TruCluster
cancel
Showing results for 
Search instead for 
Did you mean: 

TruCluster+LSM disaster recovery procedure

Miran.Lampe
Occasional Visitor

TruCluster+LSM disaster recovery procedure

Hallo,
I need an advice regarding to use of Trucluster and cluster_root in LSM configuration.
We have a customer that uses LSM on trucluster for disaster recovery purpose. They have two EVA storages and two AlphaServers in trucluster. Each server has boot disk on another storage(Node1-EVA1, Node2-EVA2). They have cluster_root, cluster_usr, cluster_var incorporated in LSM configuration and mirrored between the two EVAs, because of disaster recovery. Quorum of course exists only on one storage because it cannot be included to LSM, so they actualy don't have disaster recovery configuration.
During cluster migration to new storage we had big problems to reconfigure cluster(different LUNs/device names). At the and we decide to migrate cluster_root, cluster_usr and cluster_var form lsm to physical disk partitions.I was check literature for TruCluster or LSM disaster recovery but, there is no procedure that covers recovery of Trucluster with Cluster_root in LSM.
My question is:
Is there any strait forward recovery procedure for Trucluster with cluster_root in LSM if there is no running cluster members(e.g. boot from initial disk and LSM restoration...)?
Or is it better to leave cluster_root, cluster_usr and cluster_var on physical disk pratition and provide system backup via AdvFS disk cloning or vdump/vrestore?

I am afraid that in case of whole cluster loss, recovery will take to long or is even impossible if the cluster_root exists in LSM.

Thanks for any answer!

Regards Miran Lampe
2 REPLIES
Rob Leadbeater
Honored Contributor

Re: TruCluster+LSM disaster recovery procedure

Hi,

Personally, I would ditch LSM completely.
It just complicates things, especially when all the physical disks are (presumably) sufficiently resilient...

I've done a fair few disaster recovery tests and invocations in my time, and know that those involving LSM took significantly longer to get back to a working system.

Cheers,

Rob
Vladimir Fabecic
Honored Contributor

Re: TruCluster+LSM disaster recovery procedure

Hello Miran

Yes, what Rob said about LSM is true, it complicates.
But, this concept will work if you do not loose quorum disk (failure of second storage).
It is not a big problem to recreate other member (if you have vdump backup of member's system disk with configuration files).
The problem is when you loose quorum (quorum disk can not be LSM disk). In that case booting member is not easy, but it is possible (but gives a lot of headache).

When having two EVA-s, the best way for disaster recovery is CA.

But allways do a vdump backup!
It is possible to recreate a complete cluster from vdump backups. It takes a lot of time, but it can be done (no production lost).

The only thing I do not like in TRU64 is LSM.
In vino veritas, in VMS cluster