qemu VM does not start - kvm: -vnc 0.0.0.0:1: Failed to find an available port: Address already in use

[Pierre DEDIEU]

The concerned VM is a restore from a backup (on pve1) which was on another node (pve2) of the cluster.

It seems to be a problem on the destination node, as if I try a migration of the initial VM from pve2 to pve1, it also fails to start in the end:

Migration of the initial VM from pve2 fails, after transfering the disk, with a similar message
2021-09-06 11:05:11 starting VM 104 on remote node 'pve'
2021-09-06 11:05:16 [pve] kvm: -vnc 0.0.0.0:1: Failed to find an available port: Address already in use
2021-09-06 11:05:17 [pve] start failed: QEMU exited with code 1
2021-09-06 11:05:17 ERROR: online migrate failure - remote command failed with exit code 255
2021-09-06 11:05:17 aborting phase 2 - cleanup resources
2021-09-06 11:05:17 migrate_cancel
2021-09-06 11:05:17 scsi0: removing block-dirty-bitmap 'repl_scsi0'
2021-09-06 11:05:19 ERROR: migration finished with problems (duration 00:00:21)
TASK ERROR: migration problems

What could I check ?
(qm start 108 just returns the same message as the GUI)

 

[oguz]

hi,

2021-09-06 11:05:16 [pve] kvm: -vnc 0.0.0.0:1: Failed to find an available port: Address already in use
2021-09-06 11:05:17 [pve] start failed: QEMU exited with code 1
2021-09-06 11:05:17 ERROR: online migrate failure - remote command failed with exit code 255

here's your error, it seems you've added vnc parameter to your VM configuration, but there's another VM running VNC on the same port on the target node. if you change your VM config before migrating to use another vnc port, or change the vnc port of the VM in the target node, it should work

 

[Pierre DEDIEU]

I found a bypass

I still do not understand the reason, but I found in
https://pve.proxmox.com/wiki/VNC_Client_Access
that one could modify the VNC screen number in /etc/pve/local/qemu-server/<KVM ID>.conf

I modified
args: -vnc 0.0.0.0:1
to
args: -vnc 0.0.0.0:2

And was able to start the VM (!)

Any explaination would be appreciated

 

[oguz]

And was able to start the VM (!)

Any explaination would be appreciated

yes, as i mentioned in my previous post, when you defined the vnc arg with :1 as the port, it starts listening on port 5901.

you had another VM in the target node where it's being migrated to, which also had the same port defined for the VNC server, this caused a conflict and the newly migrated VM was thus not started.

when you changed the config to use :2, it starts the VNC server on port 5902, which fixes the conflict

please mark the thread as [SOLVED] so others know what to expect.