Fix ProcessGroupGloo crash when setting unsupported backend options#324
Fix ProcessGroupGloo crash when setting unsupported backend options#324wanna-01 wants to merge 1 commit into
Conversation
|
Hi @wanna-01! Thank you for your pull request and welcome to our community. Action RequiredIn order to merge any pull request (code, docs, etc.), we require contributors to sign our Contributor License Agreement, and we don't seem to have one on file for you. ProcessIn order for us to review and merge your suggested changes, please sign at https://code.facebook.com/cla. If you are contributing on behalf of someone else (eg your employer), the individual CLA may not be sufficient and your employer may need to sign the corporate CLA. Once the CLA is signed, our tooling will perform checks and validations. Afterwards, the pull request will be tagged with If you have received this in error or have any questions, please contact us at [email protected]. Thanks! |
|
Thank you for signing our Contributor License Agreement. We can now accept your code for this (and any) Meta Open Source project. Thanks! |
Title
Fix
ProcessGroupGloocrash when setting unsupported backend optionsSummary
This MR fixes a runtime crash in
ProcessGroupGloo._create_pg()caused by unconditional assignment to Gloo backend options that are not present in PyTorch'sProcessGroupGloo._Options.Why this change is needed
Current code in
ProcessGroupGloo._create_pg()does:backend_class.options.global_ranks_in_group = ...backend_class.options.group_name = ...For Gloo,
backend_class.optionsistorch._C._distributed_c10d._Options, which does not expose these attributes (it typically only hasbackend). Entering quorum on Gloo can therefore raiseAttributeErrorand fail immediately.NCCL path is not affected because NCCL
Optionsdoes provide these fields.Root cause
The code assumes Gloo options support the same attributes as NCCL options.
What this MR changes
In
torchft/process_group.py(ProcessGroupGloo._create_pg):global_ranks_in_groupintry/except AttributeErrorgroup_nameintry/except AttributeErrorSo unsupported optional fields are skipped on Gloo instead of crashing.
Behavioral impact
AttributeError