Issue Summary

The Runner pod is composed of two containers: the Runner container and the Docker Dind container.

Sometimes, when scaling down, the runner container is correctly shut down, but the dind container remains running, preventing the pod from being terminated, and making the pod become an unhealthy state, where it can't be assigned new jobs, but still consumes host resources.

Steps to Reproduce

1. Use runners autoscaler
2. Try to scale down the runners

• As this is an intermittent issue, it's not reproducible in all scenarios.

Expected Results

When a runner is scaled down, the entire pod is terminated, and not just the runner container.

Actual Results

The dind container keeps running, preventing the pod from being terminated.

Workaround

Currently there is no known workaround for this behavior. A workaround will be added here when available