Hello All,
I'm in the market for a couple of systems with high disk throughput. I
work in a network test lab and a number of the tests I run read and
write files to servers. I'm finding that I'm really overloading my SCSI
controllers even at Ultra-Wide speeds. I'm particularly puzzled as one
of my NT servers is currently outperforming a Linux(RH 5.1) box!? (must
be poorly tuned...) In any case I was wondering what kind of experience
some of you out there might have had with RAID controllers and your
systems. My main Linux system is running a 7200rpm 9gb Seagage U/W on
300mhz PII system with 128mb RAM using an onboard Adaptec 2940 SCSI
controller. The NT system that beats it is a dual 233mhz PII/256mb RAM
with a 4gb U/W drive and onboard Adaptec 7880. I get a message in
/var/log/messages on the Linux box which basically equates to:
<date> <name> kernel: scsi : aborting command due to timeout : pid <#>
<scsi device info> Read or Write and some hex info
What basically happens is that the Linux box will service requests from
clients (via SAMBA) and at some point apparently resets SCSI, stopping
traffic until reset is done, in a continuous cycle. Meanwhile the NT
box is just trucking along sustaining a constant load.
I've seen threads on the list recommending against Adaptec hardware.
Could some of this behaviour be attributed to driver quirkiness? Any
ideas on what I can do to help alleviate the situation or
recommendations on RAID hardware will be much appreciated. At this
point I'm looking at RAID 0 (just striping) controllers for a pure speed
boost.
js
* Jason Scherbarth
* Extreme Networks
* (408)863-2842
Gigabit Ethernet? check out http://www.extremenetworks.com
--
PLEASE read the Red Hat FAQ, Tips, Errata and the MAILING LIST ARCHIVES!
http://www.redhat.com/RedHat-FAQ /RedHat-Errata /RedHat-Tips /mailing-lists
To unsubscribe: mail [EMAIL PROTECTED] with
"unsubscribe" as the Subject.
