xfs
[Top] [All Lists]

Re: [Fwd: [Bug 150427] New: XFS internal error xfs_alloc_read_agf]

To: "Nathan Scott" <nathans@xxxxxxx>
Subject: Re: [Fwd: [Bug 150427] New: XFS internal error xfs_alloc_read_agf]
From: "Daniel Tschan" <tschan+sgi.com@xxxxxxxxxx>
Date: Thu, 10 Mar 2005 09:50:34 +0100 (CET)
Cc: linux-xfs@xxxxxxxxxxx
Importance: Normal
In-reply-to: <20050310003246.GB14665@frodo>
References: <33504.217.162.2.34.1110409767.squirrel@217.162.2.34> <20050310003246.GB14665@frodo>
Sender: linux-xfs-bounce@xxxxxxxxxxx
User-agent: SquirrelMail/1.4.3a-0.f0.9.1.legacy
Hi Nathan

> Did you build that xfs_repair/xfsprogs yourself or is that one
> shipped by Redhat?  It has debugging enabled, hence the assert,
> hence its not running to completion - if you build a non-debug
> version it should complete.
>
> If Redhat's shipping a debug version, they'll need to fix that.
This is xfsprogs 2.6.13 included wih Fedora Core 3. Building with
debugging enabled seems to be the default. I'll file another Red Hat bug
report.

I now rebuilt the rpm with:
make
changed to
make OPTIMIZER=-O3 DEBUG=-DNDEBUG

The assertions are gone but xfs_repair still isn't able to repair the
filesystem. It now aborts with:

corrupt dinode 651602685, extent total = 1, nblocks = 0.  Unmount and run
xfs_repair.
fatal error -- couldn't map inode 651602685, err = 990

A second run reported even more problems but in the end aborted with the
same fatal error.

I already rescued all data so I don't really care what happens with this
filesystem. But I guess you're interested in the results.

Kind regards
  Daniel



<Prev in Thread] Current Thread [Next in Thread>