We have a HA VRRP pair of VyOS systems using conntrack but after upgrading 1.2.2 → 1.2.4 conntrack seems broken. While running 1.2.2 we’ve had no conntrack issues.
Steps taken:
Added system image 1.2.4 and rebooted the VyOS
After the first boot everything was good:
Logging visible of all conntrack (-sync) events
Systemd shows running conntrackd service
After rebooting the VyOS (I wanted to add RAM so I shut it down again):
No logging regarding conntrack
Systemd shows failed / stopped conntrackd service
show log conntrack shows conntrackd.service entered failed state
Again reboot the situation stays the same (failed).
Performed the upgrade on the second unit and it shows the exact same behaviour: First boot on 1.2.4 all ok and after another reboot conntrackd failed.
Looking into the full boot log and journalctl is shows during boot and initial startup of the service:
conntrackd[15557]: lockfile `/var/lock/conntrack.lock’ exists, perhaps conntrackd already running?
I tried manually disabling the conntrackd.service and deleting the lock file before reboot. After the reboot this shows the same failed state and log message about the lock file.
Looks like conntrack-sync does sync the connections and works despite the errors.
This might be issue during boot between conntrack-tools and conntrackd.
I have errors:
vyos@kvmvyosf02:~$ show log conntrack-sync
Mar 26 10:34:51 kvmvyosf02 systemd[1]: conntrackd.service: main process exited, code=exited, status=1/FAILURE
Mar 26 10:34:51 kvmvyosf02 systemd[1]: Unit conntrackd.service entered failed state.
Mar 26 10:36:10 kvmvyosf02 systemd[1]: conntrackd.service: main process exited, code=exited, status=1/FAILURE
Mar 26 10:36:10 kvmvyosf02 systemd[1]: Unit conntrackd.service entered failed state.
But the standby unit does seem to be in sync with the active:
vyos@vyos02:~$ show conntrack-sync external-cache
Source Destination Protocol