Can't launch a new container because /run/docker runs out of space

[markusjevringsesame]

Hi. In one of our repos we've recently encountered an interesting issue. It's a Java project that uses testcontainers. We'll run through our tests, which launches the same image a couple of times. At some point, we get this error: "no space left on device". When checking (watch df -h) on the runner, the only mount that's actually filling up is /run/docker. As soon as that hits 100%, the build fails.
It only happens on a single one of our builds. It started happening since we upgraded to 0.27, at which point we also switched to using helm instead of manually applying yaml files in the cluster.

Any idea what we can do to fix this? It works fine on all our other repos with very similar behavior, but it fails on this one.

An example of the error is:

{"message":"failed to create task for container: failed to start shim: start failed: write /var/run/docker/containerd/daemon/io.containerd.runtime.v2.task/moby/eeb192b887d9588ec2801846de13ced8787ccc9da0c2e67721cb8df356d28888/shim-binary-path: no space left on device: unknown"}

[markusjevringsesame]

It seems like this 4-day-old fix might work for me, but then I guess I have to wait for a release, because it's not yet supported
#2624
#2621