Server freeze When seeing no disk I/O and 100% highet active time
We are running Server 2012 Hyper-V with 5 hosts and about 70 VM's Following the firmware upgrade on the equallogic to version 6 we started to see performance issues with the VM's The server would freeze anyone using the applications running on the server would report them as hanging. In resource monitor on the VM under Disk when the freeze occurs the Disk I/O Drops to 0 and the Highest Active Time jumps to 100% The queue length for a couple of the disks jumps to 1 and this could last for about 1 minute. It looks like the Iscsi for the VM can not read from SAN. This can affect several VM's at the same time but on different host's. The hosts don't show anything. We have been trying to Identify the cause but so far we have not been able to find anything or a solution to the problem. Has anyone seen this type of issue before or have any suggestions?
We have had calls logged with Dell regarding the issue but so far they have not been able to track down the issue.
We have updated network card drivers, Bios on the blade servers. The HIT pack. So far nothing has gotten to the bottom of the issue. I've logged the issue here to see if anyone has seen the issue we have before. This issue seems to be beating everyone.
@Chris Richards Any chance you were able to find a resolution to this issue? I'm currently banging my head against an old EqualLogic showing identical symptoms to what you've described. We've tried following a number of recommendations including disabling VMQ, TCP Offloads, but have not found a resolution. I realise this is a total shot in the dark as this thread is nearly 6 years old....
Chris Richards
2 Posts
1518
0
Posted April 1st, 2014 08:00
Hi Don
We have had calls logged with Dell regarding the issue but so far they have not been able to track down the issue.
We have updated network card drivers, Bios on the blade servers. The HIT pack. So far nothing has gotten to the bottom of the issue. I've logged the issue here to see if anyone has seen the issue we have before. This issue seems to be beating everyone.