2

I am trying to fix a 9 TB filesystem with e2fsck, however the scan always fails after about 10% or so with the following error:

Error storing directory block information: ... Memory allocation failed
e2fsck: aborted

The machine runs Ubuntu 10 (64 bit) and has 4 GB of memory. I have noticed that during checks the ram is getting nearly exhausted even though top only reports only 22% usage by the application and the memory can be reclaimed by a drop_caches. The swap space isn't touched.

In any case, I have approached this by giving e2fsck a scratch_files location which it fills with about 500M by the time it fails. Additionally, I have also set swapon -a, which also seem to prolong e2fsck's crashes.

How can I tune e2fsck so it doesn't crash? Is there any better way to repair my fs?

Christian
  • 411
  • 2
  • 7
  • hmm... I have not tried from a boot cd yet... i'll have to give this a go today... i was hoping to not have to take the machine out of production since the drive i'm trying to e2fsck isn't critical for the servers use. But at this point i can't keep eating up all available memory each time I try to run e2fsck so whats the difference? lol. I'll report back any findings. – Jason Jul 11 '11 at 16:51
  • Maybe a duplicate of http://serverfault.com/questions/9218/running-out-of-memory-running-fsck-on-large-filesystems ? – Giovanni Toraldo Nov 12 '11 at 15:57

1 Answers1

1

Sounds like you're running e2fsck from the local OS?

I don't know if it would make a difference or not, but what happens if you try from a LiveCD? Some have a lomem boot option, have you looked into that?

For lack of a better idea as of yet, is that an option?

NginUS
  • 468
  • 1
  • 5
  • 13