Recently I had to reinstall an existing Dell PowerEdge 2970 server.
This server was running SUSE Linux Enterprise Server 10 SP2 (32bit) with Xen 3.2.0 as virtualization software (host), but I decided to put VMWare ESXi 4.1.0 on this box.
This server has a single Dell PERC 5/I RAID Controller, with six 72GB SAS disks attached to it.
I started by destroying the existing two RAID sets, and created 2 new RAID sets. One for the OS (RAID1) and the other for the datastore (RAID10).
After the RAID sets where intialized I started the installation of VMWare ESXi and the server was up-and-running in less than 20 minutes.
As part of the post installation configuration of the VMWare ESXi server I tried to add a second (local) datastore by using the Add Storage wizard in the vSphere Client.
In the Add Storage Wizard, I specified Disk/LUN as Storage Type and selected the available disk (the RAID10 set).
After pressing Next, the Add Storage Wizard returned the following error:
An error occured during host configuration.
Call “HostDatastoreSystem.QueryVmfsDatastoreCreateOptions” for object “ha-datastoresystem” on ESXi “192.168.1.61” failed.
On a side note, if you close this error message and go a step backwards and continue forward it wil not give an error anymore.
This is not a solution or workaround, because when the Wizard gets at the Formatting property sheet, it wil give the following error:
"De objectverwijzing is niet op een exemplaar van een object ingesteld" is dutch for "
Object reference not set to an instance of an object"
I couldn’t find more detailed information about this error in the eventlog of the vSphere client, located in the directory: %USERPROFILE%\Local Settings\Application Data\VMware\vpx.
And a search on the Internet didn’t gave a answer either.
So I was forced to troubleshoot on my own.
I was certain that the volume was consistent , because I had performed a full initialization and integrity check when the RAID10 disk was created.
As shown in the second screenshot of this post, the Add Storage wizard indicated that the disk was blank. I further noticed that there was no Available space on this disk.
Normally speaking it would show how much space is available on the disk, so I assumed that there was something wrong with the partition table of this disk.
I couldn’t find an option in the vSphere client to remove the existing partition info on this volume.
So I logged on to the ESXi server by using a SSH connection and issued the command fdisk -l.
The following output was shown:
The disk /dev/disks/naa.6001e4f023eb110013ffed36d40990cb is the disk on which I tried to create a datastore.
I directly knew which disk I should look for, but it can be easily identified through the Add Storage wizard of the vSphere client. (See the first screenshot in this post)
As you can see, this disk contains 3 partitions (p1, p2 and p3)!
How is this possible?
Probably a leftover of the previous installation, although I had destroyed the previous RAID set.
Fdisk also gave some warnings, indicating a non-consistent partition table on this disk.
I decided to remove the partitions manually and used the following commands to remove them:
- fdisk /dev/disks/naa.6001e4f023eb110013ffed364d40990cb.
fdisk is started in the context of this disk.
- Press d and enter 1 for the partition number, partition 1 on this disk will be deleted
- Press d again and enter 2, which will delete partition 2
- Because partition 2 is an extended partition, partition 3 will be deleted as well. There is no need or even possible to delete the third partition.
- Press w, to write the changes to disk
To verify the deletion of the partitions on this disk, I issued the fdisk -l command again:
After this procedure, I was able to create a datastore on this disk through the Add Storage wizard!