Action failing semi-frequently with `Connect Timeout Error` in private repo

[dgolombek]

We see about a 1% failure rate of this job with the only output being:

Error: Connect Timeout Error

Our pyproject.toml has ruff = "^0.13.2"

The full output (aside from the git checkout and cleanup):

Run astral-sh/ruff-action@57714a7c8a2e59f32539362ba31877a1957dded1
  with:
    version-file: pyproject.toml
    args: check
    src: /home/runner/_work/XXXX
    github-token: ***
  env:
    DD_ENV: testing
    NODE_VERSION: 20.9.0
    DD_APM_ENABLED: false
    RUNNER_TOOL_CACHE: /opt/hostedtoolcache
    RUNS_ON_HOME: /runs-on
    RUNS_ON_VERSION: v2.8.5
    RUNS_ON_AGENT_BINARY: /runs-on/agent
    RUNS_ON_AGENT_ARCH: arm64
    RUNS_ON_AGENT_USER: runner
    RUNS_ON_AMI_NAME: runs-on-v2.2-ubuntu24-full-arm64-20251001080408
    RUNS_ON_AWS_REGION: us-east-1
    RUNS_ON_AWS_AZ: us-east-1a
    RUNS_ON_ENV_CUSTOM_FILE: /runs-on/env.custom
    RUNS_ON_INSTANCE_ID: i-XXXXX
    RUNS_ON_INSTANCE_LAUNCHED_AT: 2025-10-06T13:54:54Z
    RUNS_ON_INSTANCE_LIFECYCLE: spot
    RUNS_ON_INSTANCE_TYPE: t4g.micro
    RUNS_ON_STACK_NAME: runs-on
    RUNS_ON_RUNNER_NAME: runs-on--i-XXXX--XXXXX
    RUNS_ON_S3_BUCKET_CACHE: runs-on-s3bucketcache-XXXX
    ACTIONS_RUNNER_HOOK_JOB_STARTED: /runs-on/pre.sh
    ACTIONS_RUNNER_HOOK_JOB_COMPLETED: /runs-on/post.sh
    ACTIONS_CACHE_URL: http://XXXX:6871/
    ACTIONS_CACHE_SERVICE_V2: true
    ACTIONS_RESULTS_URL: http://XXXX:6871/
Error: Connect Timeout Error

We're seeing a far higher failure rate due to this than to actual failures, and higher random failure rate than almost any other job. Please let me know if there's anything I can do or provide to help debug this.

[eifinger]

Does it show Download action repository 'astral-sh/ruff-action.......? Before these log lines?

[dgolombek]

Yes, that's shown in the Set up job output. On a normal run, the next output line would be Added /opt/hostedtoolcache/ruff/0.13.3/aarch64 to the path (since we're running on arm64 runners).