Running Drone on k8s (starter enterprise version).
Was going to raise an issue on github but raising here as requested.
When a k8s job is started to build a pipeline, it is being allocated to a node that has a disk pressure status of True, i.e. running out of disk space.
The job starts, you can see the pipeline progress in the drone UI.
After a minute or so the Job is Evicted by k8s, and restarted on another node (as expected)
However the drone ui does not seem to detect this, stops updating the pipeline status and the pipeline never finishes.
Had the same issue on 1.0.1
K8s v 1.12.5