🏷️ Discussion Type

Question

💬 Feature/Topic Area

Actions Runner

Discussion Details

Hello GitHub Community,

I'm investigating an unexpected shutdown of a GitHub-hosted Ubuntu runner during an active GitHub Actions job.

Environment

• Runner: GitHub-hosted ubuntu-24.04
• Runner version: 2.338.0
• Runner image version: 20261012.334.1
• Workflow trigger: workflow_dispatch
• Job timeout: 25 minutes
• Step timeout: 23 minutes
• Concurrency: cancel-in-progress: false

What happened

The workflow successfully reached a Docker container execution stage and started npm ci --offline.

Approximately 10 seconds after npm started, the job reported:

The runner has received a shutdown signal.
This can happen when the runner service is stopped,
or a manually started runner is canceled.

Process completed with exit code 143.

The exit code 143 belongs to the host workflow step. No npm exit code was captured.

The last available container heartbeat showed no OOM kill, but the final container state could not be verified.

The job concluded as failure, not cancelled.

A subsequent artifact-upload step configured with if: always() was skipped.

What I investigated

I reviewed GitHub Actions Runner v2.338.0 source code and found that runner shutdown can bypass step-condition evaluation, explaining why an always() step may be skipped.

However, this does not identify what initiated the shutdown.

The job had not reached its configured timeouts, and I found no evidence of a competing workflow cancellation.

Questions

1. What conditions can cause a GitHub-hosted runner to receive a shutdown signal while a job is actively running?
2. Is there a way to distinguish user cancellation, control-plane termination, runner service shutdown, and infrastructure interruption?
3. Can billing enforcement or ephemeral-runner lifecycle events produce this behavior?
4. What diagnostic information is available to users when the runner shuts down before artifact upload?
5. Are there recommended practices for preserving minimal diagnostic evidence during an unexpected runner shutdown?

I have not retried the workflow because I would first like to understand the possible causes and the appropriate diagnostic approach.

Any guidance from the community would be appreciated. Thank you!