What would you like to be added:
Now and then we hit occurrences of https://github.com/distribution/distribution/issues/4478.
We could actually consider adding an init container that mitigates the issue. In the init container, when the issue occurs (the scheduler-state.json file exists but it is empty and the /var/lib/registry/docker dir is not empty), then we could delete the scheduler-state.json file and the /var/lib/registry/docker dir to mitigate the issue.
Why is this needed:
Reduce repetitive and recurring ops effort until the upstream issue is fixed.
How to categorize this issue?
/area ops-productivity /kind enhancement
What would you like to be added: Now and then we hit occurrences of https://github.com/distribution/distribution/issues/4478. We could actually consider adding an init container that mitigates the issue. In the init container, when the issue occurs (the
scheduler-state.json
file exists but it is empty and the/var/lib/registry/docker
dir is not empty), then we could delete thescheduler-state.json
file and the/var/lib/registry/docker
dir to mitigate the issue.Why is this needed: Reduce repetitive and recurring ops effort until the upstream issue is fixed.