pytorch / pytorch/tutorials

[BUG] Confusing machine names in Jenkins CI

Open
#3,780 1 comment 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

bug build automation build issue
Dominant language
Python
Stars
9.3k
Forks
4.4k
Avg merge
1d 21h
Merged PRs (30d)
4

Description

Add Link

My PR for a new tutorial: https://github.com/pytorch/tutorials/pull/3763/

Describe the bug

THe Jenkins CI has a few hard-coded behaviors that are confusing. The linux.16xlarge.nvidia.gpu label in Jenkins appears to be a legacy label that means "needs multi-GPU". It doesn't describe the actual hardware that is used.

get_files_to_run.py has a hard-coded check for this exact key, which routes it to shard 0. Shard 0 maps to WORKER_ID=1 in the matrix (also confusing), and then shard 1's runner is a linux.g5.12xlarge.nvidia.gpu, not a 16xl.

Reporting this since it resulted in some confusion while getting my CI for my PR to pass. If it were me, the key should be '4-gpu' instead of linux.16xlarge.nvidia.gpu.

Describe your environment

PyTorch tutorial CI

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start with .jenkins/metadata.json at line 27 and .jenkins/get_files_to_run.py at lines 43-45, then inspect the CI matrix mapping described in the issue. Verify the label and shard/WORKER_ID relationship, including the linux.g5.12xlarge.nvidia.gpu runner. Done means the machine naming and routing consistently describe the multi-GPU CI behavior without breaking tutorial CI.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
ci-cd, devops
Issue type
Bug
Difficulty
3/5
Estimated time
1-2 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
45/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.