Linux-Misc Digest #998, Volume #19 Fri, 30 Apr 99 18:13:13 EDT
Contents:
Re: Gnome Help ! (Donald)
Re: WORDS OF WISDOM!! Upgrading RedHat 5.1 to 2.2.X Kernel (Johan Kullstam)
Re: Where to get Linux OS (Alex)
Kernel bug or my stupidity? linux 2.2.6 ext2 problem? (Moritz Moeller-Herrmann)
Re: Where to get Linux OS (jik-)
Re: bttv device not created (2.2.2) (William Burrow)
Re: I have to save my system (Roger Sondermann)
Re: Printing on LaserJet: keep asking for wrong paper size (S P Arif Sahari Wibowo)
Re: Windows NT vs. Linux testing by mindcraft ([EMAIL PROTECTED])
Re: Telnet Login as ROOT ("D. Vrabel")
Re: GNU reeks of Communism (returning to %252522GNU Communism%252522) (Prins Olivier)
Re: Kernel 2.2.X rpm (Neal Barney)
Re: Porting Greg Lehey's rawio.c from FreeBSD to Linux... (Brad Knowles)
Re: BRU v. arkeia ([EMAIL PROTECTED])
----------------------------------------------------------------------------
Crossposted-To:
alt.linux,comp.os.linux.hardware,comp.os.linux.help,comp.os.linux.questions,comp.os.linux.setup,comp.os.linux.x,linux.redhat,linux.redhat.install,linux.redhat.misc,linux.redhat.rpm,nl.comp.os.linux,nl.comp.os.linux.installat
From: Donald <[EMAIL PROTECTED]>
Subject: Re: Gnome Help !
Date: Fri, 30 Apr 1999 17:01:25 -0400
create a file in your home directory called .xinitrc
then type this in:
#!/bin/bash
gnome-session
On Fri, 30 Apr 1999, John van der Zanden wrote:
> Unpacked all gnome packages without serious problems, but what to do now ??
> the manual says this
>
> To start gnome, you must edit the X startup files. (which one and how ?)
> a sample x startup file using gnome-session follows:
> #!/bin/bash
> exec gnome-session
> ## end sample
> the default Gnome session configuratio file is
> /usr/share/gnome/default.session. The user gnome session is
> $HOME/.gnome/session.
>
> All nice and well, but what must i do ? I dont have any clue what this all
> means as i am a NEWBIE !!
> Which file must i edit and what should i put in it ??
>
>
> John
>
>
>
>
------------------------------
From: Johan Kullstam <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.setup
Subject: Re: WORDS OF WISDOM!! Upgrading RedHat 5.1 to 2.2.X Kernel
Date: 30 Apr 1999 15:34:45 -0400
David Pendell <[EMAIL PROTECTED]> writes:
> I understand this. I tried the download method. Unfortunatly, I only
> have a dialup line that proved unreliable for a download of such
> magnitude.
yes, i know the feeling. however, if all you want is to move to
kernel 2.2.x, it's not all that much beyond downloading the kernel
itself.
> You are however missing the point. I am not afraid of have to work at
> upgrading. I am not afraid of having to recompile various programs in
> order to get incompatabilities solved. I am a person that likes to
> undestand the whats, whys and hows of the underlying functionality of
> the system. I am a professional programmer with a great deal of interest
> in system operations. I like to do things for myself.
>
> The point of all of this is:
>
> I did not expect RedHat to hobble my upgrade path by mixing two
> different software packages into each other.
how did they hobble the upgrade path?
i've been using new kernels since 2.1.124 back in october on my
redhat 5 system. just get the latest binutils and other sundry
kernel upgrade items. check www.linuxhq.com for their list of what
you need.
libc is a *much* bigger issue. the big difference with redhat 6 is
that they seem to be using glibc-2.1. i don't know how much trouble
it is to go to that (i hear that it is mostly compatible with
glibc-2.0 so once you get 2.1 and recompile ncurses almost everything
will continue to run).
since the library issue is fairly important, i usually get source
rpms. usually it's no big deal to recompile it and get the rpms using
the libraries which you have.
--
johan kullstam
------------------------------
Date: Fri, 30 Apr 1999 14:24:30 -0600
From: Alex <[EMAIL PROTECTED]>
Subject: Re: Where to get Linux OS
Go to:
http://www.linux.org
There is an abundance of info on Linux and where to get it.
