Friday, December 22, 2006

When Good Drives Go Bad... News at 11

For the 3rd time in 2 weeks one of our RAID5 drives has gone bad. What's worse than being woken up at 5am to work on a degraded RAID5 array? Knowing that it was avoidable.

Upon closer research it turns out that the drives themselves need a new firmware installed in order to prevent timouts from marking the disk as failed. What's most frustrating is there was no way to tell if our drives were affected until they failed, but ( LARGE OEM ) had the records of the part numbers and should have been able to warn us in advance that we might be affected.

So now I'm forced to come in on the holiday weekend in order to bring down both file servers and boot them with the patch cd in order to "fix" the firmware problems. This is yeat another unseen issues with RAID5. Since the drives usually come in one batch if there is a problem with one the likleyhood is that it will affect all the drives increasing the likleyhood of a double disk failure in the whole array.

No comments: