mdadm Raid 5 hang in Fedora 8 / Ubuntu 7.10
 

Come across an intriguing problem, was hoping someone could shed light on the situation. I recently converted my desktop system into a server running Fedora 8 (I have another Ubuntu 7.10 install on another HD.)

AMD X2 4400+, 3GB ram, Nvidia 7800GT, 430W PSU, 8x320GB HDD, 1x150GB Raptor, 1x120GB IDE and 1xDVD drive.

The 8x320 drives are configured in a Raid 5 array I built/expanded from a Raid 0 4x320, since I wanted to add redundancy. It worked perfectly until tonight; after a system reboot, I attempted to reload the array with

`sudo mdadm --assemble /dev/md0`; md0 being the array configured in the /etc/mdadm.conf file using a UUID. System hangs around 5 seconds after this command executes, telling me that 7 of the drives are now active and 1 is running as a spare.

I systematically reboot and retry this operation, unplugging a different harddrive each time until I find the spare. Lo and behold, the array constructs, and I can mount it correctly. This leads me to believe the problem is in the HDD that I pulled.

I format the HDD using `sudo mkfs.ext3 /dev/sdj,` and it formats correctly, indicating that there is nothing wrong with the drive. I then try to re-add it to the array using `sudo mdadm /dev/md0 --add /dev/sdj`; the system adds the drive, and predicting the freeze I execute `sudo mdadm --detail /dev/md0,` which informs me that the spare was activated and is rebuilding. However, the system, as I predicted, hangs.

My question to the community is this: why would a healthy, new disk, that was running the night before, suddenly kill my RAID? At one point i even received a kernel panic, indicating a fault in the processor (which is false.) Do I need to reconstruct my entire machine just to add this last drive?

Thanks.

That is a lot of load for a 430 watt PSU. The PSU may be getting old (they loose capacity as they age) and is just now starting to show problems.

Well, I considered that might be the problem; however, like I said, the configuration was working a few days ago, I can do read-write to the drive itself (wrote zeros w/ dd if=/dev/zero, formatted it xfs, mounted it, and put stuff on it with no problems), and it is powered up even when it isn't added to the array. I'm going to reload F8, then RMA the disk if that doesn't fix it. If all else fails, I'll upgrade the PSU; just getting a bit expensive.

Thanks for the input. If anyone else has any ideas, I'd be glad to hear them.