[EMAIL PROTECTED] wrote:
> I was looking for a free copy of the LINUX OS for my personal use. The
> problem I am having is, I don't know enough about LINUX as of yet to know
> what I need to start using or playing with the LINUX OS. Is there anyone
> that can help? I don't know what files need to be downloaded. Or where to
> find documentation. Let me know if it is possible to still get a free copy
> of the LINUX OS.
>
> Thanks!!
>
> ------------------ Posted via SearchLinux ------------------
> http://www.searchlinux.com
------------------------------
From: [EMAIL PROTECTED] (Moritz Moeller-Herrmann)
Crossposted-To: alt.uu.comp.os.linux.questions,comp.os.linux.hardware
Subject: Kernel bug or my stupidity? linux 2.2.6 ext2 problem?
Date: Fri, 30 Apr 1999 22:41:55 +0200
Reply-To: [EMAIL PROTECTED]
Hi,
I just had a weird experience when playing around with my PC...
I had a program crash (Lyx1.0) when importing a Tex file and failing
miserably.
It filled my /home ext2 partition on an IDE drive by filling the log file
~/.xsession-errors(120M). When I noticed this (my bash prompt shows free space
in /home) I removed the offending file and killed the hung Lyx process.
To my surprise the free disk space did not show up! I ran sync, du, df and
stuff like that. I thought I had overlooked a big file somewhere in my $HOME.
Nothing. I killed a few files and regained a couple of free MBs, but nowhere
near what I had before.
So finally I tried to unmount the partition and it claimed device busy. I did
fuser -va /deb/hda2 and it showed a mount process owned by root with PID root.
strange. I changed to single user mode (init S). I tried again. I could now
unmount. When remounting the drive I had all my space back. Phew!
Does anyone have a clue why this happened? Any idea, comment?
I do have NFS and automount, but they weren't active at the time. my NFS user
tools are two revisions off the required in the Changes file. But this was a
normal ext2 file system....
Specs:
SuSE6.1beta based...
Linux Gertrud 2.2.6 #5 Mon Apr 19 17:25:27 CEST 1999 i586 unknown
Reading specs from /usr/local/lib/gcc-lib/i586-pc-linux-gnu/pgcc-2.91.60/specs
gcc version pgcc-2.91.60 19981201 (egcs-1.1.1 release)
mount: mount-2.9i
Linux automount version 3.1.3
Universal NFS Server 2.2beta38
Thanks!
--
Moritz Moeller-Herrmann [EMAIL PROTECTED] ICQ# 3585990 # Not only
Get my public pgp / gpg key from # Open Source(TM)
http://webrum.uni-mannheim.de/jura/moritz/pubkeymoritz # but also
KDE forever! Use Linux to impress your friends! # Open Minded!
------------------------------
Date: Fri, 30 Apr 1999 14:03:55 -0700
From: jik- <[EMAIL PROTECTED]>
Subject: Re: Where to get Linux OS
Alex wrote:
>
> Go to:
> http://www.linux.org
> There is an abundance of info on Linux and where to get it.
>
> [EMAIL PROTECTED] wrote:
>
> > I was looking for a free copy of the LINUX OS for my personal use. The
> > problem I am having is, I don't know enough about LINUX as of yet to know
> > what I need to start using or playing with the LINUX OS. Is there anyone
> > that can help? I don't know what files need to be downloaded. Or where to
> > find documentation. Let me know if it is possible to still get a free copy
> > of the LINUX OS.
As I have offered several times before,....I have 5 Linux CDROM disks I
will give away for the cost of shipping. 1 of each:
RedHat 5.2
LinuxPro 5.4
SuSE 5.3
OpenLinux 1.3
Debian 2.0
------------------------------
From: [EMAIL PROTECTED] (William Burrow)
Crossposted-To: alt.os.linux
Subject: Re: bttv device not created (2.2.2)
Date: 30 Apr 1999 16:58:44 GMT
Reply-To: [EMAIL PROTECTED]
On Fri, 30 Apr 1999 16:36:29 +0100,
XXX <[EMAIL PROTECTED]> wrote:
>However, no "bttv" or "video" device is or has been created, and so none of
>the applications work.
