Fix broken component integration test due to compute_world_size app not respecting env vars set by torchrun #1029
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Outdated suggestions cannot be applied.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:
compute_world_size
is run as an integration test in a-j 2x2
configuration usingtorchrun
which setsMASTER_ADDR
andMASTER_PORT
. However, it was ignoring those env vars and overriding them with the ones in the hydra config (added to makecompute_world_size
work as a single process withouttorchrun
).Integ tests are failing in CI because
localhost:0
(pick random free port) is used as theMASTER_ADDR:MASTER_PORT
on all 4 workers, hence all 4 workers are deadlocked waiting for each other to join the job.This diff fixes this by only setting the env vars if one is not already set.
Differential Revision: D71919903