: LGTM! Parameter removal aligns with engine upgrade support.
The removal of the upgradeRequired parameter from EngineCreate call is consistent with the PR objectives to support engine live upgrades.
354-354: Skip duplicate review.
This change is identical to the previous one, removing the upgradeRequired parameter.
530-530: Skip duplicate review.
This change is identical to the previous ones, removing the upgradeRequired parameter.
pkg/spdk/engine.go (3)
37-49: LGTM: Engine struct modification for standby target support
The addition of StandbyTargetPort field is well-structured and aligns with the existing port management pattern in the Engine struct.
108-110: Add unit tests for isNewEngine function
The function logic is correct, but it requires test coverage to ensure reliability across different engine states.
2406-2426: 🛠️ Refactor suggestion
Add validation for port release operations
The port release logic should:
- Validate port numbers before release
- Handle allocation bitmap errors gracefully
- Add logging for successful releases
func (e *Engine) releaseTargetAndStandbyTargetPorts(superiorPortAllocator *commonbitmap.Bitmap) error {
+ if superiorPortAllocator == nil {
+ return fmt.Errorf("invalid nil port allocator")
+ }
+
+ // Validate port numbers
+ if e.TargetPort < 0 || e.TargetPort > 65535 {
+ return fmt.Errorf("invalid target port number: %d", e.TargetPort)
+ }
+ if e.StandbyTargetPort < 0 || e.StandbyTargetPort > 65535 {
+ return fmt.Errorf("invalid standby target port number: %d", e.StandbyTargetPort)
+ }
releaseTargetPortRequired := e.TargetPort != 0
releaseStandbyTargetPortRequired := e.StandbyTargetPort != 0 && e.StandbyTargetPort != e.TargetPort
if releaseTargetPortRequired {
if err := superiorPortAllocator.ReleaseRange(e.TargetPort, e.TargetPort); err != nil {
return errors.Wrapf(err, "failed to release target port %d", e.TargetPort)
}
+ e.log.Infof("Successfully released target port %d", e.TargetPort)
}
e.TargetPort = 0
if releaseStandbyTargetPortRequired {
if err := superiorPortAllocator.ReleaseRange(e.StandbyTargetPort, e.StandbyTargetPort); err != nil {
return errors.Wrapf(err, "failed to release standby target port %d", e.StandbyTargetPort)
}
+ e.log.Infof("Successfully released standby target port %d", e.StandbyTargetPort)
}
e.StandbyTargetPort = 0
return nil
}Likely invalid or redundant comment.
There was a problem hiding this comment.
Actionable comments posted: 3
🧹 Outside diff range and nitpick comments (5)
pkg/api/types.go (1)
43-53: LGTM! Consider adding documentation.The new NVMe-related types are well-structured and follow Go conventions. The fields in
NvmeDevicePathcorrectly capture NVMe-oF transport parameters.Consider adding GoDoc comments to document these types, especially explaining:
- The purpose and usage of each field in
NvmeDevicePath- The relationship between paths in
NvmeSubsystempkg/spdk/engine.go (4)
37-50: Add documentation for the StandbyTargetPort fieldConsider adding a comment explaining the purpose and lifecycle of the
StandbyTargetPortfield, similar to howPortandTargetPortare documented. This would help clarify when this port is used during live upgrades.TargetPort int32 // Port of the target that is used for letting initiator connect to + StandbyTargetPort int32 // Port used for the standby target during live upgrades
117-147: Improve error message clarity and add test coverageThe function has complex branching logic that determines when to create initiator and target instances. Consider:
- The error message in the else branch could be more descriptive by including the pod IP
- The function would benefit from comprehensive test coverage
- err = fmt.Errorf("invalid initiator and target addresses for engine %s creation with initiator address %v and target address %v", e.Name, initiatorIP, targetIP) + err = fmt.Errorf("invalid initiator and target addresses for engine %s creation: pod IP %v does not match initiator IP %v or target IP %v", e.Name, podIP, initiatorIP, targetIP)
Line range hint
2166-2264: Enhance error handling in target switchoverThe target switchover implementation is thorough but could benefit from:
- More detailed error messages when specific operations fail
- Structured logging for better debugging
- Consider adding metrics or events for monitoring switchover operations
if err := e.disconnectTarget(currentTargetAddress); err != nil { - return errors.Wrapf(err, "failed to disconnect target %s for engine %s", currentTargetAddress, e.Name) + return errors.Wrapf(err, "failed to disconnect target %s for engine %s: %v", currentTargetAddress, e.Name, err) } + e.log.WithFields(logrus.Fields{ + "currentTarget": currentTargetAddress, + "newTarget": newTargetAddress, + "engineName": e.Name, + }).Info("Successfully disconnected from current target")
Line range hint
2107-2114: Improve error messaging in Suspend and ResumeThe error handling in these state transition functions could be more descriptive:
- Add context about the current state when errors occur
- Include more details in log messages
- e.log.WithError(err).Warn("Failed to suspend engine") + e.log.WithFields(logrus.Fields{ + "currentState": e.State, + "error": err, + }).Error("Failed to suspend engine") - e.log.WithError(err).Warn("Failed to resume engine") + e.log.WithFields(logrus.Fields{ + "currentState": e.State, + "error": err, + }).Error("Failed to resume engine")Also applies to: 2138-2145
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
⛔ Files ignored due to path filters (5)
go.modis excluded by!go.modgo.sumis excluded by!**/*.sum,!go.sumvendor/github.com/longhorn/go-spdk-helper/pkg/nvme/initiator.gois excluded by!vendor/**vendor/github.com/longhorn/types/pkg/generated/spdkrpc/spdk.pb.gois excluded by!**/*.pb.go,!**/generated/**,!vendor/**vendor/modules.txtis excluded by!vendor/**
📒 Files selected for processing (2)
pkg/api/types.go(5 hunks)pkg/spdk/engine.go(20 hunks)
🔇 Additional comments (3)
pkg/api/types.go (2)
177-177: LGTM! Field mappings are correct.
The new field mappings in ProtoEngineToEngine are properly implemented and maintain consistency with existing patterns.
Also applies to: 186-186
144-144: LGTM! Verify field usage across the codebase.
The new fields support engine live upgrade functionality:
StandbyTargetPortfor managing the standby target during upgradesNvmeSubsystemfor tracking NVMe paths
Let's verify the usage of these new fields:
Also applies to: 153-153
✅ Verification successful
LGTM! Both fields are properly integrated across the codebase
The verification shows comprehensive and consistent usage:
-
StandbyTargetPortis well-integrated:- Properly defined in protobuf and API types
- Used in engine operations for managing standby targets
- Covered by unit tests including various scenarios
- Has proper release mechanisms in place
-
NvmeSubsystemshows complete implementation:- Defined in both API types and protobuf
- Integrated into engine operations for NVMe path management
- Used consistently for device path handling
🏁 Scripts executed
The following scripts were executed for the analysis:
Script:
#!/bin/bash
# Description: Check usage of new Engine fields across the codebase
# Check StandbyTargetPort usage
echo "=== StandbyTargetPort usage ==="
rg "StandbyTargetPort|standby_target_port" --type go
# Check NvmeSubsystem usage
echo "=== NvmeSubsystem usage ==="
rg "NvmeSubsystem|nvme_subsystem" --type go
Length of output: 6529
pkg/spdk/engine.go (1)
113-115: LGTM! Clear and concise helper function
The function provides a clear way to check if an engine is in its initial state.
derekbit
force-pushed
the
v2-control-upgrade
branch
from
752b14e
to
60c7419
Compare
There was a problem hiding this comment.
Actionable comments posted: 0
🧹 Outside diff range and nitpick comments (5)
pkg/spdk/engine.go (5)
113-115: Add documentation for isNewEngine methodConsider adding a doc comment explaining the criteria that determines a new engine state.
+// isNewEngine returns true if the engine is in its initial state, +// determined by empty IP, TargetIP and zero StandbyTargetPort. func (e *Engine) isNewEngine() bool { return e.IP == "" && e.TargetIP == "" && e.StandbyTargetPort == 0 }
117-147: Improve error messages in checkInitiatorAndTargetCreationRequirementsThe error message in the else clause could be more descriptive by including the pod IP.
- err = fmt.Errorf("invalid initiator and target addresses for engine %s creation with initiator address %v and target address %v", e.Name, initiatorIP, targetIP) + err = fmt.Errorf("invalid initiator and target addresses for engine %s creation: pod IP %v does not match initiator address %v or target address %v", e.Name, podIP, initiatorIP, targetIP)
Line range hint
2166-2263: Consider implementing transaction-like behavior for SwitchOverTargetThe method handles complex state changes that could benefit from a more structured approach to ensure atomicity and rollback capability.
Consider:
- Creating a transaction struct to track state changes
- Implementing rollback for each step
- Using defer to ensure cleanup on failure
Example approach:
type switchoverTransaction struct { oldIP string oldPort int32 oldTargetPort int32 oldStandbyPort int32 } func (t *switchoverTransaction) backup(e *Engine) { t.oldIP = e.IP t.oldPort = e.Port t.oldTargetPort = e.TargetPort t.oldStandbyPort = e.StandbyTargetPort } func (t *switchoverTransaction) rollback(e *Engine) { e.IP = t.oldIP e.Port = t.oldPort e.TargetPort = t.oldTargetPort e.StandbyPort = t.oldStandbyPort }
2107-2108: Standardize error handling in Suspend/Resume methodsThe error handling could be more consistent between the two methods.
Consider extracting common error handling logic:
func (e *Engine) handleOperationError(op string, err error) { if e.State != types.InstanceStateError { e.log.WithError(err).Errorf("Failed to %s engine", op) } e.ErrorMsg = err.Error() }Also applies to: 2138-2139
621-632: Extract NvmeSubsystem conversion to helper methodConsider extracting the NvmeSubsystem conversion logic to improve readability and reusability.
+func (e *Engine) convertNvmeSubsystem() *spdkrpc.NvmeSubsystem { + nvmeSubsystem := &spdkrpc.NvmeSubsystem{ + Paths: map[string]*spdkrpc.NvmeDevicePath{}, + } + for pathName, path := range e.NvmeSubsystem.Paths { + nvmeSubsystem.Paths[pathName] = &spdkrpc.NvmeDevicePath{ + Trtype: path.Trtype, + Traddr: path.Traddr, + Trsvcid: path.Trsvcid, + SrcAddr: path.SrcAddr, + State: path.State, + } + } + return nvmeSubsystem +}
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
⛔ Files ignored due to path filters (1)
vendor/github.com/longhorn/go-spdk-helper/pkg/nvme/initiator.gois excluded by!vendor/**
📒 Files selected for processing (2)
pkg/api/types.go(5 hunks)pkg/spdk/engine.go(20 hunks)
🚧 Files skipped from review as they are similar to previous changes (1)
- pkg/api/types.go
🔇 Additional comments (3)
pkg/spdk/engine.go (3)
37-50: LGTM: Engine struct changes
The addition of StandbyTargetPort field is well-placed and logically grouped with other port-related fields.
2396-2428: 🛠️ Refactor suggestion
Enhance error handling in DeleteTarget
The method should consider implementing cleanup tracking to ensure resources are properly released even when errors occur.
Consider:
- Tracking successful operations
- Implementing partial cleanup on failure
- Adding debug logging for cleanup steps
func (e *Engine) DeleteTarget(spdkClient *spdkclient.Client, superiorPortAllocator *commonbitmap.Bitmap) (err error) {
+ type cleanupState struct {
+ exposeBdevStopped bool
+ portsReleased bool
+ raidDeleted bool
+ replicasDisconnected map[string]bool
+ }
+ state := &cleanupState{
+ replicasDisconnected: make(map[string]bool),
+ }
+ defer func() {
+ if err != nil {
+ e.log.WithField("cleanupState", state).Error("Target deletion failed, partial cleanup state recorded")
+ }
+ }()Likely invalid or redundant comment.
2433-2453: 🛠️ Refactor suggestion
Add port validation in releaseTargetAndStandbyTargetPorts
The method should validate port numbers before attempting to release them.
func (e *Engine) releaseTargetAndStandbyTargetPorts(superiorPortAllocator *commonbitmap.Bitmap) error {
+ if superiorPortAllocator == nil {
+ return fmt.Errorf("invalid nil port allocator")
+ }
+
releaseTargetPortRequired := e.TargetPort != 0
releaseStandbyTargetPortRequired := e.StandbyTargetPort != 0 && e.StandbyTargetPort != e.TargetPort
+ // Validate port numbers
+ if releaseTargetPortRequired && (e.TargetPort < 1 || e.TargetPort > 65535) {
+ return fmt.Errorf("invalid target port number: %d", e.TargetPort)
+ }
+ if releaseStandbyTargetPortRequired && (e.StandbyTargetPort < 1 || e.StandbyTargetPort > 65535) {
+ return fmt.Errorf("invalid standby target port number: %d", e.StandbyTargetPort)
+ }Likely invalid or redundant comment.
derekbit
force-pushed
the
v2-control-upgrade
branch
from
60c7419
to
59d33f4
Compare
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Outside diff range and nitpick comments (6)
pkg/api/types.go (2)
43-53: Add documentation for the new NVMe types and fields.The new types
NvmeDevicePathandNvmeSubsystemwould benefit from documentation explaining:
- The purpose of each type
- The meaning and expected values for each field
- Any constraints or requirements for the fields
Add documentation like this:
+// NvmeDevicePath represents an NVMe over Fabrics (NVMeoF) transport path configuration type NvmeDevicePath struct { + // Trtype specifies the transport type (e.g., "tcp", "rdma") Trtype string `json:"trtype"` + // Traddr specifies the transport address (e.g., IP address) Traddr string `json:"traddr"` // ... (continue for other fields) }
144-144: Document the new Engine fields.The new fields
StandbyTargetPortandNvmeSubsystemwould benefit from documentation explaining their roles in the live upgrade process.Add documentation like this:
type Engine struct { // ... existing fields ... + // StandbyTargetPort is the port number for the standby target used during live upgrades StandbyTargetPort int32 `json:"standby_target_port"` // ... other fields ... + // NvmeSubsystem represents the NVMe subsystem configuration for this engine NvmeSubsystem NvmeSubsystem `json:"nvme_subsystem"` }Also applies to: 153-153
pkg/spdk/engine.go (4)
113-115: Add documentation for isNewEngine functionConsider adding a function comment explaining the criteria for determining a new engine state.
+// isNewEngine returns true if the engine is in its initial state, +// determined by empty IP, TargetIP and zero StandbyTargetPort. func (e *Engine) isNewEngine() bool { return e.IP == "" && e.TargetIP == "" && e.StandbyTargetPort == 0 }
117-147: Simplify complex creation requirements logicThe function has good logging but could be more maintainable with simplified conditions.
Consider refactoring into smaller, more focused functions:
+func (e *Engine) shouldCreateBothInstances(podIP, initiatorIP, targetIP string) bool { + return podIP == initiatorIP && podIP == targetIP && e.Port == 0 && e.TargetPort == 0 +} +func (e *Engine) shouldCreateTargetOnly(podIP, targetIP string) bool { + return podIP == targetIP && e.Port != 0 && e.TargetPort == 0 && e.StandbyTargetPort == 0 +} func (e *Engine) checkInitiatorAndTargetCreationRequirements(podIP, initiatorIP, targetIP string) (bool, bool, error) { initiatorCreationRequired, targetCreationRequired := false, false - if podIP == initiatorIP && podIP == targetIP { - if e.Port == 0 && e.TargetPort == 0 { - // ... existing code - } + if e.shouldCreateBothInstances(podIP, initiatorIP, targetIP) { + e.log.Info("Creating both initiator and target instances") + return true, true, nil + } + if e.shouldCreateTargetOnly(podIP, targetIP) { + e.log.Info("Creating a target instance") + return false, true, nil } // ... rest of the function }
Line range hint
2166-2264: Improve error handling in target switchoverThe target switchover logic should use a transaction-like pattern to ensure proper cleanup on failure.
