RAID 50

Concur, Lsi based adapters can not be mirrored. Not that I have tried it but you could mirror two hardware raid 5s through the Windows Disk Manager, which I am NOT recommending at all.

 

5 controller failures in a year (????????), in 17 years of using Lsi based raid adapters, I have had 1 failure. Unless you have a severe power or heat issue, 5 controller failures is virtually impossible, an unfortunate miracle; more likely your issue is array loss due to drive losses, specifically the losses of two or more drives in a very short period of time. Schedule Consistency Checks to run often on raid 5.

 

What is likely happening.....read the section "Why raid 6" about multiple bad bocks spread across multiple disks.

 

http://www.intel.com/technology/magazine/computing/RAID-6-0505.htm

 

 

Anyway this is how LSI explains how to setup for raid 50

 

http://lsionline2.lsil.com/esupport/esupport/consumer/esupport.asp?id=5a01557e-05a7-49e9-8c63-1aa2d0468fed&resource=&number=1&isExternal=0&nShowFacts=&nShowCause=&nShowChange=&nShowAddInfo=&activepage=statement.asp&bForceMatch=&strCurrentSymptom=&searchtype=normal&searchclass=&bnewsession=false&selecttype=match

 

http://lsionline2.lsil.com/esupport/esupport/consumer/esupport.asp?id=71b3913d-3cf5-4718-9060-220ffc579701&resource=&number=0&isExternal=0&nShowFacts=&nShowCause=&nShowChange=&nShowAddInfo=&activepage=statement.asp&bForceMatch=False&strCurrentSymptom=&searchtype=normal&searchclass=QuickSearch&bnewsession=false&selecttype=match

Thanks all for your suggestions.

 

Yes, we have had 5 failures this year, on two db servers. By failure, I mean that the controller card has completely lost its configuration, and I arrive to find an the server beeping desperately.

I have replaced both of these controller cards after 3 failures, 2 on one and one on the other, and have experienced 2 further failures on one of the new cards.

We do not have any environment problems, redundant aircon, 3 ups and a generator, I am coming to the conclusion that I should switch to a different raid controller manufacturer, although I have several other Dell machines

with the same configuration which do not fail.

 

Once again thanks,

 

Adam

So you don't have a miraculous raid adapter.

 

If you read the Intel paper on raid 6, you will see the references to multiple disk failures which is what you have. As one disk go out, your in degraded mode, then an error is found on one or more of the remaining "good" disks resulting in a destroyed array.

 

The fun part of this is finding the failing disks. The worst case scenario, there are no errors in Windows or any management software referring to the disks with errors on them...suppose you have two disks which fail in an array, you replace the drives, and it continues to fail..... I have had arrays in which one of the remaining "good" disks causes the array to fail, but no testing software finds errors on any of the "good" disks (multiple software packages)...this is the ultimate nightmare. The only way to find this is to get a new disk and swap out any remaining "good" disks until the array stays up. I have tested some of these "good" disks with Seagate Enterprise tools, and other SCSI tests off a standard SCSI adapter and they have passed 100% on continuous testing over a full weeks time, place on a standard SCSI adapter these disks can function flawlessly for years. Had times when Dell offers to relace all the disks in the involved arrays, not the greatest answer as disks which are sent out are often recertified/recalibrated....often disks which have failed in array and returned.

 

Now, to combat this I have all my client's servers run Consistency Checks weekly or bi-monthly on off hours which is not always possible on large shop servers. Monthly is not sufficient, particularly with large capacity arrays. Until raid 6 is fully supported, you will see more of this issue.