1 Reply Latest reply on Apr 23, 2009 12:21 AM by Osm3um

    IMS Shared LUN and ESX


      I am running an IMS MFSYS25 with the share LUN piece loaded.


      I have two storage pools.  ESX 3.5u4 is loaded on a VD in the first storage pool and a second copy of ESX 3.5u4 is loaded on a VD in the second storage pool.  There is a second VD on the second storage pool which I would like to share amongst the ESX installations.


      So everytime I assign the second VD (VMFS) on the second pool to both ESX instances, it corrupts all of the data.


      Isn't that the point of the shared LUN?  To be able to have two copies of ESX (for example) use it simultaneously?

      Is the issue occuring because I have an additonal VD on the second pool which is running ESX (which is not VMFS) although I have not assigned it to two servers?

      Does the entire Pool need to be a singel VMFS VD to be shareable?

      Could I run two VDs each containing a copy of ESX on a single pool?


      Problem is I do not have enough HD slots so I either have to run a second ESX on the same pool as the first ESX OR run an instance of ESX on a VD in the same pool as the VMFS store.


      Any help would be appreciated.



        • 1. Re: IMS Shared LUN and ESX

          I have determined that in order to use the shared LUN piece of the Intel Modular Server, it appears, that the shared volume must be the only virtual disk within a Pool.  Here is a more in depth explanation:


          In the IMS (Intel Modular Server) in order to use the storage system the user must setup Pools.  A Pool is made up of any number of drives.  Then VDs (Virtual Drives) must be carved out of the Pools.  Then the VDs are "assigned" to a server or multiple servers.  Just like a SAN.


          When I had the following setup the shared storage failed:
          Pool1: One VD containing an ESX installation.
          Pool2: One VD containing an ESX installation and a VD VMFS Store shared between both ESX servers.


          When I changed my Pools to the following it has been working fine:
          Pool1:  One VD with two VDs both containing an instance of ESX.
          Pool2:  One VD containing a single VMFS store shared between the two ESX servers in Pool1.


          As the "assignment" process occurs at the VD level (as opposed to the Pool level) it is not clear to me why the first scenario failed.


          Would it would work if there were two VDs each with  VMFS store assigned to multiple servers?  No clue.....


          If someon can clrify my findings I woudl appreciate it,