HA Proxmox 5 3node + 1 NFS server not work

[borjamf]

1. Hello
   
   We're planning to migrate part of our production environment to Proxmox if it posible. I have set up a testing env this week with 3 Proxmox nodes: 2 physical and 1 virtual and 1 basic NFS Debian Server, but I could not make HA work in any way.
   
   When I plug off the network cable or power some node off just show the node unreachable, but the VM allocated inside does not automove to another node avalaible
   
   . I have encountered some weird behaviour with HA-manager:
   
   Current conf:
   
   --- Cluster status
   
   
   
   -- HA-Manager weird issue (old timestamp - dead?)
   
   
   
   I set up watchdog module file etc/modprobe.d/ipmi_watchdog.conf with
   options ipmi_watchdog action=power_cycle panic_wdt_timeout=10
   
   After I disconnect the network cable, syslog shows this.
   
   Dec 14 16:17:38 pmtest1 pve-ha-crm[1073]: successfully acquired lock 'ha_manager_lock'
   Dec 14 16:17:38 pmtest1 pve-ha-crm[1073]: ERROR: unable to open watchdog socket - No such file or directory
   Dec 14 16:17:38 pmtest1 pve-ha-crm[1073]: server received shutdown request
   Dec 14 16:17:38 pmtest1 pve-ha-crm[1073]: server stopped
   Dec 14 16:17:38 pmtest1 systemd[1]: pve-ha-crm.service: Main process exited, code=exited, status=255/n/a
   Dec 14 16:17:39 pmtest1 pve-ha-lrm[1083]: successfully acquired lock 'ha_agent_pmtest1_lock'
   Dec 14 16:17:39 pmtest1 pve-ha-lrm[1083]: ERROR: unable to open watchdog socket - No such file or directory
   Dec 14 16:17:39 pmtest1 pve-ha-lrm[1083]: restart LRM, freeze all services
   Dec 14 16:17:39 pmtest1 pve-ha-lrm[1083]: server stopped
   Dec 14 16:17:39 pmtest1 systemd[1]: pve-ha-lrm.service: Main process exited, code=exited, status=255/n/a
   Dec 14 16:17:39 pmtest1 systemd[1]: pve-ha-crm.service: Unit entered failed state.
   Dec 14 16:17:39 pmtest1 systemd[1]: pve-ha-crm.service: Failed with result 'exit-code'.
   Dec 14 16:17:39 pmtest1 systemd[1]: pve-ha-lrm.service: Unit entered failed state.
   Dec 14 16:17:39 pmtest1 systemd[1]: pve-ha-lrm.service: Failed with result 'exit-code'

Could anybody help me? I dont know what else to do.
Thanks.

 

[wolfgang]

Hi,

As your logs tell you

Dec 14 16:17:38 pmtest1 pve-ha-crm[1073]: ERROR: unable to open watchdog socket - No such file or directory

I think your watchdog is not proper setup.
Send the output of this two commnads.

cat /etc/default/pve-ha-manager
journalctl -u watchdog-mux.service

 

[dietmar]

And where does that virtual node run?

 

[borjamf]

Thanks for the reply:

the virtual node is running on Oracle Virtual Box, but it does not contain anything. I have configured it with Proxmox just for test HA with 3 nodes min.

output of /etc/default/pve-ha-manager:
# select watchdog module (default is softdog)
WATCHDOG_MODULE=ipmi_watchdog



So there is something wrong with the watchdog

 

[borjamf]

Hello,
looks like watchdog software aka softdog runs ok. The problem was that I did not have any kind of hardware watchdog, so I'll update when I test it in another machine.