HI all,
I've got a few ESX hosts that are experiencing the same issue. All are configured with the same hardware IBM system x3650 M3 with the M5015 SAS RAID controller.
Every now and then, usually about a week apart, the guest servers become very sluggish to the point of being unresponsive and in turn the host is unreacheable through the vSphere client and the console us also unusable.
I've rolled back to ESX 4.1 on all but one host and they are since working fine, they were initially built as ESXi 5.0. The last one however was upgraded from 4.1 to 5.0 and this is when the problems occurred. Whether a new build or upgrade from 4.1 to 5.0 we experience the same issues.
Has anyone else experienced this on same or different hardware? I've configured logging to write to the datastore rather than the scratch location so I can investigate further when it occurs next. So far the only warnings I am seeing are the below.
2012-02-28T06:23:40.010Z [47AFBB90 warning 'Statssvc'] Calculated read I/O size 891448 for scsi0:1 is out of range -- 891448,prevBytes = 134233361920 curBytes = 134586375680 prevCommands = 176437curCommands = 176833
Cheers,
Luke.