Postgres major upgrade with repmgr

[IdanArad]

I have a three node cluster pg_version 10, i'm trying to upgrade it to pg_version 11. Also, that cluster is managed by repmgr 10, which should also be upgraded to version 11.

I have followed after Postgres Documentation in order to do the upgrade. for Upgrading standbys with pg_upgrade and rsync i have followed different solutions from Repmgr Documentation

I am running with ansible on the cluster and achieved an upgraded postgres cluster - the rsync and pg_upgrade are working fine. After running those upgrade commands, The slave nodes were in standby role but running as primary, in order to fix that, I ran : repmgr standby clone --replication-conf-only

The problem is that the replication slots are not active.
IMPORTANT: When pausing the daemon and restarting the postgresql services, the cluster manages to reconfigure itself - the original primary and two secondaries are active and running and the replication slots are activated as needed.
In that case, the repmgr daemon are not running.

The log from /var/log/repmgr/repmgr.log is : WARNING : standby 2 not connected to streaming replication

postgres old version: 10.7
postgres new version: 11.2
repmgr old version: 10-4.2.0
repmgr new version: 11-4.2.0
OS: rhel 19.8

Any help with how to activate the repmgr daemon?
Did I upgrade that cluster the right way?
Do you have any other suggestions how to upgrade that cluster?

Thanks!

[ibarwick]

I have a three node cluster pg_version 10, i'm trying to upgrade it to pg_version 11. Also, that cluster is managed by repmgr 10, which should also be upgraded to version 11.

I have followed after Postgres Documentation in order to do the upgrade. for Upgrading standbys with pg_upgrade and rsync i have followed different solutions from Repmgr Documentation

I am running with ansible on the cluster and achieved an upgraded postgres cluster - the rsync and pg_upgrade are working fine. After running those upgrade commands, The slave nodes were in standby role but running as primary in order to fix that, I ran : repmgr standby clone --replication-conf-only

If this means the standby nodes were actually running as primaries, repmgr standby clone --replication-conf-only won't help there. You'd need to run repmgr node rejoin to convert the nodes back to standbys, though that might not always be possible, depending on your setup.

The problem is that the replication slots are not active.

This is a problem, and would certainly be the case if your standbys were running as primaries.

IMPORTANT: When pausing the daemon and restarting the postgresql services, the cluster manages to reconfigure itself - the original primary and two secondaries are active and running and the replication slots are activated as needed.
In that case, the repmgr daemon are not running.

The log from /var/log/repmgr/repmgr.log is : WARNING : standby 2 not connected to streaming replication

postgres old version: 10.7
postgres new version: 11.2

Current PostgreSQL 11 minor release is 11.12, did you mean that?

repmgr old version: 10-4.2.0
repmgr new version: 11-4.2.0
OS: rhel 19.8

I presume you mean some RHEL 8.x version (RHEL currently goes up to 8.4)?

Any help with how to activate the repmgr daemon?

What is the output of repmgr cluster show, executed on the primary?

Did I upgrade that cluster the right way?

From the description it sounds like the standbys may have started up as primaries.

Do you have any other suggestions how to upgrade that cluster?

Upgrading PostgreSQL clusters is outside the scope of any repmgr support, so we are unable to assist further with that.

Regards

Ian Barwick

[ibarwick]

Closing as no further response received.