Esxi 5.1 - VM Inaccessible

higherho

Hello all,

Recently our network went down briefly (made our Cisco switches into a Stack switch) and when it came back online I took notice that some VM's could not access there configuration files even though the virtual switch was fine, I was able to browse the datastore, and I was able to ping everything. The only way I found around this issue was to delete the VM from the inventory and then reload it from the datastore. Once, I did this everything worked fine. However, I'm still puzzeled why it would not see it in the first place when the network connectivity came back online. Thoughts?

Essendon

I reckon a restart of the hosts could have fixed the issue too. Looks like the paths to the datastores were messed up.

higherho

Essendon wrote: »

I reckon a restart of the hosts could have fixed the issue too. Looks like the paths to the datastores were messed up.

That was the first thing I tried and same error. After that I went into the storage adapters and checked all my manage paths and all were active. I then went to the networking adapters and verified networking connectivity. It was reallyt strange =/ I even tried unmounting the datastore and re adding it, still nothing =/

I should have used the maps feature to see if anything else I was missing. The logs on the datastore did not generate anything either, the only thing I got was the generic alarm in Vcenter stating my network connectivity was lost.

what a time to end on a friday Brougt back the important VM's at least.

odysseyelite

What type of storage are you using? I woud have said the same thing as Essendon with the path to the datastore being messed up.

Essendon

Good you bumped this thread up, I hope our resident VMware experts chime in with some ideas too.

jmritenour

Like Essendon suggested, most times I've seen this issue, a host restart is the solution.

One time a few months back, however, I saw something similar when a host had a RAID controller failure that left it in a state that the actual ESXi OS & VMKernel crashed, but the hypervisor layer was still alive. The VMs were all still up and functioning, and I could even ping the host, but it was dead to the vCenter server. The VMs didn't fail over since they were still alive and happy, but I had no way of vMotioning them over to another host, so I had the customer gracefully shut them down within the guest OS then power cycled the host, replaced the RAID controller and brought it back up.

The host was fine, but all the VMs that had been running on it were listed as inaccessible in inventory. Like you, I had to remove them all, and and them back in by their VMX file. Unfortunately, I never got a chance to investigate beyond that as this was mostly a black box system that I had no access to accept when needed, and the customer was satisfied that everything was back up.

Essendon

That's an interesting one, jmritenour! Bookmarked thread for future reference.

higherho

Very interesting! We are using a 20TB ISCSI SAN (soon will be 120 TB though) . I haven't seen the issue happen again but from time to time I do see some VM's going invisible and then coming back to normal within seconds (this does not happen often).