my raid went critical: no sleep tonight

[VJ]

Typical.
At 23.33, my pc started beeping: a high tone, a low tone, a high tone, etc. I noticed the lights of my raid system blinking in an abnormal way, so I launched the web interface for the raid (pc kept beeping). I managed to disable the beeping in it.

As it turns out, for some reason it considered my drive 2 to have been unplugged: I suspect I may have touched a sata cable while checking my vga cooler (see GH), but all cables did appear to still be connected . The raid controller has now begun rebuilding the raid5, but as my drive currently needs to be accessed, it is going slowly. Still, I hope it will be ready in an hour or so, so I can get to bed...

AAAAAAAAARRRRRRRRRGGGGGHHHHHHHHH!

Sorry, needed to get that out of my system.


Jörg

[VJ]

It is now 1:40 and it is at 50 percent; just about an hour to go.
I can pause rebuilding the raid, but I' rather have everything in order by the next time I want to use the pc.

(it seems calm in here, I wonder why... )

Jörg

[VJ]

Guess my time estimate was a tad off: it took till 3:40...

To make matters worse, this morning, my WM5 device (which I also use for alarm clock) decided not to give me a "dismiss" option when the alarm went off. So I couldn't stop it! Even after a reboot, it just started with the alarm sound again.

Is this the rise of the machines (terminator 3)?


Jörg

[KeiFront]

Try rebuilding a 1.5tb array. It took 14 hours (2 weeks ago we had a disk failure on a NAS server)

[VJ]

Why does rebuilding a raid5 has to take so long?
All it has to do is XOR the data from the working disks, and write this value to the disk that is being build. I honestly don't understand the problem...

Either way, I'll probabely have to rebuild it again tonight: the failure may have been caused by a sub-optimal (or non-identeical) settings of the harddisks. To change these settings, I can use SeaTools Enterprise, but this software cannot access a hardisk that is in a raid. So I'll copy all the data off the raid to another drive (have sufficient spare space), kill the raid5, adjust the settings for all drives, re-create the raid and copy the data back to it... Part of me says I should start creating the raid tomorrow morning rather than in the evening....


Jörg

[VJ]

what is a raid???
is it just a bunch of hard drives stuck together or something?

Yup.
Can be done for different reasons: performance, single volume size, redundancy, ...
Most common are:
RAID0: striping, 2 disks that act as one logical drive (both hold different data, no redundancy)
RAID1: mirroring, 2 disks that act as one logical drive (both hold exactly the same data, if one fails, data is not lost)
RAID5: three or more disks, presented as one logical drive, but with redundancy checks (if one disk crashes, the system can keep running; there is no dataloss and if the RAID is hotswap you can simply plug in replace the faulty disk and the RAID ought to rebuild itself. You have the storage space of the added capacity of all the disks minus 1 (considering they are all the same size).

For more details:

http://en.wikipedia.org/wiki/RAID

In my case, it is a hardware SATA RAID5 with 3 400 GB disks (so I have 800 GB of fault tolerant storage).

The problem is that the rebuilding process (when a disk fails, or is thought to have failed and is replaced) is a slow process.


Jörg