fix: Fix for eviction deadlock #25
Merged
Conversation
This file contains 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
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: njhill The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
Motivation A recent fix for a cascading eviction race condition introduced the possibility of a deadlock between the model cache lock and per-cache entry locks due to eviction callback logic now performed in-line under the cache lock. Modifications Rework cache accounting logic to make an eviction/unload related adjustment unconditionally up-front within the cache lock so that it can be released before the entry removal/unload processing is performed along with state-dependent readjustment if necessary. Also: - Ensure all entry weight updates are guarded by the cache lock - Change the cache's oldestTime() method to not take the cache lock since it's used on the inference request path Result Hopefully no more deadlocks Signed-off-by: Nick Hill <[email protected]>
njhill
force-pushed
the
deadlock-fix
branch
2 times, most recently
from
696c2a2 to
7fedcb4
Compare
|
This has passed our internal deployment tests, Merging due to urgency of fix and will be performing reviews in parallel. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
Motivation
A recent fix for a cascading eviction race condition introduced the possibility of a deadlock between the model cache lock and per-cache entry locks due to eviction callback logic now performed in-line under the cache lock.
Modifications
Rework cache accounting logic to make an eviction/unload related adjustment unconditionally up-front within the cache lock so that it can be released before the entry removal/unload processing is performed along with state-dependent readjustment if necessary.
Also:
oldestTime()method to not take the cache lock since it's used on the inference request pathResult
Hopefully no more deadlocks
Signed-off-by: Nick Hill [email protected]