I’ve been running family-size NextCloud for many years, but the last 2 or 3 updates the update process has been bumpy. Last one is from 29.0.1 to 29.0.2. Each time, I’ve copied back the backup it made to keep things running.
The NC area is much larger than I remember seeing before, 852 MB in /var/www/nextcloud/, with 86 MB in apps/calendar/ for example, of which 57 MB are in js/*.map. Is this normal? I’m running the NC “stable” branch, are these debug files needed?
The error comes after the delete taking a very long time, much longer than I am used to. There are no disk errors, this is running off a 1 TB disk (magnetic, not SSD, which I still need to replace).
Could it be that the HD is simply too slow? Is there a way to extend the nginx timeouts so it completes without errors, even if this takes a bit longer?
On DietPi we can control the base OS more or less but our influence on individual app is limited. Something like this might need to be checked with application developer if and how such thinks could be configured differently. Maybe @MichaIng knows if there is a way to control these feels.
Regarding the timeout. Did you checked Nginx system logs to get more information what is timing out exactly?
Indeed, about half of the size of the Nextcloud directory consists of map files, which help debugging and reviewing JavaScript files with e.g. browser developer tools. I would consider to just delete those for our installs, since it is not really required for end users, but Nextcloud has internal integrity checks, which would print errors when files are missing.
Thanks - didn’t realise that dropping files will trigger a self-check in NC. Ok, let’s not mess with that, I agree.
I have finally been able to trim down total disk space use, and can now fit my whole HD setup back on a snappy 500 GB SSD. I do need to repartition from scratch, but failed to get it right (my setup is sdb1=vfat:boot:128M, sdb2=ext4:root:10G sdb3=btrfs:dietpi_userdata:≈490G). I know that the UUIDs must match, and adjusted root= in boot/cmdline.txt, and the btrfs mount point in /etc/fstab). I got to a booting system, but login gave me a broken/incomplete shell).
I’ve gone back to the HD-based setup for now.
This is getting a bit off topic for this thread, but is there a description somewhere on how to do migrate DietPi to a smaller drive, using the command line?
My family server is up again, much snappier and the Nextcloud updates (started from scratch) went through without a hitch this time \o/
Summary: my Nextcloud updates failed due to a (rust-based) hard drive taking too long and nginx timing out as a result. Once moved to a faster SSD, all was well again. Thanks for all your tips and help in getting this resolved.