Consider implementing a rollback mechanism:
+type switchoverState struct { + oldTargetAddress string + oldTargetPort int32 + oldStandbyPort int32 + disconnected bool +} + +func (e *Engine) rollbackSwitchover(state *switchoverState) error { + if state.disconnected { + if err := e.connectTarget(state.oldTargetAddress); err != nil { + return fmt.Errorf("rollback failed: %v", err) + } + e.TargetPort = state.oldTargetPort + e.StandbyTargetPort = state.oldStandbyPort + } + return nil +} func (e *Engine) SwitchOverTarget(spdkClient *spdkclient.Client, newTargetAddress string) (err error) { + state := &switchoverState{ + oldTargetPort: e.TargetPort, + oldStandbyPort: e.StandbyTargetPort, + } + defer func() { + if err != nil { + if rollbackErr := e.rollbackSwitchover(state); rollbackErr != nil { + e.log.WithError(rollbackErr).Error("Switchover rollback failed") + } + } + }()
2107-2108: Enhance error logging in Suspend/ResumeThe error logging could provide more context about the failure.
- e.log.WithError(err).Warn("Failed to suspend engine") + e.log.WithError(err).WithFields(logrus.Fields{ + "state": e.State, + "frontend": e.Frontend, + }).Warn("Engine suspension failed") - e.log.WithError(err).Warn("Failed to resume engine") + e.log.WithError(err).WithFields(logrus.Fields{ + "state": e.State, + "frontend": e.Frontend, + }).Warn("Engine resume failed")Also applies to: 2138-2139
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
⛔ Files ignored due to path filters (4)
go.modis excluded by!go.modgo.sumis excluded by!**/*.sum,!go.sumvendor/github.com/longhorn/go-spdk-helper/pkg/nvme/initiator.gois excluded by!vendor/**vendor/modules.txtis excluded by!vendor/**
📒 Files selected for processing (2)
pkg/api/types.go(5 hunks)pkg/spdk/engine.go(20 hunks)
🔇 Additional comments (5)
pkg/api/types.go (2)
157-168: LGTM! The NvmeSubsystem initialization and mapping look good.
The implementation:
- Properly initializes the NvmeSubsystem with an empty path map
- Correctly copies paths from the protobuf object
- Safely handles the new StandbyTargetPort field
Also applies to: 179-179, 188-188
144-144: Verify the usage of new fields across the codebase.
Let's ensure the new fields are properly handled throughout the codebase.
Also applies to: 153-153, 157-168, 179-179, 188-188
✅ Verification successful
New fields are properly integrated across the codebase
The verification shows comprehensive and consistent usage of the new fields:
StandbyTargetPortis well-integrated with proper initialization, handling, and cleanup inpkg/spdk/engine.go- Proper port management through
releaseTargetAndStandbyTargetPortsfunction - Complete test coverage in
pkg/spdk/engine_test.go NvmeSubsystemand related structures are properly defined and used across the codebase- No potential nil pointer dereference issues found
🏁 Scripts executed
The following scripts were executed for the analysis:
Script:
#!/bin/bash
# Description: Check usage of StandbyTargetPort and NvmeSubsystem fields
# Check for StandbyTargetPort field usage
echo "=== Checking StandbyTargetPort usage ==="
rg "StandbyTargetPort|standby_target_port" --type go
# Check for NvmeSubsystem related code
echo -e "\n=== Checking NvmeSubsystem usage ==="
rg "NvmeSubsystem|NvmeDevicePath|nvme_subsystem" --type go
# Check for potential nil pointer dereferences
echo -e "\n=== Checking for potential nil checks ==="
ast-grep --pattern 'if $x.NvmeSubsystem != nil { $$$ }'
Length of output: 9235
pkg/spdk/engine.go (3)
37-50: LGTM: Engine struct changes for standby target support
The addition of StandbyTargetPort field is well-structured and properly integrated with the existing fields. The field ordering and type are consistent with other port-related fields.
2433-2453: 🛠️ Refactor suggestion
Add validation for port release operations
The port release logic should validate port numbers before attempting to release them.
func (e *Engine) releaseTargetAndStandbyTargetPorts(superiorPortAllocator *commonbitmap.Bitmap) error {
+ if superiorPortAllocator == nil {
+ return fmt.Errorf("invalid nil port allocator")
+ }
+
releaseTargetPortRequired := e.TargetPort != 0
releaseStandbyTargetPortRequired := e.StandbyTargetPort != 0 && e.StandbyTargetPort != e.TargetPort
+ // Validate port numbers
+ if releaseTargetPortRequired && (e.TargetPort < 1 || e.TargetPort > 65535) {
+ return fmt.Errorf("invalid target port number: %d", e.TargetPort)
+ }
+ if releaseStandbyTargetPortRequired && (e.StandbyTargetPort < 1 || e.StandbyTargetPort > 65535) {
+ return fmt.Errorf("invalid standby target port number: %d", e.StandbyTargetPort)
+ }Likely invalid or redundant comment.
2396-2428: 🛠️ Refactor suggestion
Improve cleanup handling in DeleteTarget
The target deletion should handle partial failures better and ensure proper cleanup order.
Consider implementing a cleanup tracker:
+type cleanupTracker struct {
+ exposeBdevStopped bool
+ portsReleased bool
+ raidDeleted bool
+ replicasDisconnected map[string]bool
+}
func (e *Engine) DeleteTarget(spdkClient *spdkclient.Client, superiorPortAllocator *commonbitmap.Bitmap) (err error) {
+ tracker := &cleanupTracker{
+ replicasDisconnected: make(map[string]bool),
+ }
+ defer func() {
+ if err != nil {
+ e.rollbackFailedDeletion(tracker)
+ }
+ }()Likely invalid or redundant comment.
derekbit
force-pushed
the
v2-control-upgrade
branch
from
59d33f4
to
1774180
Compare
There was a problem hiding this comment.
Actionable comments posted: 2
🧹 Outside diff range and nitpick comments (2)
pkg/spdk/engine.go (2)
37-50: Add documentation for the newStandbyTargetPortfieldThe new field needs documentation to explain its purpose and relationship with other port fields (
Port,TargetPort). Consider adding comments explaining:
- When StandbyTargetPort is used vs TargetPort
- Valid value ranges and constraints
- State transitions during target switchover
117-147: Simplify complex branching logicThe function has nested if conditions that make it hard to follow. Consider refactoring to:
- Extract the conditions into well-named boolean variables
- Use early returns to reduce nesting
- Add debug logging for better observability
func (e *Engine) checkInitiatorAndTargetCreationRequirements(podIP, initiatorIP, targetIP string) (bool, bool, error) { + isLocalPod := podIP == initiatorIP && podIP == targetIP + isInitiatorOnly := podIP == initiatorIP && podIP != targetIP + isTargetOnly := podIP != initiatorIP && podIP == targetIP + + if !isLocalPod && !isInitiatorOnly && !isTargetOnly { + return false, false, fmt.Errorf("invalid initiator and target addresses for engine %s creation with initiator address %v and target address %v", + e.Name, initiatorIP, targetIP) + } + + if isInitiatorOnly { + e.log.Info("Creating an initiator instance") + return true, false, nil + } + + if isTargetOnly { + e.log.Info("Creating a target instance") + return false, true, nil + } + + // Handle local pod case if e.Port == 0 && e.TargetPort == 0 { e.log.Info("Creating both initiator and target instances") return true, true, nil } if e.Port != 0 && e.TargetPort == 0 { e.log.Info("Creating a target instance") if e.StandbyTargetPort != 0 { e.log.Warnf("Standby target instance with port %v is already created, will skip the target creation", e.StandbyTargetPort) return false, false, nil } return false, true, nil } e.log.Infof("Initiator instance with port %v and target instance with port %v are already created, will skip the creation", e.Port, e.TargetPort) return false, false, nil }
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
⛔ Files ignored due to path filters (4)
go.modis excluded by!go.modgo.sumis excluded by!**/*.sum,!go.sumvendor/github.com/longhorn/go-spdk-helper/pkg/nvme/initiator.gois excluded by!vendor/**vendor/modules.txtis excluded by!vendor/**
📒 Files selected for processing (2)
pkg/api/types.go(5 hunks)pkg/spdk/engine.go(20 hunks)
🚧 Files skipped from review as they are similar to previous changes (1)
- pkg/api/types.go
🔇 Additional comments (1)
pkg/spdk/engine.go (1)
2433-2453: 🛠️ Refactor suggestion
Add port validation and improve error handling
The port release logic should include validation and better error handling:
- Validate port numbers before release
- Add cleanup logic for partial failures
- Consider using constants for port ranges
func (e *Engine) releaseTargetAndStandbyTargetPorts(superiorPortAllocator *commonbitmap.Bitmap) error {
+ const (
+ minPort = 1
+ maxPort = 65535
+ )
+
+ if superiorPortAllocator == nil {
+ return fmt.Errorf("invalid nil port allocator")
+ }
+
releaseTargetPortRequired := e.TargetPort != 0
releaseStandbyTargetPortRequired := e.StandbyTargetPort != 0 && e.StandbyTargetPort != e.TargetPort
+ // Validate port numbers
+ if releaseTargetPortRequired && (e.TargetPort < minPort || e.TargetPort > maxPort) {
+ return fmt.Errorf("invalid target port number: %d", e.TargetPort)
+ }
+ if releaseStandbyTargetPortRequired && (e.StandbyTargetPort < minPort || e.StandbyTargetPort > maxPort) {
+ return fmt.Errorf("invalid standby target port number: %d", e.StandbyTargetPort)
+ }
+
// Release the target port
if releaseTargetPortRequired {
if err := superiorPortAllocator.ReleaseRange(e.TargetPort, e.TargetPort); err != nil {
return errors.Wrapf(err, "failed to release target port %d", e.TargetPort)
}
+ e.log.Infof("Released target port %d", e.TargetPort)
}
e.TargetPort = 0Likely invalid or redundant comment.
There was a problem hiding this comment.
Actionable comments posted: 2
🧹 Outside diff range and nitpick comments (2)
pkg/spdk/engine.go (2)
Line range hint
2197-2294: Add port validation in target switchoverThe target switchover logic should validate port numbers before proceeding with the operation to prevent invalid states.
Add port validation:
func (e *Engine) SwitchOverTarget(spdkClient *spdkclient.Client, newTargetAddress string) (err error) { + if newTargetPort <= 0 || newTargetPort > 65535 { + return fmt.Errorf("invalid target port %d for engine %s target switchover", newTargetPort, e.Name) + }
2427-2484: Improve error handling in DeleteTargetThe target deletion process should use a transaction-like pattern to ensure proper cleanup even in case of partial failures. Consider tracking the state of each operation and implementing rollback capabilities.
Consider implementing a cleanup tracker:
+type cleanupState struct { + exposeBdevStopped bool + portsReleased bool + raidDeleted bool + replicasDisconnected map[string]bool +} + func (e *Engine) DeleteTarget(spdkClient *spdkclient.Client, superiorPortAllocator *commonbitmap.Bitmap) (err error) { + state := &cleanupState{ + replicasDisconnected: make(map[string]bool), + } + defer func() { + if err != nil { + e.rollbackFailedDeletion(state) + } + }()
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
⛔ Files ignored due to path filters (8)
go.modis excluded by!go.modgo.sumis excluded by!**/*.sum,!go.sumvendor/github.com/jinzhu/copier/.gitignoreis excluded by!vendor/**vendor/github.com/jinzhu/copier/Licenseis excluded by!vendor/**vendor/github.com/jinzhu/copier/README.mdis excluded by!vendor/**vendor/github.com/jinzhu/copier/copier.gois excluded by!vendor/**vendor/github.com/jinzhu/copier/errors.gois excluded by!vendor/**vendor/modules.txtis excluded by!vendor/**
📒 Files selected for processing (1)
pkg/spdk/engine.go(22 hunks)
🧰 Additional context used
🪛 golangci-lint (1.62.2)
pkg/spdk/engine.go
692-692: Error return value of copier.Copy is not checked
(errcheck)
668-668: SA4000: identical expressions on the left and right side of the '!=' operator
(staticcheck)
🪛 GitHub Check: CodeFactor
pkg/spdk/engine.go
[notice] 667-667: pkg/spdk/engine.go#L667
Redefinition of the built-in function new. (redefines-builtin-id)
🔇 Additional comments (1)
pkg/spdk/engine.go (1)
39-52: LGTM: Engine struct changes are well-structured
The addition of StandbyTargetPort field with proper documentation aligns well with the live upgrade support objective.
derekbit
force-pushed
the
v2-control-upgrade
branch
from
bd9951c
to
99f98a7
Compare
There was a problem hiding this comment.
Actionable comments posted: 2
🧹 Outside diff range and nitpick comments (5)
pkg/api/types.go (1)
45-55: Consider adding documentation for the new typesThe NvmeDevicePath and NvmeSubsystem types are well-structured and align with NVMe-oF specifications. Consider adding godoc comments to describe:
- The purpose of each type
- The significance of each field
- Valid values for the State field
Example documentation:
// NvmeDevicePath represents an NVMe-oF connection path // with transport and addressing information type NvmeDevicePath struct { // Trtype specifies the NVMe transport type (e.g., "tcp", "rdma") Trtype string `json:"trtype"` // ... (document other fields) }pkg/spdk/engine.go (4)
105-108: Add test coverage for NvmeSubsystem initialization.The NvmeSubsystem initialization in NewEngine is not covered by tests. Consider adding test cases to verify proper initialization of the NvmeSubsystem and its Paths map.
Would you like me to help create test cases for the NvmeSubsystem initialization?
🧰 Tools
🪛 GitHub Check: codecov/patch
[warning] 105-108: pkg/spdk/engine.go#L105-L108
Added lines #L105 - L108 were not covered by tests
Line range hint
2211-2308: LGTM! Consider adding more detailed logging.The SwitchOverTarget implementation has good error handling and rollback capabilities. Consider adding more detailed logging for better observability during the switchover process.
Add debug logs before critical operations:
+ e.log.Debugf("Current target state - IP: %s, Port: %d, StandbyPort: %d", e.TargetIP, e.TargetPort, e.StandbyTargetPort) if newTargetIP == podIP { e.TargetPort = newTargetPort e.StandbyTargetPort = 0 } else { e.StandbyTargetPort = e.TargetPort e.TargetPort = 0 } + e.log.Debugf("New target state - IP: %s, Port: %d, StandbyPort: %d", e.TargetIP, e.TargetPort, e.StandbyTargetPort)
2478-2498: Add port validation before release.The releaseTargetAndStandbyTargetPorts function should validate port numbers before attempting to release them.
Add validation:
func (e *Engine) releaseTargetAndStandbyTargetPorts(superiorPortAllocator *commonbitmap.Bitmap) error { + if superiorPortAllocator == nil { + return fmt.Errorf("invalid nil port allocator") + } + releaseTargetPortRequired := e.TargetPort != 0 releaseStandbyTargetPortRequired := e.StandbyTargetPort != 0 && e.StandbyTargetPort != e.TargetPort + // Validate port numbers + if releaseTargetPortRequired && (e.TargetPort <= 0 || e.TargetPort > 65535) { + return fmt.Errorf("invalid target port number: %d", e.TargetPort) + } + if releaseStandbyTargetPortRequired && (e.StandbyTargetPort <= 0 || e.StandbyTargetPort > 65535) { + return fmt.Errorf("invalid standby target port number: %d", e.StandbyTargetPort) + }
214-220: Add test coverage for critical engine operations.Several critical operations lack test coverage, including:
- Initiator and target creation requirements
- Engine state management
- Port allocation and release
- Target switchover logic
Consider adding comprehensive test cases to verify these operations.
Would you like help creating test cases for these operations?
