We are in the process of setting up VMware stretch cluster using VPLEX Metro
and a cross-connect configuration. Currently the VMware cluster is setup with
six (6) hosts and one site, and six (6) hosts at another site.  There are
two (2) redundant SAN fabrics, and each fabric spans both sites via DWDM
support FC.  The link between the sites is 24 Gb/s per fabric.  We are using
PowerPath/VE for multipathing on the VMware hosts.  All hosts in the VMware
cluster see all LUNs from both VPLEX clusters.

From a VMware point of view:

     1.  VMkernel.Boot.terminateVMOnPDL changed to Yes
     2.  DRS Groups created for each site (should rules)
     3.  HA Admission Control set to 50%

This past week we attempted two (2) failover tests:

     1.  Physical host at site A running the VM.  Unzoned all hosts from
         site A from all paths between the hosts and the VPLEX clusters.
         Maintained paths for site B hosts.

     2.  Physical host at site A running the VM.  Removed all hosts from
         site A from the storage views on BOTH VPLEX clusters.  Maintained
         the hosts at site B in the storage views.

What we found is:

     Test 1.  If the physical server loses all paths to the VPLEX, HA will
              NOT occur from the VMware point of view.  Instead, the host goes
              into an APD state, and you have to manually initiate failover. 
              It appears to be how VMware designed this.  It is unfortunate.

     Test 2.  If the physical server is removed from the VPLEX storage view,
              but maintains its paths to the VPLEX, HA will occur, but it will
              question you first.  What you get in vCenter is a pop-up window
              that states: "The storage backing virtual disk
              TSTKRSW8K1v-TEST.vmdk has permanent device loss. You may be able
              to hot remove this virtual device from the virtual machine and
              continue after clicking button.retry.  Click button.abort to
              terminate this session."  And you have a choice of "Retry" or
              "Cancel".  If you select “Cancel” the HA will occur, BUT it is
              not automatic as hoped.  We thought that TerminateVMonPDL would
              suppress the question.

Has anyone else seen this? 

     1.  Is there a way to make VMware automatically HA a VM if it goes into
         an APD state?
     2.  Is there a way to make VMware automatically HA a VM if some hosts can
         still access the VPLEX cluster but cannot access the LUN?  A way to
         get around that question from Test 2 above?

Any thoughts?

Thanks,

Will