UNSOLVED Nextcloud take very long time to respond
Suddenly nextcloud takes a very long time to open the home page and clients are timing out. From what I observed in the services section, PostgreSQL is almost running with 100% usage. Also, logs show the error message. Find below screenshot for reference. I do appreciate any recommendation to solve this issue
@chaitanya since you also contacted support, I was able to look at the server directly and so far I am not sure what that postgres error about disk space is, the server has enough tmpfs or actual disk space available. I am still investigating what the root case might be.
To update here, restarting the Postgres service made that space issue go away for the memory segment. It is still unclear what this was.
There seem to be two issues here:
- One some db inconsistency which is causing duplicate errors. This has the same errors as https://forum.cloudron.io/topic/4212/how-to-fix-thes-errors-in-postgresql-logs-related-to-oc_filecache_extended
- Another is "ERROR: could not resize shared memory segment "/PostgreSQL.1655854505" to 4194304 bytes: No space left on device".
Investigating the latter. There is a
shared_bufferssetting in postgresql config which is currently at 64MB. Then docker has a default --shm-size of 64MB as well (https://github.com/moby/moby/issues/29492). The default is in fact 128MB per https://www.postgresql.org/docs/12/runtime-config-resource.html these days. I will fix our container and docker configuration.
There's also some cryptic math here - https://www.oreilly.com/library/view/postgresql-10-high/9781788474481/cfcdb792-38e6-42c9-b60b-3c0a86007aa4.xhtml
I have made a fix here - https://git.cloudron.io/cloudron/box/-/commit/949b2e2530283e1690656c2dd417e00a78b3251d . I am still monitoring @chaitanya's server to see the impact the change has.