Details
This article contains steps that allow an ESX host to access a LUN presented to other ESX hosts.
Solution
To ensure consistent LUN presentation to all ESX hosts from the EMC CLARiiON, all hosts and LUNs must be in the same storage group.
Example scenario:
- You have two ESX hosts.
- Two storage groups are configured on the EMC Clariion.
- Storage group 1 contains the first ESX host and the first LUN.
- Storage group 2 contains the second ESX host and the second LUN.
- You want both ESX hosts to use both LUNs.
It is possible to add both LUNs to both storage groups, but VMware does not recommend this as it can lead to problems like the datastore being detected as a snapshot on one or more ESX hosts. Instead, migrate one of the storage groups into the other.
To migrate one of the storage groups:
- Identify the LUN that contains the least amount of critical virtual machines and shut them down.
Note: This procedure uses LUN1 as an example of a LUN that contains non-critical virtual machines. - Log in to EMC Navisphere and remove LUN1 and ESX1 from the storage group, then delete the storage group.
- Perform a rescan on host ESX1. The LUN and datastore no don appear.
- Add LUN1 and ESX1 to the storage group occupied by LUN2 and host ESX2.
- Perform a rescan on both ESX hosts.
The LUN appears under Storage Adapters but the datastore does not appear in the Storage area. This is expected behavior as moving LUN1 to a different storage group causes its UUID to change. LUN1 shows up in the vmkernel log as being detected as a snapshot. - Note: If the ESX hosts are managed by VirtualCenter, you must remove the old datastore from the datastore view before you can change the name back to the original.
The ESX hosts and LUNs are now in the same storage group.
Based on VMware KB 1002107
