Enhance torch240 rendezvous to improve fault tolerance ability. #1454
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
BalaBalaYi
requested review from
workingloong,
samplise and
majieyue
as code owners
…rance_when_assign_workers # Conflicts: # dlrover/python/tests/test_elastic_training_agent.py
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## master #1454 +/- ##
==========================================
+ Coverage 81.61% 81.74% +0.13%
==========================================
Files 240 240
Lines 24045 24097 +52
==========================================
+ Hits 19624 19699 +75
+ Misses 4421 4398 -23 ☔ View full report in Codecov by Sentry. |
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Outdated suggestions cannot be applied.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.
What changes were proposed in this pull request?
Add 'wait' logic on 2 parts:
Why are the changes needed?
Optimization for rendezvous logic targeting torch version greater than 2.4.0.
If a worker exits during the process of assigning RANK for rendezvous, resulting in empty metadata retrieval, the process will not immediately exit due to deserialization failure. Instead, it will wait (to prevent all networking workers from encountering errors and exiting immediately). If non-empty data cannot be obtained eventually, the process will terminate with an exception due to a pending timeout.
Does this PR introduce any user-facing change?
No.
How was this patch tested?
UT and training(TODO).