Skip to content

Disable workflow telemetry and cache in all CI workflows#917

Open
ChrisRackauckas-Claude wants to merge 1 commit into
SciML:masterfrom
ChrisRackauckas-Claude:disable-ci-telemetry-cache
Open

Disable workflow telemetry and cache in all CI workflows#917
ChrisRackauckas-Claude wants to merge 1 commit into
SciML:masterfrom
ChrisRackauckas-Claude:disable-ci-telemetry-cache

Conversation

@ChrisRackauckas-Claude
Copy link
Copy Markdown
Contributor

Summary

Pass enable_telemetry: false and enable_cache: false to every LuxDL/Lux.jl/.github/workflows/CommonCI.yml@main call site — 23 occurrences across all 11 CI_*.yml workflow files.

Why

Both features have been causing spurious CI job failures on the self-hosted SciML runners, most recently observed on #910 where 5+ jobs reported FAILURE despite Testing … tests passed:

Telemetry (catchpoint/workflow-telemetry-action@v2):

##[error][Workflow Telemetry] AxiosError
##[error]AxiosError: Request failed with status code 413

Cache (julia-actions/cache@v3):

error: could not lock config file /home/chrisrackauckas/.gitconfig: File exists
##[error]Process completed with exit code 255.
##[warning]uploadCacheArchiveSDK: Server failed to authenticate the request.

The enable_telemetry and enable_cache inputs were added in LuxDL/Lux.jl#1695 (defaults true for backward compatibility with other consumers). Setting them explicitly to false here makes NonlinearSolve.jl CI resilient to the self-hosted runner environment regardless of any future default changes upstream.

Test plan

🤖 Generated with Claude Code

Pass `enable_telemetry: false` and `enable_cache: false` to every
`LuxDL/Lux.jl/.github/workflows/CommonCI.yml@main` call site (23
occurrences across 11 CI workflow files).

These two features have been causing spurious CI job failures on the
self-hosted SciML runners:

- **Telemetry** (`catchpoint/workflow-telemetry-action@v2`): AxiosError
  413 (payload too large) / 403 (auth) during post-job cleanup. The
  action annotates `##[error]` which marks the entire job as FAILURE
  even when all Julia tests pass.

- **Cache** (`julia-actions/cache@v3`): `could not lock config file
  /home/chrisrackauckas/.gitconfig: File exists` and
  `uploadCacheArchiveSDK: Server failed to authenticate the request`
  when concurrent jobs on the same runner contest the depot lock or
  cache upload token.

Both inputs were added to CommonCI.yml in LuxDL/Lux.jl#1695 (with
defaults `true` for backward compatibility). Setting them explicitly to
`false` here makes NonlinearSolve.jl CI resilient to the self-hosted
runner environment regardless of any future default changes upstream.

Co-Authored-By: Chris Rackauckas <accounts@chrisrackauckas.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
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.

2 participants