how to restore from s3 when db is re-init? about helm-charts HOT 4 OPEN

There's a few components here.

If you would have only rm -rf data directory, you should be fine. The Services, Endpoints and ConfigMaps of the Deployment would still contain enough configuration to allow you to do the restore.

However, if you fully removed all the Kubernetes Objects (you did helm delete, or kubectl delete namespace), then no references of the old deployment still exist within the kubernetes context.

Currently, the behaviour is that at this point, a new initdb is run.
As you already do have a backup, pgbackrest complains about a mismatch.

We want to include the behavior of allowing to restore a previously deployed database on a pristine Kubernetes namespace here:

#108

However it still needs some work, plus it is inclear whether or not this should be default behavior or not. (I personally am leaning to make it the default).

And If I mannally call pgbackrest restore , seems I need to run command before database startup,can I use your image and change command in yaml to do that?

You can add a debug.execStartPre command that takes a very long time, that will allow you to intervene currently before anything is started:

https://github.com/timescale/timescaledb-kubernetes/blob/master/charts/timescaledb-single/values.yaml#L415-L419

from helm-charts.

Thanks for your explain,I think whether restoring from previous backup is default or not. The function of
call pgbackrest command mannually is quite important.

In some cases people may not need a full restore,but restore some of it's table or a Point-in-Time Recovery,and most command need to shutdown pg database , which is a little difficult in a k8s pod.

from helm-charts.

It would be great if you could manually restore a specific backup.

from helm-charts.

@lonelyleaf @devopsrelband There are 2 restore scenarios described here, they do not address the topic of the issue, but may be useful for you:

https://github.com/timescale/timescaledb-kubernetes/blob/master/charts/timescaledb-single/admin-guide.md#testing-restorerecovery-from-inside-the-kubernetes-cluster

from helm-charts.