>
>What have I done wrong, and how can I get my TV card (Hauppauge WinTV PCI
>Primo) to work now?
Just create the devices:
cd /dev
mknod video0 c 81 0
mknod video1 c 81 1
mknod video1 c 81 2
mknod video1 c 81 3
ln -s video0 video
I'm working on a bttv mini-HOWTO, it has been SGMLized recently. I
guess I just have to find a way to get it on the LDP. Shouldn't be
long.
Somebody needs to update the Documentation/devices.* files as well.
--
William Burrow -- New Brunswick, Canada o
Copyright 1999 William Burrow ~ /\
~ ()>()
------------------------------
Date: Fri, 30 Apr 1999 23:07:35 +0200
From: Roger Sondermann <[EMAIL PROTECTED]>
Subject: Re: I have to save my system
Roger Sondermann wrote:
>
> At Fri, 30 Apr 1999 12:00:35 +0100, Jim Howes wrote:
>
> >tar will preserve symbolic links and file ownership. There
> >are other ways, but I always seem to come back to tar.
> >
> >If you have your old partition on /old and your new one on /new,
> >you can do...
> >
> > cd /old
> > cp /old/etc/passwd /old/etc/group /etc
> > tar cf - . | (cd /new ; tar xpf - )
> >
> >The passwd and group file copy is necessary if you want tar to
> >retain file ownership (a lot of things won't work if you don't)
> >because tar records ownership in terms of user/group name, not
> >UID/GID, so needs a valid passwd/group file to convert back on
> >the restore.
>
> Thanks a lot. Can you tell me the whole extract command, too?
I thought, TAR would create an archive. Now I know it better.
My Linux is running again, thank you very much.
> >If you had lilo on your old partition, you probably want to install
> >it on the new one too. This is best done when your new disk is set
> >up to be where it is going to be in terms of primary/secondary master/
> >slave. When you have done that, you can run lilo from your bootdisk
> >by mounting it in, say, /new, and doing..
> >
> > sync
> > lilo -r /new -C /etc/lilo.conf
>
> I use loadlin and know what to do with the *.par file. Thanks again.
--
****************
Roger Sondermann
****************
�
PGP-KeyIDs:
- 0xC7932FC7 (DH/DSS)
- 0x68B535A7 (RSA)
- 0x17E12E09 (RSA) -is invalid-
------------------------------
From: S P Arif Sahari Wibowo <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.setup,linux.redhat,linux.redhat.misc
Subject: Re: Printing on LaserJet: keep asking for wrong paper size
Date: Fri, 30 Apr 1999 16:11:18 -0500
On Fri, 23 Apr 1999, Chris MacKenzie wrote:
>S P Arif Sahari Wibowo wrote:
>
>> I use RedHat Linux 5.1 with lpr-0.33-1 and rhs-printfilters-1.46-3. I
>> setup the Linux to use a HP LaserJet 4 in the network (attached to a NT
>> Server) using the printtool Tcl/Tk script came with RedHat.
>>
>> My problem is that when I print to the printer, the printer keep asking
>> for A4 paper. I setup the prnter to use letter, I setup printtool to use
>
>Ok, make sure that the default form type is set to Letter in both NT and
>in the printer itself.
That was the first thing I checked. Yes, the default form is set to
Letter in the printer, the NT server, in Linux' printtool, and in the
application itself.
Thanks.
S P Arif Sahari Wibowo
_____ _____ _____ _____
/____ /____/ /____/ /____ [EMAIL PROTECTED]
_____/ / / / _____/ http://spas.8m.com/
------------------------------
From: [EMAIL PROTECTED]
Crossposted-To: comp.unix.bsd.freebsd.misc
Subject: Re: Windows NT vs. Linux testing by mindcraft
Date: Fri, 30 Apr 1999 21:20:10 GMT
In article <7gcvlv$[EMAIL PROTECTED]>,
Alexander Viro <[EMAIL PROTECTED]> wrote:
>In article <[EMAIL PROTECTED]>, <[EMAIL PROTECTED]> wrote:
>>In article <7gcorh$[EMAIL PROTECTED]>,
>>Alexander Viro <[EMAIL PROTECTED]> wrote:
>> ...<snip>...
