Very odd problems.

[Oliver Hoffmann]

Hi list!

My attempt is to make backups to external firewire-hds, because they are much 
cheaper and faster than tapes. I have two servers which are almost identical. 
The first one has FreeBSD 4.8 Release running on an Athlon1000 with an Asus 
board. Two ICP-Vortex controller with two raids (both 42 GB, level 5) are 
attached. fw-card is the Texas Instruments TSB12LV26. I connected an external 
5.25" case (probably oxford 911, but I couldn't find it out) with an 
mobilerack inside. In this rack is a wd80 (wd800BB-00CAA1) harddisk. This 
system works (I would say 90% stable).
The second one runs FreeBSD 4.8-RC1 on an Athlon 1500+ with an Abit 
motherboard. It has one ICP with a bigger raid (ca. 200 GB, level 5). Hence 
the external harddisks have 160 GB (Maxtor 6Y160PO). All the firewire-parts 
are the same. That means the main differences are the hardisks and the 
FreeBSD-relaease. Btw, I had to take 4.8, because there isn't a support for 
vortex-controller in 5.0.
This second server has permanently errors and sometimes automatically reboots 
or it crashes. I can see such errors (plenty of them) in the logs:

/kernel: sbp0:0:0 Obj: Data buffer, Error: Time-out error
/kernel: Unknown service addr 0x000000f0:0x16e800f0 tcode=5
/kernel: sbp0:0:0 request timeout ... agent reset

I do the backups with simple shell-scripts:

/sbin/kldload firewire
/bin/sleep 10
/sbin/kldload sbp
/bin/sleep 60
/sbin/fsck -y /dev/da1s1e
/bin/sleep 20
/sbin/mount /dev/da1s1e /mnt/fw
/bin/sleep 60
/bin/cp -R /mnt/raid1 /mnt/fw
/bin/df -h
/bin/sleep 240
/sbin/umount /dev/da1s1e
/bin/sleep 60
/sbin/kldunload sbp
/bin/sleep 60
/sbin/kldunload firewire
/sbin/kldstat

Maybe I exaggerated with the sleeptimes, but too short causes definitely 
problems. If the system reboots, it always does it after everything is 
copied.
I hope that somebody can help me to get my system reasonabily working.
Thanks for any help in advance!
Regards,

Oliver.