[Linux-cluster] Failed gfs_grow causing corrupt volume

[Bob Peterson]

On Fri, 2008-01-25 at 09:56 -0500, Steffen Plotner wrote:
> My question would be, if Bob had done a gfs_fsck before attempting to
> grow the gfs space, what would that have returned? Would that have
> prevented the gfs_grow issue?
> 
> Steffen

Hi Steffen,

I don't think gfs_fsck would have found any problems prior to the
gfs_grow.  The gfs_fsck can handle a gfs file system extended by
gfs_grow.  In fact, I put gfs_fsck through hell by doing different
combinations of gfs_grows and forcing different kinds of corruption.
In fact, to stress these different situations, I developed an elaborate
test that does many different combinations of:

  1. gfs_mkfs
  2. gfs_grow
  3. gfs_grow again sometimes
  4. deliberately damage different parts of the RGs and/or rg index.
  5. gfs_fsck to repair the damage

The problem, I think, is that the gfs_grow failed and left things in
a bad state.

Regards,

Bob Peterson
Red Hat GFS