[USMON-1083] usm: http2: Refactor handle_first_frame #32514
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
github-actions
bot
added
component/system-probe
medium review
Uncompressed package size comparisonComparison with ancestor Diff per package
Decision✅ Passed |
Test changes on VMUse this command from test-infra-definitions to manually test this PR changes on a VM: inv aws.create-vm --pipeline-id=51895814 --os-family=ubuntuNote: This applies to commit 50a6b66 |
Regression DetectorRegression Detector ResultsMetrics dashboard Baseline: 57183f8 Optimization Goals: ✅ No significant changes detected
|
| perf | experiment | goal | Δ mean % | Δ mean % CI | trials | links |
|---|---|---|---|---|---|---|
| ➖ | quality_gate_idle | memory utilization | +0.64 | [+0.61, +0.68] | 1 | Logs bounds checks dashboard |
| ➖ | tcp_syslog_to_blackhole | ingress throughput | +0.62 | [+0.56, +0.69] | 1 | Logs |
| ➖ | quality_gate_logs | % cpu utilization | +0.61 | [-2.65, +3.87] | 1 | Logs |
| ➖ | uds_dogstatsd_to_api_cpu | % cpu utilization | +0.45 | [-0.22, +1.12] | 1 | Logs |
| ➖ | file_to_blackhole_500ms_latency | egress throughput | +0.15 | [-0.63, +0.93] | 1 | Logs |
| ➖ | file_to_blackhole_0ms_latency_http2 | egress throughput | +0.04 | [-0.83, +0.91] | 1 | Logs |
| ➖ | tcp_dd_logs_filter_exclude | ingress throughput | +0.00 | [-0.01, +0.01] | 1 | Logs |
| ➖ | uds_dogstatsd_to_api | ingress throughput | -0.00 | [-0.12, +0.12] | 1 | Logs |
| ➖ | file_to_blackhole_0ms_latency | egress throughput | -0.00 | [-0.85, +0.84] | 1 | Logs |
| ➖ | file_to_blackhole_1000ms_latency_linear_load | egress throughput | -0.03 | [-0.49, +0.43] | 1 | Logs |
| ➖ | file_to_blackhole_100ms_latency | egress throughput | -0.03 | [-0.72, +0.65] | 1 | Logs |
| ➖ | file_to_blackhole_0ms_latency_http1 | egress throughput | -0.04 | [-0.95, +0.87] | 1 | Logs |
| ➖ | file_to_blackhole_300ms_latency | egress throughput | -0.07 | [-0.71, +0.57] | 1 | Logs |
| ➖ | quality_gate_idle_all_features | memory utilization | -0.08 | [-0.16, +0.00] | 1 | Logs bounds checks dashboard |
| ➖ | file_tree | memory utilization | -0.12 | [-0.24, +0.01] | 1 | Logs |
| ➖ | file_to_blackhole_1000ms_latency | egress throughput | -0.52 | [-1.29, +0.25] | 1 | Logs |
Bounds Checks: ❌ Failed
| perf | experiment | bounds_check_name | replicates_passed | links |
|---|---|---|---|---|
| ❌ | file_to_blackhole_0ms_latency_http2 | lost_bytes | 9/10 | |
| ✅ | file_to_blackhole_0ms_latency | lost_bytes | 10/10 | |
| ✅ | file_to_blackhole_0ms_latency | memory_usage | 10/10 | |
| ✅ | file_to_blackhole_0ms_latency_http1 | lost_bytes | 10/10 | |
| ✅ | file_to_blackhole_0ms_latency_http1 | memory_usage | 10/10 | |
| ✅ | file_to_blackhole_0ms_latency_http2 | memory_usage | 10/10 | |
| ✅ | file_to_blackhole_1000ms_latency | memory_usage | 10/10 | |
| ✅ | file_to_blackhole_1000ms_latency_linear_load | memory_usage | 10/10 | |
| ✅ | file_to_blackhole_100ms_latency | lost_bytes | 10/10 | |
| ✅ | file_to_blackhole_100ms_latency | memory_usage | 10/10 | |
| ✅ | file_to_blackhole_300ms_latency | lost_bytes | 10/10 | |
| ✅ | file_to_blackhole_300ms_latency | memory_usage | 10/10 | |
| ✅ | file_to_blackhole_500ms_latency | lost_bytes | 10/10 | |
| ✅ | file_to_blackhole_500ms_latency | memory_usage | 10/10 | |
| ✅ | quality_gate_idle | memory_usage | 10/10 | bounds checks dashboard |
| ✅ | quality_gate_idle_all_features | memory_usage | 10/10 | bounds checks dashboard |
| ✅ | quality_gate_logs | lost_bytes | 10/10 | |
| ✅ | quality_gate_logs | memory_usage | 10/10 |
Explanation
Confidence level: 90.00%
Effect size tolerance: |Δ mean %| ≥ 5.00%
Performance changes are noted in the perf column of each table:
- ✅ = significantly better comparison variant performance
- ❌ = significantly worse comparison variant performance
- ➖ = no significant change in performance
A regression test is an A/B test of target performance in a repeatable rig, where "performance" is measured as "comparison variant minus baseline variant" for an optimization goal (e.g., ingress throughput). Due to intrinsic variability in measuring that goal, we can only estimate its mean value for each experiment; we report uncertainty in that value as a 90.00% confidence interval denoted "Δ mean % CI".
