Why did the RAID controller miss a disk error?

T

Throwable2012-06-13 11:52:48

linux

Throwable, 2012-06-13 11:52:48

Good afternoon!

Some time ago, we bought a Dell server for a project with two disks combined in RAID-1. The disks were small at 250G, so more space was needed.
In order not to buy expensive drives from Dell, we made one more additional RAID-1 from two regular 1T drives that we bought in a store. We keep non-critical information on it - logs, statistics, backups.

Yesterday one of the "cheap" drives failed and the RAID controller now only works with one drive. However, at the time of the crash, a bug infiltrated the OS and crashed the file system. Linux, as expected, remounted this partition to read-only, which caused the service that writes logs there to crash. The question is why did this happen? Shouldn't the RAID controller detach drives transparently to the system? I remembered on Habré about the articleabout SCT. Is it possible that the drives had SCT enabled?

PS OC Ubuntu 10.04 LTS, PERC 5/i controller, Seagate drives.

Reply

Answer the question

In order to leave comments, you need to log in

4 answer(s)

G

grossws, 2012-06-13
@grossws

Are the disks desktop or at least raid-edition? View the options with smartctl -l scterc /dev/sdX.

A

AlekseyPolyakov, 2012-06-13
@AlekseyPolyakov

Your situation is more like a problem with controllers. Are you sure that the designated hard drive is really out of order?
Or a variant of a specific failure of the hard drive comes to mind, potentially it could somehow break down in a non-standard way, as a result of which the controller logic temporarily hung up or went into a non-standard state.

C

cepera_ang, 2012-06-14
@cepera_ang

It’s not really clear from the question anything, except that you stepped on a typical rake - for some reason you took branded hardware and stuffed components that were not intended for this. Well, it’s also not very clear on the problem itself - what kind of error arose with the disk? What exactly and how "penetrated" into the OS? What kind of file system failure has occurred?

C

cepera_ang, 2012-06-14
@cepera_ang

It’s not really clear from the question anything, except that you stepped on a typical rake - for some reason you took branded hardware and stuffed components that were not intended for this. Well, it’s also not very clear on the problem itself - what kind of error arose with the disk? What exactly and how "penetrated" into the OS? What kind of file system failure has occurred?