Multiple HD crashes.. why?

[Vincent Deblois_1]

We got 3 ultra scsi 18gb hotswap HD running in a Proliant DL580 and over the last 2 days, 5 times we had to replace one of them. It wasnt the same slot everytime. What could possibly be the cause of these HD failures.

P.S. This thread has been moved from Disk to ProLiant Servers (ML,DL,SL). - Hp forum Moderator

[Simon Hargrave]

Most likely causes: -

Environment : is the aircon sufficient in the room where the server is? What temperature is the room running at?

Power : Possible power spikes. Is the server on a UPS, which will smooth out power surges.

Other : Do cleaners go in your server room? Could have a clumsy cleaner either knocking the server, or even (don't laught we've had this) unplugging the server to plug the vacuum cleaner in!

[Michael Lampi]

You do not indicate what prompted the replacement of these drives.

Were you getting read/write failures on them?

Were these configured as part of a RAID, and informed by the RAID s/w or h/w that a particular drive had failed?

Having a UPS will definitely make things more stable.

Making sure that your DL580's fans are all working, and the environment is adequately cool will also help.

Making sure that the system is not subject to excessive shock or vibration will also help to reduce drive problems.

[Marino Meloni_1]

Unfortunately I didn't know the DL580, but i will point to a bus problem, in fact the Raid HBA usualy inplement some diag routine based on the time the disk respond and the quality of data he receive (Parity check).
if something from one firve is not working well, the disk is put off line.
So as you have multiple off-line on several disk, the cable, teminator or eventualy if there is, the Hot swapp cage can cause this behaviour

[Vincent Deblois_1]

Yesterday we had the opportunity to stop the server and make a firmware and bios upgrade. Since then we had no HD failures at all so we're keeping our fingers crossed. Air con is more than ok and its not subject to vibration or anything because other servers (we have 30 something) would show similar problems. I'll let you all know how it turns out after a couple more days. Thanks for all the replys