Skip to content

Conversation

@MarcusSorealheis
Copy link
Collaborator

@MarcusSorealheis MarcusSorealheis commented Dec 12, 2025

Description

NativeLink uploads outputs concurrently and uses update_with_whole_file, which moves the file into CAS. If two uploads race on the same path, the second one hits NotFound on rename even though the digest is already in CAS. This PR fixes that issue.

Fixes #2092

Type of change

Please delete options that aren't relevant.

  • Bug fix (non-breaking change which fixes an issue)

How Has This Been Tested?

It's being tested by our users and our customers.

Checklist

  • Updated documentation if needed
  • Tests added/amended
  • bazel test //... passes locally
  • PR is contained in a single commit, using git amend see some docs

This change is Reviewable

@MarcusSorealheis
Copy link
Collaborator Author

This bugfix is based on this section of the RE spec:

https://github.com/bazelbuild/remote-apis/blob/3051389c06348307437e92e3a1d3c6d6566094b4/build/bazel/remote/execution/v2/remote_execution.proto#L1178-L1181

I'm pointing it out because the user who reported this issue was an AOSP user, although other users may encounter this issue because we were not compliant before.

@MarcusSorealheis
Copy link
Collaborator Author

/build-image

@github-actions
Copy link

Image built and pushed!

ghcr.io/TraceMachina/nativelink:caae987

@amankrx
Copy link
Collaborator

amankrx commented Dec 16, 2025

/build-image

@github-actions
Copy link

Image built and pushed!

ghcr.io/TraceMachina/nativelink:bfa4fea

@amankrx amankrx merged commit 44ada84 into TraceMachina:main Dec 16, 2025
27 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Cascade of Disconnects from Overloaded Worker

2 participants