Also applies to: 222-226, 229-237
🧰 Tools
🪛 GitHub Check: codecov/patch
[warning] 214-220: pkg/spdk/engine.go#L214-L220
Added lines #L214 - L220 were not covered by tests
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
⛔ Files ignored due to path filters (12)
go.modis excluded by!go.modgo.sumis excluded by!**/*.sum,!go.sumvendor/github.com/jinzhu/copier/.gitignoreis excluded by!vendor/**vendor/github.com/jinzhu/copier/Licenseis excluded by!vendor/**vendor/github.com/jinzhu/copier/README.mdis excluded by!vendor/**vendor/github.com/jinzhu/copier/copier.gois excluded by!vendor/**vendor/github.com/jinzhu/copier/errors.gois excluded by!vendor/**vendor/github.com/longhorn/go-spdk-helper/pkg/nvme/initiator.gois excluded by!vendor/**vendor/github.com/longhorn/go-spdk-helper/pkg/nvme/nvme.gois excluded by!vendor/**vendor/github.com/longhorn/go-spdk-helper/pkg/nvme/nvmecli.gois excluded by!vendor/**vendor/github.com/longhorn/go-spdk-helper/pkg/types/types.gois excluded by!vendor/**vendor/modules.txtis excluded by!vendor/**
📒 Files selected for processing (2)
pkg/api/types.go(6 hunks)pkg/spdk/engine.go(22 hunks)
🧰 Additional context used
🪛 GitHub Check: codecov/patch
pkg/spdk/engine.go
[warning] 105-108: pkg/spdk/engine.go#L105-L108
Added lines #L105 - L108 were not covered by tests
[warning] 135-137: pkg/spdk/engine.go#L135-L137
Added lines #L135 - L137 were not covered by tests
[warning] 151-151: pkg/spdk/engine.go#L151
Added line #L151 was not covered by tests
[warning] 214-220: pkg/spdk/engine.go#L214-L220
Added lines #L214 - L220 were not covered by tests
[warning] 222-226: pkg/spdk/engine.go#L222-L226
Added lines #L222 - L226 were not covered by tests
[warning] 229-237: pkg/spdk/engine.go#L229-L237
Added lines #L229 - L237 were not covered by tests
[warning] 269-270: pkg/spdk/engine.go#L269-L270
Added lines #L269 - L270 were not covered by tests
[warning] 275-275: pkg/spdk/engine.go#L275
Added line #L275 was not covered by tests
[warning] 277-277: pkg/spdk/engine.go#L277
Added line #L277 was not covered by tests
[warning] 288-291: pkg/spdk/engine.go#L288-L291
Added lines #L288 - L291 were not covered by tests
[warning] 298-298: pkg/spdk/engine.go#L298
Added line #L298 was not covered by tests
[warning] 308-309: pkg/spdk/engine.go#L308-L309
Added lines #L308 - L309 were not covered by tests
[warning] 312-320: pkg/spdk/engine.go#L312-L320
Added lines #L312 - L320 were not covered by tests
[warning] 406-406: pkg/spdk/engine.go#L406
Added line #L406 was not covered by tests
[warning] 416-418: pkg/spdk/engine.go#L416-L418
Added lines #L416 - L418 were not covered by tests
[warning] 427-434: pkg/spdk/engine.go#L427-L434
Added lines #L427 - L434 were not covered by tests
[warning] 436-447: pkg/spdk/engine.go#L436-L447
Added lines #L436 - L447 were not covered by tests
[warning] 449-449: pkg/spdk/engine.go#L449
Added line #L449 was not covered by tests
[warning] 453-467: pkg/spdk/engine.go#L453-L467
Added lines #L453 - L467 were not covered by tests
[warning] 469-471: pkg/spdk/engine.go#L469-L471
Added lines #L469 - L471 were not covered by tests
[warning] 473-473: pkg/spdk/engine.go#L473
Added line #L473 was not covered by tests
[warning] 476-478: pkg/spdk/engine.go#L476-L478
Added lines #L476 - L478 were not covered by tests
[warning] 481-481: pkg/spdk/engine.go#L481
Added line #L481 was not covered by tests
[warning] 484-487: pkg/spdk/engine.go#L484-L487
Added lines #L484 - L487 were not covered by tests
[warning] 489-500: pkg/spdk/engine.go#L489-L500
Added lines #L489 - L500 were not covered by tests
[warning] 506-507: pkg/spdk/engine.go#L506-L507
Added lines #L506 - L507 were not covered by tests
[warning] 515-516: pkg/spdk/engine.go#L515-L516
Added lines #L515 - L516 were not covered by tests
[warning] 520-520: pkg/spdk/engine.go#L520
Added line #L520 was not covered by tests
[warning] 522-524: pkg/spdk/engine.go#L522-L524
Added lines #L522 - L524 were not covered by tests
[warning] 623-634: pkg/spdk/engine.go#L623-L634
Added lines #L623 - L634 were not covered by tests
🪛 GitHub Check: CodeFactor
pkg/spdk/engine.go
[notice] 667-667: pkg/spdk/engine.go#L667
Redefinition of the built-in function new. (redefines-builtin-id)
🔇 Additional comments (5)
pkg/api/types.go (3)
7-8: LGTM: Import changes are appropriate
The addition of helpertypes import is necessary for the NVMeControllerState type used in the new NvmeDevicePath struct.
159-172: LGTM: Robust conversion implementation
The conversion logic:
- Properly initializes the NvmeSubsystem map
- Includes appropriate nil checks
- Creates deep copies of path objects
- Correctly handles the new StandbyTargetPort field
Also applies to: 183-183, 192-192
146-146: Verify port allocation strategy for StandbyTargetPort
The addition of StandbyTargetPort suggests a high-availability setup. Please ensure:
- Port allocation doesn't conflict with existing TargetPort
- Port range validation is implemented in the engine creation logic
Also applies to: 155-155
✅ Verification successful
Based on the results, I can now generate the final response as there's sufficient evidence about the port allocation and validation strategy.
Port allocation for StandbyTargetPort is properly implemented
The codebase shows proper handling of StandbyTargetPort:
- Port allocation uses the same
superiorPortAllocatorbitmap as the main TargetPort - Ports are properly released when no longer needed via
ReleaseRange - Port conflicts are prevented as the code checks for existing port allocations
- Port switchover logic correctly manages the transition between primary and standby ports
🏁 Scripts executed
The following scripts were executed for the analysis:
Script:
#!/bin/bash
# Search for port allocation and validation logic
rg -A 5 "StandbyTargetPort.*=|standbyTargetPort.*="
rg -A 5 "func.*Create.*Engine"
Length of output: 6047
pkg/spdk/engine.go (2)
115-117: LGTM! Clean and focused helper function.
The isNewEngine helper function has clear logic and follows the single responsibility principle.
130-134: 🛠️ Refactor suggestion
Remove redundant condition check.
The condition check for e.StandbyTargetPort != 0 is redundant as it's already checked in the parent if statement.
Apply this diff to simplify the logic:
- if e.StandbyTargetPort != 0 {
- e.log.Warnf("Standby target instance with port %v is already created, will skip the target creation", e.StandbyTargetPort)
- } else {
- targetCreationRequired = true
- }
+ if e.StandbyTargetPort != 0 {
+ e.log.Warnf("Standby target instance with port %v is already created, will skip the target creation", e.StandbyTargetPort)
+ return false, false, nil
+ }
+ targetCreationRequired = trueLikely invalid or redundant comment.
derekbit
force-pushed
the
v2-control-upgrade
branch
from
99f98a7
to
6e0cb47
Compare
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Outside diff range and nitpick comments (3)
pkg/spdk/engine.go (3)
37-46: Add documentation for the newStandbyTargetPortfield.The new field needs documentation to explain its purpose and usage in the context of live upgrades.
TargetPort int32 // Port of the target that is used for letting initiator connect to - StandbyTargetPort int32 + StandbyTargetPort int32 // Port used for the standby target during live upgrades
112-142: Simplify complex branching logic.The method has nested if conditions that make it hard to follow. Consider restructuring to reduce complexity and improve readability.
func (e *Engine) checkInitiatorAndTargetCreationRequirements(podIP, initiatorIP, targetIP string) (bool, bool, error) { - initiatorCreationRequired, targetCreationRequired := false, false - var err error + if podIP != initiatorIP && podIP != targetIP { + return false, false, fmt.Errorf("invalid initiator and target addresses for engine %s creation with initiator address %v and target address %v", + e.Name, initiatorIP, targetIP) + } - if podIP == initiatorIP && podIP == targetIP { - if e.Port == 0 && e.TargetPort == 0 { - e.log.Info("Creating both initiator and target instances") - initiatorCreationRequired = true - targetCreationRequired = true - } else if e.Port != 0 && e.TargetPort == 0 { - // ... existing code - } - } else if podIP == initiatorIP { - // ... existing code - } + initiatorCreationRequired := podIP == initiatorIP && e.Port == 0 + targetCreationRequired := podIP == targetIP && e.TargetPort == 0 && e.StandbyTargetPort == 0 + + if initiatorCreationRequired && targetCreationRequired { + e.log.Info("Creating both initiator and target instances") + } else if initiatorCreationRequired { + e.log.Info("Creating an initiator instance") + } else if targetCreationRequired { + e.log.Info("Creating a target instance") + } else { + e.log.Infof("Initiator instance with port %v and target instance with port %v are already created, will skip the creation", + e.Port, e.TargetPort) + } - return initiatorCreationRequired, targetCreationRequired, err + return initiatorCreationRequired, targetCreationRequired, nil }
Line range hint
2139-2236: Improve error handling and cleanup in target switchover.The error handling should ensure proper cleanup of resources in all failure scenarios. Consider using a deferred cleanup function.
func (e *Engine) SwitchOverTarget(spdkClient *spdkclient.Client, newTargetAddress string) (err error) { + type switchState struct { + oldTargetDisconnected bool + newTargetConnected bool + deviceReloaded bool + } + state := &switchState{} + + defer func() { + if err != nil && state.oldTargetDisconnected && !state.newTargetConnected { + if errRollback := e.connectTarget(currentTargetAddress); errRollback != nil { + e.log.WithError(errRollback).Error("Failed to rollback target switchover") + } + } + }() // ... existing code if err := e.disconnectTarget(currentTargetAddress); err != nil { return errors.Wrapf(err, "failed to disconnect target %s for engine %s", currentTargetAddress, e.Name) } + state.oldTargetDisconnected = true if err := e.connectTarget(newTargetAddress); err != nil { return errors.Wrapf(err, "failed to connect target %s for engine %s", newTargetAddress, e.Name) } + state.newTargetConnected = true
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
⛔ Files ignored due to path filters (8)
go.modis excluded by!go.modgo.sumis excluded by!**/*.sum,!go.sumvendor/github.com/longhorn/go-spdk-helper/pkg/nvme/initiator.gois excluded by!vendor/**vendor/github.com/longhorn/go-spdk-helper/pkg/nvme/nvme.gois excluded by!vendor/**vendor/github.com/longhorn/go-spdk-helper/pkg/nvme/nvmecli.gois excluded by!vendor/**vendor/github.com/longhorn/go-spdk-helper/pkg/types/types.gois excluded by!vendor/**vendor/github.com/longhorn/types/pkg/generated/spdkrpc/spdk.pb.gois excluded by!**/*.pb.go,!**/generated/**,!vendor/**vendor/modules.txtis excluded by!vendor/**
📒 Files selected for processing (6)
pkg/api/types.go(2 hunks)pkg/client/client.go(1 hunks)pkg/spdk/engine.go(18 hunks)pkg/spdk/engine_test.go(1 hunks)pkg/spdk/server.go(1 hunks)pkg/spdk_test.go(6 hunks)
🚧 Files skipped from review as they are similar to previous changes (4)
- pkg/api/types.go
- pkg/client/client.go
- pkg/spdk/engine_test.go
- pkg/spdk/server.go
🔇 Additional comments (7)
pkg/spdk_test.go (6)
354-354: Consistent API update for EngineCreate
530-530: Consistent API update for EngineCreate
1228-1228: Consistent API update for EngineCreate
1243-1243: Consistent API update for EngineCreate
1354-1354: Consistent API update for EngineCreate
271-271: API change: Simplified EngineCreate signature
The EngineCreate function signature has been updated to remove one boolean parameter, aligning with the PR's objective to support engine live upgrades.
Let's verify the consistency of this change across the codebase:
✅ Verification successful
Let me verify the function signature change by checking the actual implementation.
Let me try a different approach to find the EngineCreate function definition.
Let me check the proto file to understand the API changes better.
Based on the evidence gathered from the codebase:
- The client implementation in
pkg/client/client.goshows the currentEngineCreatesignature with only one boolean parametersalvageRequested - All test calls in
pkg/spdk_test.gohave been updated to use the new signature with a single boolean parameter - No instances of the old signature (with two boolean parameters) were found in the codebase
API change verified: EngineCreate signature update is consistent
The removal of one boolean parameter from the EngineCreate function has been consistently applied across the codebase. All test cases have been properly updated to match the new signature, and there are no remaining instances of the old two-boolean parameter signature.
🏁 Scripts executed
The following scripts were executed for the analysis:
Script:
#!/bin/bash
# Description: Verify all EngineCreate calls have been updated
# Expected: No instances of old signature with two boolean parameters
# Search for any remaining old-style calls
rg "EngineCreate.*,.*,.*,.*false.*false" --type go
# Search for updated calls to ensure consistency
rg "EngineCreate.*,.*,.*,.*false[^,]" --type go
Length of output: 1091
Script:
#!/bin/bash
# Search for EngineCreate function definition
ast-grep --pattern 'func $_EngineCreate($$$) {
$$$
}'
Length of output: 57
Script:
#!/bin/bash
# Search for EngineCreate function definition using ripgrep
rg "func.*EngineCreate" -A 5 --type go
Length of output: 10623
Script:
#!/bin/bash
# Search for EngineCreateRequest definition in proto files
rg "message EngineCreateRequest" -A 15 --type proto
Length of output: 83
pkg/spdk/engine.go (1)
2406-2426: 🛠️ Refactor suggestion
Add port validation in releaseTargetAndStandbyTargetPorts.
The method should validate port numbers before attempting to release them.
func (e *Engine) releaseTargetAndStandbyTargetPorts(superiorPortAllocator *commonbitmap.Bitmap) error {
+ if superiorPortAllocator == nil {
+ return fmt.Errorf("invalid nil port allocator")
+ }
+
releaseTargetPortRequired := e.TargetPort != 0
releaseStandbyTargetPortRequired := e.StandbyTargetPort != 0 && e.StandbyTargetPort != e.TargetPort
+ // Validate port numbers
+ if releaseTargetPortRequired && (e.TargetPort < 1 || e.TargetPort > 65535) {
+ return fmt.Errorf("invalid target port number: %d", e.TargetPort)
+ }
+ if releaseStandbyTargetPortRequired && (e.StandbyTargetPort < 1 || e.StandbyTargetPort > 65535) {
+ return fmt.Errorf("invalid standby target port number: %d", e.StandbyTargetPort)
+ }
// ... existing codeLikely invalid or redundant comment.
derekbit
force-pushed
the
v2-control-upgrade
branch
from
6e0cb47
to
5b164d0
Compare
There was a problem hiding this comment.
Actionable comments posted: 0
🧹 Outside diff range and nitpick comments (4)
pkg/spdk_test.go (1)
1354-1354: Consider adding test coverage for live upgrade scenariosWhile the code correctly updates the
EngineCreatecall, this test focuses only on target-related scenarios. Given that this PR implements live upgrade support, consider adding specific test cases to verify the live upgrade functionality.Would you like me to help create additional test cases specifically for live upgrade scenarios?
pkg/spdk/engine.go (3)
37-49: Add documentation for the StandbyTargetPort field.Consider adding a comment explaining the purpose and lifecycle of the
StandbyTargetPortfield, particularly its relationship withTargetPortduring live upgrades.TargetPort int32 // Port of the target that is used for letting initiator connect to - StandbyTargetPort int32 + StandbyTargetPort int32 // Port used for the standby target during live upgrades
112-142: Simplify complex branching logic.The function has nested if conditions that make the logic hard to follow. Consider refactoring to:
- Extract the conditions into well-named boolean variables
- Use early returns to reduce nesting
- Add debug logging for better observability
func (e *Engine) checkInitiatorAndTargetCreationRequirements(podIP, initiatorIP, targetIP string) (bool, bool, error) { + isLocalPod := podIP == initiatorIP && podIP == targetIP + isInitiatorPod := podIP == initiatorIP + isTargetPod := podIP == targetIP + + if !isInitiatorPod && !isTargetPod { + return false, false, fmt.Errorf("invalid initiator and target addresses for engine %s creation with initiator address %v and target address %v", + e.Name, initiatorIP, targetIP) + } + + if !isLocalPod { + e.log.Infof("Creating %s instance", map[bool]string{true: "an initiator", false: "a target"}[isInitiatorPod]) + return isInitiatorPod, isTargetPod, nil + } + + if e.Port == 0 && e.TargetPort == 0 { + e.log.Info("Creating both initiator and target instances") + return true, true, nil + } + + if e.Port != 0 && e.TargetPort == 0 { + if e.StandbyTargetPort != 0 { + e.log.Warnf("Standby target instance with port %v is already created, will skip the target creation", e.StandbyTargetPort) + return false, false, nil + } + e.log.Info("Creating a target instance") + return false, true, nil + } + + e.log.Infof("Initiator instance with port %v and target instance with port %v are already created, will skip the creation", + e.Port, e.TargetPort) + return false, false, nil }
Line range hint
2139-2240: Improve state tracking during switchover.Consider implementing a transaction-like pattern to track the switchover state and ensure proper cleanup on failures. This would make the rollback logic more robust and easier to maintain.
