Proxmox

Useful commands and tasks

  • Check storage status: pvesm status
  • Check quorum in cluster: pvecm status
  • Test performance: pveperf
  • Restart services: systemctl restart corosync.service && systemctl restart pve-cluster.service && systemctl restart pvedaemon.service && systemctl restart pvestatd.service && systemctl restart pveproxy.service (1, 2, 3)
  • Forcefully kill LXC container: First try lxc-stop [id] -k. If that doesn’t work, find the PID using ps aux | grep lxc and then kill -9 [pid]
  • Start LXC container in foreground (useful for debugging start problems): lxc-start [id] -F
  • Stop an unresponsive backup: vzdump -stop (1) or ps aux | grep vzdump, kill -9 [pid] and then pct unlock [vmid] (2)

Adding a new disk

In the interface, go to the node and the Disks tab. Select the disk and click Initialize disk with GPT. Next, go to the LVM-Thin tab and click Create: Thinpool. Select the disk, give it a name, make sure Add Storage is checked and click Create.

NFS mounts

See reference/linux for how to deal with unresponsive NFS mounts.

To set the soft mount option in Proxmox, edit /etc/pve/storage.cfg and set options vers=3,soft for the applicable mounts. (1)

LXC container not starting

Failed to create veth pair

Error message when starting in foreground:

lxc-start: 1001015: network.c: instantiate_veth: 106 Operation not permitted - Failed to create veth pair "veth1001015i0" and "veth6AIO43"
lxc-start: 1001015: network.c: lxc_create_network_priv: 2462 Failed to create network device
lxc-start: 1001015: start.c: lxc_spawn: 1646 Failed to create the network
lxc-start: 1001015: start.c: __lxc_start: 1989 Failed to spawn container "1001015"
lxc-start: 1001015: tools/lxc_start.c: main: 330 The container failed to start
lxc-start: 1001015: tools/lxc_start.c: main: 336 Additional information can be obtained by setting the --logfile and --logpriority options

Try (1):

# Lookup the name of the veth associated with the affected VM.
cat /var/lib/lxc/[id]/config
# There will be a line like this: `lxc.net.0.veth.pair = veth1001015i0`

# Delete that veth.
ip link delete [veth1001015i0]

If that doesn’t work, restart the node.

Failed to create cgroup

Error message when starting in foreground:

lxc-start: 1001015: cgroups/cgfsng.c: mkdir_eexist_on_last: 1301 File exists - Failed to create directory "/sys/fs/cgroup/systemd//lxc/1001015"
lxc-start: 1001015: cgroups/cgfsng.c: container_create_path_for_hierarchy: 1353 Failed to create cgroup "/sys/fs/cgroup/systemd//lxc/1001015"
lxc-start: 1001015: cgroups/cgfsng.c: cgfsng_payload_create: 1526 Failed to create cgroup "/sys/fs/cgroup/systemd//lxc/1001015"

Try: rm -rf /sys/fs/cgroup/systemd/lxc/[id]
Note that there may also be additional directories ([id]-1 etc.)

If that doesn’t work, restart the node.

Removing a node from the cluster

This will also work for removing the second node from a two-node cluster and thus effectively reverting to an un-clustered setup (1).

  1. List the nodes using pvecm nodes to find out the name of the node to remove.
  2. Shutdown the node to remove. Make sure it never gets reconnected.
  3. On the remaining node: Regain quorum via pvecm expected 1.
  4. Remove the node from the cluster using pvecm delnode [nodename].
  5. If the node still shows up in the UI, run rm -rf /etc/pve/nodes/[nodename] (2), then reload the webpage.