Hardware

Ask something
Discussion Locked

Mirror madness?

0 helpful
4 replies

by choppit

Closed to new repliesThis discussion is closed. It stays readable, but it takes no new replies.
I had an incident recently involving failure of a drive in a RAID 1 array. The failed drive was replaced, the rebuild started but failed at 99%. I narrowed this down to a number of corrupt files on the shadow copy , removed the offending files, replaced and rebuilt the offending drives. On looking at the RAID controller documentation (Storageworks RAID 230) it appears that drives are not failed on soft errors.Surely it's better to mirror the 99.99% of good files than refusing to rebuild the mirror due to a handful of(in this case expendable)corrupt files?

4 replies

by grandhighdruid

The problem I forsee with it going ahead and finishing the mirror, is that despite all the wonderous and amazing advances with software in recent years is that even though they may put in a list of files so that it 'KNOWS' what it absolutely needs and what it can live without, software vendors tend to err on the side of caution. This means that they don't want to get angry phone calls and having customers accusing them of their software deleting their contact database in the customer's old, archaic software just because the vendor's software didn't recognize it as vital. You get the idea. Hope this helps!!

by choppit

I think you've misunderstood my point. Consider a mirrored volume, lets call the two drives A and B. The contents of B should always be the same as A. Lets assume the volume contains critical and non-critical data.

Scenario: A non-critical file becomes corrupt on the volume, and there is no warning that this has occurred. Weeks or months go by, meaning that the original uncorrupted file is no longer present on the backup set.

Drive B then fails and a replacement drive is hot swapped. The corrupted file prevents the mirror from rebuilding, leaving drive A on its own and in an unknown state (maybe the file was corrupted due to bad sectors). Until the offending file is removed from drive A, drive B will not rebuild.

Now, supposing drive A fails before I can resolve the problem.....the volume is now gone and the OS didn't give me any warnings.

After doing some digging around I found an article which suggests that this particular controller never communicates with the OS regarding soft errors.

I don't know about you, but I'd rather have 99% of my files on a repalced drive than none at all.

This discussion is closed to new replies.