Error after relating to S3 integrator and the Juju leader unit is not the PostgreSQL primary

[marceloneppel]

Steps to reproduce

1. Deploy 3 units of the PostgreSQL and 1 unit of the S3 integrator charms.
2. Execute curl -X POST {ip-of-primary-unit}:8008/switchover -d '{"leader": "primary-unit-name-replacing-the-slash-to-hyphen", "candidate": "sync_standby-unit-name-replacing-the-slash-to-hyphen"}'.
3. Configure the S3 integrator and relate it to PostgreSQL.

Expected behavior

No error (stanza should be initialised correctly).

Actual behavior

unit-postgresql-k8s-0: 11:52:51 ERROR unit.postgresql-k8s/0.juju-log s3-parameters:62: non-zero exit code 56 executing ['pgbackrest', '--stanza=prod-data-visualization.patroni-postgresql-k8s', 'stanza-create'], stdout='', stderr='ERROR: [056]: unable to find primary cluster - cannot proceed\n       HINT: are all available clusters in recovery?\n'

Versions

Charm revision: 177

Log output

Juju debug log:
https://pastebin.canonical.com/p/7xrhhG8n4Z/

Additional context

Matrix discussion.

[github-actions]

https://warthogs.atlassian.net/browse/DPE-3349

[marceloneppel]

Revision 190 (from the 14/stable channel) contains the fix for this issue.

[marceloneppel]

Reopened the issue due to revision 190 containing the fix for another issue, not this one.

[marceloneppel]

Revision 198 from the 14/edge channel contains the fix for this issue.

[Zvirovyi]

UPD: moved to the new issue.