+type switchoverState struct { + oldTargetDisconnected bool + newTargetConnected bool + deviceReloaded bool +} + func (e *Engine) SwitchOverTarget(spdkClient *spdkclient.Client, newTargetAddress string) (err error) { + state := &switchoverState{} + + defer func() { + if err != nil && state.oldTargetDisconnected && !state.newTargetConnected { + if errRollback := e.rollbackSwitchover(currentTargetAddress); errRollback != nil { + e.log.WithError(errRollback).Error("Failed to rollback target switchover") + } + } + }()
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
📒 Files selected for processing (6)
pkg/api/types.go(2 hunks)pkg/client/client.go(1 hunks)pkg/spdk/engine.go(18 hunks)pkg/spdk/engine_test.go(1 hunks)pkg/spdk/server.go(1 hunks)pkg/spdk_test.go(6 hunks)
🚧 Files skipped from review as they are similar to previous changes (4)
- pkg/api/types.go
- pkg/client/client.go
- pkg/spdk/engine_test.go
- pkg/spdk/server.go
🔇 Additional comments (6)
pkg/spdk_test.go (4)
271-271: LGTM: Parameter removal aligns with live upgrade support
The removal of the upgradeRequired parameter from EngineCreate is consistent with the PR's objective to support live upgrades.
354-354: LGTM: Consistent parameter update in test
The change maintains consistency with the updated EngineCreate signature during replica rebuilding tests.
530-530: LGTM: Consistent parameter update in snapshot test
The change maintains consistency with the updated EngineCreate signature in snapshot operation tests.
1228-1228: LGTM: Consistent parameter updates in revert operations
The changes maintain consistency with the updated EngineCreate signature in snapshot revert operations, both for empty frontend and previous frontend cases.
Also applies to: 1243-1243
pkg/spdk/engine.go (2)
482-498:
Add port range validation.
The port allocation logic should validate that allocated ports are within valid ranges (1-65535) before using them.
port, _, err = superiorPortAllocator.AllocateRange(portCount)
if err != nil {
return errors.Wrapf(err, "failed to allocate port for engine %v", e.Name)
}
+if port <= 0 || port > 65535 {
+ return fmt.Errorf("invalid port %d allocated for engine %v: must be between 1 and 65535", port, e.Name)
+}
e.log.Infof("Allocated port %v for engine", port)Likely invalid or redundant comment.
2410-2430: 🛠️ Refactor suggestion
Add port validation before release.
The port release logic should validate port numbers before attempting to release them to prevent potential issues with invalid port numbers.
func (e *Engine) releaseTargetAndStandbyTargetPorts(superiorPortAllocator *commonbitmap.Bitmap) error {
+ if superiorPortAllocator == nil {
+ return fmt.Errorf("invalid nil port allocator")
+ }
+
releaseTargetPortRequired := e.TargetPort != 0
releaseStandbyTargetPortRequired := e.StandbyTargetPort != 0 && e.StandbyTargetPort != e.TargetPort
+ // Validate port numbers
+ if releaseTargetPortRequired && (e.TargetPort < 1 || e.TargetPort > 65535) {
+ return fmt.Errorf("invalid target port number: %d", e.TargetPort)
+ }
+ if releaseStandbyTargetPortRequired && (e.StandbyTargetPort < 1 || e.StandbyTargetPort > 65535) {
+ return fmt.Errorf("invalid standby target port number: %d", e.StandbyTargetPort)
+ }Likely invalid or redundant comment.
derekbit
force-pushed
the
v2-control-upgrade
branch
from
5b164d0
to
8fd6c25
Compare
There was a problem hiding this comment.
Actionable comments posted: 2
🧹 Outside diff range and nitpick comments (14)
pkg/spdk/replica.go (1)
1325-1327: Clarify error message for better readability.The error message in the
fmt.Errorfcall is repetitive and may confuse readers. Consider rephrasing it to improve clarity.Suggested change:
- return fmt.Errorf("found mismatching between the required dst bdev NVMe controller name %s and the expected dst controller name %s for replica %s rebuilding src attach", dstRebuildingLvolName, controllerName, r.Name) + return fmt.Errorf("mismatch between expected dst bdev NVMe controller name %s and actual controller name %s for replica %s during rebuilding source attach", dstRebuildingLvolName, controllerName, r.Name)pkg/spdk/disk/nvme/nvme.go (1)
18-26: Add documentation for timeout constants and verify values for live upgrades.The newly added constants lack documentation explaining their purpose and impact. Since these timeouts are crucial for NVMe controller behavior during live upgrades, please consider:
Adding detailed comments explaining:
- The purpose of each timeout
- The implications of their values
- The relationship between related timeouts (e.g., FastIOFail and TransportAck)
Verifying if these timeout values are optimal for live upgrade scenarios
const ( - // Timeouts for disk bdev + // NVMe controller timeout configurations critical for disk operations and live upgrades + + // Maximum time to wait before declaring a controller as lost diskCtrlrLossTimeoutSec = 30 + // Delay between reconnection attempts diskReconnectDelaySec = 2 + // Maximum time to wait before failing I/O operations when controller is unresponsive diskFastIOFailTimeoutSec = 15 + // Maximum time to wait for transport layer acknowledgment diskTransportAckTimeout = 14 + // Interval for NVMe keep-alive messages in milliseconds diskKeepAliveTimeoutMs = 10000 + // Multipath configuration for NVMe devices diskMultipath = "disable" )pkg/spdk/util.go (1)
Line range hint
80-99: LGTM! The timeout parameterization enhances upgrade reliability.The changes improve the function's flexibility by allowing configurable timeouts, which is crucial for reliable live upgrades. The parameterization of
ctrlrLossTimeoutandfastIOFailTimeoutSecenables fine-tuned control over connection handling during upgrade scenarios.Consider documenting the recommended timeout values for different scenarios:
- Normal operations
- Live upgrade operations
- Recovery scenarios
This will help operators configure appropriate values based on their use case.
pkg/spdk/restore.go (3)
Line range hint
124-143: Add error handling and cleanup for NVMe initiator operationsThe NVMe initiator creation and startup lacks proper cleanup in error paths. Consider the following improvements:
- Add cleanup in error paths to prevent resource leaks
- Add timeout handling for initiator operations
- Document or make configurable the
trueparameter ininitiator.StartHere's a suggested improvement:
initiator, err := nvme.NewInitiator(lvolName, helpertypes.GetNQN(lvolName), nvme.HostProc) if err != nil { + // Clean up any partial initialization + if err := r.spdkClient.StopExposeBdev(helpertypes.GetNQN(lvolName)); err != nil { + r.log.WithError(err).Error("Failed to cleanup after initiator creation failure") + } return nil, "", errors.Wrapf(err, "failed to create NVMe initiator for lvol bdev %v", lvolName) } + +// Use context with timeout for initiator operations +ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second) +defer cancel() + -if _, err := initiator.Start(r.ip, strconv.Itoa(int(r.port)), true); err != nil { +if _, err := initiator.Start(r.ip, strconv.Itoa(int(r.port)), true /* waitForConnection */); err != nil { + if err := initiator.Stop(true, true, false); err != nil { + r.log.WithError(err).Error("Failed to cleanup after initiator start failure") + } return nil, "", errors.Wrapf(err, "failed to start NVMe initiator for lvol bdev %v", lvolName) }
160-162: Improve error handling in device closureThe error handling in the NVMe device closure could be more robust. Consider adding context and documenting the cleanup order.
Here's a suggested improvement:
+// CloseVolumeDev performs cleanup in the following order: +// 1. Close the NVMe device file handle +// 2. Stop the NVMe initiator +// 3. Unexpose the lvol bdev if needed r.log.Infof("Closing NVMe device %v", r.initiator.Endpoint) + +var errs []error if err := volDev.Close(); err != nil { - return errors.Wrapf(err, "failed to close NVMe device %v", r.initiator.Endpoint) + errs = append(errs, errors.Wrapf(err, "failed to close NVMe device %v", r.initiator.Endpoint)) }
Line range hint
165-167: Document NVMe initiator stop parametersThe
initiator.Stopcall uses hardcoded boolean parameters without clear documentation of their purpose.Consider adding comments or using named parameters:
-if _, err := r.initiator.Stop(true, true, false); err != nil { +// Stop the initiator with: +// - waitForCompletion: true to ensure graceful shutdown +// - removeDevice: true to clean up device nodes +// - forceRemove: false to avoid forced cleanup +if _, err := r.initiator.Stop( + waitForCompletion: true, + removeDevice: true, + forceRemove: false, +); err != nil {pkg/spdk/engine_test.go (1)
99-100: Improve test iteration and logging.Replace index-based iteration with value-based iteration for better readability and maintainability.
- for testName, testCase := range testCases { - c.Logf("testing checkInitiatorAndTargetCreationRequirements.%v", testName) + for _, tc := range testCases { + c.Logf("testing %s", tc.name)pkg/spdk/disk.go (4)
Line range hint
89-91: Consider improving error message clarityWhen diskUUID is not provided, the warning message could be more descriptive about the implications and potential risks.
- log.Warn("Disk UUID is not provided, trying to get lvstore with disk name") + log.Warn("Disk UUID not provided: falling back to disk name for lvstore lookup. This may be less reliable for disk identification.")Also applies to: 96-97
Line range hint
279-282: Consider adding retry mechanism for disk ID retrievalDuring live upgrades, disk device numbers might temporarily change. Consider implementing a retry mechanism with backoff.
func getDiskID(filename string) (string, error) { + maxRetries := 3 + var lastErr error + for i := 0; i < maxRetries; i++ { executor, err := spdkutil.NewExecutor(commontypes.ProcDirectory) if err != nil { - return "", err + lastErr = err + continue } dev, err := spdkutil.DetectDevice(filename, executor) if err != nil { - return "", errors.Wrapf(err, "failed to detect disk device %v", filename) + lastErr = errors.Wrapf(err, "failed to detect disk device %v", filename) + continue } return fmt.Sprintf("%d-%d", dev.Major, dev.Minor), nil + } + return "", errors.Wrap(lastErr, "failed to get disk ID after retries") }
Line range hint
366-368: Critical: Validate lvstore state during creationThe error handling for lvstore creation should be more robust, especially during live upgrades where partial states might exist.
if diskUUID == "" { log.Infof("Creating a new lvstore %v", lvstoreName) - return spdkClient.BdevLvolCreateLvstore(bdev.Name, lvstoreName, defaultClusterSize) + uuid, err := spdkClient.BdevLvolCreateLvstore(bdev.Name, lvstoreName, defaultClusterSize) + if err != nil { + return "", errors.Wrapf(err, "failed to create lvstore %v", lvstoreName) + } + // Verify the lvstore was created successfully + if _, err := spdkClient.BdevLvolGetLvstore(lvstoreName, uuid); err != nil { + return "", errors.Wrapf(err, "failed to verify newly created lvstore %v", lvstoreName) + } + return uuid, nil }
Line range hint
372-401: Consider caching disk information for performanceThe
lvstoreToDiskfunction makes multiple RPC calls which could impact performance during live upgrades with multiple concurrent operations.Consider implementing a simple cache with TTL for disk information to reduce RPC overhead during high-concurrency scenarios. This would be particularly beneficial during live upgrades where multiple components might request disk information simultaneously.
pkg/spdk_test.go (1)
Line range hint
1-1354: Consider adding upgrade-specific test casesWhile the parameter removal is consistent, there appears to be no explicit test coverage for the new live upgrade functionality that this PR aims to support.
Would you like me to help create test cases that specifically verify:
- Live upgrade behavior
- Upgrade state transitions
- Error handling during upgrades
pkg/spdk/engine.go (2)
Line range hint
1310-1315: Improve error handling in replica operationsThe error handling for replica operations should be more robust and provide better cleanup.
+ defer func() { + if err != nil { + e.log.WithError(err).Error("Failed to handle replica operation") + if errCleanup := e.cleanupReplicaOperation(replicaName); errCleanup != nil { + e.log.WithError(errCleanup).Error("Failed to cleanup after replica operation failure") + } + } + }()
Line range hint
2149-2250: Add test coverage for target switchover functionalityThe target switchover functionality lacks comprehensive test coverage. Consider adding tests for:
- Successful switchover
- Failed switchover with rollback
- Edge cases in port allocation
- Error conditions in target connection/disconnection
Would you like me to help create these test cases?
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
⛔ Files ignored due to path filters (8)
go.modis excluded by!go.modgo.sumis excluded by!**/*.sum,!go.sumvendor/github.com/longhorn/go-spdk-helper/pkg/nvme/initiator.gois excluded by!vendor/**vendor/github.com/longhorn/go-spdk-helper/pkg/nvme/nvme.gois excluded by!vendor/**vendor/github.com/longhorn/go-spdk-helper/pkg/nvme/nvmecli.gois excluded by!vendor/**vendor/github.com/longhorn/go-spdk-helper/pkg/types/types.gois excluded by!vendor/**vendor/github.com/longhorn/types/pkg/generated/spdkrpc/spdk.pb.gois excluded by!**/*.pb.go,!**/generated/**,!vendor/**vendor/modules.txtis excluded by!vendor/**
📒 Files selected for processing (15)
pkg/api/types.go(2 hunks)pkg/client/client.go(2 hunks)pkg/client/types.go(1 hunks)pkg/spdk/backup.go(3 hunks)pkg/spdk/disk.go(1 hunks)pkg/spdk/disk/nvme/nvme.go(2 hunks)pkg/spdk/engine.go(28 hunks)pkg/spdk/engine_test.go(1 hunks)pkg/spdk/replica.go(3 hunks)pkg/spdk/restore.go(2 hunks)pkg/spdk/server.go(3 hunks)pkg/spdk/types.go(2 hunks)pkg/spdk/util.go(3 hunks)pkg/spdk_test.go(6 hunks)pkg/util/block.go(1 hunks)
✅ Files skipped from review due to trivial changes (3)
- pkg/client/types.go
- pkg/spdk/backup.go
- pkg/util/block.go
🚧 Files skipped from review as they are similar to previous changes (3)
- pkg/api/types.go
- pkg/client/client.go
- pkg/spdk/server.go
🔇 Additional comments (21)
pkg/spdk/disk/nvme/nvme.go (1)
58-58: Consider enabling multipath for live upgrade reliability.
The current implementation disables multipath support by default. For live upgrade scenarios, enabling multipath could provide better reliability and seamless transitions. Please verify if this is intentional and consider:
- Making multipath configurable based on the upgrade context
- Adding documentation explaining the implications of the multipath setting during live upgrades
pkg/spdk/util.go (1)
92-92: Verify the replicaReconnectDelaySec constant
The code uses replicaReconnectDelaySec constant but it's not visible in the provided code. Please ensure this constant is properly defined and accessible.
✅ Verification successful
Constant replicaReconnectDelaySec is properly defined
The constant replicaReconnectDelaySec is correctly defined in pkg/spdk/types.go with a value of 2 seconds, alongside other related RAID base bdev timeout constants. The usage in connectNVMfBdev is valid.
