From: Wynne, Conor (Conor.Wynne at domain compaq.com)
Date: Mon 13 Aug 2001 - 08:48:41 IST
Yep, it was the processor speed that caused a major bottle neck. I moved it
to a p200 and works much better. Created a RAID5 and a RAID1, you can even
create online spares!!! I failed drives with the software utility - for the
craic, in fact all the features of RAID controllers, except it uses 25
percent cpu. The RAM usuage is minimal - the p200 has 256MB anyway.
The RAM is perfect, system is super stable, I really didn't wan't to reboot
with the newly compiled kernel as I had tons of uptime, now I'm back to 1
day :(
Unfortunately, I also managed to "kill" 2 - 3 drives, DAMN! I may have a
controller issue, *gonna wait for a new one* not sure. Further testing
needed as they format fine in the controller utility, but fail with "scsi
device disconnect" errors, and the fdisk fails.
Out of 180GB of drives, only 20GB remains, ah well suppose I could not
expect all the drives to have been perfect from a skip!
All said and done, I am really impressed with linux raid. Once I stick a
graphics card back in my dual mobo, its gonna be transformed into a RAID
server for NFS/SAMBA and go into the attick.
Thanks Paul for your input,
regards
CW
ok.. p120 will really crawl. esp. for RAID5.
> first and has redundant power suplies - hot-plug naturally! I'm off to
> Dunsink tip tomorrow :)
to throw out the drive cages???? no.. give 'em to me!
(oh.. you're going scavenging for /more/ kit. damm).
> I also checked the pins on the cables for abnormalities, may be a
> termination issue as one drive negotiates at 10Mhz rather than 20 like the
> others.
ah.. definitely not good.
i remember someone making the point once (on this list?) that SCSI
can not guarantee data integrity cause there is no check (eg CRC) on
the data in a scsi frame. if that's true and you have cabling problems
then that'd explain the problems you're having.
This archive was generated by hypermail 2.1.6 : Thu 06 Feb 2003 - 13:11:34 GMT