Take the 2-minute tour ×
Server Fault is a question and answer site for system and network administrators. It's 100% free, no registration required.

I'm in the process of upgrading a pair of physical servers to 3 virtual machines. The 2 physical boxes (HP DL380 G5 with Smart Array P400 disk controller) will run VMware ESXi 4 update 1. Each physical machine has 8 72GB drives originally configured as follows:

  • Disk 1 and 2, RAID 1 (mirror), operating system
  • Disks 3 - 8, RAID 5 (striped), data

After doing a physical to virtual conversion/migration of one of the physical machines to a temporary VM on another existing ESX host I pulled the mirror set (OS) disks out to preserve the original OS until I was comfortable with the temporary VM. 6 disks remained and were reconfigured as a new RAID 5 array. I then rebuilt the box as a new ESX host and created virtual machines.

Later, I decided to add the 2 saved disks back into the system. After physically adding, I used the Array Configuration Utility (ACU) to add the 2 disks to the array, then expanded and extended it. However, the ESX environment will not recognize the expanded disk. The datastore properties dialog shows the local physical disk capacity as 473 GB, but the datastore shows a capacity of 336 GB. When I click the "Increase..." button, there are no paths to go down since there are no new, unallocated LUNs from which to choose. I rebooted the system several times during these steps.

Having struggled with some other issues, I decided to start from scratch. Using the ACU, I deleted the logical volume and started the host installation again. When prompted for the volume to which ESX would be installed, it showed the single volume with the expected full disk size (~470 GB). I chose that and installed. After starting the ESX host, I looked at the datastore and saw that it is still only 336 GB. However, like before, the datastore properties shows the physical file system having a capacity of 473 GB. Furthermore, and most surprising, when I browse the datastore all the files (virtual disks and such) that existed before I supposedly deleted and recreated the logical volume are still there as if nothing had happened.

I'm sure the datastore size is not limited by ESX since I have another system running version 4 that has a logical volume of around 410 GB.

Does anyone know what I'm doing wrong or if I have to do something specific within the Linux environment to get it to recognize the larger volume?

share|improve this question

2 Answers 2

up vote 2 down vote accepted

Basically you shouldn't have used the two 'new' disks to expand the existing logical drive, VMFS3 doesn't usually work that way, given you have a (pretty capable) P400 adapter you should have just added them new disks to the existing array but then created a new logical disk from the unused space, still with R5 (I know that sounds mental to have what sounds like a 2 disk R5 disk but it's aggregated across the array), this new logical drive could then be added to the datastore as a second extent.

As it stands now you have a 470GB disk with a single 336GB VMFS3 partition, personally I'd be tempted to copy off the VMs temporarily and rebuild from scratch, pretty sure you can't shrink back the logical drive, I wouldn't try it anyway.

edit - oh and I just spotted another of your posts on another question, serverfault's not a forum, don't 'answer' with a comment, use (when you can) the comment option, I've converted it to a comment but normally they get deleted ok :)

share|improve this answer
    
I would have, but commenting wasn't available on that post. Probably not enough rep. I just signed up. –  Peter Apr 19 '11 at 19:44
    
I suspected that was the answer I'd see after doing a bit more reading. However, it doesn't explain why the reinstall on the recreated volume didn't see the new disk size. –  Peter Apr 19 '11 at 20:02
    
I'm having problems copying the VMs off the machine so I was just going to start over entirely. Not much lost work fortunately. –  Peter Apr 19 '11 at 20:03
1  
By default the installer leaves VMFS partitions intact with no changes at all. –  Chopper3 Apr 19 '11 at 20:53
1  
Ah, ok. So I should go into the ACU and blow away the whole array and just start clean. I finally got one of the VM files downloaded. Will try the others before I commit to toasting the whole thing. Thanks. –  Peter Apr 20 '11 at 15:04

The final outcome of this was that I started over completely.

I needed to move one of the VMs I created off to another host, so I just downloaded all the VM files. Then I was able to start from scratch.

As @Chopper3 suggested, the installer left the VMFS partitions in place. First I tried just a simple reinstall and all the data remained. Then I deleted the (now expanded) array and recreated it. This apparently didn't change anything and the volumes and partitions still remained. Finally, I powered down, swapped all the disks around, used the HP utility to "Erase Disks". Then I was able to create the new array and volume. The ESXi installer created the new partitions using all of the available space.

Honestly, I've never found it so hard to destroy data. Usually my systems do that for me.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.