Business Copy fsck Problem on Backup Server

[David Bosiljevac]

Hi,

I've been trying to mount some business copies of volumes on our backup server. I'm having a problem with one particular logical volume, though. The business copy contains volume groups 01 through 04, with a bunch of logical volumes in each volume group. The fsck runs on every single logical volume except for the last one. When the command:

/sbin/fsck -y -F vxfs -o full /dev/_vg04/lvol4

... is run, I get the following error:

log replay in progress
vxfs fsck: malloc failed!, add more swap space
file system check failure, aborting ...

... which means we can't actually get the business copies mounted in order to do a backup of them.

I've checked, and the memory utilization is very low when running the fsck. In fact, none of the swap is actually used when the command is run.

Can anybody shed some light onto why this might be happening?

Thanks in advance,

--db

[A. Clay Stephenson]

You probably need to increase maxdsiz and/or its 64-bit cousin maxdsiz_64bit.

[Tom Danzig]

I agree with Clay. The fsck process itself cannot allocate sufficient memory. Increase the maxdsiz in the kernel.

This file system most likely has a lot more files in it than the others that were successful and hence fsck requires more memory to run.

[David Bosiljevac]

I've determined that the kernel parameters you suggested (mazdsiz and mazdsiz_64bit) are currently set to 67108864 and 1073741824, respectively.

So you think these should be raised a bit more in order to get fsck to work on that filesystem?

Thanks,

--db

[Emmanuelle]

hi,

fsck -F vxfs -o full ....
malloc failed! add more swap space

swapinfo -t .... is ok

increased maxdsiz from 64mb to 128mb solved the problem

bye
Emmanuelle

[Paula J Frazer-Campbell]

David

Are you sure the filesystem is vxfs?

What error does the mount command give you?

Try unmounting all the other and then mounting vgo4/lvol4

Have you tried a reboot?

Paula