Hi Brian!

 Thanks for your time!, i've just to run xfs_repair -n the result you can see --> http://pastebin.centos.org/16106/ and as you can see there was problems, but why this problems occurs?, we have xfs partition used for one KVM where running Amanda with Centos 7, perhaps it was a bad configuration inside KVM or perhaps a bug?, the host server is running centos 6

[root@vtl ~]# uname -a
Linux vtl.areaprod.b2b 2.6.32-504.8.1.el6.x86_64 #1 SMP Wed Jan 28 21:11:36 UTC 2015 x86_64 x86_64 x86_64 GNU/Linux


But the KVM running centos 7

Linux amanda 3.10.0-123.8.1.el7.x86_64 #1 SMP Mon Sep 22 19:06:58 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux

The KVM config is this --> http://pastebin.centos.org/16111/

And /etc/fstab is this---> http://pastebin.centos.org/16116/


Thanks in advance, for any hint.

-Pablo


On Mon, Feb 16, 2015 at 11:10 AM, Brian Foster <bfoster@redhat.com> wrote:
On Fri, Feb 13, 2015 at 03:44:57PM -0300, Pablo Silva wrote:
> Hi !
>
>     We have a server with centos 6.6, kernel version:
> 2.6.32-431.17.1.el6.x86_64, where we got the following message:
>
> Feb 12 19:22:15 vtl kernel:
> Feb 12 19:22:15 vtl kernel: Pid: 3502, comm: touch Not tainted
> 2.6.32-431.17.1.el6.x86_64 #1
> Feb 12 19:22:15 vtl kernel: Call Trace:
> Feb 12 19:22:15 vtl kernel: [<ffffffffa041ae5f>] ?
> xfs_error_report+0x3f/0x50 [xfs]
> Feb 12 19:22:15 vtl kernel: [<ffffffffa0422980>] ? xfs_ialloc+0x60/0x6e0 [xfs]
> Feb 12 19:22:15 vtl kernel: [<ffffffffa041ec2e>] ? xfs_dialloc+0x43e/0x850 [xfs]
> Feb 12 19:22:15 vtl kernel: [<ffffffffa0422980>] ? xfs_ialloc+0x60/0x6e0 [xfs]
> Feb 12 19:22:15 vtl kernel: [<ffffffffa044007a>] ?
> kmem_zone_zalloc+0x3a/0x50 [xfs]
> Feb 12 19:22:15 vtl kernel: [<ffffffffa043b814>] ?
> xfs_dir_ialloc+0x74/0x2b0 [xfs]
> Feb 12 19:22:15 vtl kernel: [<ffffffffa043d900>] ? xfs_create+0x440/0x640 [xfs]
> Feb 12 19:22:15 vtl kernel: [<ffffffffa044aa5d>] ? xfs_vn_mknod+0xad/0x1c0 [xfs]
> Feb 12 19:22:15 vtl kernel: [<ffffffffa044aba0>] ? xfs_vn_create+0x10/0x20 [xfs]
> Feb 12 19:22:15 vtl kernel: [<ffffffff81198086>] ? vfs_create+0xe6/0x110
> Feb 12 19:22:15 vtl kernel: [<ffffffff8119bb9e>] ? do_filp_open+0xa8e/0xd20
> Feb 12 19:22:15 vtl kernel: [<ffffffff811a7ea2>] ? alloc_fd+0x92/0x160
> Feb 12 19:22:15 vtl kernel: XFS: Internal error
> XFS_WANT_CORRUPTED_GOTO at line 990 of file fs/xfs/xfs_ialloc.c.
> Caller 0xffffffffa0422980
>

        /*
         * None left in the last group, search the whole AG
         */
        error = xfs_inobt_lookup(cur, 0, XFS_LOOKUP_GE, &i);
        if (error)
                goto error0;
        XFS_WANT_CORRUPTED_GOTO(i == 1, error0);

        for (;;) {
                error = xfs_inobt_get_rec(cur, &rec, &i);
                if (error)
                        goto error0;
                XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
                if (rec.ir_freecount > 0)
                        break;
                error = xfs_btree_increment(cur, 0, &i);
                if (error)
                        goto error0;
--->            XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
        }

That corresponds to the check above. This code is part of the inode
allocator where we expect an AG to have free inodes and we're doing a
brute force search for a record. Apparently we go off the AG or some
other problem occurs before we find a free inode record.

Does 'xfs_repair -n' report any problems with this fs? Have there been
any other storage errors reported in the logs? Is the problem
reproducible or was it a one off occurrence?

Brian

> I can't find more information for this..., perhaps a bug or other
> thing ..., welcome any hint for to research..
>
> Thanks in advance!
>
> -Pablo

> _______________________________________________
> xfs mailing list
> xfs@oss.sgi.com
> http://oss.sgi.com/mailman/listinfo/xfs