Conversation
This was referenced Oct 1, 2026
SumanthRH
added this pull request to stack #2378
October 1, 2026 14:18
SumanthRH
force-pushed
the
metrics/generation-activity
branch
from
October 1, 2026 14:36
fbca2af to
75d8248
Compare
SumanthRH
force-pushed
the
metrics/generation-activity
branch
from
October 1, 2026 18:46
75d8248 to
849cf57
Compare
Signed-off-by: Codex <codex@users.noreply.github.com> Signed-off-by: SumanthRH <sumanthrh99@gmail.com>
Signed-off-by: Codex <codex@users.noreply.github.com> Signed-off-by: SumanthRH <sumanthrh99@gmail.com>
SumanthRH
force-pushed
the
metrics/generation-activity
branch
from
October 1, 2026 19:54
849cf57 to
a99c068
Compare
This branch was successfully deployed
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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?
Measure the union of outstanding generation calls for async throughput denominators and publish their count through Ray metrics.
TLDR: Overlapping calls count once; gaps with no outstanding call contribute no generation time.
How it works
Wrap sync, async producer and evaluation generation calls in a shared activity context. Environment/tool waits within a call count as active time. The
skyrl_generation_active_callsgauge provides timeline correlation; gauge failures leave duration accounting intact. Zero active duration does not fall back to wall time.This layer is optional in principle: elapsed-time async throughput is valid without it. Removing it would change the async denominator and require adapting the later summary layer. It remains enabled pending the user's explicit removal decision.
Validation
Overlap/error cleanup, failed gauge publication and zero-duration sampling are covered in the final combined CPU suite recorded in PR #2368. No GPU performance claim.
Second layer of stack #2378; depends on #2364. All commits have the configured user's sign-off.