I have 2 Dell PowerEdge servers (PE2850, PE 2950) that are running ESXi 4.0 Update 1, and host a variety of VM's using the Essentials package from VMWare. No HA, no VMotion, etc. I use local storage only for my ESXi hosts.

Recently my 2850 through up 2 predictive drive failures, and I ended up losing 2 out of 6 drives in the array. With Dell's help, I was able to limp back into service, but the VM's suffered catastrophic Windows failures, and had to be deleted & re-created. No data was lost, and these VM's were simply gateways or servers for other systems, but it was still a pain to re-create everything.

My confusion lies in the setup of the discs through the Dell PERC setup utility. It looks like all of the drives are members of the same single Array (A00-00, A00-01, A00-02, A00-03, A00-04, A00-05). I think that should be set up differently, to be a hardware-based RAID configuration, to protect me in case of future problems like this. Is that correct, and if it is, what is the best way to set up the RAID configuration for maximum protection? I was lucky this time in that the failure hit a server without my core VM's, but I'm guessing the 2950 may have the same flaw in configuration, and that server does contain VM's that I can't afford to lose.

I will blow out the current ESXi install when the new drives arrive, and start from scratch to rebuild the host. First, though, I need some advice to help ensure that this doesn't happen again.

Thanks for any advice you can provide!

With the RAID 5 array that you had setup, you're OK with one drive failure but not two. For 2 drives to fail and still be OK, you would need RAID 6.

Couple suggestions
1) Go RAID 6 - you'll have all the drives in the array working and you could loose up to 2 drives
2) Stick with RAID 5 but use a hotspare. In this case you could loose up to 2 drives but not at the same time (i.e. if one drive failed there would have to be sufficient time for the hotspare to be brought online and the data written to it before the 2nd failure).
3) RAID 10 - with 6 drives you could loose up to 3, but if the wrong 2 fail your array is gone.

Was Dell able to explain why the 2 failed about the same time?

No clue as to what happened, and the tech didn't do any investigating, just tried to help me recover what I could.

By forcing one, then the other drive back online, I was able to boot into ESXi, and my VM's were there, but they were corrupt. They would boot into Windows, but had all kinds of weird issues, so it was easiest to just blow them out & re-create them from scratch. I was lucky this time.

I was totally impressed with the help I got from Dell. I didn't buy this server from them, I got it on eBay. The failure happened on a Saturday morning. I went to Dell's website and initiated a chat session, and a tech walked me through the steps I needed to try, and even stood by while I went and tried various steps. Then, I got a phone call follow-up a few days later from the tech, asking how the recovery was going, and then a couple of e-mail follow-ups as well. Very impressive tech support.


Being somewhat unfamiliar with the steps I need to take next, do I go into my current array, wipe it out & re-initialize it as a different RAID level, or can the change be made to the array that's there now? I plan to re-install ESXi, so if I have to wipe everything clean and start over, I'm not worried about it.

With the above options you would be looking at rebuild the array so you would be starting from scratch. If you were adding another disk then the RAID 5 (and probably RAID 6) option would not require a rebuild.