Skip to content

Scope pybind11 functions to torch.distributed.{autograd,rpc} #27529

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Closed
pietern wants to merge 5 commits into from
Closed

Scope pybind11 functions to torch.distributed.{autograd,rpc} #27529

pietern wants to merge 5 commits into from

Conversation

@pietern
Copy link
Contributor

@pietern pietern commented Oct 8, 2019 •
edited
Loading

Stack from ghstack:

Differential Revision: D17808209

@pytorchbot pytorchbot added the oncall: distributed Add this issue/PR to distributed oncall triage queue label Oct 8, 2019
This was referenced Oct 8, 2019
Closed
Closed
Closed
Closed
Closed
Closed
Closed
Closed
Closed
Closed
mrshenli
mrshenli approved these changes Oct 8, 2019
View reviewed changes
@mrshenli
Copy link
Contributor

mrshenli commented Oct 8, 2019

Test failure is real:

Oct 08 10:55:52 Running test_dist_autograd_fork ... [2019-10-08 10:55:52.553930]
Oct 08 10:55:52 Traceback (most recent call last):
Oct 08 10:55:52   File "test_dist_autograd_fork.py", line 4, in <module>
Oct 08 10:55:52     from dist_autograd_test import DistAutogradTest
Oct 08 10:55:52   File "/var/lib/jenkins/workspace/test/dist_autograd_test.py", line 9, in <module>
Oct 08 10:55:52     from dist_utils import INIT_METHOD_TEMPLATE, dist_init
Oct 08 10:55:52   File "/var/lib/jenkins/workspace/test/dist_utils.py", line 8, in <module>
Oct 08 10:55:52     from torch.distributed.rpc.api import RpcBackend
Oct 08 10:55:52   File "/opt/python/2.7.9/lib/python2.7/site-packages/torch/distributed/rpc/api.py", line 1, in <module>
Oct 08 10:55:52     from . import invoke_rpc_builtin, invoke_rpc_python_udf
Oct 08 10:55:52 ImportError: cannot import name invoke_rpc_builtin
Oct 08 10:55:52 Traceback (most recent call last):
Oct 08 10:55:52   File "test/run_test.py", line 458, in <module>
Oct 08 10:55:52     main()
Oct 08 10:55:52   File "test/run_test.py", line 450, in main
Oct 08 10:55:52     raise RuntimeError(message)
Oct 08 10:55:52 RuntimeError: test_dist_autograd_fork failed!

@pietern
Copy link
Contributor Author

pietern commented Oct 9, 2019

The failure is real, but the problem is that test_dist_autograd_fork shouldn't even run for Python 2.

The imports it does are assumed to work, but this commit makes initialization of the RPC module (and the pybind11 functions) conditional on using Python 3.

@pietern pietern mentioned this pull request Oct 9, 2019
Closed
@pietern pietern added the module: rpc Related to RPC, distributed autograd, RRef, and distributed optimizer label Oct 9, 2019
@pietern pietern mentioned this pull request Nov 4, 2019
Closed
@pietern
Copy link
Contributor Author

pietern commented Nov 5, 2019

The commit that's landing has a fix for the build failures in the most recent commit in this PR.

This was referenced Nov 5, 2019
Closed
Closed
@facebook-github-bot
Copy link
Contributor

facebook-github-bot commented Nov 5, 2019

@pietern merged this pull request in b4df413.

@facebook-github-bot facebook-github-bot deleted the gh/pietern/51/head branch November 8, 2019 15:16
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@mrshenli mrshenli mrshenli approved these changes

@apaszke apaszke Awaiting requested review from apaszke

Assignees

No one assigned

Labels

Merged module: rpc Related to RPC, distributed autograd, RRef, and distributed optimizer oncall: distributed Add this issue/PR to distributed oncall triage queue

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

6 participants

@pietern @mrshenli @facebook-github-bot @pytorchbot @mruberry