My tables are partitioned daily and I still have had issues. The problem is they all still need vacuumed but at least the individual table vacuums are shorter and it makes it easier to drop older data.
Essentially a large volume of sensor data. Tons of inserts. Some updates in other tables.
Reducing the autovacuum_vacuum_cost_delay seems to have helped in my case. The autovacuums just run too slow.
I hit this issue all the time and it is never easy or fast to fix once postgres stops accepting writes. If you have a write heavy postgres database, the autovacuum settings are not sufficient. The entire database needs…
My tables are partitioned daily and I still have had issues. The problem is they all still need vacuumed but at least the individual table vacuums are shorter and it makes it easier to drop older data.
Essentially a large volume of sensor data. Tons of inserts. Some updates in other tables.
Reducing the autovacuum_vacuum_cost_delay seems to have helped in my case. The autovacuums just run too slow.
I hit this issue all the time and it is never easy or fast to fix once postgres stops accepting writes. If you have a write heavy postgres database, the autovacuum settings are not sufficient. The entire database needs…