🏁 Scripts executed
The following scripts were executed for the analysis:
Script:
#!/bin/bash
# Search for replicaReconnectDelaySec definition
rg -l 'replicaReconnectDelaySec.*=.*'
Length of output: 57
Script:
#!/bin/bash
# Let's check the actual definition of replicaReconnectDelaySec in types.go
rg 'replicaReconnectDelaySec' pkg/spdk/types.go -A 2 -B 2
Length of output: 234
pkg/spdk/restore.go (2)
13-14: LGTM: Clean addition of NVMe initiator support
The addition of the nvme package import and initiator field to the Restore struct is well-structured and aligns with the PR's objective of supporting live upgrades.
Also applies to: 41-41
Line range hint 124-143: Verify upgrade process with NVMe device handling
The NVMe device handling changes look good, but we should verify the behavior during live upgrades.
Let's check for any existing upgrade-related tests:
Also applies to: 160-167
✅ Verification successful
Based on the analysis of the codebase, I can now generate the final response:
NVMe device handling during restore is properly implemented and verified
The code shows a robust implementation of NVMe device handling during restore operations with proper state management:
- The restore process correctly handles device exposure with
IsExposedstate tracking - Proper cleanup is implemented with
StopExposeBdevwhen needed - Error handling and rollback mechanisms are in place for NVMe device operations
- The restore implementation includes status tracking and progress monitoring
- The code supports both full and incremental restore operations with proper device management
🏁 Scripts executed
The following scripts were executed for the analysis:
Script:
#!/bin/bash
# Search for upgrade-related tests
rg -t go "TestUpgrade|upgrade.*Test"
# Look for upgrade-related configuration
rg -t go "upgrade.*Config|UpgradeConfig"
Length of output: 81
Script:
#!/bin/bash
# Let's search for NVMe-related tests and functionality
rg -t go "func.*Test.*NVMe" -A 5
# Search for live migration or upgrade-related code
rg -t go "(Live(Migration|Upgrade)|Migration|Upgrade)" -A 3
# Look for NVMe initiator usage
ast-grep --pattern 'nvme.NewInitiator($$$)'
# Check for any restore-related tests
rg -t go "func.*Test.*Restore" -A 5
Length of output: 5808
Script:
#!/bin/bash
# Let's look for the restore implementation and its usage
rg -t go "func.*Restore" -B 2 -A 10
# Search for NVMe device handling in the codebase
rg -t go "IsExposed|exposeSnapshotLvolBdev" -B 2 -A 5
# Look for any error handling or recovery scenarios
ast-grep --pattern 'func ($$$) {
$$$
if err != nil {
$$$
return $$$, errors.Wrapf(err, "failed to $$$")
}
$$$
}'
Length of output: 100384
pkg/spdk/types.go (2)
10-10: LGTM! Import statement is correctly placed.
The re-addition of the spdkrpc import is necessary for the ServiceLvolToProtoLvol function's type conversions.
35-43: Verify timeout values for high-availability scenarios.
The timeout constants for RAID base bdev operations are critical for ensuring reliable replica management during live upgrades. Let's verify these values:
replicaCtrlrLossTimeoutSec = 15: Time to wait before failing I/O after losing connectionreplicaReconnectDelaySec = 2: Delay between reconnection attemptsreplicaFastIOFailTimeoutSec = 10: Time before failing I/O operations quicklyreplicaTransportAckTimeout = 14: Transport layer acknowledgment timeoutreplicaKeepAliveTimeoutMs = 10000: Keep-alive timeout for connection health
Let's search for any existing timeout configurations or related discussions:
Consider adding inline documentation explaining:
- The rationale behind these specific timeout values
- How these values affect failover behavior during live upgrades
- Any potential impact on system availability
✅ Verification successful
The timeout values are consistent with the vendor defaults and well-documented
The timeout values in pkg/spdk/types.go exactly match the default values from the vendor package go-spdk-helper, which includes detailed documentation about their purpose:
replicaCtrlrLossTimeoutSec = 15: Matches vendor'sDefaultCtrlrLossTimeoutSecreplicaReconnectDelaySec = 2: Matches vendor'sDefaultReconnectDelaySecwith a constraint that it can't exceedFastIOFailTimeoutreplicaFastIOFailTimeoutSec = 10: Matches vendor'sDefaultFastIOFailTimeoutSecreplicaTransportAckTimeout = 14: Matches vendor's documented value, which translates to 16.384 seconds (2^14 ms)replicaKeepAliveTimeoutMs = 10000: Matches vendor'sDefaultKeepAliveTimeoutMs
These values are properly tuned for TCP transport error detection and connection management during failover scenarios.
🏁 Scripts executed
The following scripts were executed for the analysis:
Script:
#!/bin/bash
# Search for timeout configurations in the codebase
echo "Searching for existing timeout configurations..."
rg -A 2 "CtrlrLossTimeout|ReconnectDelay|FastIOFailTimeout|TransportAckTimeout|KeepAliveTimeout" --type go
echo "Searching for timeout-related discussions in docs..."
rg -A 5 "timeout|reconnect|failover" docs/
Length of output: 7662
pkg/spdk/engine_test.go (4)
1-11: LGTM!
The package structure and imports are well-organized and include all necessary dependencies.
165-167: Improve test iteration and logging.
Replace index-based iteration with value-based iteration for better readability and maintainability.
- for testName, testCase := range testCases {
- c.Logf("testing isNewEngine.%v", testName)
+ for _, tc := range testCases {
+ c.Logf("testing %s", tc.name)222-232: 🛠️ Refactor suggestion
Enhance test execution and add bitmap verification.
- Replace index-based iteration with value-based iteration
- Add verification that ports are actually released in the bitmap
- for testName, testCase := range testCases {
- c.Logf("testing releaseTargetAndStandbyTargetPorts.%v", testName)
+ for _, tc := range testCases {
+ c.Logf("testing %s", tc.name)
bitmap, err := commonbitmap.NewBitmap(0, 100000)
c.Assert(err, IsNil)
- err = testCase.engine.releaseTargetAndStandbyTargetPorts(bitmap)
- c.Assert(err, DeepEquals, testCase.expectedError)
- c.Assert(testCase.engine.TargetPort, Equals, testCase.expectedTargetPort)
- c.Assert(testCase.engine.StandbyTargetPort, Equals, testCase.expectedStandbyTargetPort)
+ // Setup initial bitmap state
+ if tc.engine.TargetPort > 0 {
+ bitmap.Set(uint64(tc.engine.TargetPort))
+ }
+ if tc.engine.StandbyTargetPort > 0 {
+ bitmap.Set(uint64(tc.engine.StandbyTargetPort))
+ }
+
+ err = tc.engine.releaseTargetAndStandbyTargetPorts(bitmap)
+ c.Assert(err, DeepEquals, tc.expectedError,
+ Commentf("Test case '%s': unexpected error result", tc.name))
+ c.Assert(tc.engine.TargetPort, Equals, tc.expectedTargetPort,
+ Commentf("Test case '%s': unexpected target port", tc.name))
+ c.Assert(tc.engine.StandbyTargetPort, Equals, tc.expectedStandbyTargetPort,
+ Commentf("Test case '%s': unexpected standby target port", tc.name))
+
+ // Verify ports are released in bitmap
+ if tc.expectedError == nil {
+ if tc.engine.TargetPort > 0 {
+ c.Assert(bitmap.Test(uint64(tc.engine.TargetPort)), Equals, false,
+ Commentf("Test case '%s': target port not released", tc.name))
+ }
+ if tc.engine.StandbyTargetPort > 0 {
+ c.Assert(bitmap.Test(uint64(tc.engine.StandbyTargetPort)), Equals, false,
+ Commentf("Test case '%s': standby target port not released", tc.name))
+ }
+ }Likely invalid or redundant comment.
13-98: 🛠️ Refactor suggestion
Add test cases for critical live upgrade scenarios.
Given that this PR's primary objective is to support live upgrades, consider adding these essential test cases:
- Port validation
- Live upgrade state transitions
}{
+ {
+ name: "Invalid negative standby target port",
+ podIP: "192.168.1.1",
+ initiatorIP: "192.168.1.1",
+ targetIP: "192.168.1.1",
+ port: 100,
+ targetPort: 8000,
+ standbyTargetPort: -1,
+ expectedInitiatorCreationRequired: false,
+ expectedTargetCreationRequired: false,
+ expectedError: fmt.Errorf("invalid standby target port: must be positive"),
+ },
+ {
+ name: "Port conflict between active and standby targets",
+ podIP: "192.168.1.1",
+ initiatorIP: "192.168.1.1",
+ targetIP: "192.168.1.1",
+ port: 100,
+ targetPort: 8000,
+ standbyTargetPort: 8000,
+ expectedInitiatorCreationRequired: false,
+ expectedTargetCreationRequired: false,
+ expectedError: fmt.Errorf("port conflict: standby target port must be different from active target port"),
+ },Likely invalid or redundant comment.
pkg/spdk/disk.go (3)
9-10: LGTM: Import changes are well-organized and necessary
The new imports support the enhanced disk management functionality and are logically grouped.
Also applies to: 15-16, 17-17, 22-22
Line range hint 98-108: Verify error handling during live upgrades
The current error handling for lvstore lookup needs verification in the context of live upgrades, particularly when transitioning between versions.
✅ Verification successful
Based on the search results, I can now provide a final response as we have sufficient context about the error handling patterns and lvstore operations in the codebase.
Error handling during live upgrades is properly implemented
The current error handling for lvstore lookup is consistent with the codebase's patterns and is safe during live upgrades because:
- The code properly handles the "no such device" error case using
jsonrpc.IsJSONRPCRespErrorNoSuchDevice(err), which is a standard pattern used throughout the codebase for handling transient states. - The implementation has a fallback mechanism that tries both UUID and name-based lookups, making it resilient during version transitions:
- First attempts with UUID:
lvstores, err = spdkClient.BdevLvolGetLvstore("", diskUUID) - Falls back to name:
lvstores, err = spdkClient.BdevLvolGetLvstore(diskName, "")
- First attempts with UUID:
- The error handling includes appropriate logging and wrapping of errors, maintaining observability during upgrades.
🏁 Scripts executed
The following scripts were executed for the analysis:
Script:
#!/bin/bash
# Description: Check for potential race conditions during lvstore lookups
# Test: Search for related error handling patterns across the codebase
# Look for similar lvstore error handling patterns
rg -A 5 "BdevLvolGetLvstore.*jsonrpc\.IsJSONRPCRespErrorNoSuchDevice"
# Check for upgrade-related error handling
rg -A 5 "upgrade.*lvstore"
Length of output: 100
Script:
#!/bin/bash
# Let's broaden our search to understand the error handling patterns better
# Search for all BdevLvolGetLvstore usages
rg -A 10 "BdevLvolGetLvstore"
# Look for error handling patterns with IsJSONRPCRespErrorNoSuchDevice
rg -A 5 "IsJSONRPCRespErrorNoSuchDevice"
# Search for upgrade-related code in SPDK
fd -e go -e md . | grep -i spdk | xargs rg -A 5 "upgrade"
# Check for any lvstore-related error handling
rg -A 5 "lvstore.*err"
Length of output: 25931
Line range hint 1-401: Verify live upgrade compatibility end-to-end
While the individual components appear sound, we should verify the entire disk management flow during live upgrades.
✅ Verification successful
Let me gather more specific information about disk operations and error handling.
Let me check for more specific disk operation patterns and state management.
Based on the analysis of the code, I can now provide a final response regarding the live upgrade compatibility:
Disk operations are safe for live upgrades
The codebase implements robust error handling and cleanup mechanisms that make it safe during upgrades:
- All disk operations (Create/Delete/Get) have proper deferred cleanup handlers
- State transitions are properly managed with rollback capabilities
- Driver-specific operations (virtio-blk, virtio-scsi, nvme, aio) include proper cleanup on failures
- Disk detection and validation occurs before any state changes
The code includes:
- Proper state validation before operations
- Graceful handling of missing components
- Cleanup of partially completed operations
- Consistent error propagation
🏁 Scripts executed
The following scripts were executed for the analysis:
Script:
#!/bin/bash
# Description: Analyze potential upgrade paths and error scenarios
# Test: Look for upgrade-related patterns and potential race conditions
# Check for upgrade-related error handling patterns
rg -A 5 "upgrade.*error"
# Look for potential race conditions in disk operations
ast-grep --pattern 'func.*Disk.*\(.*\).*error.*{
$$$
if.*err.*!=.*nil.*{
$$$
}
$$$
}'
# Check for proper cleanup handling
rg -A 5 "defer.*cleanup"
Length of output: 1421
Script:
#!/bin/bash
# Check disk operation error handling and state management
# Look for disk state transitions and cleanup
rg -A 10 "disk\.(Create|Delete|Get)" pkg/
# Check for transaction-like patterns in disk operations
ast-grep --pattern 'func.*Disk.*\(.*\).*error.*{
$$$
defer.*{
$$$
}
$$$
}'
# Look for version or compatibility checks
rg -A 5 "(version|compatibility)" pkg/
Length of output: 871
Script:
#!/bin/bash
# Check for disk state management and error handling patterns
rg -A 10 "svcDisk(Create|Delete|Get)" pkg/
# Look for any disk-related error handling and recovery
rg -A 5 "failed to.*disk" pkg/
# Check for any disk driver registration or initialization
ast-grep --pattern 'func init() {
$$$
disk.Register$$$
$$$
}'
Length of output: 19615
pkg/spdk_test.go (2)
530-530: Consistent parameter removal across test functions
The upgradeRequired parameter has been consistently removed from all EngineCreate calls in:
- TestSPDKMultipleThreadSnapshotOpsAndRebuilding
- revertSnapshot helper function
- TestSPDKEngineOnlyWithTarget
Also applies to: 1228-1228, 1243-1243, 1354-1354
271-271: LGTM: Parameter removal aligns with upgrade changes
The removal of the upgradeRequired parameter from EngineCreate is consistent with the PR's objective to support live upgrades.
Let's verify this change is consistent across the codebase:
pkg/spdk/engine.go (6)
39-52: LGTM: Engine struct changes
The addition of StandbyTargetPort field is well-organized with other port-related fields and properly documented.
117-119: LGTM: Clean implementation of isNewEngine
The method provides a clear way to check if an engine is in its initial state by verifying all relevant fields are empty/zero.
418-420: 🛠️ Refactor suggestion
Add validation for standby target creation
The standby target creation check should include validation of the port value and current engine state.
+ if e.Port <= 0 {
+ return fmt.Errorf("invalid port %d for standby target creation", e.Port)
+ }
standbyTargetCreationRequired := false
if e.Port != 0 && e.TargetPort == 0 {
standbyTargetCreationRequired = true
}Likely invalid or redundant comment.
2149-2188: 🛠️ Refactor suggestion
Improve error handling and cleanup in target switchover
The error handling in the target switchover could be improved:
- Consider using a transaction-like pattern for cleanup
- Add more detailed logging for each step
- Ensure proper cleanup on partial failures
+ type switchoverState struct {
+ oldTargetDisconnected bool
+ newTargetConnected bool
+ deviceReloaded bool
+ }
+ state := &switchoverState{}
+
+ defer func() {
+ if err != nil && state.oldTargetDisconnected && !state.newTargetConnected {
+ if errRollback := e.rollbackSwitchover(currentTargetAddress); errRollback != nil {
+ e.log.WithError(errRollback).Error("Failed to rollback target switchover")
+ }
+ }
+ }()Likely invalid or redundant comment.
2383-2400: 🛠️ Refactor suggestion
Improve cleanup handling in DeleteTarget
The deletion process should attempt to clean up all resources even if some operations fail. Consider implementing a cleanup tracker to ensure thorough cleanup.
+ type cleanupState struct {
+ exposeBdevStopped bool
+ portsReleased bool
+ raidDeleted bool
+ replicasDisconnected bool
+ }
+ state := &cleanupState{}
+
+ defer func() {
+ if err != nil {
+ e.log.WithError(err).Error("Failed to delete target, cleanup state: %+v", state)
+ }
+ }()Likely invalid or redundant comment.
