My Promox server shutdown cleanly, without me asking

[marcosscriven]

I don't understand why my Promox server shutdown. In the logs I see the 'replication runner' over and over, until then I see things are shutting down cleanly.

Thing is, I don't understand what triggered that?

103 is a Windows VM.

I see some UPS messages, but UPS didn't lose power, and the messages seem to tell me the UPS processes are reacting to the shutdown, not asking for it?

May 27 12:58:00 pve systemd[1]: Started Proxmox VE replication runner.
May 27 12:59:00 pve systemd[1]: Starting Proxmox VE replication runner...
May 27 12:59:00 pve systemd[1]: pvesr.service: Succeeded.
May 27 12:59:00 pve systemd[1]: Started Proxmox VE replication runner.
May 27 13:00:00 pve systemd[1]: Starting Proxmox VE replication runner...
May 27 13:00:00 pve systemd[1]: pvesr.service: Succeeded.
May 27 13:00:00 pve systemd[1]: Started Proxmox VE replication runner.

---^ Get this over and over

---> Then this happens
May 27 13:00:30 pve systemd[1]: 103.scope: Succeeded.
May 27 13:00:30 pve systemd[1]: Stopped 103.scope.
May 27 13:00:30 pve systemd[1]: Stopped target Timers.
May 27 13:00:30 pve systemd[1]: systemd-tmpfiles-clean.timer: Succeeded.
May 27 13:00:30 pve systemd[1]: Stopped Daily Cleanup of Temporary Directories.
May 27 13:00:30 pve systemd[1]: lvm2-lvmpolld.socket: Succeeded.
May 27 13:00:30 pve systemd[1]: Closed LVM2 poll daemon socket.
May 27 13:00:30 pve systemd[1]: Stopping LVM event activation on device 259:3...
May 27 13:00:30 pve systemd[1]: Unmounting RPC Pipe File System...
May 27 13:00:30 pve systemd[1]: 102.scope: Succeeded.
May 27 13:00:30 pve systemd[1]: Stopped 102.scope.
May 27 13:00:30 pve systemd[1]: Removed slice qemu.slice.
May 27 13:00:30 pve systemd[1]: logrotate.timer: Succeeded.
May 27 13:00:30 pve systemd[1]: Stopped Daily rotation of log files.
May 27 13:00:30 pve systemd[1]: man-db.timer: Succeeded.
May 27 13:00:30 pve systemd[1]: Stopped Daily man-db regeneration.
May 27 13:00:30 pve systemd[1]: session-9.scope: Killing process 27163 (upsd) with signal SIGTERM.
May 27 13:00:30 pve systemd[1]: session-9.scope: Killing process 18891 (usbhid-ups) with signal SIGTERM.
May 27 13:00:30 pve usbhid-ups[18891]: Signal 15: exiting
May 27 13:00:30 pve upsd[27163]: mainloop: Interrupted system call
May 27 13:00:30 pve systemd[1]: Stopping Session 9 of user root.
May 27 13:00:30 pve systemd[1]: pvesr.timer: Succeeded.
May 27 13:00:30 pve systemd[1]: Stopped Proxmox VE replication runner.
May 27 13:00:30 pve systemd[1]: pve-daily-update.timer: Succeeded.
May 27 13:00:30 pve systemd[1]: Stopped Daily PVE download activities.
May 27 13:00:30 pve upsd[27163]: Signal 15: exiting
May 27 13:00:30 pve systemd[1]: Stopped target Bluetooth.
May 27 13:00:30 pve systemd[1]: Stopped target RPC Port Mapper.
May 27 13:00:30 pve systemd[1]: Stopped target Sound Card.
May 27 13:00:30 pve systemd[1]: systemd-rfkill.socket: Succeeded.

 

[oguz]

hi,

* is it a cluster or standalone machine? if HA is configured that can cause reboot because of the watchdog

I see some UPS messages, but UPS didn't lose power, and the messages seem to tell me the UPS processes are reacting to the shutdown, not asking for it?

yes it looks like the process is getting terminated because of the shutdown.

are you the sole owner of the server? maybe someone just ran shutdown command or pressed the power button. does this happen regularly?

 

[marcosscriven]

hi,

* is it a cluster or standalone machine? if HA is configured that can cause reboot because of the watchdog

Thanks for response.

It's only standalone. If it were part of a cluster though, would there be some indication in the logs that was the cause?

are you the sole owner of the server? maybe someone just ran shutdown command or pressed the power button. does this happen regularly?

Yes - I'm the sole owner, and it's a machine on my desktop. The power button is located such that's it's not easy to accidentally press.

Forunately, this has only happened once - but was obviously curious why.

 

[oguz]

If it were part of a cluster though, would there be some indication in the logs that was the cause?

yes if HA watchdog kills the server you would see watchdog entries in the log

Yes - I'm the sole owner, and it's a machine on my desktop. The power button is located such that's it's not easy to accidentally press.

okay just making sure

Forunately, this has only happened once - but was obviously curious why.

if you post the log here we can take a detailed look, but it's hard to tell anything without knowing more about the setup