howto reboot properly with hardware watchdog enabled?

[sadai]

Hi All,

I am using the latest PVE 5.4 and the builtin hardware watchdog in a Supermicro X11 board with the ipmi_watchdog kernel module. After rebooting the machine the baseboard management controller (AST2500) complains about a watchdog timeout and tries to reset the machine somewhere between the shutdown and boot process.

Shouldn't the watchdog-mux service prepare the watchdog for the reboot or do I have to tell it before? If yes,
then this is not documented anywhere.
Could it be related to a communication issue between the watchdog service through the ipmi kernel module
or is the timeout of 10s just too short? Does anybody know how this is supposed to work?

Here it is circumvented by setting a longer timeout before the actual reboot:
https://lore.kernel.org/patchwork/patch/785703/

 

[Binary Bandit]

I've never had to do anything but reboot ...works fine. Latest PVE as of this writing on 3 x Dell r510s in an HA cluster.

 

[sb-jw]

You could try to the two Services first:

systemctl stop pve-ha-lrm.service pve-ha-crm.service

I used this too on a PVE HA Cluster, and i do not have any futher Problems with the Watchdog.

 

[sadai]

You could try to the two Services first:

systemctl stop pve-ha-lrm.service pve-ha-crm.service

I used this too on a PVE HA Cluster, and i do not have any futher Problems with the Watchdog.

Thanks for this hint, I will give it a try at the next opportunity and report if it helped.

 

[sadai]

I have tried but didn't help, I will stick to turning it off this way:

systemctl stop watchdog-mux.service ; sleep 1; ipmitool mc watchdog off