Distro: Slackware 10.2
Drive: 3Ware RAID (5)
Partition: /dev/sda1 (ext3)

The system restarted due to a brown-out and tried to do a forced fsck. It gets to about 70% and resets. I booted with a boot CD and when I manually run fsck.ext3, it barfs with the following error:

"Error allocating inode bitmap (2): Memory allocation failed"<br>
"e2fsk aborted"

Any idea what this means and HOW TO FIX IT?

"Memory allocation failed" <--- I'd say that's a red flag, you might want to check your memory modules.
Got a Knoppix cd lying around? If you do you can easily issue 'memtest' at the boot prompt and you'll know very quickly whether your RAM is bad or not.

Doesn't seem to be a RAM problem. Other ideas?

Is that '3Ware RAID' a RAID controller card or an on-board controller?
In case it's a card, I don't suppose you have a spare to try out in it's place
Does the RAID BIOS have any on-board diagnostics? What state is the RAID set reported as? (Functional, Degraded, etc.) Is there a logging function on the 3Ware?

There's also the S.M.A.R.T. Monitor Tools (Also available readily installed on a Knoppix cd), I believe these should work on the disk(s) attached to the controller, you could check the disk(s) for errors that way to see if there is physical damage, but it won't help sort out the filesystem in case it's still intact, we'll need to find another way around that one.
Any chance you can get to the logs on your machine? perhaps something was written there that might help locate the real problem.

It's a 3ware 7506-4LP card.


I wish!


Not really. It has a "MAINTAIN ARRAY" and a "REBUILD ARRAY" function but, both are acting squirrely. The 3ware status screen shows everything is fine... When I use MAINATIN ARRAY (with VERIFY ARRAY checked), it reports that the array is degraded an cannot continue. When I run MAINTAIN ARRAY without VERIFY ARRAY checked, it shows no problem. I tried REBUILDING ARRAY but it complains that the array is not degraded and cannot continue.


The main status page shows no problems but, the MAINTAIN ARRAY VERIFY function reports that it is degraded (with no further details as to the problem).


There is but you need to boot the thing into DOS and then run 3ware's program (which they were supposed to send me but as of yet, hasn't) which will pull the logs off of the card.


I don't think the SMART tools will work since these discs are in an array through a RAID card.



I have the machine booting. I just booted off of my Slackware disk and then set the Mount Count up to 50. It boots and runs fine now with only a few quirks. The logs show squat.


The problem I have it that I REALLY don't want to have to wipe the whole server and reload everything if I can fix the disk errors.

I ran fsck.ext3 last night and checked for bad blocks and found none. When I run fsck.ext3 in forced check, I get the:

Error allocating inode bitmap (2): Memory allocation failed<br>
e2fsk aborted

Which is what I need to fix...

That's really bothersome indeed. Those logs from the card itself would be really interesting right about now, since you mention the card acts quirky when checking/rebuilding the array. If the card itself can't determine correctly if the array is functional or degraded i'm starting to suspect the card itself is dying rather than the attached disks.
Pity that 3Ware tool for reading the logs isn't up for download :/

Okay, it's fixed.

1) I booted off the Slackware CD.
2) I ran fsck.ext3 -f (forced mode) and it failed.
3) I tried running fsck.ext3 -p (preen mode) and it failed.
4) So, out of desperation, I ran it in -c ('read-only but check for bad blocks' mode) and it ran all night and completed successfully with no errors.


After the sucessful fsck.ext3 -c run, it caused two nice things:
1) It reset the mount count and doesn't force fsck on start-up any more.
2) The 3ware raid will now actually do a raid verification without barfing!

Why this is is beyond me!