Re: [ext2resize] Load testing: ext2resize complains about a bug

[Andreas D. <ad...@cl...>]

On Dec 09, 2002  15:01 -0500, Brian J. Murrell wrote:
> On Mon, Dec 09, 2002 at 11:33:51AM -0800, Poul Petersen wrote:
> > However, the ext3 file-system stopped extending at about 16GB.

You probably should have gotten messages to the console or syslog
saying that the filesystem couldn't be expanded further...

> Did you "prepare" the filesystem after creating it initially?  By
> default, there is only enough space in the block bitmaps to support
> 16GB worth of blocks (assuming 4k blocks).  "Preparing" (man
> ext2prepare) allocates more block bitmaps to support larger number of
> blocks in the filesystem.

Sadly, ext2prepare does not work with the new ext3 code.  I am such a
procrastinator, shoot me.  We do have a patch for mke2fs which will
prepare a new filesystem for a 1000x growth from the initial size, but
we _really_ need ext2prepare for the new ext3 format to handle existing
filesystems also.

> > [root@nfs1node1 ext2resize-1.1.18_pre]# ext2resize /dev/vg00/sa_test
> > ext2resize v1.1.18 - 2001/03/18 for EXT2FS 0.5b
> > BUG! block 32782 shouldn't have been marked!
> 
> Now this, this does not look good.  I will have to leave this up to a
> real expert (I am just playing one in this episode :-) to answer.

Not sure about the source of this, but in general such an error will
not cause any filesystem corruption, just prevent you from resizing...
Could you run "ext2resize -v -d /dev/vg00/sa_test" and send that +
"dumpe2fs -h /dev/vg00/sa_test" to me and Robert Walsh
<rj...@us...>?

Cheers, Andreas
--
Andreas Dilger
http://sourceforge.net/projects/ext2resize/
http://www-mddsp.enel.ucalgary.ca/People/adilger/