For sharded deployments with TLS cluster sometimes goes into "maintenance" mode with "enabling TLS" message

[dmitry-ratushnyy]

Steps to reproduce

All steps to reproduce are not clear, however these steps were performed when issue was discovered

1. Deploy sharded cluster on LXD + Juju 3.1.7
2. Deploy self-signed-certificates operator
3. Integrate all shards and config server with self-signed-certificates operator
4. Remove relation between self-signed-certificates and all shards and config server
5. Integrate all shards and config server with self-signed-certificates operator for the second time.

Note:
You can adjust juju update-status hook time to speedup reproducibility with
juju model-config update-status-hook-interval=10s

Expected behavior

Once TLS is enabled, if the certificate did not expired cluster should not re-enable TLS once it is enabled

Actual behavior

Versions

Operating system:

Juju CLI:

Juju agent: 3.1.7

Charm revision:

LXD:

Log output

Juju debug log:

Additional context

juju.log

[github-actions]

https://warthogs.atlassian.net/browse/DPE-3831

[dmitry-ratushnyy]

Upd: same happens on AWS deployment

[MiaAltieri]

@dmitry-ratushnyy can you send the exact commands you sent to achieve this?

[dmitry-ratushnyy]

Upd: This also might be an issue with "self-signed-certificates" charm and is require additional investigation