Sorry for the copy paste from stackoverflow, but this is probably a better place to post this.
I’ve seen some chat online about this from a while back, but I haven’t found a solve.
I have 2 pgpool servers running streaming replication and load balancing. They connect to 3 postgres servers. 2 are virtual machines and 1 Physical machine.
I’d prefer the physical machine to be the master server. I have done test using the pcp_promote_node command, and I see that the show pool_nodes; shows the node I select as primary, however on postgres side the command
SELECT pg_is_in_recovery(); show that server still in recovery.
pcp_promote_node command does, is detach master node. Which then prevents insert into the DB.
Secondly If the master fails, another server does become master, however the other slave is clueless of this and does not replicate any new data until I recover it from the new master.
Is there a work around for the above 2 problems. Also please let me know if u need more info, such as logs etc etc.