5 Replies Latest reply on Feb 23, 2007 7:59 AM by neil_murphy

    ESX 3.0.1 with HP EVA and Continuous Access

    neil_murphy Enthusiast

      Hi,

       

      I have a pilot project going where we have two ESX 3.0.1 servers in separate buildings, say building A and building B. There are two HP EVA SANs, one in each building, that are replicated using HP Continuous Access. The servers normally access the SAN in building A.

       

      My question is, if we loose building A completely i.e. server and SAN, how do we restart the VMs that were running on the building A server? They Appear in VC as "disconnected" I cannot delete them from the inventory and add them back while the server in building A is down as I am not given that option by VC.

       

      I'm sure this problem pertains to anyone with a replicated SAN. Has anyone seen this scenario or got any thoughts.

       

      Thanks in advance.

        • 1. Re: ESX 3.0.1 with HP EVA and Continuous Access
          CWedge@Amsa Expert

          Since you haven't got any replies. I throw a stab at it being a San Admin.

           

           

          I do also have 2 EVA 5000's same location though..

           

          Consceptually. You would need VC w/HA (not just ESX HA)

           

          I wonder if you granted access to the replicated volume in Site B to the ESX server over there and added it to the cluster, then should Site A go down Vcenter HA should kick in at Site B and turn on all the VM's against thier locally attached EVA at site B.

           

          Don't know if that'd work...but Alteast u get a free bump.

           

          GL. I'm also interested in this.

          • 2. Re: ESX 3.0.1 with HP EVA and Continuous Access
            neil_murphy Enthusiast

            Hi,

            Maybe I should describe what is happening at present. If we initiate a manual failover from one EVA to the other, all VMs will hang and the VMFS volume can no longer be seen even though Continuous Access has presented the replicated volume back to the ESX servers. In order to see the volume we have to rescan the storage adapters and then refresh the datastores in VC. Then I can restart the VMs. The problem arises when one of the ESX servers goes down at the same time as a Continuous Access failover. Then, the VMs that were running on that server appear grayed out in VC and say disconnected. Most right-click options are also grayed out including power on/off, remove from inventory, etc.

            Glad to have your input on this.

            • 3. Re: ESX 3.0.1 with HP EVA and Continuous Access
              frankdenneman Expert

              When you have a site failover but you can connect to the vc.

              You can (cold) migrate the virtual machines to the running ESX servers.

              HA clusters can do this for you.

               

              Vmware has posted the following document about HA:

              Automating High Availability (HA) Services with VMware HA

              http://www.vmware.com/pdf/vmware_ha_wp.pdf

              • 4. Re: ESX 3.0.1 with HP EVA and Continuous Access
                epping Expert

                OK this is no problem, doing the same thing with Clariions.

                 

                present the LUNs to you a esx box in you DR site.

                 

                i VC delete the cluster that has failed, this will remove all disconected hosts and VMs.

                 

                browse the data store from the DR esx host and add the vms back into the inventory and power on.

                • 5. Re: ESX 3.0.1 with HP EVA and Continuous Access
                  neil_murphy Enthusiast

                  This would pose a problem for me as my customer is looking for an automated disaster tolerant solution for maybe 100 VM's.