codeflare-sdk icon indicating copy to clipboard operation
codeflare-sdk copied to clipboard

feat(RHOAIENG-25241): disable ray usage metrics by default

Open kryanbeane opened this issue 6 months ago • 3 comments

Issue link

What changes have been made

Made RAY_USAGE_STATS_ENABLED default to = 0, which == disabled. Added unit tests to ensure enabledment and disablement. Also did manual testing and verified that a deployed custom build image has a default env var of RAY_USAGE_STATS_ENABLED = 0 on both head and worker pods

I verified that the env was present on the cluster

I also deployed a custom codeflare-sdk build on cluster and verified that /tmp/ray/session_xxx/usage_stats.json file does not exist. This is where ray temp writes usage stats to on the head.

Verification steps

  • Build custom image of codeflare-sdk (which goes to .dist folder)
  • Drag that to jupyter notebook in your cluster or locally, use it to install codeflare in notebook
  • Create a cluster (don't specify envs) and ensure head and workers have the env in place with a value of 0

Checks

  • [ ] I've made sure the tests are passing.
  • Testing Strategy
    • [ ] Unit tests
    • [ ] Manual tests
    • [ ] Testing is not required for this change

kryanbeane avatar May 15 '25 20:05 kryanbeane