rbird97 wrote:
We have come to the conclusion that there is something wrong with the one host in the file cluster. We are still waiting for HP to come back with some kind of idea of what is wrong.
Hello,
have you checked the Windows MPIO settings on the host for the affected LUN ?
because we had a similar problem on a Windows 2012R2 hyper-V cluster and two replicated 3par arrays with Peer Persistence ,since the end of June ( some Microsoft securty patch bundle included a MPIO change?) :
in our case, it turns out on some LUNs sometimes the Windows policy wrongly changed itself from RRSW (roundrobin with subset) to plain round robin :
so it tried to use the passive path , with a ~3 000 ms timeout before going to the next path. the I/O doesnt truly fail , it is just slow...
also, if your LUN is disk8 type the powershell command
mpclaim -s -d 8
and check for any discerpancy bettwed the "actual" and "desired" state for each of its paths
(sorry, my servers are in french)