Raid Wont Rebuild, One Core Overloading

[Crushdepth]

Got some strange problems with my PC today:

i) My RAID 1 is degraded, one of the member disks says 'error occurred'. This actually happens quite often (usually if shutdown is not clean) but this time when I reset the affected disk it didn't rebuild. So I replaced the affected disk with a spare, and the RAID still won't rebuild. Usually while rebuilding the drive light comes on continuously for an hour or so when rebuilding starts, but this time it isn't. The drives are also making a faint, repetetive click every 3-4 seconds.

ii) My computer keeps locking up for a few seconds at a time. When I look in the performance monitor, one of the cores is repeatedly hitting 100% every 3-4 seconds in a consistent cycle. The other appears to be working normally. However, none of my applications appear to be using much processor time, so I don't know where all this 'work' is coming from.

I have a feeling that these two problems are related since they both arrived together, but no idea what is the cause. I ran spinrite in maintenance mode about 2 weeks ago and no problems were reported with the disks. I have a disk image from last week, before anything went wrong, but I'm not sure if this is a hardware or software problem.

Anyone got any ideas?

[dave_boo]

Got some strange problems with my PC today:

i) My RAID 1 is degraded, one of the member disks says 'error occurred'. This actually happens quite often (usually if shutdown is not clean) but this time when I reset the affected disk it didn't rebuild. So I replaced the affected disk with a spare, and the RAID still won't rebuild. Usually while rebuilding the drive light comes on continuously for an hour or so when rebuilding starts, but this time it isn't. The drives are also making a faint, repetetive click every 3-4 seconds.

It sounds like your other drive has taken a crap also. Sorry about that, but if you have been using it for a while, especially rebuilding arrays from it, that's not terribly surprising. As a sidenote, they're not Maxtors are they?

ii) My computer keeps locking up for a few seconds at a time. When I look in the performance monitor, one of the cores is repeatedly hitting 100% every 3-4 seconds in a consistent cycle. The other appears to be working normally. However, none of my applications appear to be using much processor time, so I don't know where all this 'work' is coming from.

Assuming you're using Windows, open the processes tab of Task Manager and click on the "CPU" part of the taskbar. This sorts the columns by cpu usage. Hopefull that will make something rise to the top.

I have a feeling that these two problems are related since they both arrived together, but no idea what is the cause. I ran spinrite in maintenance mode about 2 weeks ago and no problems were reported with the disks. I have a disk image from last week, before anything went wrong, but I'm not sure if this is a hardware or software problem.

They could be related if it's an onboard controller trying to read bad sectors/rebuild your array.

Anyone got any ideas?

[dsys]

agree the other disk has gone wandering as well.

Only one core overloaded. Are we talking dual(multi) core or a multiprocessor system here.

Have you set the affinity on some app only to one proc/core? I remember a problem on 64 systems where Raid1 controiller fail caused a similar fault.

Due to the way the RAID code was inplemented in 2003 servers there was a problem with affintiy when the controller took a dive. Can't for the life of me find it in my notes though. Guess if you are not using 2003/64 then its not that. If you are I'd try swapping the controllers out.

[Crushdepth]

It's a dual core, motherboard is an Asus P5W-DH Deluxe. I'm using the built in raid functionality of the board. I haven't messed with core affinities, and looking at the CPU usage doesn't show anything using much resources. The drives are Seagate 320GB. I'm pretty disappointed with the reliability of the built in raid, I've had it in raid 0,1 and 5 configurations and anytime vista hangs or the power goes out it degrades. I think I was happier with an ordinary disk and regular external backups!

Update: I tried reattaching the power/SATA cables to the drives and now the clicking and CPU usage seems to be back to normal, but it still looks like the RAID isn't rebuilding. Grrr...

Edited February 3, 2008 by Crushdepth

[TopDogger]

Update: I tried reattaching the power/SATA cables to the drives and now the clicking and CPU usage seems to be back to normal, but it still looks like the RAID isn't rebuilding. Grrr...

Have you tried re-building it manually? I've been using 1 & 0+1 on Asus for years and have never had a problem re-building or with crashes effecting the RAIDs integrity.

Edited February 3, 2008 by TopDogger

[dsys]

Update: I tried reattaching the power/SATA cables to the drives and now the clicking and CPU usage seems to be back to normal, but it still looks like the RAID isn't rebuilding. Grrr...

Have you tried re-building it manually? I've been using 1 & 0+1 on Asus for years and have never had a problem re-building or with crashes effecting the RAIDs integrity.

I'd go with a manual rebuild on this as well.

Very interesting problem with the processor , (sad techy comment coming) wish I was there to have a look at this one. If you find out what was causing it could you post the details, please keep us posted either way.

[dave_boo]

I really don't want to slam Windows, but if you had been using Linux's built in software RAID, you'd have been pleasently surprised at the results. See for example this graph:

(Granted I took the most favorable one from Linux.com, but even the other ones weren't too shabby for RAID1).

Another thing you might want to do is make sure you have write caching correctly set. Doing it correctly will be a bit slower (however on a RAID-1 array that really shouldn't be a problem), but will be so much safer.

[Crushdepth]

Thanks all. After a few more reboots the raid finally rebuilt and now appears to be working normally. Still no idea what went wrong though.