Permanent postgres writes for pleroma database
I see a strange behaviour for a few days now: It is permanent disk writes by the postgres database engine. It's not a lot and in the beginning it was only one process, but meanwhile it's all pleroma connections, I guess.
There a several databases running within this cluster, so I started to investigate: There were no writes into the syslog by pleroma and also no writes to the postgres log itself. So I increased the postgres log level. First I only saw autovacuum messages, after disabling autovacuum I finally saw messages correlating to my issue:
There are registered, degistered and stopped background worker processes permanently, see a screenshot of top and the postgres log output:
The pids in the postgres log are the pids of the pleroma postgres processes. There is also a constant load of roughly 10% of one core by beam which I believe, I never saw before.
- OS is Ubuntu 18.04
- Postgres is version 12
- Pleroma is running on commit
commit 095635453ac58b9e01a32ad226c0b61466c16da0 (HEAD -> develop, origin/develop, origin/HEAD)
Author: lain <lain@soykaf.club>
Date: Sat May 2 18:10:50 2020 +0000
Replace deprecated Roma by Fedi.
Elixir's version is:
Erlang/OTP 22 [erts-10.7.1] [source] [64-bit] [smp:2:2] [ds:2:2:10] [async-threads:1] [hipe]
Elixir 1.10.3 (compiled with Erlang/OTP 21)
Any ideas, what is going on here? Is this desired behaviour by pleroma? On should I provide more information?