Use this issue to track bug notes (if it is a bug):
Problem:
Postgres pod will sometimes not start back up after system shutdown, entire cluster has to be deleted and rebuilt, losing data.
Possible cause: cannot connect to persistent volume.
Possible root cause: any changes to k8s/local/secrets.yml, causes the pv name to change [redacted secret secrets-config:POSTGRES_HOST]-0.
Possible solutions:
Instead of persistent volume, use a mounted volume
Document the behaviour: Any changes to the secrets.yml file will result in loss of pv
Write a script or document how to change the claim of a new pvc (which gets created when secrets.yml is changed) so that the orphaned pv can be attached to a new pvc
I think this is really just a documentation need: if k8s/local/secrets.yml changes it's possible the postgres container won't be able to reconnect to the pvc/pv
Use this issue to track bug notes (if it is a bug):
Problem:
Postgres pod will sometimes not start back up after system shutdown, entire cluster has to be deleted and rebuilt, losing data.
Possible cause: cannot connect to persistent volume. Possible root cause: any changes to
k8s/local/secrets.yml
, causes the pv name to change[redacted secret secrets-config:POSTGRES_HOST]-0
.Possible solutions:
secrets.yml
file will result in loss of pv