Deduplicate lockstep MLA pending job cleanup - #11
Closed
procr1337 wants to merge 1 commit into
Closed
Conversation
Lockstep MLA groups share physical block IDs, so a transfer job may contain the same block ID once per logical group. Iterate the deduplicated set of block IDs in _remove_pending_job() so each (block, job) pair is only removed once from the set-backed pending-job index, instead of crashing with KeyError on the second occurrence. Fixes a KeyError in _remove_pending_job() observed when KV cache offloading (--kv-offloading-size) is enabled while serving GLM-5.2 (MLA + decode context parallelism), reproduced by offload-fix-glm52/probe_gpu_cache.py.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Lockstep MLA groups share physical block IDs, so a transfer job may contain the same block ID once per logical group. Iterate the deduplicated set of block IDs in _remove_pending_job() so each (block, job) pair is only removed once from the set-backed pending-job index, instead of crashing with KeyError on the second occurrence.
Fixes a KeyError in _remove_pending_job() observed when KV cache offloading (--kv-offloading-size) is enabled while serving GLM-5.2 (MLA + decode context parallelism), reproduced by
offload-fix-glm52/probe_gpu_cache.py.