[ClusterLabs] PAF recover master after crash

[damiano giuliani]

Ehy,

Take in account when a master node crash, you should re-allign the old
master into the slave using pg_basebackup/pg_rewind and then rejoin the
node into the cluster as a slave. This is the only way to avoid data
corruption and be sure the new slave is correcly synchronised with the new
master.
Diskless sbd need at least 3 nodes so i expect you are using classic sbd
disk shared

On Fri, 21 Jan 2022, 17:15 Miro Igov, <miro.igov at gmail.com> wrote:

> Hello,
>
>
>
> I have a 2 node cluster with pgsqlms and sbd fencing.
>
> I tested the failover by resetting the VM running the postgresql master,
> it promoted the other node.
>
> When i started pacemaker on fenced node i get this error:
>
> Instance "pgsqld" controldata indicates a running primary instance, the
> instance has probably crashed
>
> One second later it changes to :
>
>                 Unexpected state for instance "pgsqld" (returned 9)
>
> Then node is fenced again.
>
>
>
>
>
> What is the correct way of recovering failed master?
> _______________________________________________
> Manage your subscription:
> https://lists.clusterlabs.org/mailman/listinfo/users
>
> ClusterLabs home: https://www.clusterlabs.org/
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.clusterlabs.org/pipermail/users/attachments/20220121/50e4b652/attachment.htm>