RAID 5 on Windows 2K SP4

[jshepherd]

I know this is really a hardware question, but I'm sure someone can shed some light. I've just reconfigured an old Dell 2400 with Windows 2K SP4. I'm running a standard PERC 2/si RAID with updated BIOS firmware and Windows drivers. When the computer is running, only two disks appear to have their activity lights on when data is writen (all have their power lights on) but the RAID does appear to be configured correctly as Windows is reporting a container of 72GB where each of the five drives is 18GB. However, on tests, I can remove any of the three drives (one at a time only) who's activity lights don't appear to be lighting up and the system carries on as normal. When I re-add the removed drive, the system appears to write the data back to the drive (with activity light). If I remove either of the two drives those activity lights do appear to light up all the time, I get Windows disk write system errors. Anyone have any experience with actually testing a RAID 5 array?

John

 

[markdmac]

Well, just by definition your removing the disk shows that the RAID is working.

Question of course is have you tried removing each of the drives? Normally you should see some activity on each drive unless it is a hot spare.

Since you have a PERC2 I assume this is a DELL system? Make sure you have loaded Dell Array Manager and check in there. It will give you the status of each disk in the array.

I hope you find this post helpful.

Regards,

Mark

 

[Davetoo]

It's a Dell 2400 ;-)

I'm Certifiable, not cert-ified.
It just means my answers are from experience, not a book.

 

[jshepherd]

Hi Mark,

Thanks for taking the time to answer my query. I agree that by removing the a drive shows that they are probably working, but I should also be able to remove all the drives (individually) without causing any data loss or Windows write errors. I hadn't installed the Array manager, but am downloading that now and will install. The other strange thing that happened is when I set the drives up in the BIOS. Normally when I've done this before, the BIOS takes some time to prepare the array (this can sometimes take two hours or so). However, this time, the array status was <ok> immediately. I'll try to the Array manager and see what that shows, but I also feel it's worth rebuilding the array again and testing it then. It just that it worries me about two other primary servers I have that run RAID 5 and whether they will actually work should a drive fail. I don’t think we test this technology enough and that we just take it for granted Hopefully others reading this may want to check their critical servers to see if they actually work!

John

 

[markdmac]

The array manager is the best way to manage this. It will let you see under the hood as well as repair problems.

I hope you find this post helpful.

Regards,

Mark

 

[jshepherd]

Hi Mark,

Just thought I'd let you know I found the problem. It was very interesting, but it appeared to be a problem with the back-plane in the server. I was using slots no’s 0,1,2,3,4 while 5 was empty. The problems encountered were with drives in slots 0 & 2.

I moved the drives around and the same problem existed with the different drives in slots 0 & 2 which pointed to the back-plane and not the drives. I removed drives from slots 0 & 2 and added just one of them to slot 5. This worked a treat, so I changed the drives back to their original locations and the all worked perfectly.

It just goes to show we shouldn’t too readily accept RAID configurations without testing – I certainly won’t.

Thanks for your help.

John

 

[markdmac]

Sounds like a drive was just not seated properly then, but I would contact Dell and get them to replace the backplane just in case.

I hope you find this post helpful.

Regards,

Mark

 

[Freestone]

If two of the five drives were not functional in slots 0 and 2, then by definition your RAID5 was broke. Have you checked your current configuration with Array Manager as has been suggested to ensure you truly are configured for RAID5?