QA (public) » openQA Project (public) » openQA Infrastructure (public)

Did sudo su and found with df -h that / is full. Did snapper ls

   # | Type   | Pre # | Date                     | User | Used Space | Cleanup | Description           | Userdata     
-----+--------+-------+--------------------------+------+------------+---------+-----------------------+--------------
  0  | single |       |                          | root |            |         | current               |              
  1* | single |       | Thu Oct 19 19:40:57 2017 | root | 105.70 MiB |         | first root filesystem |              
184  | pre    |       | Mon May 17 12:45:23 2021 | root |  93.90 MiB | number  | zypp(zypper)          | important=yes
185  | post   |   184 | Mon May 17 12:46:46 2021 | root |   8.89 MiB | number  |                       | important=yes
208  | pre    |       | Sun Jun  6 12:45:22 2021 | root |   4.31 MiB | number  | zypp(zypper)          | important=yes
209  | post   |   208 | Sun Jun  6 12:46:14 2021 | root |  15.76 MiB | number  |                       | important=yes
216  | pre    |       | Sun Jun 13 03:00:07 2021 | root |   2.08 MiB | number  | zypp(zypper)          | important=no 
217  | post   |   216 | Sun Jun 13 03:00:25 2021 | root |   2.05 MiB | number  |                       | important=no 
218  | pre    |       | Sun Jun 20 03:00:24 2021 | root |  88.91 MiB | number  | zypp(zypper)          | important=yes

Did snapper rm 184-217 to get some space. After that did mount -o rw,remount / and triggered a zypper dup to ensure a consistent upgraded step.

I also encountered a process running wild:

 6958 jenkins   B4  20   0 3510M 1646M   640 D  845         N/A 23.9 83.2  1:27.14 /usr/bin/perl /usr/share/openqa/script/client --host https://openqa.opensuse.org jobs/

eating up all memory.

The problem about the accumulating space could be mainly due to big updates especially kernel updates where old versions are never deleted because we do not automatically reboot so we need automatic reboots triggered.

Using https://github.com/okurz/scripts/blob/master/opensuse-install-auto-update I extended /etc/cron.d/auto-update now with && needs-restarting --reboothint >/dev/null || (command -v rebootmgrctl >/dev/null && rebootmgrctl reboot ||:)' > /etc/cron.d/auto-update to trigger a reboot when updates demand it. Also did systemctl enable --now rebootmgr and triggered an explicit reboot now.

After the reboot the older kernel versions have been removed, so that only two versions (current and previous) remain. But / is still 99% used. I think it's problematic that the filesystem layout is still old and /var/lib/jenkins is not on a subvolume so it's content will be included in every snapshot.

So creating a new btrfs subvol and moving the existing content there:

systemctl stop jenkins
mount /dev/sda2 -o subvol=@ /mnt
btrfs subvol create /mnt/var/lib/jenkins
echo 'UUID=0b8a4aea-1f96-4b60-b305-57a7317d0067 /var/lib/jenkins btrfs subvol=@/var/lib/jenkins 0 0' >> /etc/fstab
mv /var/lib/jenkins/* /mnt/var/lib/jenkins/
umount /mnt

and now trying to reclaim more space, e.g. with systemctl start btrfs-balance btrfs-scrub btrfs-trim