Hi, I am a storage admin trying to determine why LUN ID's presented from different storage arrays behave differently.
We have several ESX 5.1 clusters, and they were all getting storage from an IBM V7000. We recently added a new IBM XIV for additional capacity.
There is a significant difference in functionality from the same ESX Clusters and the two storage arrays. This is all connected using the same SAN.
We recently had an issue where an XIV LUN was presented to an ESX cluster, with non-uniform LUN ID's. For example, servers 1-9 were presented with one particular LUN as LUN ID 1. New server 10 was added to the cluster, however this one LUN was errantly presented with LUN ID 2. This confused the cluster and ultimately the LUN was remounted by ESX as a snapshot, which crashed all running VM's on that lun/datastore. Best practices and VMWARE support say to present the LUN with the same LUN ID to all servers in the cluster. This snapshot remount has happened twice.
After getting this issue resolved, we have been reviewing LUN presentations across the board. In our reviews, we have found several luns coming from the IBM V7000 with non-uniform LUN ID's. In three years, we have not had this snapshot remount issue happen one time. For example, one LUN is presented to some servers in a cluster with LUN ID 1, some with LUN ID 5, some with LUN ID 23.
Why would ESX seemingly not care how luns are presented with a V7000 storage array, but seem to be extremely sensitive how they presented on XIV storage ?
I appreciate any insight anyone may offer here.