Question Locked

Compaq Proliant Server SCSI 'disk failure' light: Drives really dead?

0 helpful
8 replies
Closed to new repliesThis discussion is closed. It stays readable, but it takes no new replies.

I've got an ancient Proliant DL380 server and a stack of ten old hot-swappable 18.2Gig SCSI hard drives that all light up the 'disk failed' red light when you plug them into the server.

I have some drives that work just fine, but more 'failed' drives than good drives.

It seems to me that if you even look at one of these Compaq drives the wrong way, it lights the 'fail light'.

Anyone know what type of failure typically triggers the fail light?

Are these drives really dead?

Does this 'predictive failure' system result in false-failures? Like a bad sector or something that can be reversed by low-level format?

8 replies

by cg it

the only way to tell is to run diagnostics on the drives.

by robo_dev

Years ago, I had a problem where predictive-failure was hosed, such that it showed that all five drives in the array had failed at the same time (yikes!)

Compaq tech support walked me through hot-swapping in all five drives AFTER the array controller initialized. Something about the microcode on the drives getting corrupted....

I'm trying to figure out if some of these drives have just hiccupped, and are not really dead.

by cg it

the controller diagnostics.

SCSI are pretty dependable in that if they die, their pretty much dead with only a rebuild able to get to the data.

I've only had problems with terminators not doing what they are supposed to do, or is bad, thus the whole array looks fried.

If your hotswap backplane slots have problems, that can make drives look fried. or if you use a caddy and the caddy has connection problems, that can make the drive looked fried or not even power on thus look bad.

I'd first try the known good in all the slots and all the caddies and see if a backplane slot or caddy is bad.

Then go from there.

by robo_dev

the HP/Compaq predictive failure is all based on SMART, the low-level diags built into the drive. There are thresholds for 20 or so paramaters, e.g.:

Read Error Rate
Spin-Up Time
Recalibration Retries
Vibration During Write

Some utils let you view the smart data, but it is not meant to be edited/changed (like resetting the odometer on a car).

Some PCs let you ignore the smart data by shutting it off in bios, but most servers do not.

One potential issue is that SMART sometimes gives a false failure due to temperature, as it assumes that the drive getting too hot is a sign of imminent failure.

As a learning exercise I plan to do a post-mortem on the failed drives with a SMART data viewing utility to see what killed them.

by macaries2

Just curious for my stack of failed 36GB drives After using SMART did yours recover?

What software do you use? Compaq diagnostic cd?

by robo_dev

As I understand it, SMART data is designed to be Read-only and tamper-proof.

If I get some xtra time I may see if I can find a utility to modify the smart data.

I suspect that software to modify smart data may only work with special disk controller hardware since some special drive functions, such as drive password security, are not visible or modifiable thru standard methods or interfaces.

by cg it

in the SCSI controller utilities. You can with PATA/SATA in the motherboard BIOS. The SCSI controller health monitor or RAID array health monitor relies on SMART data then.... think your out of luck.

This discussion is closed to new replies.