disk i/o problems

[Nico -telmich- Schottelius]

Hello everyone,

I did some tests on our Dell Poweredge SC 1425, because our new
mailserver had one outage (reason unknown) and was onetime running
extremly slow.

So I took another brand new sc1425 and run
http://home.schottelius.org/~nico/unix/freebsd/testdisks.sh on it.

The result is that the machine is practically dead, when the script is
running. I started the script in a screen, in which also top, gstat,
systat -iostat and vmstat -a runs in. I can navigate through the screen,
but if I type 'ls' in a open shell it will never return until I stop
the testscript. Connecting to the machine via ssh is impossible within
this time, existing connections are still there.

Configuration of the machine:

   - 1x dual core 3.2ghz xeon
   - 2x scsi disks (u320) combined via gmirror to raid1
   - RAID bus controller: Adaptec ASC-39320(B) U320 w/HostRAID (rev 10)
   - mount: /dev/mirror/raid1s1a on / (ufs, local, soft-updates)

My question regarding this issue:

   - it seems not to be normal to me, that a system is practically dead
     when running some i/o heavy processes and it also does not look to
     me like a general freebsd issue; what could be the reason for it?
   - is it possible to choose i/o schedulars in freebsd like linux
     offers it with sfq/deadline/cbq?
   - what's the general / normal behaviour of freebsd when doing i/o
     scheduling?


I'll also post a more detailled description about the problem in the
other server on -questions in the next minutes.

Sincerly

Nico

-- 
Think about Free and Open Source Software (FOSS).
http://nico.schottelius.org/documentations/foss/the-term-foss/

PGP: BFE4 C736 ABE5 406F 8F42  F7CF B8BE F92A 9885 188C
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://lists.freebsd.org/pipermail/freebsd-hackers/attachments/20070521/a5091500/attachment.pgp