>
>>Gary Kildall and for the hardware it worked on it was pretty good. I
>
> Nice way to flame the hardware...
>
I wasn't flaming the hardware. The hardware was limited, but it
was what was possible for a home user with the technology of the time, and
so was CP/M.
--
Praeterea censeo Micromolle non esse utendum.
("Moreover, I maintain that Microsoft should not be used." With apologies
to Cato the Elder)
---- Remove "UhUh" and "Spam" to get my real email address -----
------------------------------
From: "D. Vrabel" <[EMAIL PROTECTED]>
Subject: Re: Telnet Login as ROOT
Date: Fri, 30 Apr 1999 18:35:09 +0100
On Fri, 30 Apr 1999, TurkBear wrote:
>
> For security reasons telneting as root is a very bad idea....The password is
> passed unencoded over the net, so any malicious ( or just curious ) sniffer out
> there could potentially gain control over your server....
>
>
> You can, however, telnet in as some other non-root user then issue a su command
> to become root....
Your password is still passed unencrypted over the network. Use ssh for
greater security.
> This should allow you to do all your admin stuff....
David
--
David Vrabel
Engineering Undergraduate at University of Cambridge, UK.
------------------------------
From: Prins Olivier <[EMAIL PROTECTED]>
Crossposted-To: comp.os.ms-windows.advocacy,comp.os.linux.advocacy,gnu.misc.discuss
Subject: Re: GNU reeks of Communism (returning to %252522GNU Communism%252522)
Date: Fri, 30 Apr 1999 23:37:31 +0200
>Craig Dowell wrote:
>In that context, your "cutting corners" is not always a bad thing. You
>may call it cutting a corner, someone else may call it a tradeoff or
>product definition or design decision. Unix and Linux certainly weren't
>built without any of this. Some people just like the result of the
>tradeoffs and design decisions made for Unix better than those made for
>Windows; and, of course, vice versa.
What do you get for your tradeoff when using Windows???...
The only advantage i could think would be that you'd have to pay less taxes,
because you've got to pay money for windows....
------------------------------
Crossposted-To: alt.os.linux,comp.os.linux,comp.os.linux.setup
Subject: Re: Kernel 2.2.X rpm
From: Neal Barney <[EMAIL PROTECTED]>
Date: Fri, 30 Apr 1999 15:23:43 -0600
Dan Nguyen wrote:
>
> In comp.os.linux.setup Aaron Dershem <[EMAIL PROTECTED]> wrote:
> : Any word on when Red Hat will release a 2.2.X kernel RPM? I downloaded the
> : source files from kernel.org, but I'd rather have a painless, no-brainer
> : upgrade.
>
> Why upgrade your kernel? Shouldn't it becuz you want your system to
> run more effienciently. So compile your own kernel. Kernel RPMs are dumb!!!!
This actually, isn't too far from the truth. I just compiled the 2.2.7
kernel on my machine two days ago. Before I was running standard
RH5.2. I can't believe the speed increase. I figured it would improve,
but I had no idea. Oh, and be careful with a kernel RPM. Don't use rpm
-U {kernel rpm here} That doesn't work like you think it would. Trust
me. :) I was a newbie once.
--
*----------------------------------------------*
Neal J. Barney Phone: (435) 797-4340
Space Dynamics Laboratory/USU
Software Engineering and Networking Group
*----------------------------------------------*
------------------------------
From: [EMAIL PROTECTED] (Brad Knowles)
Crossposted-To: comp.unix.bsd.freebsd.misc
Subject: Re: Porting Greg Lehey's rawio.c from FreeBSD to Linux...
Date: Fri, 30 Apr 1999 23:32:21 +0200
In article <[EMAIL PROTECTED]>,
[EMAIL PROTECTED] (Brad Knowles) wrote:
> I don't want to get into any OS wars here, but I'd like to do some
> low-level disk benchmarking of a Linux system using the same tool I used
> for that under FreeBSD, namely Greg Lehey's "rawio" (see
> <ftp://ftp.lemis.com/pub/rawio.tar.gz>).
Some people have recommended that I check out "hdparm". Let me quote
you the only section from the man page on this program that appears to be
relevant:
>> -T Perform timings of cache reads for benchmark and
>> comparison purposes.
