Predictive failure/online spare

[Andrew Horne]

Hi,
We have a disk showing a predictive failure. It is a disk in a 5 disk raid 5 array in a DL380. There is also an online spare in the DL380.

The documented advice from HP is to replace the disk as soon as possible. As there is an online spare I would have expected behaviour to be that the array controller would start duplicating the prefailed disk on to the spare. It is not possible to manually trigger a rebuild to the spare and given that it has not started automatically rebuilding to the spare disk, is the problem:

A - I am mistaken in my understanding.
B - There is a problem with the array controller.
C - There is a problem with the online spare.
D - Something else.

If A then why bother to predict failure, the feature becomes a bit of a 'chocolate teapot'!

Regards,

Andy

[Torsten.]

This is a predictive alert, but the disk is still working. You can either wait until the disk is dead (maybe not the best idea) or pull the disk, wait for the rebuilt, insert the new disk.

[Andrew Horne]

Hi,
Thanks for the reply, my point was - shouldn't it automatically pre-fail to the online spare, this would be more elegeant than waiting for full failure or manual intervention.

Are you saying that from your experience that predictive failure does not automatically rebuild to an online spare?

I know that full failure would cause immediate rebuild, but while the disk is still running it would be less hazardous and and less stressful on the server to start the rebuild in advance. Having not seen predictive failure on a server with an online spare available I thought that it would be a more intelligent process.


Cheers,

Andy

[Uwe Zessin]

"predictive failure" is a misnomer. It just means that the disk does no longer conform to some parameters the vendor has specified - it could fail the next minute or it _could_ keep running for _years_.

We have also seen the other extreme: too narrow parameters and disk drives get declared "failed" although they are working fine.