2 node HA cluster don't work

[Ilia Popov]

Hi!

I've created a 2 node cluster with PVE 4.2 and iSCSI storage.
Everything looks good exept HA functionality.
VMs, appointed to be HA, stay "queued" and do not migrate.

What's wrong with my config?

 

[dietmar]

What's wrong with my config?

2 nodes (no quorum)...

 

[Ilia Popov]

Thanks!
I've added a 3rd node and HA started to work.

Extra question: how to move HA master role?

 

[dietmar]

Extra question: how to move HA master role?

master election is fully automatic ...

 

[Ilia Popov]

During testing I've got a trouble:

Aug 30 17:42:38 node2 pve-ha-crm[1224]: successfully acquired lock 'ha_manager_lock'
Aug 30 17:42:38 node2 pve-ha-crm[1224]: ERROR: unable to open watchdog socket - no such file or folder
Aug 30 17:42:38 node2 pve-ha-crm[1224]: server received shutdown request
Aug 30 17:42:38 node2 pve-ha-crm[1224]: server stopped
Aug 30 17:42:38 node2 systemd[1]: pve-ha-crm.service: main process exited, code=exited, status=255/n/a
Aug 30 17:42:39 node2 systemd[1]: Unit pve-ha-crm.service entered failed state.

nmi_watchdog=0 is added to /etc/init.d/grub
I had Intel AMT enabled in BIOS, no changes after disabling one.

 

[dietmar]

What is the output of

# systemctl status watchdog-mux.service

 

[Ilia Popov]

What is the output of

# systemctl status watchdog-mux.service

● watchdog-mux.service - Proxmox VE watchdog multiplexer
Loaded: loaded (/lib/systemd/system/watchdog-mux.service; static)
Active: failed (Result: exit-code) since Ср 2016-08-31 09:43:12 MSK; 11min ago
Process: 6889 ExecStart=/usr/sbin/watchdog-mux (code=exited, status=1/FAILURE)
Main PID: 6889 (code=exited, status=1/FAILURE)

авг 31 09:43:12 node2 watchdog-mux[6889]: watchdog set timeout: Invalid argument
авг 31 09:43:12 node2 systemd[1]: watchdog-mux.service: main process exited, code=exited, ...LURE
авг 31 09:43:12 node2 systemd[1]: Unit watchdog-mux.service entered failed state.
Hint: Some lines were ellipsized, use -l to show in full.


also:

root@node2:~# ps -ax | grep mei
336 ? S 0:00 [irq/123-mei_me]
7644 pts/0 S+ 0:00 grep mei


Looks like mei_me need to be disabled. How to switch one off?

 

[dietmar]

This module is already blacklisted in lastest kernel from pvetest, so you can fix with:

# wget ftp://download1.proxmox.com/debian/dists/jessie/pvetest/binary-amd64/pve-kernel-4.4.16-1-pve_4.4.16-63_amd64.deb
# dpkg -i pve-kernel-4.4.16-1-pve_4.4.16-63_amd64.deb

 

[Ilia Popov]

Done. No changes.

 

[Ilia Popov]

Finally, I've reinstalled node with proxmox 4.2 iso distr. The same problem.