[ ... deletia ... ]
>> This displays the speed of reading directly from
>> the Linux buffer cache without disk access. This
>> measurement is essentially an indication of the
>> throughput of the processor, cache, and memory of
>> the system under test.
[ ... deletia ... ]
>> -t Perform timings of device reads for benchmark and
>> comparison purposes.
[ ... deletia ... ]
>> This displays the speed of reading through the
>> buffer cache to the disk without any prior caching
>> of data. This measurement is an indication of how
>> fast the drive can sustain sequential data reads
>> under Linux, without any filesystem overhead. To
>> ensure accurate measurments, the buffer cache is
>> flushed during the processing of -t using the BLK-
>> FLSBUF ioctl.
[ ... deletia ... ]
So far as I can tell, all other options with this program are useful
only for (E)IDE drives, and would not be applicable to SCSI drives or SCSI
drive arrays (either managed through a software-only
striping/mirroring/RAID device driver, or an intelligent caching host
adaptor such as the DPT SmartRAID V, or a SCSI host adaptor such as an
Adaptec 2940 attached to a SCSI-SCSI controller such as the Mylex
DAC960SXI).
Okay, now let me quote from the man page of "rawio":
> By default, rawio spawns eight processes, each of which
> performs the same test. Four tests are available:
>
> Random Read
> The random read test reads varying length records
> from the specified device special, starting at
> random positions within the file. The offset is
> necessary to protect the disk label and any
> possible future extensions.
>
> Sequential Read
> The sequential read test reads constant length
> records from the specified device special,
> starting at offset 32 sectors the beginning
> of the file.
>
> Random Write
> The random read test writes varying length
> records to the specified device special,
> starting at random positions within the file.
> THIS TEST OVERWRITES DATA ON THE SPECIAL DEVICE.
> DO NOT USE IT ON A SYSTEM WHERE YOU HAVE
> IMPORTANT DATA.
>
> Sequential Write
> The sequential read test writes constant length
> records to the specified device special,
> starting at offset 32 sectors the beginning of
> the file. The offset is necessary to protect
> the disk label and any possible future extensions.
> THIS TEST OVERWRITES DATA ON THE SPECIAL DEVICE.
> DO NOT USE IT ON A SYSTEM WHERE YOU HAVE
> IMPORTANT DATA
[ ... deletia ... ]
> OPTIONS
[ ... deletia ... ]
> -c transfer-count
> Specify the length of sequential transfers or
> the maximum length of random transfers. The
> length of a random transfer is, on average,
> about half of this value. It may be
> specified either in sectors (<512) or in
> bytes (>= 512), and must be an integral number
> of sectors. This value defaults to 32768
> bytes (64 sectors). The maximum value is
> system-dependent. On FreeBSD it is 256
> sectors (131072 bytes).
[ ... deletia ... ]
> -n record-count
> Specify the total number of records to
> transfer. The default value is 16384. If
> the number is not divisible by the number of
> processes, the first remainder processes
> transfer one extra record.
>
> The sequential transfer tests will stop at
> the end of the device, which may result in
> fewer records than indicated being transferred.
>
> -p process-count
> Specify the number of processes to start.
> The default value is 8.
As you can see, rawio obviously has a lot more knobs that can be
twiddled to try and make sure that you thrash the disks as much as
physically possible, and hopefully manage to find some really good
performance numbers (or at least, the best you can get with that hardware
and software config, and it might point out weaknesses that can be
corrected).
Okay, now for some output of some tests. This first table (sorry
about the wrapping, folks, but this looks okay in 80 columns) shows a
drive array (striped with vinum), with five Quantum Viking II drives, with
a 512KB stripe size, and varying numbers of maximum I/O size and number of
child processes (the final -#K and -# parameters, as shown in the "ID"
field).
