🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/151355

• 📄 Preview Python docs built from this PR
• 📄 Preview C++ docs built from this PR
• ❓ Need help or want to give feedback on the CI? Visit the bot commands wiki or our office hours

Note: Links to docs will display an error until the docs builds have been completed.

❌ 7 New Failures

As of commit 9452804 with merge base daf2ccf ():

This comment was automatically generated by Dr. CI and updates every 15 minutes.

@huydhn Is this something you can help with?
https://github.com/pytorch/pytorch/actions/runs/14479904535/job/40614420540

denied: User: arn:aws:sts::391835788720:assumed-role/ghci-lf-github-action-runners-runner-role/i-05b0cda8626afb94c is not authorized to perform: ecr:InitiateLayerUpload on resource: arn:aws:ecr:us-east-1:308535385114:repository/pytorch/manylinux2_28-builder-rocm6.4 because no resource-based policy allows the ecr:InitiateLayerUpload action

The only difference between the previous passed build and current failing build seems to be that the former used non-lf CI runners while the latter uses lf CI runners lf.ephemeral.linux.9xlarge.ephemeral

@pytorchbot rebase -b viable/strict

@pytorchbot started a rebase job onto refs/remotes/origin/viable/strict. Check the current status here

Successfully rebased rocm64_nightly onto refs/remotes/origin/viable/strict, please pull locally before adding more changes (for example, via git checkout rocm64_nightly && git pull --rebase)

A bit unrelated: why are we using new ECRs here instead of tags? We stopped creating new ECRs and just use tags for different builds

A bit unrelated: why are we using new ECRs here instead of tags? We stopped creating new ECRs and just use tags for different builds.

But current failure are due to the fact that LF runners do not have push permissions to new ECRs

@pytorchbot rebase

@pytorchbot started a rebase job onto refs/remotes/origin/viable/strict. Check the current status here

Successfully rebased rocm64_nightly onto refs/remotes/origin/viable/strict, please pull locally before adding more changes (for example, via git checkout rocm64_nightly && git pull --rebase)

A bit unrelated: why are we using new ECRs here instead of tags? We stopped creating new ECRs and just use tags for different builds

@malfet Yes, we'd like to move away from using ECRs to just tags. But I think that requires changing the naming convention for the docker repo to be ROCm-version agnostic, and have the ROCm version as part of the tag:

@pytorchbot merge -f "only failures are due to rocm 6.4 builder images not refreshed w/ magma package and available for the 6.4 wheels; images built fine, should work out okay"

Merge started

Your change will be merged immediately since you used the force (-f) flag, bypassing any CI checks (ETA: 1-5 minutes). Please use -f as last resort and instead consider -i/--ignore-current to continue the merge ignoring current failures. This will allow currently pending tests to finish and report signal before the merge.

Learn more about merging in the wiki.

Questions? Feedback? Please reach out to the PyTorch DevX Team