xfs
[Top] [All Lists]

Serious XFS trouble: umount filesystem -> corrupt (on IDE drive)

To: "Linux XFS Mailing List" <linux-xfs@xxxxxxxxxxx>
Subject: Serious XFS trouble: umount filesystem -> corrupt (on IDE drive)
From: "Ralf G. R. Bergs" <rabe@xxxxxxxxxxxxxx>
Date: Sat, 23 Nov 2002 11:21:54 +0100
Cc: "Maciej.BOROWKA@xxxxxxxxxxxxxxxxxxx" <Maciej.BOROWKA@xxxxxxxxxxxxxxxxxxx>
Priority: Normal
Sender: linux-xfs-bounce@xxxxxxxxxxx
Hi there,

yesterday I replaced the motherboard of my home server (running Debian/GNU 
Linux 3.0) for a CPU upgrade (486-133 -> K6-III+ 400) and the network card. 
Therefore (to support the card) I had to recompile the kernel, and so I 
upgraded the kernel source from 2.4.19-xfs to yesterday's 2.4.20-rc2-xfs ("SGI 
XFS CVS-2002-11-22_06:00_UTC with ACLs, quota, no debug enabled.")

Suddenly I noticed strange filesystem corruption. Even after a clean "halt" or 
"reboot" upon mounting the XFS filesystems the system would give me error 
messages like the following:

Nov 23 00:41:33 Server kernel: Starting XFS recovery on filesystem: ide0(3,9) 
(dev: 3/9)
Nov 23 00:41:33 Server kernel: XFS: xlog_recover_process_data: bad clientid
Nov 23 00:41:33 Server kernel: XFS: log mount/recovery failed
Nov 23 00:41:33 Server kernel: XFS: log mount failed

or even

Nov 22 23:39:04 Server kernel: Starting XFS recovery on filesystem: ide0(3,9) 
(dev: 3/9)
Nov 22 23:39:04 Server kernel: XFS: log mount/recovery failed
Nov 22 23:39:04 Server kernel: XFS: log mount failed
Nov 22 23:39:04 Server kernel: XFS mounting filesystem ide0(3,10)
Nov 22 23:39:04 Server kernel: XFS mounting filesystem ide0(3,9)
Nov 22 23:39:04 Server kernel: XFS: nil uuid in log - IRIX style log
Nov 22 23:39:04 Server kernel: XFS: failed to locate log tail
Nov 22 23:39:04 Server kernel: XFS: log mount/recovery failed
Nov 22 23:39:04 Server kernel: XFS: log mount failed
Nov 22 23:39:04 Server kernel: XFS mounting filesystem ide0(3,9)
Nov 22 23:39:04 Server kernel: XFS: nil uuid in log - IRIX style log
Nov 22 23:39:04 Server kernel: XFS: failed to locate log tail
Nov 22 23:39:04 Server kernel: XFS: log mount/recovery failed
Nov 22 23:39:05 Server kernel: XFS: log mount failed
Nov 22 23:39:05 Server kernel: XFS mounting filesystem ide0(3,9)

When I booted into single user mode, the same problems occured. xfs_repair 
failed and told me to zap the log. I zapped the log by invoking "xfs_repair -
L," and after that I was able to successfully mount the filesystem.

I then entered "exit" to boot into multi-user mode, and since that moment the 
system is absolutely stable. So I seriously doubt it's a hardware problem (on-
board IDE controller broken, cabling damaged, etc.)

I seem to remember a problem that had to do with flushing the cache on IDE 
drives, but I can't find anything about that on Google. But I found the 
following post to this list that EXACTLY describes my problem -- obviously the 
poster didn't receive a reply:

http://oss.sgi.com/projects/xfs/mail_archive/200202/msg00446.html

Note that I don't use RAID, just a plain XFS filesystem on a physical 
partition.

One thing that I'd like to mention (altho I don't think it's important) is 
that the hard drive had been partitioned using OnTrack's drive manager 
software because the bios of the old mobo didn't support hard drives larger 
than 2.5 gig. I'm pretty sure this isn't the root of my problems because the 
partition table is detected properly:

============================ 8x =================================
Nov 22 23:59:23 Server kernel: hda: 16514064 sectors (8455 MB) w/467KiB Cache, 
CHS=1027/255/63, (U)DMA
Nov 22 23:59:23 Server kernel: Partition check:
Nov 22 23:59:23 Server kernel:  hda: [DM6:DDO] [remap +63] [1027/255/63] hda1 
hda2 < hda5 hda6 hda7 hda8 hda9 hda10 hda11 >



                                 cfdisk 2.11n

                             Disk Drive: /dev/hda
                            Size: 8455168512 bytes
             Heads: 255   Sectors per Track: 63   Cylinders: 1027

   Name        Flags      Part Type  FS Type          [Label]        Size (MB)
 -----------------------------------------------------------------------------
   hda1        Boot        Primary   FAT12            [MS-DOS     ]       8.23
   hda5                    Logical   Linux ext3       [root]             32.91
   hda6                    Logical   Linux XFS                         1579.26
   hda7                    Logical   Linux XFS                          526.42
   hda8                    Logical   Linux XFS                         2097.45
   hda9                    Logical   Linux XFS                         3676.71
   hda10                   Logical   Linux XFS                          477.07
   hda11                   Logical   Linux ext2                          49.36



Server:/usr/src# fdisk /dev/hda 

The number of cylinders for this disk is set to 1027.
There is nothing wrong with that, but this is larger than 1024,
and could in certain setups cause problems with:
1) software that runs at boot time (e.g., old versions of LILO)
2) booting and partitioning software from other OSs
   (e.g., DOS FDISK, OS/2 FDISK)

Command (m for help): p

Disk /dev/hda: 255 heads, 63 sectors, 1027 cylinders
Units = cylinders of 16065 * 512 bytes

   Device Boot    Start       End    Blocks   Id  System
/dev/hda1   *         1         1      8001    1  FAT12
/dev/hda2             2      1027   8241345    f  Win95 Ext'd (LBA)
/dev/hda5             2         5     32098+  83  Linux
/dev/hda6             6       197   1542208+  83  Linux
/dev/hda7           198       261    514048+  83  Linux
/dev/hda8           262       516   2048256   83  Linux
/dev/hda9           517       963   3590496   83  Linux
/dev/hda10          964      1021    465853+  83  Linux
/dev/hda11         1022      1027     48163+  83  Linux

============================ 8x =================================

Any idea where to look?

Thanks,

Ralf


-- 
   L I N U X       .~.
  The  Choice      /V\
   of a  GNU      /( )\
  Generation      ^^-^^



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