> Random read Sequential read Random write Sequential write
>ID K/sec /sec K/sec /sec K/sec /sec K/sec /sec
>s5x512K-1K-8 254.0 992 5615.5 5484 256.6 997 1277.7 1248
>s5x512K-2K-8 504.0 662 5973.9 2917 515.6 669 2353.2 1149
>s5x512K-4K-8 993.5 553 9293.2 2269 1027.7 572 4242.9 1036
>s5x512K-8K-8 1959.2 514 13275.0 1620 1989.3 518 7835.4 956
>s5x512K-16K-8 3712.3 464 19517.1 1191 3806.2 479 14594.0 891
>s5x512K-32K-8 6882.8 425 25788.7 787 6900.8 425 23564.3 719
>s5x512K-64K-8 11738.0 350 13527.0 206 11494.3 358 9765.6 149
>s5x512K-128K-8 11909.5 179 13056.1 100 14312.9 218 15068.5 115
>s5x512K-1K-16 297.0 1165 5595.2 5464 285.1 1109 1285.2 1255
>s5x512K-2K-16 602.9 784 6672.5 3258 564.5 737 2315.9 1131
>s5x512K-4K-16 1157.1 647 10848.4 2649 1098.9 612 4351.7 1062
>s5x512K-8K-16 2319.6 604 16468.1 2010 2176.3 569 8207.8 1002
>s5x512K-16K-16 4262.2 545 26277.9 1604 4167.2 531 17632.3 1076
>s5x512K-32K-16 8043.1 496 30302.0 925 7457.8 475 23183.9 708
>s5x512K-64K-16 12826.8 390 18056.1 276 12923.7 393 13158.5 201
>s5x512K-128K-16 16150.1 240 19701.4 150 15671.6 244 18997.9 145
>s5x512K-1K-32 325.9 1274 5568.2 5438 299.2 1182 1292.7 1262
>s5x512K-2K-32 650.7 852 6804.0 3322 589.7 767 2429.2 1186
>s5x512K-4K-32 1243.2 694 11752.3 2869 1178.6 666 4873.8 1190
>s5x512K-8K-32 2550.8 660 21334.4 2604 2295.9 603 10423.0 1272
>s5x512K-16K-32 4657.2 589 27842.7 1699 4178.7 530 16339.1 997
>s5x512K-32K-32 8306.0 523 31467.9 960 8048.7 484 25893.4 790
>s5x512K-64K-32 13288.3 416 22006.7 336 13606.4 430 15799.3 241
>s5x512K-128K-32 17559.3 270 19408.0 148 16489.4 267 16512.2 126
>s5x512K-1K-64 339.7 1330 5345.6 5220 299.2 1168 1259.1 1230
>s5x512K-2K-64 675.1 878 7638.2 3730 615.7 797 2357.6 1151
>s5x512K-4K-64 1332.9 750 14969.9 3655 1198.6 664 5565.9 1359
>s5x512K-8K-64 2591.2 681 22761.7 2779 2436.5 622 10467.1 1278
>s5x512K-16K-64 4799.9 614 28184.2 1720 4514.3 570 17694.3 1080
>s5x512K-32K-64 8629.2 535 30985.3 946 8132.6 509 24329.3 742
>s5x512K-64K-64 13762.8 423 20020.5 305 12830.1 390 13859.4 211
>s5x512K-128K-64 15606.6 238 19103.6 146 14207.7 216 14313.4 109
>s5x512K-1K-128 341.1 1333 1468.6 1434 308.7 1211 634.0 619
>s5x512K-2K-128 675.1 880 2486.5 1214 611.5 793 918.2 448
>s5x512K-4K-128 1336.0 746 4810.8 1175 1225.2 683 2828.9 691
>s5x512K-8K-128 2598.2 678 16836.3 2055 2315.1 604 6087.6 743
>s5x512K-16K-128 4885.8 609 27262.6 1664 4403.2 558 12698.7 775
>s5x512K-32K-128 8902.7 552 29494.3 900 8143.8 506 17290.7 528
>s5x512K-64K-128 14375.6 443 18649.2 285 12518.1 386 10897.2 166
>s5x512K-128K-128 15345.9 229 17728.1 135 15776.9 242 9773.2 75
The best performance for sequential read operations appears to be with
32KB maximum I/O sizes (pretty much regardless of number of child
processes, although that does have some influence), while maximum random
read and write throughput appears to be with 128KB maximum I/O sizes (and
with a very definite peak at 32 child processes), and sequential write
appears to have best throughput at 32KB maximum I/O transfers, with a
definite peak at 32 child processes.
All of this was striped with vinum, and connected through a single
Adaptec 2940UW host adaptor.
