Hi All,
I have the following setup :
2 x HP EVA5000's
1 x HP MSA
4 x ESX hosts with FC Connectivity.
2 of the ESX hosts have 2 dual port HBA's (LPe11000) and the other 2 have 2 x dual port HBA's LPe1150's.
The storage has an Active Active config and therefore we are using fixed paths.
Below is a run through of events that may or may not be giving us the issue but I thought it best to list them all.
We have only just moved to vSphere and for the first couple of days everything looked ok. I then had to upgrade our VCB from 1.0.3 to 1.5 Update 1 as the VCB still needs to backup 3.5 vm's.
The backups took a little while to get working but we had some successful ones. Around this time we started seeing an issue in ESX. The initial error was regarding MKS - unable to find config file vmx errors when opening the console on a VM and users not being able to access the servers.When I went to browse the datastore where the vm was locate, the datastore had disappeared from 2 of the ESX hosts (both with different HBA's), but the other 2 hosts could still see the datastore. All other datastores where viewable from all hosts.
In order to resolve the problem I logged a call with VMWare and we changed the preffered path to the volume and the ESX host was able to see the datastore. So all was good and I went home on Friday not having to worry about multipathing !! Today, the problem has happened again on a different datastore. It has happened with a host I have changed the preferred path on and one I didnt. Again if I change the path everything can be viewable again and the vm's work.
VMWare have said to me that this is something to do with the storage array but does anyone here know how I can prove that ? The storage guys are giving me the teflon cold shoulder and are saying its VMware multipathing not working properly. We never saw this issue on 3.5.
Thanks for looking at this problem.
Annie