2420-2440: 🛠️ Refactor suggestion
Add validation for port release operations
The port release logic should validate port numbers before attempting to release them.
+ if superiorPortAllocator == nil {
+ return fmt.Errorf("invalid nil port allocator")
+ }
+
+ // Validate port numbers
+ if releaseTargetPortRequired && (e.TargetPort < 1 || e.TargetPort > 65535) {
+ return fmt.Errorf("invalid target port number: %d", e.TargetPort)
+ }
+ if releaseStandbyTargetPortRequired && (e.StandbyTargetPort < 1 || e.StandbyTargetPort > 65535) {
+ return fmt.Errorf("invalid standby target port number: %d", e.StandbyTargetPort)
+ }Likely invalid or redundant comment.
pkg/spdk/replica.go
Outdated
| r.rebuildingSrcCache.dstRebuildingBdevName, err = connectNVMfBdev(spdkClient, dstRebuildingLvolName, dstRebuildingLvolAddress, | ||
| replicaCtrlrLossTimeoutSec, replicaFastIOFailTimeoutSec) |
There was a problem hiding this comment.
Undefined variables replicaCtrlrLossTimeoutSec and replicaFastIOFailTimeoutSec.
The variables replicaCtrlrLossTimeoutSec and replicaFastIOFailTimeoutSec are not defined within the scope of the file, leading to a compilation error.
Please define these variables or ensure they are properly imported or accessible in this context.
pkg/spdk/replica.go
Outdated
| externalSnapshotBdevName, err := connectNVMfBdev(spdkClient, externalSnapshotLvolName, externalSnapshotAddress, | ||
| replicaCtrlrLossTimeoutSec, replicaFastIOFailTimeoutSec) |
There was a problem hiding this comment.
Undefined variables replicaCtrlrLossTimeoutSec and replicaFastIOFailTimeoutSec.
The variables replicaCtrlrLossTimeoutSec and replicaFastIOFailTimeoutSec are used but not defined, which will cause a compilation error.
Define these variables or pass the correct parameters to the connectNVMfBdev function.
derekbit
force-pushed
the
v2-control-upgrade
branch
from
8fd6c25
to
5236c00
Compare
There was a problem hiding this comment.
Actionable comments posted: 2
🧹 Outside diff range and nitpick comments (5)
pkg/spdk/engine_test.go (2)
99-119: Improve test execution and add cleanup.The test execution can be enhanced for better maintainability and reliability:
- Use range over test cases by value instead of index
- Add cleanup after each test
- for testName, testCase := range testCases { - c.Logf("testing checkInitiatorAndTargetCreationRequirements.%v", testName) + for _, tc := range testCases { + c.Logf("testing %s", tc.name) + + // Setup engine := &Engine{ - Port: testCase.port, - TargetPort: testCase.targetPort, - StandbyTargetPort: testCase.standbyTargetPort, + Port: tc.port, + TargetPort: tc.targetPort, + StandbyTargetPort: tc.standbyTargetPort, Name: "test-engine", log: logrus.New(), } + + // Cleanup + defer func() { + if engine != nil { + engine.Close() + } + }()
165-169: Improve test execution.Replace index-based iteration with value-based iteration for better maintainability.
- for testName, testCase := range testCases { - c.Logf("testing isNewEngine.%v", testName) - result := testCase.engine.isNewEngine() - c.Assert(result, Equals, testCase.expected, Commentf("Test case '%s': unexpected result", testCase.name)) + for _, tc := range testCases { + c.Logf("testing %s", tc.name) + result := tc.engine.isNewEngine() + c.Assert(result, Equals, tc.expected, Commentf("Test case '%s': unexpected result", tc.name))pkg/spdk_test.go (1)
1354-1354: Consider enhancing test coverage for live upgrade scenarios.While the test verifies target-only configuration, consider adding explicit test cases to verify:
- Live upgrade behavior when only target is present
- State transitions during upgrade process
- Error handling during upgrade operations
This would strengthen test coverage for the new live upgrade feature.
Would you like me to help draft additional test cases for these scenarios?
pkg/spdk/engine.go (2)
39-52: Add documentation for the StandbyTargetPort field.The new
StandbyTargetPortfield would benefit from documentation explaining its purpose, valid values, and relationship withTargetPort. This helps maintainers understand when and how this field is used during live upgrades.TargetPort int32 // Port of the target that is used for letting initiator connect to - StandbyTargetPort int32 + StandbyTargetPort int32 // Port of the standby target used during live upgrades
121-151: Simplify complex branching logic.The function has nested conditions that could be simplified for better readability and maintainability. Consider extracting the logic into smaller, focused functions.
func (e *Engine) checkInitiatorAndTargetCreationRequirements(podIP, initiatorIP, targetIP string) (bool, bool, error) { + if podIP != initiatorIP && podIP != targetIP { + return false, false, fmt.Errorf("invalid initiator and target addresses for engine %s creation with initiator address %v and target address %v", + e.Name, initiatorIP, targetIP) + } + + initiatorCreationRequired := podIP == initiatorIP + targetCreationRequired := false + + if podIP == targetIP { + if e.Port == 0 && e.TargetPort == 0 { + e.log.Info("Creating both initiator and target instances") + targetCreationRequired = true + } else if e.Port != 0 && e.TargetPort == 0 && e.StandbyTargetPort == 0 { + e.log.Info("Creating a target instance") + targetCreationRequired = true + } else { + e.log.Infof("Instances already exist (port: %v, target: %v), skipping creation", + e.Port, e.TargetPort) + } + } + + return initiatorCreationRequired, targetCreationRequired, nil }
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
⛔ Files ignored due to path filters (4)
go.modis excluded by!go.modgo.sumis excluded by!**/*.sum,!go.sumvendor/github.com/longhorn/types/pkg/generated/spdkrpc/spdk.pb.gois excluded by!**/*.pb.go,!**/generated/**,!vendor/**vendor/modules.txtis excluded by!vendor/**
📒 Files selected for processing (6)
pkg/api/types.go(2 hunks)pkg/client/client.go(2 hunks)pkg/spdk/engine.go(28 hunks)pkg/spdk/engine_test.go(1 hunks)pkg/spdk/server.go(3 hunks)pkg/spdk_test.go(6 hunks)
🚧 Files skipped from review as they are similar to previous changes (3)
- pkg/api/types.go
- pkg/client/client.go
- pkg/spdk/server.go
🔇 Additional comments (9)
pkg/spdk/engine_test.go (3)
1-12: LGTM!
The package structure and imports are well-organized and follow Go conventions.
222-232: 🛠️ Refactor suggestion
Enhance test execution and add bitmap verification.
- Replace index-based iteration with value-based iteration
- Add verification that ports are actually released in the bitmap
13-98: 🛠️ Refactor suggestion
Add test cases for critical live upgrade scenarios.
Given that this PR's primary objective is to support live upgrades, consider adding these essential test cases:
-
Live upgrade scenarios:
- Transition from active to standby target
- Concurrent active and standby targets
- Failed upgrade scenarios
-
Port validation:
- Invalid port ranges (negative, zero, out of range)
- Port conflicts between active and standby targets
}{
+ {
+ name: "Port conflict between active and standby targets",
+ podIP: "192.168.1.1",
+ initiatorIP: "192.168.1.1",
+ targetIP: "192.168.1.1",
+ port: 100,
+ targetPort: 8000,
+ standbyTargetPort: 8000,
+ expectedInitiatorCreationRequired: false,
+ expectedTargetCreationRequired: false,
+ expectedError: fmt.Errorf("port conflict: standby target port must be different from active target port"),
+ },
+ {
+ name: "Live upgrade transition",
+ podIP: "192.168.1.1",
+ initiatorIP: "192.168.1.1",
+ targetIP: "192.168.1.1",
+ port: 100,
+ targetPort: 8000,
+ standbyTargetPort: 8001,
+ expectedInitiatorCreationRequired: false,
+ expectedTargetCreationRequired: true,
+ expectedError: nil,
+ },Likely invalid or redundant comment.
pkg/spdk_test.go (3)
271-271: LGTM! Test coverage maintained after parameter removal.
The modification to remove the upgradeRequired parameter aligns with the PR's objective of supporting live upgrades while maintaining test coverage for core engine functionality.
530-530: LGTM! Snapshot and rebuilding test coverage preserved.
The parameter removal maintains the test's comprehensive coverage of snapshot operations and rebuilding scenarios.
1228-1228: LGTM! Snapshot reversion handling remains robust.
The modifications preserve the helper function's ability to manage frontend state transitions during snapshot reversion operations.
Also applies to: 1243-1243
pkg/spdk/engine.go (3)
117-119: LGTM!
The isNewEngine() helper function is well-implemented and correctly checks all relevant fields.
Line range hint 2149-2255: Consider potential race conditions in rollback logic.
The rollback logic in the deferred function could face race conditions if multiple goroutines attempt to switch targets simultaneously. Consider:
- Adding synchronization for the rollback operations
- Documenting the concurrency assumptions
- Adding logging for rollback operations
2420-2440: 🛠️ Refactor suggestion
Add port validation before release.
The port release logic should validate port numbers before attempting to release them to prevent potential issues with invalid port values.
func (e *Engine) releaseTargetAndStandbyTargetPorts(superiorPortAllocator *commonbitmap.Bitmap) error {
+ if superiorPortAllocator == nil {
+ return fmt.Errorf("invalid nil port allocator")
+ }
+
releaseTargetPortRequired := e.TargetPort != 0
releaseStandbyTargetPortRequired := e.StandbyTargetPort != 0 && e.StandbyTargetPort != e.TargetPort
+ // Validate port numbers
+ if releaseTargetPortRequired && (e.TargetPort < 1 || e.TargetPort > 65535) {
+ return fmt.Errorf("invalid target port number: %d", e.TargetPort)
+ }
+ if releaseStandbyTargetPortRequired && (e.StandbyTargetPort < 1 || e.StandbyTargetPort > 65535) {
+ return fmt.Errorf("invalid standby target port number: %d", e.StandbyTargetPort)
+ }Likely invalid or redundant comment.
| func (s *TestSuite) TestReleaseTargetAndStandbyTargetPorts(c *C) { | ||
| testCases := []struct { | ||
| name string | ||
| engine *Engine | ||
| expectedTargetPort int32 | ||
| expectedStandbyTargetPort int32 | ||
| expectedError error | ||
| }{ | ||
| { | ||
| name: "Release both target and standby target ports", | ||
| engine: &Engine{ | ||
| TargetPort: 2000, | ||
| StandbyTargetPort: 2005, | ||
| }, | ||
| expectedTargetPort: 0, | ||
| expectedStandbyTargetPort: 0, | ||
| expectedError: nil, | ||
| }, | ||
| { | ||
| name: "Release target port only but standby target port is not set", | ||
| engine: &Engine{ | ||
| TargetPort: 2000, | ||
| StandbyTargetPort: 0, | ||
| }, | ||
| expectedTargetPort: 0, | ||
| expectedStandbyTargetPort: 0, | ||
| expectedError: nil, | ||
| }, | ||
| { | ||
| name: "Release target and standby ports when they are the same", | ||
| engine: &Engine{ | ||
| TargetPort: 2000, | ||
| StandbyTargetPort: 2000, | ||
| }, | ||
| expectedTargetPort: 0, | ||
| expectedStandbyTargetPort: 0, | ||
| expectedError: nil, | ||
| }, | ||
| { | ||
| name: "Release snapshot target port only", | ||
| engine: &Engine{ | ||
| TargetPort: 0, | ||
| StandbyTargetPort: 2000, | ||
| }, | ||
| expectedTargetPort: 0, | ||
| expectedStandbyTargetPort: 0, | ||
| expectedError: nil, | ||
| }, | ||
| } |
There was a problem hiding this comment.
🛠️ Refactor suggestion
Add test cases for port validation.
Consider adding these test cases to improve validation coverage:
- Invalid port values (negative ports)
- Port values exceeding valid ranges
- Error scenarios
}{
+ {
+ name: "Invalid negative ports",
+ engine: &Engine{
+ TargetPort: -1,
+ StandbyTargetPort: -2,
+ },
+ expectedTargetPort: -1,
+ expectedStandbyTargetPort: -2,
+ expectedError: fmt.Errorf("invalid port values"),
+ },
+ {
+ name: "Port exceeding valid range",
+ engine: &Engine{
+ TargetPort: 100001,
+ StandbyTargetPort: 0,
+ },
+ expectedTargetPort: 100001,
+ expectedStandbyTargetPort: 0,
+ expectedError: fmt.Errorf("port value exceeds valid range"),
+ },Committable suggestion skipped: line range outside the PR's diff.
| // DeleteTarget deletes the target instance | ||
| func (e *Engine) DeleteTarget(spdkClient *spdkclient.Client, superiorPortAllocator *commonbitmap.Bitmap) (err error) { | ||
| e.log.Infof("Deleting target") | ||
| e.log.Infof("Deleting target with target port %d and standby target port %d", e.TargetPort, e.StandbyTargetPort) | ||
|
|
||
| if err := spdkClient.StopExposeBdev(e.Nqn); err != nil { | ||
| return errors.Wrapf(err, "failed to stop expose bdev after engine %s target switchover", e.Name) | ||
| err = spdkClient.StopExposeBdev(e.Nqn) | ||
| if err != nil { | ||
| return errors.Wrapf(err, "failed to stop expose bdev while deleting target instance for engine %s", e.Name) | ||
| } | ||
|
|
||
| if e.TargetPort != 0 { | ||
| if err := superiorPortAllocator.ReleaseRange(e.TargetPort, e.TargetPort); err != nil { | ||
| return err | ||
| } | ||
| e.TargetPort = 0 | ||
| err = e.releaseTargetAndStandbyTargetPorts(superiorPortAllocator) | ||
| if err != nil { | ||
| return errors.Wrapf(err, "failed to release target and standby target ports while deleting target instance for engine %s", e.Name) | ||
| } | ||
|
|
||
| e.log.Infof("Deleting raid bdev %s before target switchover", e.Name) | ||
| if _, err := spdkClient.BdevRaidDelete(e.Name); err != nil && !jsonrpc.IsJSONRPCRespErrorNoSuchDevice(err) { | ||
| return errors.Wrapf(err, "failed to delete raid bdev after engine %s target switchover", e.Name) | ||
| e.log.Infof("Deleting raid bdev %s while deleting target instance", e.Name) | ||
| _, err = spdkClient.BdevRaidDelete(e.Name) | ||
| if err != nil && !jsonrpc.IsJSONRPCRespErrorNoSuchDevice(err) { | ||
| return errors.Wrapf(err, "failed to delete raid bdev after engine %s while deleting target instance", e.Name) | ||
| } | ||
|
|
||
| for replicaName, replicaStatus := range e.ReplicaStatusMap { | ||
| e.log.Infof("Disconnecting replica %s after target switchover", replicaName) | ||
| if err := disconnectNVMfBdev(spdkClient, replicaStatus.BdevName); err != nil { | ||
| e.log.WithError(err).Warnf("Engine failed to disconnect replica %s after target switchover, will mark the replica mode from %v to ERR", replicaName, replicaStatus.Mode) | ||
| e.log.Infof("Disconnecting replica %s while deleting target instance", replicaName) | ||
| err = disconnectNVMfBdev(spdkClient, replicaStatus.BdevName) | ||
| if err != nil { | ||
| e.log.WithError(err).Warnf("Engine failed to disconnect replica %s while deleting target instance, will mark the replica mode from %v to ERR", | ||
| replicaName, replicaStatus.Mode) | ||
| replicaStatus.Mode = types.ModeERR | ||
| } | ||
| replicaStatus.BdevName = "" | ||
| } | ||
| return nil | ||
| } | ||
|
|
There was a problem hiding this comment.
🛠️ Refactor suggestion
Add error handling for replica disconnection.
