NAS1200s - RAID disk errors/need to reactivate

[IDO_2]

Hello all,

we use a hp nas1200s (windows storage server 2003) with 4 maxtor ide hdd (hotplug), dynamic disks RAID5 for data, and mirroring for system (pre-configured and installed, we changed nothing in partitioning), Gigabit Ethernet connection.
Some of the disks frequently go into error state, a warning sign appears in the logical disk manager, and we have to reactive the failed disks... it works for a while then goes into error state again. Sometimes we have to stop the machine, extract the disks, reinsert them and restart it. Could there be a problem with hot plug/software raid support of Windows server 2003?

What is strange is that it is the second server of this type we have and the same problems occur (we initially purchased a 1000s, hp intervened onsite and replaced the 1000s with a 1200s, we copied our data onto the new server). We first thought it might be a problem due to excessive heat, but we allow more air flow around the server and it does not seem to make a difference (the server is in a wooden case open on the front and the back).

Has anyone encountered this kind of problems with Windows RAID5/mirroring?

Thanks for your help and best regards
IDO

[Greg Carlson]

IDO,

What kind of work load is being placed on this NAS? It sounds like you have a failed or failing hdd. Keep in mind these were designed for small business usage with about 8 hours use per day and if they are used more intensively you'll see a higher failure rate of hdds.

Ciao,
Greg

[Clemens Sibon]

Hi there,

Last year we have had numerous Maxtor disk failures in other systems and therefore we swapped all the Maxtor disks in our NAS 1200s and changed them for Hitachi I think.

You can do this without interrupting the system . To do this, swap 1 disk, repair the software RAID volumes (the mirrored boot partitions and the RAID 5 partitions) and after that is finished (will take quite a while) you can swap the next and so on.

When the system is under heavy load, we do see mesages in the system log about delaying of non-paging IO requests but hey, it's software RAID so you get what you pay for.