After upgrading incus to 7, can’t apply any Resource limits, CPU, Mem …
Containers with Resource limits in profile or config won’t start, creating new containers fail as well:
Created an account to +1 this. I just did an upgrade last night to Incus 7.0 and encountered the same Cgroup controller is missing error affecting all my containers.
Results of systemctl status incus:
level=warning msg="Failed auto start instance attempt" attempt=1 err="Load go-lxc struct: Cgroup controller is missing" instance=<INSTANCE> maxAttempts=3 project=default
Partial output of incus info –show-log <instance>:
Previously also had an error message mentioning incus conflict regards cgroup hybrid cgroup1/cgroup2.
Disabled cgroup1 in grub, yet it hasn’t resolved the issue.
Can you show the full container config (incus config show --expanded NAME) and also show the exact version of the packages being used? (dpkg -l | grep incus)
config:
boot.autostart: "true"
boot.autostart.delay: "5"
environment.TZ: America/Phoenix
limits.cpu: "2"
limits.memory: 4GB
limits.memory.swap: "false"
description: Arch linux small LXD profile
devices:
aurbuilds:
path: /opt/aurbuilds/
source: /opt/aurbuilds/
type: disk
eth0:
name: eth0
nictype: bridged
parent: br0
type: nic
root:
path: /
pool: zfsstorage
type: disk
name: archsmall
used_by:
- <OTHER_INSTANCES>
project: default
Output of pacman -Q | grep incus (because in my case I’m on Arch Linux):
incus 7.0.0-1
incus-tools 7.0.0-1
Also the output of pacman -Q | grep lxc (because I use LXCFS):
lxc 1:7.0.0-1
lxcfs 7.0.0-1
In my case I never had any cgroups-related kernel parameters, but my system did migrate years ago from LXD to Incus in case there’s anything legacy breaking because of that.
Confirmed, that works! (@vice I didn’t mean to hijack your thread, please do add on to this if your result is different.)
For posterity, I think what I encountered was Incus issue #3300. I swear I searched the Github issues, but somehow didn’t find that one until you added more detail. Thank you!