Hello,
I have a new ESXi host with a single VM I am testing. The single VM has an extremely high "Average response time" for its hard drive and extremely high "Active time". Basically the hard drive in the VM is at 100% nearly 24/7 and the VM is so sluggish it is near unusable.
See here: https://i.imgur.com/b3uzOYB.png
HOST: AMD Threadripper 32 core CPU / 128 GB RAM / H740p RAID of 8x SSD running ESXi 6.7u3. It is extremely fast.
VM: 6 core 16GB RAM 250GB HD Windows 10 LTSB with VMTools
esxtop u shows this
11:19:53am up 1:41, 797 worlds, 1 VMs, 6 vCPUs; CPU load average: 0.04, 0.02, 0.01
DEVICE PATH/WORLD/PARTITION DQLEN WQLEN ACTV QUED %USD LOAD CMDS/s READS/s WRITES/s MBREAD/s MBWRTN/s DAVG/cmd KAVG/cmd GAVG/cmd Q
naa.6d0946608826be0025b455c6ea6eabe0 - 192 - 0 0 0 0.00 250.40 80.13 170.27 0.17 1.84 0.11 0.00 0.12
Notice the CMD/s is extremely high @ 250. This is for a VM that is doing nothing. Task manager shows 3% cpu usage (idle) memory 1.8GB/16 and Disk 0 @ 100%.
The ESXi host is a 32 core Threadripper, 128GB RAM, RAID6 of SSDs. I know the host datastore isn't slow as I've just benched it @ 6.98GB/s.
The VM has GPU pass-thru (though no keyboard/mouse yet). 16GB RAM. 250GB disk. LSI SAS. 6x processor cores. If it matters, when I play a Youtube video, the sound is extremely fragmented and the whole thing is extremely choppy, possibly due to disk queues being whack. The Disk 0 has 2550ms average response time, with 100% Active time, nailed solid, yet read/write is something like 800kb/s or so. I have confirmed the VM is not doing anything, just shows read/write to the NTLogFile. That is it. Yet disk is pegged at 100%.
My VM is thick eager zeroed and I've tested setting the powercli command of "fsutil behavior set DisableDeleteNotify 1" as well.
So far, nothing seems to help.
Any ideas?
Thanks