The replica disconnection logic in DeleteTarget should handle errors more gracefully by:
- Attempting to disconnect all replicas even if some fail
- Collecting all errors and returning them together
- Ensuring proper cleanup of resources
func (e *Engine) DeleteTarget(spdkClient *spdkclient.Client, superiorPortAllocator *commonbitmap.Bitmap) (err error) {
+ var errors []error
+
e.log.Infof("Deleting target with target port %d and standby target port %d", e.TargetPort, e.StandbyTargetPort)
err = spdkClient.StopExposeBdev(e.Nqn)
if err != nil {
- return errors.Wrapf(err, "failed to stop expose bdev while deleting target instance for engine %s", e.Name)
+ errors = append(errors, errors.Wrapf(err, "failed to stop expose bdev"))
}
err = e.releaseTargetAndStandbyTargetPorts(superiorPortAllocator)
if err != nil {
- return errors.Wrapf(err, "failed to release target and standby target ports while deleting target instance for engine %s", e.Name)
+ errors = append(errors, errors.Wrapf(err, "failed to release ports"))
}
+ if len(errors) > 0 {
+ return fmt.Errorf("multiple errors during target deletion: %v", errors)
+ }Committable suggestion skipped: line range outside the PR's diff.
|
This pull request is now in conflict. Could you fix it @derekbit? 🙏 |
derekbit
force-pushed
the
v2-control-upgrade
branch
from
5236c00
to
5fa89fb
Compare
There was a problem hiding this comment.
Actionable comments posted: 0
🧹 Outside diff range and nitpick comments (4)
pkg/spdk_test.go (1)
354-354: Consider adding test coverage for live upgrade scenariosWhile the basic engine recreation is tested here, there's an opportunity to add specific test cases for live upgrade scenarios.
Consider adding test cases that verify:
- Engine behavior during live upgrades
- State preservation during upgrades
- Rollback scenarios
pkg/spdk/engine.go (3)
112-142: Simplify nested conditions in creation requirements check.The method has complex nested conditions that could be simplified. Consider extracting the port validation logic into a separate method for better readability.
func (e *Engine) checkInitiatorAndTargetCreationRequirements(podIP, initiatorIP, targetIP string) (bool, bool, error) { + if err := e.validatePorts(); err != nil { + return false, false, err + } initiatorCreationRequired, targetCreationRequired := false, false - var err error if podIP == initiatorIP && podIP == targetIP { - if e.Port == 0 && e.TargetPort == 0 { - e.log.Info("Creating both initiator and target instances") - initiatorCreationRequired = true - targetCreationRequired = true - } else if e.Port != 0 && e.TargetPort == 0 { - e.log.Info("Creating a target instance") - if e.StandbyTargetPort != 0 { - e.log.Warnf("Standby target instance with port %v is already created, will skip the target creation", e.StandbyTargetPort) - } else { - targetCreationRequired = true - } - } else { - e.log.Infof("Initiator instance with port %v and target instance with port %v are already created, will skip the creation", e.Port, e.TargetPort) - } + return e.determineLocalCreationRequirements() } else if podIP == initiatorIP { e.log.Info("Creating an initiator instance") initiatorCreationRequired = true } else if podIP == targetIP { e.log.Info("Creating a target instance") targetCreationRequired = true } else { - err = fmt.Errorf("invalid initiator and target addresses for engine %s creation with initiator address %v and target address %v", e.Name, initiatorIP, targetIP) + return false, false, fmt.Errorf("invalid initiator and target addresses for engine %s creation with initiator address %v and target address %v", + e.Name, initiatorIP, targetIP) } - return initiatorCreationRequired, targetCreationRequired, err + return initiatorCreationRequired, targetCreationRequired, nil } +func (e *Engine) determineLocalCreationRequirements() (bool, bool, error) { + if e.Port == 0 && e.TargetPort == 0 { + e.log.Info("Creating both initiator and target instances") + return true, true, nil + } + if e.Port != 0 && e.TargetPort == 0 { + e.log.Info("Creating a target instance") + if e.StandbyTargetPort != 0 { + e.log.Warnf("Standby target instance with port %v is already created, will skip the target creation", + e.StandbyTargetPort) + return false, false, nil + } + return false, true, nil + } + e.log.Infof("Initiator instance with port %v and target instance with port %v are already created, will skip the creation", + e.Port, e.TargetPort) + return false, false, nil +}
Line range hint
2139-2240: Add transaction-like handling for switchover operations.The switchover operation involves multiple steps that should be handled in a more transactional way to ensure proper cleanup on failure.
func (e *Engine) SwitchOverTarget(spdkClient *spdkclient.Client, newTargetAddress string) (err error) { + type switchoverState struct { + oldTargetDisconnected bool + newTargetConnected bool + deviceReloaded bool + } + state := &switchoverState{} + + defer func() { + if err != nil && state.oldTargetDisconnected && !state.newTargetConnected { + if errRollback := e.rollbackSwitchover(); errRollback != nil { + e.log.WithError(errRollback).Error("Failed to rollback target switchover") + } + } + }() // ... rest of the implementation
2373-2405: Improve error handling in DeleteTarget.The target deletion process should handle errors more gracefully by attempting to complete all cleanup operations even if some fail.
func (e *Engine) DeleteTarget(spdkClient *spdkclient.Client, superiorPortAllocator *commonbitmap.Bitmap) (err error) { + var errors []error + e.log.Infof("Deleting target with target port %d and standby target port %d", e.TargetPort, e.StandbyTargetPort) err = spdkClient.StopExposeBdev(e.Nqn) if err != nil { - return errors.Wrapf(err, "failed to stop expose bdev while deleting target instance for engine %s", e.Name) + errors = append(errors, errors.Wrapf(err, "failed to stop expose bdev")) } err = e.releaseTargetAndStandbyTargetPorts(superiorPortAllocator) if err != nil { - return errors.Wrapf(err, "failed to release target and standby target ports while deleting target instance for engine %s", e.Name) + errors = append(errors, errors.Wrapf(err, "failed to release ports")) } + if len(errors) > 0 { + return fmt.Errorf("multiple errors during target deletion: %v", errors) + }
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
⛔ Files ignored due to path filters (5)
go.modis excluded by!go.modgo.sumis excluded by!**/*.sum,!go.sumvendor/github.com/longhorn/go-spdk-helper/pkg/types/types.gois excluded by!vendor/**vendor/github.com/longhorn/types/pkg/generated/spdkrpc/spdk.pb.gois excluded by!**/*.pb.go,!**/generated/**,!vendor/**vendor/modules.txtis excluded by!vendor/**
📒 Files selected for processing (6)
pkg/api/types.go(2 hunks)pkg/client/client.go(1 hunks)pkg/spdk/engine.go(18 hunks)pkg/spdk/engine_test.go(1 hunks)pkg/spdk/server.go(1 hunks)pkg/spdk_test.go(6 hunks)
🚧 Files skipped from review as they are similar to previous changes (4)
- pkg/api/types.go
- pkg/client/client.go
- pkg/spdk/engine_test.go
- pkg/spdk/server.go
🔇 Additional comments (8)
pkg/spdk_test.go (4)
271-271: LGTM: Engine creation parameter update aligns with live upgrade support
The removal of the upgradeRequired parameter from EngineCreate is consistent with the PR's objective of supporting live upgrades. The test case properly verifies the basic engine creation functionality.
530-530: LGTM: Snapshot operations properly tested with updated engine creation
The test case thoroughly verifies snapshot operations with the updated engine creation parameters, including complex scenarios with multiple snapshots and rebuilding operations.
1228-1228: LGTM: Frontend handling during snapshot revert is well tested
The test properly verifies engine recreation with different frontend configurations during snapshot revert operations, which is crucial for maintaining system stability during upgrades.
Also applies to: 1243-1243
1354-1354: LGTM: Target-only configuration properly tested
The test case verifies the engine creation with separate initiator and target IPs, which is important for distributed deployments and upgrade scenarios.
Let's verify the test coverage for target-related scenarios:
✅ Verification successful
Target IP configuration is properly tested with comprehensive assertions
The verification confirms that the test suite includes thorough validation of target-related scenarios:
- Separate initiator and target IP configurations are tested (
"127.0.0.1"vsip) - Target IP and port validations are implemented with explicit assertions
- Multiple test cases cover different target configurations including:
- Same-node scenarios (where initiator IP equals target IP)
- Distributed scenarios (where initiator and target IPs differ)
- Target port availability checks
🏁 Scripts executed
The following scripts were executed for the analysis:
Script:
#!/bin/bash
# Description: Check test coverage for target-related scenarios
# Look for target-related test cases
rg -l "TestSPDKEngine.*Target" --type go
# Check for target IP validation in tests
rg "TargetIP.*Assert" --type go
Length of output: 93
Script:
#!/bin/bash
# Let's examine the actual test implementations for target-related scenarios
# Check the content of the test file focusing on target IP configurations
rg -B 2 -A 2 "TargetIP|targetIP|target.*IP" pkg/spdk_test.go
# Look for engine creation test cases with IP configurations
rg -B 2 -A 2 "EngineCreate.*ip" pkg/spdk_test.go
# Search for any target-related validations or assertions
rg -B 2 -A 2 "Assert.*target|target.*Assert" pkg/spdk_test.go
Length of output: 2373
pkg/spdk/engine.go (4)
37-49: LGTM: Engine struct fields are well organized.
The new StandbyTargetPort field is appropriately placed with other port-related fields and follows the same type pattern.
108-110: LGTM: Clear and concise helper method.
The isNewEngine() method provides a clear way to check if an engine instance is newly created by verifying all connection-related fields are empty.
482-485: 🛠️ Refactor suggestion
Add port range validation after allocation.
The port allocation should validate that the allocated port is within a valid range (1-65535).
port, _, err = superiorPortAllocator.AllocateRange(portCount)
if err != nil {
return errors.Wrapf(err, "failed to allocate port for engine %v", e.Name)
}
+ if port <= 0 || port > 65535 {
+ if err := superiorPortAllocator.ReleaseRange(port, port); err != nil {
+ e.log.WithError(err).Warnf("Failed to release invalid port %d", port)
+ }
+ return fmt.Errorf("allocated port %d is outside valid range (1-65535) for engine %v", port, e.Name)
+ }Likely invalid or redundant comment.
2410-2430: 🛠️ Refactor suggestion
Add port range validation in releaseTargetAndStandbyTargetPorts.
The port release logic should validate port numbers before attempting to release them.
func (e *Engine) releaseTargetAndStandbyTargetPorts(superiorPortAllocator *commonbitmap.Bitmap) error {
+ if superiorPortAllocator == nil {
+ return fmt.Errorf("invalid nil port allocator")
+ }
+
releaseTargetPortRequired := e.TargetPort != 0
releaseStandbyTargetPortRequired := e.StandbyTargetPort != 0 && e.StandbyTargetPort != e.TargetPort
+ // Validate port numbers
+ if releaseTargetPortRequired && (e.TargetPort < 1 || e.TargetPort > 65535) {
+ return fmt.Errorf("invalid target port number: %d", e.TargetPort)
+ }
+ if releaseStandbyTargetPortRequired && (e.StandbyTargetPort < 1 || e.StandbyTargetPort > 65535) {
+ return fmt.Errorf("invalid standby target port number: %d", e.StandbyTargetPort)
+ }
// Rest of the implementationLikely invalid or redundant comment.
Longhorn 9104 Signed-off-by: Derek Su <[email protected]>
Longhorn 9104 Signed-off-by: Derek Su <[email protected]>
… or resume engine No need to mark engine as error after failing to suspend or resume engine. Longhorn manager will retry the operation later. Longhorn 9104 Signed-off-by: Derek Su <[email protected]>
derekbit
force-pushed
the
v2-control-upgrade
branch
from
5fa89fb
to
1f1a32b
Compare
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Outside diff range and nitpick comments (8)
pkg/spdk/engine_test.go (1)
13-119: Add test cases for edge cases and improve error messages.The test coverage is good but could be enhanced by adding:
- Edge cases for port validation (negative ports, max port value)
- Error cases for invalid IP addresses
- Test case for concurrent active and standby target ports
Add these test cases:
}{ + { + name: "Invalid negative port values", + podIP: "192.168.1.1", + initiatorIP: "192.168.1.1", + targetIP: "192.168.1.1", + port: -1, + targetPort: -2, + standbyTargetPort: -3, + expectedInitiatorCreationRequired: false, + expectedTargetCreationRequired: false, + expectedError: fmt.Errorf("invalid port values"), + }, + { + name: "Invalid IP address format", + podIP: "invalid", + initiatorIP: "192.168.1.1", + targetIP: "192.168.1.1", + port: 0, + targetPort: 0, + standbyTargetPort: 0, + expectedInitiatorCreationRequired: false, + expectedTargetCreationRequired: false, + expectedError: fmt.Errorf("invalid IP address format"), + },pkg/spdk/engine.go (4)
39-51: Add documentation for the StandbyTargetPort field.The new field
StandbyTargetPortshould be documented to explain its purpose and usage in the context of engine live upgrades.TargetPort int32 // Port of the target that is used for letting initiator connect to - StandbyTargetPort int32 + StandbyTargetPort int32 // Port used by the standby target during live upgrades
121-151: Simplify complex creation requirements logic.The method contains complex nested conditions that could be simplified for better maintainability. Consider:
- Extracting the port validation logic into a separate method
- Using early returns to reduce nesting
- Adding debug logging for better observability
func (e *Engine) checkInitiatorAndTargetCreationRequirements(podIP, initiatorIP, targetIP string) (bool, bool, error) { + // Validate inputs + if podIP == "" || initiatorIP == "" || targetIP == "" { + return false, false, fmt.Errorf("invalid empty IP addresses") + } + + // Check if pod matches neither initiator nor target + if podIP != initiatorIP && podIP != targetIP { + return false, false, fmt.Errorf("invalid initiator and target addresses for engine %s creation with initiator address %v and target address %v", + e.Name, initiatorIP, targetIP) + } + + // Handle case where pod is both initiator and target if podIP == initiatorIP && podIP == targetIP { - if e.Port == 0 && e.TargetPort == 0 { - e.log.Info("Creating both initiator and target instances") - initiatorCreationRequired = true - targetCreationRequired = true - } else if e.Port != 0 && e.TargetPort == 0 { - e.log.Info("Creating a target instance") - if e.StandbyTargetPort != 0 { - e.log.Warnf("Standby target instance with port %v is already created, will skip the target creation", e.StandbyTargetPort) - } else { - targetCreationRequired = true - } - } + return e.checkLocalCreationRequirements() + } + + // Handle initiator-only or target-only cases + if podIP == initiatorIP { + e.log.Info("Creating an initiator instance") + return true, false, nil + } + + e.log.Info("Creating a target instance") + return false, true, nil } +func (e *Engine) checkLocalCreationRequirements() (bool, bool, error) { + if e.Port == 0 && e.TargetPort == 0 { + e.log.Info("Creating both initiator and target instances") + return true, true, nil + } + + if e.Port != 0 && e.TargetPort == 0 { + if e.StandbyTargetPort != 0 { + e.log.Warnf("Standby target instance with port %v is already created, will skip the target creation", + e.StandbyTargetPort) + return false, false, nil + } + e.log.Info("Creating a target instance") + return false, true, nil + } + + e.log.Infof("Initiator instance with port %v and target instance with port %v are already created, will skip the creation", + e.Port, e.TargetPort) + return false, false, nil +}
Line range hint
407-529: Add port validation and improve error handling.The port allocation and validation logic should be enhanced:
- Add validation for port numbers
- Improve error handling for port allocation failures
- Add cleanup in error cases
func (e *Engine) handleFrontend(spdkClient *spdkclient.Client, superiorPortAllocator *commonbitmap.Bitmap, portCount int32, targetAddress string, initiatorCreationRequired, targetCreationRequired bool) (err error) { + // Validate inputs + if superiorPortAllocator == nil { + return fmt.Errorf("invalid nil port allocator") + } + if portCount <= 0 { + return fmt.Errorf("invalid port count: %d", portCount) + } // ... existing code ... port, _, err = superiorPortAllocator.AllocateRange(portCount) if err != nil { + // Clean up any previously allocated resources + if e.Port != 0 { + _ = superiorPortAllocator.ReleaseRange(e.Port, e.Port) + } + if e.StandbyTargetPort != 0 { + _ = superiorPortAllocator.ReleaseRange(e.StandbyTargetPort, e.StandbyTargetPort) + } return errors.Wrapf(err, "failed to allocate port for engine %v", e.Name) } + if port <= 0 || port > 65535 { + return fmt.Errorf("invalid allocated port %d for engine %v", port, e.Name) + } // ... rest of the code ...
