[pve-devel] Race with PBS Backup of GW running as VM in PVE

[Tom Weber <pve@junkyard.4t2.com>]

Hi,

with a Setup like

PBS <--->| GW VM |<--> PVE Host

where the GW VM is running on the PVE Host and connecting the PVE Host with 
the PBSever, I seem to run into a race condition when trying to 
snapshot-backup the GW VM to the PBS Server.

The snapshot appears to freeze the GW VM just the moment the PVE Host tries 
to establish the connection to the PBS:

INFO: starting new backup job: vzdump 121001254 --storage pbs-XXX --remove 0 
--node XXX-01 --mode snapshot
INFO: Starting Backup of VM 121001254 (qemu)
INFO: Backup started at 2021-07-27 13:48:12
INFO: status = running
INFO: VM Name: XXX.gw-XXX
INFO: backup mode: snapshot
INFO: ionice priority: 7
INFO: creating Proxmox Backup Server archive 'vm/121001254/2021-07-27T11:48:12Z'
ERROR: VM 121001254 qmp command 'backup' failed - backup connect failed: 
command error: error trying to connect: tcp connect error: No route to host 
(os error 113)
ERROR: Backup of VM 121001254 failed - VM 121001254 qmp command 'backup' 
failed - backup connect failed: command error: error trying to connect: tcp 
connect error: No route to host (os error 113)
INFO: Failed at 2021-07-27 13:48:47
INFO: Backup job finished with errors
TASK ERROR: job errors


on a "Fast" Setup meaning fast Hosts and LAN, it works sometimes, sometimes 
it fails.
on a "Slow" Setup meaning non powerfull Host Hardware and VPNs via LTE, this 
always fails.

This is with PVE 6 and PBS 1.1

Any ideas to work around this? Not sure if I'd consider this a bug.

Regards,
   Tom