topic Re: Remove quorum requirement from cluster in StoreVirtual Storage

Remove quorum requirement from cluster

Brian Pither — Thu, 06 Jul 2017 00:11:50 GMT

I have two 4730 connected by 10Gb link with a FOM on separate hardware. The quorum is two.

When the FOM and one 4730 are lost the remaining 4730 crashes.

Is it possible in planned maintenace to remove the quorum requirement to enable the remaining 4730 to keep running?

This would also be a requiement in a disaster recovery situation where one 4730 and the FOM are destroyed but the remaining 4730 is operational.

I have not been able to find an answer to this in the documentation or forum. If it is available please advise.

Re: Remove quorum requirement from cluster

giladzzz — Thu, 06 Jul 2017 08:08:04 GMT

you can use a virtual manager. the diffrence between FOM and virtual manager is with FOM failover

is automatic but with virtual manager failover is manual.

you can not use FOM and virtual manager together.

virtual manager meens one not has two votes so when the other node is down it keeps working

if it is planned downtime you start virtual manger before shuting down the other node.

if unplanned you start it after the node is done manually to get things working on the node

that remains.

Regards

give a KUDO if this helps

Re: Remove quorum requirement from cluster

oikjn — Thu, 06 Jul 2017 23:03:17 GMT

if the FOM is on seporate hardware, are the two 4730's at two different site? If you lose 2/3 of your votes, the last 1/3 is stopping specifically to prevent a split brain situation as it doesn't know if the other two nodes are gone or if it just lost network connectivity to both nodes..

Multi-site configurations get complicated and there are tradeoffs for either, but if everything is in the same datacenter, but on different hardware, worrying about an FOM and a 4730 going down at the same time is a bit overkill... if you are that worried about a dual node falure, you should buy two more 4730's so you can keep 51+% of your votes online when two nodes go offline.