Issues with HA on Intel Modular Server

davidekema

New Member
Oct 2, 2012
1
0
1
Hi all

we are testing the proxmox HA confiuration on a dual server IMS configuration.

Everything seem configured correctly cluster is created and fencing domain created.
But if we try to power off one the first node that has 1 VM protected by HA the second other one hangs and does not start the VM.

The strange thing are that when the other first node re-cames up the machine are moved to the second node.
We have also seen that doing a "fence_node intel1" we get a success but the node is rebooted and not keeped off by the chassis.

Here is the output of fence_tool ls:

root@intel1:~# fence_tool ls
fence domain
member count 2
victim count 0
victim now 0
master nodeid 1
wait state none
members 1 2

and here is the log of the rgmanager.log on the node that remains active, the second one:

Oct 02 11:02:55 rgmanager State change: intel1 UP
Oct 02 11:02:56 rgmanager Migration: pvevm:100 is running on 1
Oct 02 11:07:45 rgmanager #1: Quorum Dissolved
Oct 02 11:07:45 rgmanager [pvevm] VM 100 is already stopped

and here hangs.

Any help is kindly appreciated.

Thanks

Davide
 
Hi Davide, I was curious if I can get assistance from you on how you setup the snmp fencing agent with the modular server ? I seem to have a snag - following what appears to be a pretty standard procedure, and I keep getting errors like this when trying to test:

Code:
root@prox1:~# /usr/sbin/fence_intelmodular --action=status --ip=192.168.10.10 --username=snmpv3user --password=SuperSecretPass --snmp-version=3 --snmp-auth-prot=SHA --verbose --plug=1

/usr/bin/snmpget -m '' -Oeqn  -v '3' -a 'SHA' -A 'SuperSecretPass' -u 'snmpv3user' '192.168.10.10:161' '.1.3.6.1.4.1.343.2.19.1.2.10.202.1.1.6.1'
Error in packet
Reason: authorizationError (access denied to that object)

Clearly I've configured the modular server for an snmpv3user with SuperSEcretPass and SHA based auth.

Any pointers are greatly appreciated!

Many thanks,

Tim Chipman