Mikhail f. Shiryaev

It's launched as `run.sh` in the cloud-init, not as a service. I wonder, how would it affect the result? According to the service template, it sends SIGTERM to the process....

``` /home/ubuntu/actions-runner/bin/Runner.Listener run √ Connected to GitHub Current runner version: '2.304.0' 2023-05-05 13:11:14Z: Listening for Jobs 2023-05-05 13:11:34Z: Running job: CheckLabels (1, c) ^CExiting... 2023-05-05 13:11:41Z: Job CheckLabels (1, c)...

@mochja thank you for the hint, but `ephemeral` does not work for us at all If we are talking about a fresh instance per runner, then we obviously lose a...

Once again, there's no guaranteed way to avoid racing there in case of restarting an `ephemeral` runner. - runner restarts, it's not instant. At the moment, it deletes itself from...

However, thank you for your attempt to help. Let's don't spam the ticket regarding the signal processing bug with a conversation about a theoretically possible workaround.

I'll repeat here the thing I've posted to the ticket with GH support. I've done some tests to determine if the ephemeral runners would work for us. The outcome is...

The situation with ephemeral runners is a bit better than with normal. See the discussion in https://github.com/ClickHouse/ClickHouse/pull/49283. It's not necessary to shut down the host, the process could restart. It...

> Would `config.sh remove` be a workaround? At least chatGPT says it would deregister the runner, while continuing with any running jobs. We use `config.sh remove` as the signal for...

Does it wait for the process to finish the current job, or just abort whatever it makes, and then mark the workflow job/run as failed/canceled/whatever else?

https://github.com/actions/runner/issues/2582#issuecomment-2882917181 That's how we solve the issue. If we find the runner on EC2 instance is chilling around, we check if any jog is assigned to it, but hasn't yet...