Skip to content

[Core] Throttle infeasible resource warning#59790

Merged
dayshah merged 2 commits intoray-project:masterfrom
win5923:infeasible-message-warning
Jan 4, 2026
Merged

[Core] Throttle infeasible resource warning#59790
dayshah merged 2 commits intoray-project:masterfrom
win5923:infeasible-message-warning

Conversation

@win5923
Copy link
Member

@win5923 win5923 commented Dec 31, 2025

Description

The "infeasible resource requests" warning is published repeatedly during normal autoscaler scale-up operations, treating expected transient states as error conditions. This creates significant log noise in autoscaling clusters where workers scale to zero when idle.

Add rate limiting to infeasible resource request warnings using the RAY_LOG_EVERY_MS macro, with a 60s interval.

Related issues

Closes #59151

Additional information

Optional: Add implementation details, API changes, usage examples, screenshots, etc.

@win5923 win5923 requested a review from a team as a code owner December 31, 2025 19:42
Copy link
Contributor

@gemini-code-assist gemini-code-assist bot left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Code Review

This pull request introduces throttling for infeasible resource warnings to prevent log spam, which is a valuable improvement. The implementation correctly leverages the Throttler utility and adds a new configuration option for the interval. The code is clear and follows the intended logic. I have one minor suggestion to improve the robustness of the time unit conversion.

@ray-gardener ray-gardener bot added question Just a question :) core Issues that should be addressed in Ray Core observability Issues related to the Ray Dashboard, Logging, Metrics, Tracing, and/or Profiling community-contribution Contributed by the community labels Jan 1, 2026
@win5923 win5923 force-pushed the infeasible-message-warning branch from 7a684ce to 77d0c91 Compare January 2, 2026 16:53
"of hanging, set the 'RAY_enable_infeasible_task_early_exit=true'. "
"This feature will be turned on by default in a future release of Ray.";
RAY_LOG(WARNING) << error_message;
RAY_LOG_EVERY_MS(WARNING, 60000) << error_message;
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

PublishError call not throttled alongside log message

The log message is rate-limited to once every 60 seconds using RAY_LOG_EVERY_MS, but the gcs_publisher_->PublishError() call remains unthrottled. This means while the log noise is reduced, errors are still published to subscribers at full rate every time the callback executes. The PR description mentions wanting to address the warning being "published repeatedly," but only the logging portion is throttled. The published errors to GCS subscribers will continue at the same frequency as before.

Fix in Cursor Fix in Web

@dayshah
Copy link
Contributor

dayshah commented Jan 2, 2026

can you update the pr description

@win5923
Copy link
Member Author

win5923 commented Jan 3, 2026

Thanks! Updated.

@dayshah dayshah added the go add ONLY when ready to merge, run all tests label Jan 4, 2026
@dayshah dayshah enabled auto-merge (squash) January 4, 2026 08:06
@dayshah dayshah merged commit 9be6588 into ray-project:master Jan 4, 2026
8 checks passed
@win5923 win5923 deleted the infeasible-message-warning branch January 4, 2026 10:28
AYou0207 pushed a commit to AYou0207/ray that referenced this pull request Jan 13, 2026
Signed-off-by: win5923 <ken89@kimo.com>
Signed-off-by: jasonwrwang <jasonwrwang@tencent.com>
lee1258561 pushed a commit to pinterest/ray that referenced this pull request Feb 3, 2026
ryanaoleary pushed a commit to ryanaoleary/ray that referenced this pull request Feb 3, 2026
peterxcli pushed a commit to peterxcli/ray that referenced this pull request Feb 25, 2026
Signed-off-by: win5923 <ken89@kimo.com>
Signed-off-by: peterxcli <peterxcli@gmail.com>
peterxcli pushed a commit to peterxcli/ray that referenced this pull request Feb 25, 2026
Signed-off-by: win5923 <ken89@kimo.com>
Signed-off-by: peterxcli <peterxcli@gmail.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

community-contribution Contributed by the community core Issues that should be addressed in Ray Core go add ONLY when ready to merge, run all tests observability Issues related to the Ray Dashboard, Logging, Metrics, Tracing, and/or Profiling question Just a question :)

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[Autoscaler v2] Infeasible resource warning spams during normal autoscaler scale-up

2 participants