Re: updatedb segfaults, and find dies...

[Seth Mos <seth.mos@xxxxxxxxx>]

Ken D'Ambrosio wrote:
> Hello, all.  I recently installed a new server, with XFS, kernel 2.6.6, 
> on a 3Ware 9500 controller (using the 3Ware stand-alone driver).  When I 
> attempt to do an updatedb, it gets about 200 files in (shown by watching 
> strace), and segfaults.  When I do a find from root, it dies -- in the 
> same place -- with "find: ./.. changed during execution of find".  If I 
> delete the file that they are dying on, they then die on a different 
> file.  Someone suggested it might have something to do with 
> de-referencing symlinks, so, for the hell of it, I tried "find -follow", 
> and it worked... even though none of the files that were being looked at 
> were symlinks.

Odd, I don't have any issues on my box with a 3ware 9500.

Linux version 2.4.27-divas-3w (root@lsintra) (gcc version 2.95.4 
20011002 (Debian prerelease)) #10 SMP Thu Aug 19 15:37:00 CEST 2004
Kernel command line: auto BOOT_IMAGE=2427-divas-3w ro root=808 
BOOT_FILE=/boot/vmlinuz-2.4.27-divas-3w
3ware Storage Controller device driver for Linux v1.02.00.037.
3w-xxxx: No cards found.
3ware 9000 Storage Controller device driver for Linux v2.24.00.009fw.
3w-9xxx: scsi0: Found a 3ware 9000 Storage Controller at 0xf5044000, 
IRQ: 28.
3w-9xxx: scsi0: Firmware FE9X 2.02.00.011, BIOS BE9X 2.02.01.037, Ports: 8.
scsi0 : 3ware 9000 Storage Controller
  Vendor: 3ware     Model: Logical Disk 00   Rev: 1.00
  Vendor: 3ware     Model: Logical Disk 01   Rev: 1.00

/dev/sdb1 on /users type xfs (rw,usrquota,grpquota)
/dev/sdb1             745G  422G  323G  57% /users

You can find a 2.4.x kernel for it on my site.
http://iserv.nl/files/debian/

The 3ware 9xxx driver is compiled in so no need for a ramdisk.

I don't have 2.6 images yet because other driver availability issues 
prevent me from using it.

Any specific reason you are using 2.6?

> Is it possible this is an FS issue, or...?

Could be hard or software. Might be bad ram. Try memtest86 or the + variant.

> Aside from this odd behavior, everything else on the system (*knocks 
> wood*) appears to be fine.

You might be lucky :-/

Cheers,

Seth