0

I have two postgres nodes with a shared storage (only for data, not for config), works simply by stopping postgres on one side and starting it on the other. Pacemaker (PAF) handles stop/start between the nodes.

Questions:

What type of PostgreSQL configuration is recommended in this scenario?

Should I setup WAL archiving for some PITR and snapshots of the SAN volume, WAL archiving on both nodes etc?

I am just after some tips and guidelines here as it seems it's not a very common use case.

Thanks!

Andy Thompson
  • 284
  • 1
  • 3
  • 21

1 Answers1

-1

first Postgres Pro cannot be more reliable than the file system it's using. But if you want use SAN you can try DRBD. If you have a licanse on disk array massive you can create san replica or metrocluster. About snapshots. I think you need VEEAM.

noute
  • 114
  • 3
  • 13
  • The question is not about snapshots and whatnot, it's purely about how to best configure Postgres to use a SAN (shared storage) between two nodes. Should I still use WAL archiving or other Postgres-things? – Andy Thompson Dec 13 '19 at 14:16
  • Question is not correct (need know which storage you use). But I understand what you want. You have a two ways. First - where need to configure storage I mean block replica. In HPe 3par it is peer persistence, in Huawei OceanStor it is Hyper Metro. Second it is Active-Passive you can configure and use HP ServiceGuard or pacemaker + corosync. – noute Dec 18 '19 at 05:51
  • Im not looking for SAN-specific information. I have a single volume shared between two nodes and I am simply asking what the recommended way of dealing with e.g. WAL/PITR for such a scenario is. I use pacemaker/corosync to stop/start Postgres on the two nodes. – Andy Thompson Dec 18 '19 at 06:49