Can some of the files in /var/lib/pgsql/data/, like postgresql.conf and pg_hba.conf be made persistent?

[ppanon2022]

In particular I think I need to increase the shared_buffers = 1GB value. While our index hit ratio is pretty decent (99.96%), the cache hit ratio when doing LCM version promotions is not great.

uyuni=# SELECT
uyuni-#   sum(heap_blks_read) AS heap_read,
uyuni-#   sum(heap_blks_hit)  AS heap_hit,
uyuni-#   100 * sum(heap_blks_hit) / (sum(heap_blks_hit) + sum(heap_blks_read)) AS cache_hit_rate
uyuni-# FROM
uyuni-#   pg_statio_user_tables;
 heap_read  |  heap_hit  |   cache_hit_rate
------------+------------+---------------------
 1527727040 | 9154685807 | 85.6986706853495193
(1 row)

How would we go about tuning parameters like that (and the memory allocated to the container, if it's currently fixed)? This used to be possible with the non-container versions.

[mcalmer]

The whole /var/lib/pgsql/ directory is persistant

[ppanon2022]

Doh! Of course it is. Sorry, I should have thought of that. What about the container memory usage? Does that dynamically grow if I change the shared_buffers value to 2GB or 3GB, or would that reduce container memory available to the web/application services?

Looking at /etc/systemd/system/uyuni-server.service on the host, I only see --shm-size=0 and --shm-size-systemd=0 as memory-related podman options, so I'm not sure if that means there's a default memory size in the container metadata that I need to override with -m or if it just grows to whatever is available on the host. Only ~1/3 of the memory on the container host is currently in use so there's memory available if the container can use it.

[rjmateus]

You can change the postgresql memory and re-start the service. It should grow as long as the machine has memory. If not then it should a bug in our configuration.