(torch/elastic) fix scale down bug caused by calling rdzv_handler.shutdown() on premature agent failures #67749
Conversation
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
🔗 Helpful links
💊 CI failures summary and remediationsAs of commit e3dc0d9 (more details on the Dr. CI page):
🕵️ 1 new failure recognized by patternsThe following CI failures do not appear to be due to upstream breakages:
|
CI Flow Status⚛️ CI FlowRuleset - Version:
You can add a comment to the PR and tag @pytorchbot with the following commands: # ciflow rerun, "ciflow/default" will always be added automatically
@pytorchbot ciflow rerun
# ciflow rerun with additional labels "-l <ciflow/label_name>", which is equivalent to adding these labels manually and trigger the rerun
@pytorchbot ciflow rerun -l ciflow/scheduled -l ciflow/slowFor more information, please take a look at the CI Flow Wiki. |
|
This pull request was exported from Phabricator. Differential Revision: D32129005 |
|
This pull request was exported from Phabricator. Differential Revision: D32129005 |
d8879aa to
a985cb5
Compare
|
This pull request was exported from Phabricator. Differential Revision: D32129005 |
a985cb5 to
135bcea
Compare
|
This pull request was exported from Phabricator. Differential Revision: D32129005 |
135bcea to
7265756
Compare
|
This pull request was exported from Phabricator. Differential Revision: D32129005 |
7265756 to
48be767
Compare
…tdown() on premature agent failures (pytorch#67749) Summary: Pull Request resolved: pytorch#67749 Fixes: pytorch#67742 Test Plan: Added unittests. Validated manually: ``` # start agent 0 $ torchrun --rdzv_backend c10d --rdzv_id 123 --rdzv_endpoint localhost:29500 --nnodes 1:2 --nproc_per_node 1 --monitor_interval 1 test.py # start agent 1 torchrun --rdzv_backend c10d --rdzv_id 123 --rdzv_endpoint localhost:29500 --nnodes 1:2 --nproc_per_node 1 --monitor_interval 1 test.py # kill agent 0 CTRL+C (SIGINT) or kill -15 (SIGTERM) # restart it torchrun --rdzv_backend c10d --rdzv_id 123 --rdzv_endpoint localhost:29500 --nnodes 1:2 --nproc_per_node 1 --monitor_interval 1 test.py ``` Reviewed By: cbalioglu Differential Revision: D32129005 fbshipit-source-id: 4e695d0b3397951d375ecee321add5faf0cfa3ea
|
This pull request was exported from Phabricator. Differential Revision: D32129005 |
48be767 to
e3dc0d9
Compare
|
This pull request has been merged in f6402c4. |
georgkaleido
added a commit
to georgkaleido/pytorch
that referenced
this pull request
Apr 30, 2025
In pytorch#117066, shutdown of the rendezvous was added if a worker shuts down. This is incorrect, because the rendezvous is actually shutdown in [this file](https://github.com/pytorch/pytorch/blob/fa6f9eb2be07f6289d2ab4e781077f7fc75dbe55/torch/distributed/launcher/api.py#L290) but should not be shutdown if a signal is received. See also [this pull request](pytorch#67749). pytorch#124819 then tried to remediate the situation by fixing the faulty shutdown for the restart case. But this is only triggered if the agent restarts the training, but not if the shutdown of the rendezvous happened before. Removing both these changes restores the original behavior. The rendezvous should only be shutdown if a run completes or fails, not for a single worker leaving.
pytorchmergebot
pushed a commit
to georgkaleido/pytorch
that referenced
this pull request
May 13, 2025
In pytorch#117066, shutdown of the rendezvous was added if a worker shuts down. This is incorrect, because the rendezvous is actually shutdown in [this file](https://github.com/pytorch/pytorch/blob/fa6f9eb2be07f6289d2ab4e781077f7fc75dbe55/torch/distributed/launcher/api.py#L290) but should not be shutdown if a signal is received. See also [this pull request](pytorch#67749). pytorch#124819 then tried to remediate the situation by fixing the faulty shutdown for the restart case. But this is only triggered if the agent restarts the training, but not if the shutdown of the rendezvous happened before. Removing both these chang 8000 es restores the original behavior. The rendezvous should only be shutdown if a run completes or fails, not for a single worker leaving.
pytorchmergebot
pushed a commit
that referenced
this pull request
May 14, 2025
In #117066, shutdown of the rendezvous was added if a worker shuts down. This is incorrect, because the rendezvous is actually shutdown in [this file](https://github.com/pytorch/pytorch/blob/fa6f9eb2be07f6289d2ab4e781077f7fc75dbe55/torch/distributed/launcher/api.py#L290) but should not be shutdown if a signal is received. See also [this pull request](#67749). #124819 then tried to remediate the situation by fixing the faulty shutdown for the restart case. But this is only triggered if the agent restarts the training, but not if the shutdown of the rendezvous happened before. Removing both these changes restores the original behavior. The rendezvous should only be shutdown if a run completes or fails, not for a single worker leaving. Fixes #150916 Fixes #147064 Pull Request resolved: #152525 Approved by: https://github.com/kiukchung
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.
Summary: Fixes: #67742
Test Plan: deferring the testing to whoever is actually going to land this code - this diff is only for demonstrative purposes
Differential Revision: D32129005
cc @pietern @mrshenli @pritamdamania87 @zhaojuanmao @satgera @rohan-varma @gqchen @aazzolini @osalpekar @jiayisuse @SciPioneer @H-Huang