Lost access to volume (xyz123) due to connectivity issues.. Local Disks!

[WebOps]

Hi

I'm getting the error "Lost access to volume (xyz123) due to connectivity issues" but I'm using local disks (no SAN) it also comes back with a "successfully restored access to volume"

I don't want to pre-empt the answer by guessing but can this be caused by increased I/O latency? All of the answers I find are related to actual loss of connectivity such as where a SAN is used (network issues on iSCSI or FC) where this is not appropriate in this case because the disks are local to the esxi.

Thanks

WO

[CoolRam]

you installed the ESXi using the same hard disk or it is booted from the USB/card. If you booted with the same HD than the issue should not occur . If you booted with USB than please do configure the proper scratch.

The other reason may be VMware KB: Host Connectivity Degraded in ESX/ESXi

[GaneshNetworks]

What is the ESXi version that you are running?

Is it Vanilla or vendor specific one (eg., HP)?

Any error messages in RAID level, If there are multiple local HDDs?

We may see these errors, If the ESXi server went down abruptly before its last boot. If it is so, reboot the server gracefully.

[WebOps]

Ok so a bit more info to answer your questions

The server boots from Dual SD cards in RAID1 and it's ESXi5.5 as supplied by vmware. The ESXi shows no real issues, it runs as expected and the VM's that run on it appear to be all fine in respect of the ESXi.

The problem I'm investigating is that some of the 3rd party apps that are running on the VM's (Windows 2008 DC) are manifesting occasional networking issues all at the same time. The only issues I can see at the moment are these storage disconnects which would explain it because presumably the VM's will be disconnecting from the datastore for a micro second and therefore the vm's base disks? I imagine this is not too healthy for the activity of the VM.

So, can high I/O cause a disconnect of the volume? My thinking is that if the I/O becomes high enough then perhaps the ESXi treats the datastore as non-responsive and therefore "lost" - but this is a guess / theory - can anyone confirm or deny that as an explanation?

Thanks a lot

[WebOps]

Hmm, scratch? That sounds like a very possible issue.. I will investigate that as a further issue, have you a link to help me with that? I will of course Google it now

[maaca]

I'd say, it is not just "slow" storage - you would see high-latency errors, not disconnected storage.

Could you be more exact about the raid controller? What is exact controller model, firmware and driver version? ESXi is sometimes very sensitive about this constellation.