ESXi 6 fails/becomes unresponsive when attempting to create Datastore on OCZ SSD

[Solidbrass]

I am trying to bring a Mac Mini 5,3 online with vSphere 6.  The system originally came with two 500GB 7.2kRPM drives, and I swapped one with a brand new OCZ ARC100-25SAT3-480G SSD.  I am booting from a SanDisk 64GB USB plugged into the back.

I brought the system online, added it to my vCenter, created a new datastore using the remaining 7200RPM drive, and all was well up to that point.  As soon as I attempt to create a new datastore on the OCZ drive, the host becomes unresponsive to either vCenter or the old thick client.  At the console I am still able to log in and look at the logs but that is about it, and it will even hang if I attempt a restart from the console so I have to manually power cycle it. If I leave it running, after a very long time (more than 10 minutes) the thick client will eventually throw the following useless error:

Call "HostDatastoreSystem.CreateVmfsDatastore" for object "ha-datastoresystem" on ESXi "octo" failed.

So I tried using partedUtil to remove partitions and do it again, with no change as soon as I tried to create the datastore (UI becomes unresponsive). Then I booted the machine using an external drive with OS X 10.10.3 on it and tried partitioning the drive MBR/FAT32, wrote and read a few files to make sure the drive actually seemed to work, and booted it back on on vSphere 6.  When I attempted to create a Datastore again, I still got the same behavior.

After a reboot, this is what partedUtil reports for the disk:

partedUtil get /dev/disks/t10.ATA_____OCZ2DARC100______________________________A22L2061437000076___

58369 255 63 937703088

1 2048 937697984 0 0

[Alistar]

Hello,

it seems the storage subsystem of ESXi does not want to play with the SSD properly. Have you checked if the firmware is up-to-date on your drive and that device drivers/firmware for your storage controller are up-to-date as well? Try this KB to see where you would fail during the steps that are given to manually create a datastore: VMware KB: Manually creating a VMFS volume using vmkfstools -C

[RichardBush]

Hey,

Could you try creating the partition in sections, for example, 100GB first, then extend it to 200, 300 400 and then 480 ?

I had to do this on my SSD, else id have the same problems you are having.

R

[Solidbrass]

Doesn't work, in fact ESXi seems to just hate this disk.  Since my original post, I've tried a few more things.

First, I tried to use both partedUtil and disk to delete the partition on the drive.  This does not work.

partedUtil delete /vmfs/devices/disks/t10.ATA_____OCZ2DARC100______________________________A22L2061437000076___ 1

Error: Read-only file system during write on /dev/disks/t10.ATA_____OCZ2DARC100______________________________A22L2061437000076___

Unable to delete partition 1 from device /vmfs/devices/disks/t10.ATA_____OCZ2DARC100______________________________A22L2061437000076___

So I yanked the drive out, which is not trivial on a Mac Mini, and attached it to a windows box, then used diskpart to run the clean command on it.  I stuck back in the Mac Mini and had the same freezing and uselessness.  So I thought I'd try a different system.  I have an older SuperMicro X8SIL system in the rack that was easy to pop the top off of and hot insert the drive after another round using clean with diskpart.  After refreshing the storage controllers, the X8SIL machine saw it, so I told it to try to partition the device, and I got the exact same behavior.  Since this machine has a disk activity light, I can see that the disk light is pegged on.

[Solidbrass]

Well hell, after about 9 minutes the device now appears as a datastore on the X8SIL machine.  I tried swapping it into the Mac Mini and that magically works.

[JtheDBA]

I am having a similar issue with a Silicon Power SLL-S60, thought it was the SSD, removed it and tried a Sandisk SSD and getting the same exact error; Call "PropertyCollector.RetrieveContents" for object "ha-property-collector" on ESXi "ip.address" failed.

Took about 4 minutes, got the error message twice, and then the device appear. Uploaded a live .ISO to the SSD, created a VM on it, booted. All seems fine.

Scary though. When I formatted the SLL-S60 and the client failed, I went to the console and the console was unresponsive as well.