Line range hint
2149-2255: Enhance error handling and rollback in target switchover.The target switchover logic should be improved to handle failures more gracefully:
- Add transaction-like pattern for rollback
- Improve error context
- Add more detailed logging
func (e *Engine) SwitchOverTarget(spdkClient *spdkclient.Client, newTargetAddress string) (err error) { + // Track switchover state for cleanup + type switchoverState struct { + oldTargetDisconnected bool + newTargetConnected bool + deviceReloaded bool + } + state := &switchoverState{} + + defer func() { + if err != nil { + if state.oldTargetDisconnected && !state.newTargetConnected { + if errRollback := e.rollbackSwitchover(currentTargetAddress); errRollback != nil { + e.log.WithError(errRollback).Error("Failed to rollback target switchover") + } + } + } + }() // ... existing validation code ... if err := e.disconnectTarget(currentTargetAddress); err != nil { return errors.Wrapf(err, "failed to disconnect target %s for engine %s", currentTargetAddress, e.Name) } + state.oldTargetDisconnected = true if err := e.connectTarget(newTargetAddress); err != nil { return errors.Wrapf(err, "failed to connect target %s for engine %s", newTargetAddress, e.Name) } + state.newTargetConnected = true // ... rest of the code ...pkg/spdk_test.go (3)
Line range hint
271-354: Add test coverage for standby target scenarios.The test should be enhanced to cover standby target functionality:
- Add test cases for standby target creation
- Verify port allocation and release
- Test error scenarios
Consider adding these test cases:
// Test standby target creation func TestStandbyTargetCreation(t *testing.T) { // Setup code... // Test cases: // 1. Create engine with standby target // 2. Verify port allocation // 3. Test error scenarios } // Test port release func TestPortRelease(t *testing.T) { // Setup code... // Test cases: // 1. Release target port // 2. Release standby target port // 3. Test error scenarios }
Line range hint
530-1243: Add test coverage for error scenarios in snapshot operations.The test should be enhanced to cover error scenarios:
- Test snapshot operations during target switchover
- Verify behavior when port allocation fails
- Test cleanup after failures
Consider adding these test cases:
// Test snapshot operations during target switchover func TestSnapshotDuringTargetSwitchover(t *testing.T) { // Setup code... // Test cases: // 1. Create snapshot during switchover // 2. Delete snapshot during switchover // 3. Verify data integrity } // Test cleanup after failures func TestCleanupAfterFailures(t *testing.T) { // Setup code... // Test cases: // 1. Port allocation failure // 2. Target creation failure // 3. Verify resource cleanup }
1354-1354: Add assertions for target-only scenarios.The test should include additional assertions to verify:
- Standby target port allocation
- Port release behavior
- Error scenarios
Add these assertions:
// Add after engine creation c.Assert(engine.StandbyTargetPort, Equals, int32(0)) c.Assert(engine.TargetPort, Not(Equals), int32(0)) // Add after target deletion c.Assert(engine.StandbyTargetPort, Equals, int32(0)) c.Assert(engine.TargetPort, Equals, int32(0))
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
⛔ Files ignored due to path filters (4)
go.modis excluded by!go.modgo.sumis excluded by!**/*.sum,!go.sumvendor/github.com/longhorn/types/pkg/generated/spdkrpc/spdk.pb.gois excluded by!**/*.pb.go,!**/generated/**,!vendor/**vendor/modules.txtis excluded by!vendor/**
📒 Files selected for processing (6)
pkg/api/types.go(2 hunks)pkg/client/client.go(1 hunks)pkg/spdk/engine.go(18 hunks)pkg/spdk/engine_test.go(1 hunks)pkg/spdk/server.go(1 hunks)pkg/spdk_test.go(6 hunks)
🚧 Files skipped from review as they are similar to previous changes (3)
- pkg/api/types.go
- pkg/client/client.go
- pkg/spdk/server.go
🔇 Additional comments (4)
pkg/spdk/engine_test.go (2)
121-170: LGTM! Comprehensive test coverage for isNewEngine.
The test cases thoroughly validate the isNewEngine function, including:
- New engine with empty fields
- Existing engine with various field combinations
- Proper handling of
StandbyTargetPort
Line range hint 271-271: LGTM! Engine creation test cases updated correctly.
The test cases have been properly updated to reflect the removal of the upgradeRequired parameter from EngineCreate calls.
Also applies to: 354-354, 530-530, 1228-1228, 1243-1243
pkg/spdk/engine.go (2)
2383-2415: 🛠️ Refactor suggestion
Improve error handling in DeleteTarget.
The target deletion should continue cleanup even if some operations fail:
- Collect all errors and return them together
- Continue cleanup even after non-critical failures
- Add detailed logging for each step
func (e *Engine) DeleteTarget(spdkClient *spdkclient.Client, superiorPortAllocator *commonbitmap.Bitmap) (err error) {
+ var errors []error
+
e.log.Infof("Deleting target with target port %d and standby target port %d", e.TargetPort, e.StandbyTargetPort)
err = spdkClient.StopExposeBdev(e.Nqn)
if err != nil {
- return errors.Wrapf(err, "failed to stop expose bdev while deleting target instance for engine %s", e.Name)
+ errors = append(errors, errors.Wrapf(err, "failed to stop expose bdev"))
}
err = e.releaseTargetAndStandbyTargetPorts(superiorPortAllocator)
if err != nil {
- return errors.Wrapf(err, "failed to release target and standby target ports while deleting target instance for engine %s", e.Name)
+ errors = append(errors, errors.Wrapf(err, "failed to release ports"))
}
+ if len(errors) > 0 {
+ return fmt.Errorf("multiple errors during target deletion: %v", errors)
+ }Likely invalid or redundant comment.
2420-2440: 🛠️ Refactor suggestion
Add validation for port release operations.
The port release logic should validate port numbers before attempting to release them:
- Add validation for port numbers
- Improve error handling
- Add logging for port release operations
func (e *Engine) releaseTargetAndStandbyTargetPorts(superiorPortAllocator *commonbitmap.Bitmap) error {
+ if superiorPortAllocator == nil {
+ return fmt.Errorf("invalid nil port allocator")
+ }
+
releaseTargetPortRequired := e.TargetPort != 0
releaseStandbyTargetPortRequired := e.StandbyTargetPort != 0 && e.StandbyTargetPort != e.TargetPort
+ // Validate port numbers
+ if releaseTargetPortRequired && (e.TargetPort < 1 || e.TargetPort > 65535) {
+ return fmt.Errorf("invalid target port number: %d", e.TargetPort)
+ }
+ if releaseStandbyTargetPortRequired && (e.StandbyTargetPort < 1 || e.StandbyTargetPort > 65535) {
+ return fmt.Errorf("invalid standby target port number: %d", e.StandbyTargetPort)
+ }
// ... rest of the code ...Likely invalid or redundant comment.
| func (s *TestSuite) TestReleaseTargetAndStandbyTargetPorts(c *C) { | ||
| testCases := []struct { | ||
| name string | ||
| engine *Engine | ||
| expectedTargetPort int32 | ||
| expectedStandbyTargetPort int32 | ||
| expectedError error | ||
| }{ | ||
| { | ||
| name: "Release both target and standby target ports", | ||
| engine: &Engine{ | ||
| TargetPort: 2000, | ||
| StandbyTargetPort: 2005, | ||
| }, | ||
| expectedTargetPort: 0, | ||
| expectedStandbyTargetPort: 0, | ||
| expectedError: nil, | ||
| }, | ||
| { | ||
| name: "Release target port only but standby target port is not set", | ||
| engine: &Engine{ | ||
| TargetPort: 2000, | ||
| StandbyTargetPort: 0, | ||
| }, | ||
| expectedTargetPort: 0, | ||
| expectedStandbyTargetPort: 0, | ||
| expectedError: nil, | ||
| }, | ||
| { | ||
| name: "Release target and standby ports when they are the same", | ||
| engine: &Engine{ | ||
| TargetPort: 2000, | ||
| StandbyTargetPort: 2000, | ||
| }, | ||
| expectedTargetPort: 0, | ||
| expectedStandbyTargetPort: 0, | ||
| expectedError: nil, | ||
| }, | ||
| { | ||
| name: "Release snapshot target port only", | ||
| engine: &Engine{ | ||
| TargetPort: 0, | ||
| StandbyTargetPort: 2000, | ||
| }, | ||
| expectedTargetPort: 0, | ||
| expectedStandbyTargetPort: 0, | ||
| expectedError: nil, | ||
| }, | ||
| } | ||
|
|
||
| for testName, testCase := range testCases { | ||
| c.Logf("testing releaseTargetAndStandbyTargetPorts.%v", testName) | ||
|
|
||
| bitmap, err := commonbitmap.NewBitmap(0, 100000) | ||
| c.Assert(err, IsNil) | ||
|
|
||
| err = testCase.engine.releaseTargetAndStandbyTargetPorts(bitmap) | ||
| c.Assert(err, DeepEquals, testCase.expectedError, Commentf("Test case '%s': unexpected error result", testCase.name)) | ||
| c.Assert(testCase.engine.TargetPort, Equals, testCase.expectedTargetPort, Commentf("Test case '%s': unexpected target port", testCase.name)) | ||
| c.Assert(testCase.engine.StandbyTargetPort, Equals, testCase.expectedStandbyTargetPort, Commentf("Test case '%s': unexpected standby target port", testCase.name)) | ||
| } |
There was a problem hiding this comment.
🛠️ Refactor suggestion
Enhance test coverage for port release validation.
The test cases should verify:
- Bitmap state after port release
- Error handling for bitmap operations
- Port range validation
Add bitmap state verification:
err = testCase.engine.releaseTargetAndStandbyTargetPorts(bitmap)
c.Assert(err, DeepEquals, testCase.expectedError)
+ // Verify ports are properly released in bitmap
+ if testCase.expectedError == nil {
+ if testCase.engine.TargetPort > 0 {
+ c.Assert(bitmap.Test(uint64(testCase.engine.TargetPort)), Equals, false,
+ Commentf("Target port %d not released", testCase.engine.TargetPort))
+ }
+ if testCase.engine.StandbyTargetPort > 0 {
+ c.Assert(bitmap.Test(uint64(testCase.engine.StandbyTargetPort)), Equals, false,
+ Commentf("Standby target port %d not released", testCase.engine.StandbyTargetPort))
+ }
+ }📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| func (s *TestSuite) TestReleaseTargetAndStandbyTargetPorts(c *C) { | |
| testCases := []struct { | |
| name string | |
| engine *Engine | |
| expectedTargetPort int32 | |
| expectedStandbyTargetPort int32 | |
| expectedError error | |
| }{ | |
| { | |
| name: "Release both target and standby target ports", | |
| engine: &Engine{ | |
| TargetPort: 2000, | |
| StandbyTargetPort: 2005, | |
| }, | |
| expectedTargetPort: 0, | |
| expectedStandbyTargetPort: 0, | |
| expectedError: nil, | |
| }, | |
| { | |
| name: "Release target port only but standby target port is not set", | |
| engine: &Engine{ | |
| TargetPort: 2000, | |
| StandbyTargetPort: 0, | |
| }, | |
| expectedTargetPort: 0, | |
| expectedStandbyTargetPort: 0, | |
| expectedError: nil, | |
| }, | |
| { | |
| name: "Release target and standby ports when they are the same", | |
| engine: &Engine{ | |
| TargetPort: 2000, | |
| StandbyTargetPort: 2000, | |
| }, | |
| expectedTargetPort: 0, | |
| expectedStandbyTargetPort: 0, | |
| expectedError: nil, | |
| }, | |
| { | |
| name: "Release snapshot target port only", | |
| engine: &Engine{ | |
| TargetPort: 0, | |
| StandbyTargetPort: 2000, | |
| }, | |
| expectedTargetPort: 0, | |
| expectedStandbyTargetPort: 0, | |
| expectedError: nil, | |
| }, | |
| } | |
| for testName, testCase := range testCases { | |
| c.Logf("testing releaseTargetAndStandbyTargetPorts.%v", testName) | |
| bitmap, err := commonbitmap.NewBitmap(0, 100000) | |
| c.Assert(err, IsNil) | |
| err = testCase.engine.releaseTargetAndStandbyTargetPorts(bitmap) | |
| c.Assert(err, DeepEquals, testCase.expectedError, Commentf("Test case '%s': unexpected error result", testCase.name)) | |
| c.Assert(testCase.engine.TargetPort, Equals, testCase.expectedTargetPort, Commentf("Test case '%s': unexpected target port", testCase.name)) | |
| c.Assert(testCase.engine.StandbyTargetPort, Equals, testCase.expectedStandbyTargetPort, Commentf("Test case '%s': unexpected standby target port", testCase.name)) | |
| } | |
| func (s *TestSuite) TestReleaseTargetAndStandbyTargetPorts(c *C) { | |
| testCases := []struct { | |
| name string | |
| engine *Engine | |
| expectedTargetPort int32 | |
| expectedStandbyTargetPort int32 | |
| expectedError error | |
| }{ | |
| { | |
| name: "Release both target and standby target ports", | |
| engine: &Engine{ | |
| TargetPort: 2000, | |
| StandbyTargetPort: 2005, | |
| }, | |
| expectedTargetPort: 0, | |
| expectedStandbyTargetPort: 0, | |
| expectedError: nil, | |
| }, | |
| { | |
| name: "Release target port only but standby target port is not set", | |
| engine: &Engine{ | |
| TargetPort: 2000, | |
| StandbyTargetPort: 0, | |
| }, | |
| expectedTargetPort: 0, | |
| expectedStandbyTargetPort: 0, | |
| expectedError: nil, | |
| }, | |
| { | |
| name: "Release target and standby ports when they are the same", | |
| engine: &Engine{ | |
| TargetPort: 2000, | |
| StandbyTargetPort: 2000, | |
| }, | |
| expectedTargetPort: 0, | |
| expectedStandbyTargetPort: 0, | |
| expectedError: nil, | |
| }, | |
| { | |
| name: "Release snapshot target port only", | |
| engine: &Engine{ | |
| TargetPort: 0, | |
| StandbyTargetPort: 2000, | |
| }, | |
| expectedTargetPort: 0, | |
| expectedStandbyTargetPort: 0, | |
| expectedError: nil, | |
| }, | |
| } | |
| for testName, testCase := range testCases { | |
| c.Logf("testing releaseTargetAndStandbyTargetPorts.%v", testName) | |
| bitmap, err := commonbitmap.NewBitmap(0, 100000) | |
| c.Assert(err, IsNil) | |
| err = testCase.engine.releaseTargetAndStandbyTargetPorts(bitmap) | |
| c.Assert(err, DeepEquals, testCase.expectedError) | |
| // Verify ports are properly released in bitmap | |
| if testCase.expectedError == nil { | |
| if testCase.engine.TargetPort > 0 { | |
| c.Assert(bitmap.Test(uint64(testCase.engine.TargetPort)), Equals, false, | |
| Commentf("Target port %d not released", testCase.engine.TargetPort)) | |
| } | |
| if testCase.engine.StandbyTargetPort > 0 { | |
| c.Assert(bitmap.Test(uint64(testCase.engine.StandbyTargetPort)), Equals, false, | |
| Commentf("Standby target port %d not released", testCase.engine.StandbyTargetPort)) | |
| } | |
| } | |
| c.Assert(testCase.engine.TargetPort, Equals, testCase.expectedTargetPort, Commentf("Test case '%s': unexpected target port", testCase.name)) | |
| c.Assert(testCase.engine.StandbyTargetPort, Equals, testCase.expectedStandbyTargetPort, Commentf("Test case '%s': unexpected standby target port", testCase.name)) | |
| } |
Longhorn 9104 Signed-off-by: Derek Su <[email protected]>
Which issue(s) this PR fixes:
Issue longhorn/longhorn#9104
Signed-off-by: Derek Su [email protected]
What this PR does / why we need it:
Special notes for your reviewer:
Additional documentation or context