Good morning,
We run postgres 9.4. Early Saturday morning we had a production postgres outage because our pg_xlog directory ran out of space. Tracing the cause points to a scheduled reboot of the 3 database servers by our IT team to install some monitoring software. Because we did a simple stop/restart of the database our replication slots did not get reset and pg_logs queued up until the out-of-space condition
Can someone please verify that the following WOULD HAVE BEEN the appropriate action:
(Or offer corrections)
1. Stop the postgres database (on both slaves)
2. Run pg_drop_replication_slot (rep1, rep2 slotnames) (on Master)
3. Stop the postgres database (on master)
4. Reboot Linux Server to bring in monitoring software
5. Start the postgres database (on master)
6. Run pg_create_replication_slot (rep1, rep2 slotnames) (On Master)
7. Start the postgres database (on both slaves)