What I wasn't able to test was how varying the stripe size affected
things, how varying the number of devices affected things, how using
RAID-5 instead of pure striping would change the results, how doing a
mirroring+striping configuration would modify the behaviour, or how
testing different drives (say, the other four drives I had in the cabinet,
all Western Digital WDE9100 models) would affect things. There are lots
of other variables I wanted to take into account, but I just didn't have
the time.
Now, for the results of some hdparm tests (per the man page, each of
these was on an otherwise quiescent system and after a couple of "warm-up"
runs of the same command):
>># /usr/sbin/hdparm -T -t /dev/sdc1
>>
>>/dev/sdc1:
>> Timing buffer-cache reads: 64 MB in 0.59 seconds =108.47 MB/sec
>> Timing buffered disk reads: 32 MB in 5.20 seconds = 6.15 MB/sec
>># /usr/sbin/hdparm -T -t /dev/sdb1
>>
>>/dev/sdb1:
>> Timing buffer-cache reads: 64 MB in 0.59 seconds =108.47 MB/sec
>> Timing buffered disk reads: 32 MB in 1.91 seconds =16.75 MB/sec
Hardware wise, the only thing that changed was the addition of a DPT
SmartRAID V intelligent caching controller with 16MB cache on-board.
Software wise, what changed was that we removed FreeBSD 3.1-STABLE+vinum
and replaced it with Debian Linux 2.0.36 which was then upgraded to 2.2.6
so that we could compile and install the source version of the DPT
SmartRAID V device drivers for Linux.
The machine itself is a PII@400Mhz (although I don't know how much L2
cache or at what speed), 256MB RAM, an internal Western Digital 4GB SCSI
hard drive (connected off the same Adaptec 2940 controller we were using
under FreeBSD), externally connected through the DPT SmartRAID V to a
Kingston drive array with five Quantum Viking II 9GB drives and four
Western Digital WDE9100 9GB drives.
Now, based on these results, I guess I could claim that FreeBSD
3.1-STABLE+vinum is about twice as fast with the striping done in software
as Debian Linux 2.0.36->2.2.6 is with the striping done on a brand-new
high-speed intelligent caching controller with 16MB of cache.
Of course, if I were so stupid as to make that kind of a claim, I'd
expect to be laughed out of the building, and in fact would laugh myself
out of the building.
Nevertheless, if I can't manage to find some way to get some more
useful benchmark numbers out of Linux (preferably using the same tool as
I've previously used under FreeBSD, so that the tests have at least some
validity), I may be forced to make statements based on my best guess of
what real system performance is like using the only numbers I've got.
--
Brad Knowles <[EMAIL PROTECTED]> <http://www.shub-internet.org/brad/>
<http://wwwkeys.pgp.net:11371/pks/lookup?op=get&search=0xE38CCEF1>
Are you looking for a news feed from a site in the Freenix Top 130?
If so, contact me via private e-mail for details.
------------------------------
From: [EMAIL PROTECTED]
Subject: Re: BRU v. arkeia
Date: Fri, 30 Apr 1999 17:40:28 +0059
1) I have not tried BRU yet although I shall be doing so in the next few
days.
2) You can download a demo copy of Arkeia. I have tried it and it is a very
capable package. On the basis of emails with the manufacturers of BRU it
seems that Arkeia is much more capable in many areas. Its ability to backup
NT boxes over the network worked very well for me. Overall, I was impressed
by Arkeia and would have bought it but for the fact that there was no way to
schedule an automatic incremental backup based on a previous manual backup
(I confirmed this in the course of several emails with Knox).
Unfortunately, that is a capability which I require and which BRU claims to
provide.
Hope that helps,
--
============================================================================
Richard Simpson
Farnborough, Hants, Uk Fax: 01252 392118
[EMAIL PROTECTED]
I am not aware of any views shared by myself and my employers.
------------------------------
** FOR YOUR REFERENCE **
The service address, to which questions about the list itself and requests
to be added to or deleted from it should be directed, is:
Internet: [EMAIL PROTECTED]
You can send mail to the entire list (and comp.os.linux.misc) via:
Internet: [EMAIL PROTECTED]
Linux may be obtained via one of these FTP sites:
ftp.funet.fi pub/Linux
tsx-11.mit.edu pub/linux
sunsite.unc.edu pub/Linux
End of Linux-Misc Digest
******************************
