LXC Ubuntu 14.04 not starting properly with PVE 7.2 (cgroupsv2?)

[NoMed]

Hello,

I installed a new server with PVE 7.2 and migrated a backup of an LXC container from old server with PVE 6.2.

This container is still running Ubuntu 14.04, which worked perfect with PVE 6.2. Now on PVE 7.2 there is a problem with the start of the container. It is starting without error, I can enter with pct enter 100 from the host, but container has no network in this state. Checked with top, there are only 7 tasks running. If I then go manually to runlevel 3 with init 3, network and all services are starting and working well.

I found equal problem and possible solution with downgrading cgroupsv2 here:
https://forum.proxmox.com/threads/l...rly-after-upgrade-to-pve7.103598/#post-446883
https://pve.proxmox.com/pve-docs/chapter-sysadmin.html#sysboot_edit_kernel_cmdline

I did this, rebooted host and checked with cat /proc/cmdline on the host:
BOOT_IMAGE=/vmlinuz-5.15.39-1-pve root=UUID=88cad700-1471-4438-a27e-7f1072b1ecfa ro nomodeset iommu=pt systemd.unified_cgroup_hierarchy=0

Problem is, that it is still not working with this old Ubuntu, I have to enter runlevel 3 manually to get the server up properly. Same problem with fresh installation of Ubuntu 14.04 from official proxmox website.

Hoping you can help me and give some hints or possible root causes or solutions.

Thank you.

 

[Neobin]

Only an absolute wild guess, but maybe try with enabled nesting (nesting=1 under Features in the LXC-Options).

 

[Dunuin]

See the PVE 7.0 changelog:

Known Issues:

• Container:
  • cgroupv2 support by the container’s OS is needed to run in a pure cgroupv2 environment. Containers running systemd version 231 or newer support cgroupv2 [1], as do containers that do not use systemd as init system in the first place (e.g., Alpine Linux or Devuan).
  CentOS 7 and Ubuntu 16.10 are two prominent examples for Linux distributions releases, which have a systemd version that is too old to run in a cgroupv2 environment, for details and possible fixes see: https://pve.proxmox.com/pve-docs/chapter-pct.html#pct_cgroup_compat

I guess with Ubuntu 16.10 being too old to support cgoup2 then 14.04 too.

PVE 6.X got cgoup as well as cgroup2 support. PVE 7.X cgroup2 only.

 

[Neobin]

See the PVE 7.0 changelog:

I guess with Ubuntu 16.10 being too old to support cgoup2 then 14.04 too.

PVE 6.X got cgoup as well as cgroup2 support. PVE 7.X cgroup2 only.

But he already applied the "fix"/workaround: systemd.unified_cgroup_hierarchy=0. For what I understand, this should bring cgroupv1 support back?

 

[NoMed]

But he already applied the "fix"/workaround: systemd.unified_cgroup_hierarchy=0. For what I understand, this should bring cgroupv1 support back?

That's exactly my question. I did the fix to get cgroupv1 support back and checked it with cat /proc/cmdline , but it is still not working properly.

About nesting: Yes, this is importent but was still acitivated. Perhaps not correct, here is the configuration?

 

[NoMed]

Sorry, I forgot to reply: The Problem is solved, but I don't know why. It is working properly after one or some more restarts of the hostsystem. After I realized that it is working now, I repeated the restart of host and PVE several times and it is runs like clockwork.

Could be marked as solved. Sorry that I could not exactly say what fixed it.