LaunchDarkly (https://launchdarkly.com/) provides SaaS based Feature Flag management at scale, with SDKs for all major languages.
This unofficial module provides NodeJs functions wrapping the LaunchDarkly API. This is not the sdk for implementing flags in your app - it is the api to manage your account/flags.
We need a way to manage flags as part of CI/CD pipelines, and there is not another project currently providing a simple interface to manage flags and other objects in LaunchDarkly via API in nodejs.
There is a swagger.yaml available to generate bindings (https://launchdarkly.github.io/ld-openapi/swagger.yaml), so we use the swagger-js module to generate a client (https://github.com/swagger-api/swagger-js), and add some extra features around logging, error handling, and chaining of operations. In addition we expose apis as a commandline tool.
npm install launchdarkly-nodeutils --save
or clone this repo.- Generate an access token with the permissions for the operations you will use. Please read: https://docs.launchdarkly.com/v2.0/docs/api-access-tokens
The command line tool or nodejs module can be used for many things. Here are some examples.
Automated management of ACLs
Users of Enterprise LaunchDarkly can create Custom Roles for fine-grained RBAC. These can be maintained in git, and applied to LaunchDarkly when changes are committed via bulkUpsertCustomRoles
or bulkUpsertCustomRoleFolder
.
Setting flag state for automated tests
Automated test suites can call the apis to toggle features and set targetting rules before test executions. With care, a matrix of flag based test scenarios could defined and executed.
Releasing flags in bulk
A typical release workflow to move a set of flag attributes from test to prod may consist of CI jobs calling ldutils to:
- create a backup of flags to be migrated using
getFeatureFlags my-proj > ./backup.json
- use
bulkMigrateFeatureFlags my-proj flag-x,flag-y test prod
to copy flag attrs from test to prod - if issues, to rollback the flags use
restoreFeatureFlags my-proj flag-x,flag-y prod ./backup.json
Scheduled backup of flags
A cron/ci job could call getFeatureFlags for each project, and commit the json to private git repo in order to keep a versioned record of change. This could also be used to recover environment state.
Refreshing Environments
Copying flag attributes from prod back to preprod for testing. Also newly created environments can be primed with flag targetting rules using bulkMigrateFeatureFlags
.
After cloning this repo you can make ldutils
executable, and use it to make api calls based on passed in parameters.
please read the API documentation for examples.
chmod 755 ./ldutils
./ldutils
The above will display a help screen of instructions, thanks to https://github.com/tj/commander.js/
Optionally you can add ldutils
to your PATH:
# cloned repo
sudo ln -s /<clonepath>/launchdarkly-nodeutils/ldutils /usr/local/bin/ldutils
# or after 'npm install launchdarkly-nodeutils --save'
sudo ln -s /<installpath>/node_modules/launchdarkly-nodeutils/ldutils /usr/local/bin/ldutils
Make sure you have env var LAUNCHDARKLY_API_TOKEN set, and if piping output to another command, ensure that LAUNCHDARKLY_API_LOGLEVEL is not set to 'debug' to ensure only the json result of the command is returned.
Here are some examples of commandline usage (if you have not added ldutils to PATH, prefix with ./
:
export LAUNCHDARKLY_API_TOKEN=<api-token>
// collect all flags for a project
ldutils getFeatureFlags <myProjectId>
// update or create a customRole with array of policies
ldutils upsertCustomRole <customRoleKey> <customRoleName> '[{"resources":["proj/*"],"actions":["*"],"effect":"allow"}]'
If you are executing ldutils via a proxy, just set the https_proxy env var, eg:
export https_proxy=http://localhost:3128
The command line modes and parameters map directly to the functions exposed for use in nodejs apps.
Each call to ldutils
takes a 'mode', and associated parameters as below:
ldutils <mode> [parameter values space separated]
The following modes are supported. This info is also available via: ldutils -help
Mode | parameters |
---|---|
getFeatureFlags | projectKey |
getFeatureFlag | projectKey, featureFlagKey, environmentKeyQuery |
getFeatureFlagState | projectKey, featureFlagKey, environmentKeyQuery |
updateFeatureFlag | projectKey, featureFlagKey, patchComment |
toggleFeatureFlag | projectKey, featureFlagKey, environmentKeyQuery, enabled |
migrateFeatureFlag | projectKey, featureFlagKey, fromEnv, toEnv, includeState |
bulkMigrateFeatureFlags | projectKey, featureFlagKeys, fromEnv, toEnv, includeState |
restoreFeatureFlags | projectKey, featureFlagKeys, targetEnv, backupJsonFile, includeState |
migrateFeatureFlag
mode is used to copy flag attributes between environments. This covers: targets, rules, fallthrough, offVariation, prerequisites and optionally the flag on/off state. eg. to migrate a flag from dev to test env.
ldutils migrateFeatureFlag my-project my-flag dev test
use
bulkMigrateFeatureFlags
with a comma separated list of flags ids to migrations multiple flags in one operation.
Custom role functionality requires an Enterprise LaunchDarkly account.
Mode | parameters |
---|---|
getCustomRoles | none |
getCustomRole | customRoleKey |
getCustomRoleById | customRoleId |
createCustomRole | customRoleKey, customRoleName, customRolePolicyArray, customRoleDescription(optional) |
updateCustomRole | customRoleKey, customRoleName, customRolePolicyArray, customRoleDescription(optional) |
upsertCustomRole | customRoleKey, customRoleName, customRolePolicyArray, customRoleDescription(optional) |
bulkUpsertCustomRoles | roleBulkLoadFile |
bulkUpsertCustomRoleFolder | roleFolder |
bulkUpsertCustomRoles
mode iterates over a json file containing an array of role json and either creates or updates each. Promises are resolved sequentially to avoid rate limiting.bulkUpsertCustomRoleFolder
mode does the same on a folder of json files.
ldutils bulkUpsertCustomRoles ./exampleRoleBulkLoad.json
ldutils bulkUpsertCustomRoleFolder ./companyRoles
For details on role policy object structures, please see: https://docs.launchdarkly.com/docs/custom-roles
Mode | parameters |
---|---|
getTeamMembers | none |
getTeamMember | memberId |
getTeamMemberByEmail | emailAddress |
getTeamMemberCustomRoles | emailAddress |
inviteTeamMember | emailAddress, initialRole |
./ldutils inviteTeamMember [email protected] reader
Assumes that you have set the LAUNCHDARKLY_API_TOKEN environment var.
let ldUtils = await new LaunchDarklyUtils().create(process.env.LAUNCHDARKLY_API_TOKEN);
// get flag state
let flagEnabled = await ldUtils.getFeatureFlagState('myProject', 'feature-abc', 'dev');
// update flag state to on=true
await ldUtils.toggleFeatureFlag('myProject', 'feature-def', 'dev', true);
please read the API documentation for further details.
Here is a summary. For more detail see Contributing.
- Fork this repo and work on your enhancements. See note on Commitizen below.
- Ensure that you include unit tests and jsdoc annotations.
- Ensure that
yarn test
passes. - Use
yarn commit
for conventional commits. - Raise a pull request.
This project uses commitizen for conventional commit messages via git cz
instead of git commit
.
The reason for this is that it attaches meaning to each commit that is useful to others, and that is used to automatically version new releases based on http://semver.org
Just use
yarn commit
instead of git commit and follow instructions.