[ClusterLabs] Corosync rrp or bonding

[Jan Friesse]

Adam

> Thank you very much.I really appreciate your help!
> 
> These are 2 virtual machines ESXi do you see any possible hardware
> limitations ? (Don't know much about VMware).

I don't know much about VMware ether. The only limitation of bonding is 
switch (evento some bonding modes don't need hw support).

Honza

> 
> czw., 14.03.2019, 17:13 użytkownik Jan Friesse <jfriesse at redhat.com>
> napisał:
> 
>> Adam,
>>
>>> What is more recommended for providing corosync/pacemaker link
>> redundancy?
>>>
>>>   >
>>> a)    Configure two rings and rrp within corosync, if yes:
>>>
>>> -       can we use two separate VLANs connected to two separate physical
>>> NIC for the heartbeat connections
>>
>> Yes
>>
>>>
>>> -       do they have to be the same networks
>>
>> No, it actually must be different one
>>
>>>
>>> -       which mode should I use for a two node cluster passive or active
>>
>> Passive. Active doesn't make progress till timeout when one link is
>> disconnected
>>
>>>
>>> b)    bonding – if yes, why ?
>>>
>>
>> B is correct. RRP is just seriously broken - that is the reason why it
>> was replaced in Corosync 3. When you have a bonding capable HW, use it.
>>
>> Regards,
>>     Honza
>>
>>>
>>>
>>> Thank you !
>>>
>>>
>>>
>>> _______________________________________________
>>> Users mailing list: Users at clusterlabs.org
>>> https://lists.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
>>>
>>
>>
>