[Linux-cluster] errors with GFS2 and DRBD. Please help..

[Koustubha Kale]

---- "Koustubha Kale" <koustubha_kale at yahoo.com> wrote:
| This is what I get on every run one one of the LV's i.e. Stu LV. On
| the other LV i.e. Fac, this new fsck_gfs2 did a bunch of stuff and
| exited normally.
| 
(snip)
| Starting pass4
| Found unlinked inode at 380614 (0x5cec6)
| Unlinked inode has zero size
| Block 380614 (0x5cec6) seems to be free space, but is marked as inode
| in the bitmap.
| The bitmap was fixed.
| Segmentation fault

>Hi,
>
>The fsck.gfs2 should not segfault like that; that's a bug.
>But now I'm confused.  Does the new fsck.gfs2 segfault as well?
>You said it did a bunch of stuff and exited normally.
>If the new fsck.gfs2 segfaults, I definitely want to get
>more info, possibly your metadata so I can recreate, find and fix
>the bug.

It was the new fsck.gfs2 which segfaulted. After that I could not mount the Stu LV. I had to do fsck with original fsck.gfs2 to get it to mount. The original fsck.gfs2 has never segfaulted on me so far.
As I said the new fsck.gfs2 completed the Fac LV properly but segfaulted on the Stu LV each time I tried it. I tried it on both nodes with segfault each time for the Stu LV. 
(Not surprisingly it's the Stu LV on which is the problem most of the time as I have mentioned in my earlier post.)
The Stu LV's metadata file is huge about  846Mb. Not sure how to give it to you. Any other info I might supply?

Regards
Koustubha Kale


      The INTERNET now has a personality. YOURS! See your Yahoo! Homepage. http://in.yahoo.com/