Having some issues with lxd again with snapshots

[robe]

Secure might be down for a couple of minutes while I shuffle things arounds

[robe]

Issue is described here - ​https://discuss.linuxcontainers.org/t/mount-fails-when-starting-container-zfs-filesystem-already-mounted-lxd-3-22/7080/3

But only seems to be an issue with some containers. Short of rebooting the host suggested solution is below which didn't seem to work

for m in $(grep lxcfs /proc/mounts | grep proc | awk {'print $2'}); do umount ${m}; done

[robe]

FYI:

containers currently affected by this issue are

tracsvn, secure (I renamed this to secure2 and was able to snapshot, but can't rename it back), old-web - (renamed this to old-web2 and was able to snapshot but again can't rename cause it seems some snapshots got stock on old name), old-projects, old-webextra, old-wiki

Other containers are fine (download, grass, wordpress, nginx, dronie-server, collabora, nextcloud-ubuntu)

A reboot I think fixes this issue, but lxd 4.0.0 is coming out tomorrow, so figured I'd wait since it does a restart of lxd services anyway so might fix it too.

[robe]

okay was able to rename secure back by doing this:

lxc cp secure2 secure --instance-only

Unfortunate that leaves all the old snapshots on the secure2 I had moved from

[robe]

Did the same for old-webextra and old-web.

[robe]

okay it upgraded to lxd 4.0.0 but still had same issue.

This worked ​https://discuss.linuxcontainers.org/t/requesting-for-help-lxd-3-23-zfs-mount-fails-filesystem-already-mounted/7233

lxc exec old-projects -- shutdown -P -h now
#get list of mounts
 grep default/containers/old-projects /proc/*/mountinfo
#unmount
 nsenter -t 4273 -m umount  osgeo7/containers/old-projects
lxc start old-projects
lxc snapshot old-projects #works now