[ClusterLabs] RGManager: no migration on kernel panic

[Digimer]

"Works" and "safe" are not the same thing. You can drive a car without a
seatbelt, too, doesn't mean it won't end badly if something goes wrong.

digimer

On 01/09/15 12:07 PM, Stanislav Kopp wrote:
> Hi Digimer,
> 
> I may be wrong but if set "stonith-enabled=false" and
> "no-quorum-policy=ignore " in pacemaker, resource migration works fine
> with kernel panic.
> 
> Best,
> Stan
> 
> 2015-09-01 18:03 GMT+02:00 Digimer <lists at alteeve.ca>:
>> You must have fencing configured. DLM blocks until fenced confirms
>> successful fence and rgmanager uses DLM locks. Even in pacemaker,
>> fencing is required to prevent split-brains.
>>
>> On 01/09/15 11:55 AM, Stanislav Kopp wrote:
>>> Hi all,
>>>
>>> I'm testing now "corosync+cman+rgmanager" on Debian Jessie (since
>>> Pacemaker is not in repos), I'm running two nodes cluster with simple
>>> VIP resource. Resource migration work pretty well if I gracefully
>>> shutdown the active node, but if try to kill the node with simulated
>>> kernel panic (echo c > /proc/sysrq-trigger) the VIP resource will not
>>> be migrated even if "clustat" says the node is "offline". I've fome
>>> some errors in fenced.log
>>>
>>> "fence rgmanager02 dev 0.0 agent none result: error no method"
>>>
>>> So if I understand it right, since fencing is not working rgmanager
>>> does nothing in order to not make it worse, is that correct? If yes,
>>> can I force it do not care about fencing? (I know that is general not
>>> a good idea)
>>>
>>> My cluster.conf for more information:
>>> ########################################
>>> <?xml version="1.0"?>
>>> <cluster name="mycluster" config_version="15">
>>> <cman two_node="1" expected_votes="1" transport="udpu"/>
>>> <clusternodes>
>>> <clusternode name="rgmanager01" nodeid="1">
>>> <fence/>
>>> </clusternode>
>>> <clusternode name="rgmanager02" nodeid="2">
>>> <fence/>
>>> </clusternode>
>>> </clusternodes>
>>> <fencedevices/>
>>> <rm>
>>> <failoverdomains>
>>> <failoverdomain name="domainarg" ordered="0" restricted="0">
>>> <failoverdomainnode name="rgmanager02" priority="1"/>
>>> <failoverdomainnode name="rgmanager02" priority="1"/>
>>> </failoverdomain>
>>> </failoverdomains>
>>> <resources>
>>> <ip address="192.168.122.10" monitor_link="1"/>
>>> </resources>
>>> <service autostart="1" domain="domainarg" exclusive="1" name="HA_IP"
>>> recovery="relocate">
>>> <ip ref="192.168.122.10"/>
>>> </service>
>>> </rm>
>>> </cluster>
>>> ################################################
>>>
>>>
>>> Thanks,
>>> Stan
>>>
>>> _______________________________________________
>>> Users mailing list: Users at clusterlabs.org
>>> http://clusterlabs.org/mailman/listinfo/users
>>>
>>> Project Home: http://www.clusterlabs.org
>>> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
>>> Bugs: http://bugs.clusterlabs.org
>>>
>>
>>
>> --
>> Digimer
>> Papers and Projects: https://alteeve.ca/w/
>> What if the cure for cancer is trapped in the mind of a person without
>> access to education?
>>
>> _______________________________________________
>> Users mailing list: Users at clusterlabs.org
>> http://clusterlabs.org/mailman/listinfo/users
>>
>> Project Home: http://www.clusterlabs.org
>> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
>> Bugs: http://bugs.clusterlabs.org
> 
> _______________________________________________
> Users mailing list: Users at clusterlabs.org
> http://clusterlabs.org/mailman/listinfo/users
> 
> Project Home: http://www.clusterlabs.org
> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
> Bugs: http://bugs.clusterlabs.org
> 


-- 
Digimer
Papers and Projects: https://alteeve.ca/w/
What if the cure for cancer is trapped in the mind of a person without
access to education?