dump/mksnap_ffs - ufs2 deadlockss

[Lachlan O'Dea]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 02 Feb 2004, at 10:51, Andre Guibert de Bruet wrote:

> Hi,
>
> I guess I'm running into a string of problems with ufs2 across my 
> systems.
> This deadlock is 100% reproduceable across all of my systems with 
> Adaptec
> and LSI RAID controllers. I tried to dump a live ufs2 system using:
>
> /sbin/dump -0uaL -f "/usr/backups/`hostname -s`-raid5.dump" /mnt/amrd0a
>
> and I get a mksnap_ffs process in the following state after about a 
> minute
> of moderate disk activity:
>
> 18312 root      -4    0  1196K   744K ufs    1   0:04  0.00%  0.00% 
> mksnap_ffs

[snip db output]

> This looks strangely like the getblk() hangs that I have seen on this
> system but it's the first time that I've come across the ufs() state on
> the hung process(es).
>
> Any ideas? :-)

I believe the getblk and ufs hangs are related. I get find/makewhatis 
hanging on getblk as previously described and also see exactly the same 
problem you do with dump. It think find sometimes hangs on ufs as well. 
I have a PERC 3/Di, which is the aac device. The volume is RAID 5. So 
it seems that SCSI RAID is a common factor for this problem, not a 
particular driver.

- --
Lachlan O'Dea
Computer Associates
Senior Software Engineer
tel: +61 3 8416 5627
fax: +61 3 8416 5810
mobile: +61 412 390 650
odela01 at ca.com

Relax and enjoy your shoes!

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.3 (Darwin)

iD8DBQFAHyLLoBZahcUBONoRAs2aAKCcELGdgGCQn6L2EGACd+q20yip/gCgjP4a
DmXWDLcwzl6Veg+Ia0IM440=
=F1+K
-----END PGP SIGNATURE-----