[Linux-cluster] Journal 0 locked on GFS2? gfs2_fsck gives no results!

Bob Peterson rpeterso at redhat.com
Thu Jul 24 13:50:55 UTC 2008

On Thu, 2008-07-24 at 15:44 +0300, Theophanis Kontogiannis wrote:
> GFS2: fsid=tweety:gfs2-00.0: fatal: filesystem consistency error
> GFS2: fsid=tweety:gfs2-00.0:   inode = 4 25
> GFS2: fsid=tweety:gfs2-00.0:   function = jhead_scan, file =
> fs/gfs2/recovery.c, line = 239

Hi Theophanis,

I haven't seen this error before.  It indicates a bad entry in the
first journal.  The gfs2_fsck program rejected it for the same
reason that the GFS2 file system rejected it.

I've been doing a lot of work on gfs2_fsck this week, so it would
be an interesting for me to get a copy of your file system
metadata (not any of the data) and run it through my latest fsck
on one of my test systems.

I'd also kind of like to examine the journal to see what's wrong
with it and possibly give gfs2_fsck the ability to repair the damage.
I can't make any promises though.

If you're interested in doing this, run this command:

gfs2_edit savemeta /dev/vg0/data0 /tmp/theophanis.metadata
bzip2 /tmp/theophanis.metadata

Then put the resulting .bz2 file on a server where I can get it.
You can try this command on the pre-existing gfs2_edit program,
but it might not save all of the metadata I need.  I don't know
how "up to date" Centos is in regards to gfs2_edit.  You can also
download the latest cluster git tree from source code, compile
it, and run the latest version to make sure I get everything.

If you're not willing to send me your metadata, you could run
this command and email the output:

gfs2_edit -p journal0 /dev/vg0/data0 > /tmp/journal0.txt

Then I could at least try to determine what's wrong with the
bad journal.


Bob Peterson
Red Hat Clustering & GFS

More information about the Linux-cluster mailing list