Raid Problem Fedora Core 3, RAID LOST DISKS ALWAYS

[icatalan]

Hi all,

I've a big problem with RAID1 and RAID5 in a FEDORA CORE 3 server. We have a server with a SuperMicro Serverboard X5DLP-TGM with a Marvel Sata controller and 8 Maxtor SATA HD with 300GB. We installed this distro in a ATA HD, and followed the instructions of the web: http://www.keffective.com/mvsata/ to add support for the SATA disks.

Now we are experiencing lots of problems with the disks. Often we can see with the command: /sbin/mdadm --detail /dev/md0, that 1 disk has been lost of the raid and appears with state REMOVED. To fix this I use /sbin/raidahotadd /dev/md0 /dev/sdb and in two days we return at the same situation.

Thanks for all your comments!!

Isidro

[CrystalCowboy]

It sounds like you are using software RAID, is that correct?

Possibly the drives are encountering & mapping out bad blocks, which takes a bit longer and causes the controller to time out. Once the bad block is mapped out though, the drive is fine.

Which specific Maxtor drive are you using, and what is the MTBF (mean time between failure, listed in the manufacturer's specs). Is this an 'enterprise' class drive (MTBF > 1,000,000 hours) or a 'desktop' drive?

I built some ATA RAID units a couple years ago using Maxtor 536DX drives and it was a big mistake. The drives were simply not up to the job. For SATA RAID use I would recommend the Seagate Barracuda drives or the Western Digital RE2 (RAID Edition) drives.

Does your RAID configuration have hot spare drives? That way when a disk failed, the hot spare could be put into use automatically.

Is it always the same drive that fails? If so, replace it!

Please, more details on the specifics of your RAID configuration. Are the RAID5 and RAID1 sets separate, or are you attempting a two tier RAID config? How many drives in each set? Two tier RAID would mean you probably don't have spares.