For each experiment, we decide whether a change in performance is a "regression" -- a change worth investigating further -- if all of the following criteria are true:
-
Its estimated |Δ mean %| ≥ 5.00%, indicating the change is big enough to merit a closer look.
-
Its 90.00% confidence interval "Δ mean % CI" does not contain zero, indicating that if our statistical model is accurate, there is at least a 90.00% chance there is a difference in performance between baseline and comparison variants.
-
Its configuration does not mark it "erratic".
CI Pass/Fail Decision
✅ Passed. All Quality Gates passed.
- quality_gate_idle_all_features, bounds check memory_usage: 10/10 replicas passed. Gate passed.
- quality_gate_logs, bounds check lost_bytes: 10/10 replicas passed. Gate passed.
- quality_gate_logs, bounds check memory_usage: 10/10 replicas passed. Gate passed.
- quality_gate_idle, bounds check memory_usage: 10/10 replicas passed. Gate passed.
eBPF complexity changesSummary result: ❗ - significant complexity increases
usm detailsusm [programs with changes]
usm [programs without changes]
This report was generated based on the complexity data for the current branch guy.arbitman/USMON-1083-incomplete-frame-type (pipeline 51895814, commit 50a6b66) and the base branch main (commit 57183f8). Objects without changes are not reported. Contact #ebpf-platform if you have any questions/feedback. Table complexity legend: 🔵 - new; ⚪ - unchanged; 🟢 - reduced; 🔴 - increased |
guyarb
force-pushed
the
guy.arbitman/USMON-1083-incomplete-frame-type
branch
4 times, most recently
from
65f180a to
3b9934d
Compare
The new enum represents the different options for an incomlete frame. Either we have an incomplete frame header or an incomplete frame payload. Having also a default state 'unknown' to avoid adding bugs of relying on the default value of the type as either incomlete frame header or an incomplete frame payload
Introducing types incomplete frame header and incomplete frame payload types. Those types will be later used to store the correct status of the incompletion
In a followup commit, we will need to have the last frame header for the case of incomplete frame payload. For that matter, we save the last frame header from the function 'pktbuf_find_relevant_frames'. In case we do have an incomplete frame payload, we need to know whether the frame is interesting (for further processing) and how many bytes we need to read from future packets.
The code was duplicated, and in the future commits we will use it further
We distinguish between incomplete frame header and incomplete frame payload Now the logic treat each of them differentaly and makes the code much clearer
Now we do support handling incomplete frame payload
Added 2 tests, the first to check we are able to consume correctly frame that is split to many packets. The second to check an unsupported scenario (headers frame that its payload split on multiple packets) does not prevent us from continue capturing valid requests on the connection
guyarb
force-pushed
the
guy.arbitman/USMON-1083-incomplete-frame-type
branch
from
3b9934d to
b185c46
Compare
…first_frame The program 'filter_frame' assumes it reads a frame from the current offset, and knows how to generate incomplete frame state. So if we don't have an existing incomplete_frame, we can rely on 'filter_frame' to do the same as we used to do in 'handle_first_frame'
A preliminary change for the followup commit. Allow to decide dynamically what will be the next tail call we use after handle_first_frame. In the followup PR, we will have a flow in which handle_first_frame will call itself with modified state
Added a 'processed' flag to the frame_payload struct. We want to avoid from processing some frames multiple times, and the flag is meant to ensure it. For data frames we only process the header and we don't care about their payloads, so we need to ensure we processed the frame header only once. But for header frames we do process the payload, and we still need to do so as a best effort. A common practice is to send only the frame header for Headers frame, and in a separate packet to send the payload. In such a case, we will have a payload remainder, and we want to process it
Now we simplify the processing and just and only handling incomplete frames, and not generic handling first frame.
guyarb
changed the title
guyarb
added
changelog/no-changelog
team/usm
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What does this PR do?
Refactors the way we maintain and process incomplete frames to better handle edge cases we've found.
The PR introduces two types of an incomplete frame -
The PR also changes the responsibility of
handle_first_frametohandle_incomplete_frame.We had a redundant duplication between
handle_first_frameandfilter_frame, which contributed to the complex conditions we had, and to the edge cases we missed. Now the probe is just consuming the remainder (if exists).Motivation
An edge case (USMON-1083) found while working on better dealing with TCP out of order.
Describe how you validated your changes
Possible Drawbacks / Trade-offs
Additional Notes
Suggested to review it in a commit-by-commit approach and read the commit messages.