*BSD News Article 86583


Return to BSD News archive

Path: euryale.cc.adfa.oz.au!newshost.carno.net.au!harbinger.cc.monash.edu.au!news.mel.connect.com.au!munnari.OZ.AU!news.ecn.uoknor.edu!feed1.news.erols.com!news.nl.innet.net!INnl.net!feed1.news.innet.be!INbe.net!tank.news.pipex.net!pipex!news.mathworks.com!fu-berlin.de!news.belwue.de!lf.net!news.gni.net!rnima!subloch.swb.de!maulwurf
Date: 08 Jan 1997 07:07:00 +0100
From: maulwurf@subloch.swb.de (Stefan Huerter)
Newsgroups: comp.unix.bsd.freebsd.misc
Message-ID: <6OUBii$UoRB@subloch.swb.de>
References: <6JJPs3kUoRB@subloch.swb.de>
	<199610221028.DAA08386@silvia.HIP.Berkeley.EDU>
Subject: Re: Question to ccd and FreeBSD
X-Newsreader: CrossPoint v3.11 R/C2188
Organization: die wahre Antwort: 42
Lines: 96
Xref: euryale.cc.adfa.oz.au comp.unix.bsd.freebsd.misc:33896

Guckux Satoshi

takes a long time for me, to install ccd :-)

>  * My question:
>  * I am using FreeBSD 2.1.5, installed 2 Barracuda disks.
>  * Can I use ccd for mirroring partitions and stripping the same time?
>  * (The idea: RAID1 e.g. /dev/sd0a /dev/sd1a
>  *            RAID0 e.g. /dev/sd0e /dev/sd1e)
> Yes, that should work fine.  You just need to write
> ccd0 <interleave> CCDF_MIRROR /dev/sd0a /dev/sd1a
> ccd1 <interleave> 0 /dev/sd0e /dev/sd1e
> in your /etc/ccd.conf.  The ccd's are configured independently of each
> other.

> thinking about booting from your SCSI disk though.  The root partition
> is mounted before ccdconfig is run, so I don't think it's possible to
> create a ccd involving sd0a.  (Never tried it though.)

You're right, it isn't possible to ccd sd0a. No attack to you :-). but  
with Solaris and Disksuit or vxva it is possible to mirror the root  
partition, too... seems, they make s.th.

to my configuration:
ASUS PI55T2P4, i586-133, 96MB, 2 x 2940-PCI Controller, one Seagate  
ST32550N (Barracuda) on each of the 2940.
Just installed yesterday (yes finished this year! :) following  
configuration:
ccd0 RAID 0 of home
ccd1 RAID 1 of /usr
and some more of ccd1-type

question:
trying "disklabel -e ccd[>0]" doesn't work, wrong magic number (or s.th.  
else) is the result, for ccd0 (the RAID0) it works.
It works now fine, with mounting partitions of /dev/ccd[x]c (x>0).
is this correct so far?

I am using following configuration of /etc/ccd.conf
# Just the mirror and stripping set for hole.swb.de
# implemented by Frank Roeske and Stefan Huerter
# 30-12-1996
#
# following should be done:
# /dev/sd0a RAID 1 to 	/dev/sd1a for root
# /dev/sd0d RAID 0 with /dev/sd1d for home
# /dev/sd0e RAID 1 to 	/dev/sd1e for usr
# /dev/sd0f RAID 1 to 	/dev/sd1f for quark
# /dev/sd1g RAID 1 to 	/dev/sd0g for var
# /dev/sd1h RAID 1 to 	/dev/sd0h for tmp

# root
# ccd0	32	CCDF_MIRROR	/dev/sd0a /dev/sd1a
# home
ccd0	32	0		/dev/sd0d /dev/sd1d
# usr
ccd1	32	CCDF_MIRROR	/dev/sd0e /dev/sd1e
# quark
ccd2	32	CCDF_MIRROR	/dev/sd0f /dev/sd1f
# var
ccd3	32	CCDF_MIRROR	/dev/sd1g /dev/sd0g
# tmp
ccd4	32	CCDF_MIRROR	/dev/sd1h /dev/sd0h

so, there are some results of bonnie:
/dev/ccd3c
              -------Sequential Output-------- ---Sequential Input-- --Random--
              -Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks---
Machine    MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU  /sec %CPU
          100  1447 31.3  1461  9.9   869  6.2  2651 51.5  2964 11.5 229.0  8.6

not really fast, but the security is important for me (redundance).



/dev/ccd0c
              -------Sequential Output-------- ---Sequential Input-- --Random--
              -Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks---
Machine    MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU  /sec %CPU
          100  2526 52.8  2551 13.7  1387 10.2  4054 79.0  5305 21.5 281.7 10.8


it seems, that the speed is really decreased by the RAID1 but doesn't win  
a lot of RAID0.

this is the normal single Barracuda system, no RAID:
              -------Sequential Output-------- ---Sequential Input-- --Random--
              -Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks---
Machine    MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU  /sec %CPU
          100  5086 98.5  5309 22.1  1719 10.7  3836 72.0  5255 19.3 128.6  5.1

the output performance is "slow". Have I to use another interleave factor?  
Or what's wrong? Any hints are welcome!

Bye
     Stefan