Host Issue

tstrip007

We recently moved a few hosts and SAN to a datacenter. Yesterday I noticed a host and its vm's would show disconnected for a few minutes in vcenter and when it came back, most if not all of the vm's are powered off. This has happened several times now.

I am seeing one alarm on the host "Network uplink redundancy lost"

The net. engineer mentioned that the vm's seem to be losing connection to the SAN every x amount of minutes.

We've done the very basics here and gracefully rebooted host and rescanned everything. Switch net cables.. The issue still persists.

Any ideas on where to start with this? Thanks

TheProf

Without getting more info, it's hard to say. But I would assume that your virtual machine isolation response is set to power off VMs which is why they're powered off when the host is partitioned (isolated from the network due to network issues). You can change the host isolation response within the cluster settings if I recall correctly. That should prevent your VMs from shutting down.

Regarding the other issues, it's hard to say, we'd need more info, like SAN logs, vSphere logs, etc.

Lexluethar

Are you able to ping the host when it says disconnected? We've seen this on a few Hosts every now and again when there is high I/O and processing on the Host itself. I've also personally seen it when a host is busy doing something (IE if we have a lot of datastores and tell the host to rescan for vmfs volumes). When we do that every now and again the Host will drop (not really, just from vCenter) but won't miss a beat when it comes to processing or the vm's. The Host just isn't responding in a timely manner to vCenter notifications so it shows as disconnected.

Not sure if that's your issue or now - but thought I would share. When it disconnects see if you really do have a networking issue. I would also run a test vm on there and when it gets disconnected see if the VM is still responding to pings and RDP.