[ClusterLabs] CCIB migration from Pacemaker 1.x to 2.x

[Andrei Borzenkov]

23.01.2021 19:41, Sharma, Jaikumar пишет:
> Thanks Andrei for quick reply.
> 
> 
>> Which cable? There is corosync network, there is application network,
>> there may be backend network, and they may use different adapters.
> 
>  only one network where application is running - same corosync layer network.
> 
>> Assuming "network cable" means corosync network - you have
>>        stonith-enabled="false" \
>  >       no-quorum-policy="ignore" \
> 
> Yes right, i was also puzzled after reading pacemaker manual but this was working in Debian 9.x based solution.
> 
>> so when connection between nodes is interrupted, both become master and
>> database goes out of sync.
> 
> Yes, you are right I can see error on slave:
> error: Resource vip-master is active on 2 nodes (attempting recovery)
> https://wiki.clusterlabs.org/wiki/FAQ#Resource is Too Active
> 
> How to handle it?
> 

Setup STONITH or add more nodes and rely on quorum.