serious share on RAID issue

[manxdan]

currently restoring from backup . . . again.

We have a Promise RAID with two partitions providing volumes for student network homes (a2k & l2z).

These are connected via Fibrechannel to two xserves running OSX Server 10.6 for AFP

We are repeatedly having problems with just one of these.

The failure seems to occur gradually over a few days starting with users experiencing slow access, later they lose write access to their own directories, and finally the whole volume becomes un-mountable.

Disk Utility find some problems and makes some repairs with varying degrees of success, but ultimately fails while:

"Updating boot support partitions for the volume as required"

Disk Warrior one one occasion managed to rebuild the directory, but it was still unusable. generally Disk Warrior reports:

"the directory cannot be rebuilt due to a disk hardware failure"

The RAID's own management software reports everything is A ok hardware wise.

Erasing and restoring the volume from backup gets us running again- but not for long it would appear.

The xserve in question is the older of the 2

anybody have any ideas where the problem may lie?

many thanks

[DMcCoy]

You don't have two servers accessing a non cluster file system on the same direct attached storage box do you? Either both connections should be plugged into the same server or you need a cluster FS.

[manxdan]

You don't have two servers accessing a non cluster file system on the same direct attached storage box do you? Either both connections should be plugged into the same server or you need a cluster FS.

 

no there's two independent RAID controllers - It's two separate RAIDs just in the same chassis.

[DMcCoy]

Is it a vtrak? I assume the LUNs are masked so each server can only see one of the LUNs