Postgres docker github12/28/2023 ![]() Acquiring the VIP in the context of a swarm is a bit strange: the pgpool instance will connect to the host via ssh. When the system starts, the 3 pgpool will elect a leader and this one will acquire the VIP. The same for each pgpool service (pgpool01, pgpool02, pgpool03). In this case each postgres service (pg01, pg02, pg03) is sticky to a swarm node. In docker swarm together with pgpool watchdog. There are two different ways to use those docker images: There is a graphical monitoring/operational interface available (written in nodejs / react). There were simply too much edge cases otherwise. ![]() After much experimentation I have settled for using pgpool in watchdog mode (VIP), even when using docker swarm. ![]() The recommandation for now is to use pgpool for automatic failover. Both options seem to have some pros and cons. Automatic failover of postgres can be done either by repmgr (repmgrd) or by pgpool. Postgres is replicated with streaming replication, repmgr is used because it brings well documented and tested scripts and it adds some metadata about the cluster that ease monitoring.
0 Comments
Leave a Reply.AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |