fix(longlive): add logger + warning when KV cache write is skipped (#921)#924
Open
livepeer-tessa wants to merge 1 commit intomainfrom
Open
fix(longlive): add logger + warning when KV cache write is skipped (#921)#924livepeer-tessa wants to merge 1 commit intomainfrom
livepeer-tessa wants to merge 1 commit intomainfrom
Conversation
) The `if write_len > 0` guards in `CausalWanSelfAttention.forward()` already prevent the RuntimeError (empty tensor slice when write_start_index == local_end_index), but the skip was silent, making it impossible to diagnose cache-state drift from Grafana logs. Changes: - Add `import logging` and `logger = logging.getLogger(__name__)` - Add `else: logger.warning(...)` branch on both the roll_and_insert and direct_insert write paths so that stale-index events are surfaced as WARNING-level log entries (rather than ~310 silent skips/session) The WARNING includes write_start_index, local_end_index, and a reference to #921 to aid future investigation of cache index drift after rapid video↔text mode transitions. Fixes #921 Signed-off-by: Tessa (livepeer-tessa) <tessa@livepeer.org>
|
Important Review skippedAuto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
Contributor
🚀 fal.ai Preview Deployment
Livepeer Runner
Testing Livepeer Mode |
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
1 participant
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.
Summary
Addresses #921.
The
if write_len > 0guards on both KV cache write paths inCausalWanSelfAttention.forward()already prevent the RuntimeError that caused 310+ chunk failures per session (empty tensor slice whenwrite_start_index == local_end_index). However, silent skips made it impossible to detect cache-state drift from Grafana logs.Changes
import loggingandlogger = logging.getLogger(__name__)tocausal_model.pyelse: logger.warning(...)on both the roll_and_insert and direct_insert write pathsThe WARNING message includes
write_start_index,local_end_index, and a reference to #921.Why this matters
When rapid video↔text mode transitions leave the cache index in a stale state, the skip now surfaces in logs as a WARNING instead of silently passing through. This:
KV cache write skippedif this starts happening frequentlyTesting
python3 -c 'import ast; ast.parse(...)'✅cc @mjh1 @emranemran