Re: Disk geommetry, was Re: Kernel Upgrade: Why?

1999-04-24 Thread Marsh Ray
On Fri, Apr 23, 1999 at 08:50:40PM -0400, Dan Willard wrote:
> 
> Think of it as a record (ya' know those old odd looking vinyl things).
> It spins at 33.3 rpm but the sound/music doesn't change from outer to inner.
> Same deal with hard drives although the outside is 'spinning' faster, it
> still picks up the same amount of data per rotation as it would near
> the center.

Actually, this is a good analogy, but the conclusion is wrong.

Hi-fi buffs (back when it was called hi-fi) knew that there was better
sound on the first tracks of LPs (near the outside).  This was because
the vinyl was moving faster under the needle, which gave a better frequency
response at the high end i.e., more data about the music.  In fact,
audio CDs slow their rotation on the outside tracks so they can play
back the data at a constant rate.

Hard drive manufacturers work in an incredibly competitive market. They
do their best to maximize the bits-per-square-centimeter density of the
platters in your drive. One great way they have found to do this is to
store more data on the outer tracks.

Older hard drives, MFM, RLL, early SCSIs and PC-format floppies don't
take advantage of this, however.  My recollection is that the original
Mac 400K floppy was the first "personal" data recording device to use this
technique. I'm sure someone else can point out something the ancient
Egyptians did with it.

As a result, you get a much higher data rate on the outer tracks, with
the same angular latency as on the inner tracks.

- Marsh


Re: Disk geommetry, was Re: Kernel Upgrade: Why?

1999-04-22 Thread Marsh Ray
On Wed, Apr 21, 1999 at 03:56:31PM -0700, debian-user list wrote:
> The HOWTO, and the person responding to the Mindcraft survey, both assert 
> that the part of the disk furthest from the spindle is the fastest, and that 
> one can use this to optimise performance.  While I don't doubt the first part 
> of the statement, I was under the impression that once you hit a drive's 
> onboard logic, geometry is pretty much up for grabs these days, and 
> attempting to put something on the "outer edge" of the disk is an excerise in 
> self-delusion.  The HOWTO is dated 1997, and the information therein may be 
> older.  Does anyone out there have some knowledge of current hard drive 
> manufacturing?  

It's true that you can't gather much from the sector and head numbers on
a modern drive, but I think common drives can be expected to access mostly
sequentially across the cylinders.  I've always heard that the lower-numbered
cylinders were on the outer tracks.  How many filesystems start packing
data in at the beginning?  Would a drive manufacturer want to optimize or
decrease performance in that situation?


> Or better yet, some numbers from formal or informal experiments in drive 
> partition performance? 
Yes, I'd like to see this too.  Wouldn't be hard to do, but I don't have a 
spare drive at the moment.

- Marsh


Re: spam & debian-user?

1999-04-20 Thread Marsh Ray
I use a e-mail alias specific to the debian lists:
([EMAIL PROTECTED])
If it ever starts picking up too much spam, I'll cancel this alias and
resubscribe with a new one.

- Marsh

On Mon, Apr 19, 1999 at 05:09:17AM -0700, Lev Lvovsky wrote:
> I noticed thast this list gets posted to usenet...what do ppl on the list
> do to avoid gettting their e-mail addresses getting picked up by spam
> bots?  does the system get rid of this possiblity somehow?


Re: Some email bounces = misconfigured exim?

1999-04-17 Thread Marsh Ray
Maybe you have a ppp connection like me?
I run qmail from home, and I have a file in the ip-up directory like this:
This sets the 'HELO' response to the correct reverse lookup is for my IP.
You may want to try something similar/different for your MTA.

--begin '/etc/ppp/ip-up.d/6_set_qmail_hostname'--
#!/bin/sh

# copies the new hostname to the qmail control file
QCD=/etc/qmail
ME=`host $PPP_LOCAL | head -1 | cut -d" " -f2`
echo "$ME" > $QCD/helohost
---end '/etc/ppp/ip-up.d/6_set_qmail_hostname'--

A similar file in ip-down.d/ removes the helohost file on disconnect.

I found this technique in a page linked off the qmail pages.
Hope this is useful.

- Marsh


On Fri, Apr 16, 1999 at 03:48:11AM -0400, Arcady Genkin wrote:
> George Bonser <[EMAIL PROTECTED]> writes:
> > Here is your problem:
> > corsica# nslookup main.wgaf.net
> 
> Thanks. I kinda understood that. However, I don't have a permanent IP, 
> forget the domain name. wgaf.net - made up domain name for my home
> LAN. But I still want to have this address for the internal mail (from 
> system services and other users).


Re: /dev/nst0 hosed

1999-04-11 Thread Marsh Ray
On Sun, Apr 11, 1999 at 02:49:30AM -0500, Chris Hoover wrote:
> Somehow my /dev/nst0 device has become messed up.  Could someone
> please tell me what it should be?

bash-2.01$ ls -l /dev/nst0
crw-rw   1 root tape   9, 128 Jul 20  1998 /dev/nst0

> Also, how do I use /dev/MAKEDEV to recreate it.

/dev/MAKEDEV st0


Re: 2 Questions

1999-04-11 Thread Marsh Ray
On Sun, Apr 11, 1999 at 02:30:05AM -0500, Chris Hoover wrote:
> 1.  How can I get my machine to automatically start fetchmail after
> a reboot?  Also, how do I have it ran by a user other than root
> (it is only setup to run as me right now)?

Debian's boot process is fairly well-described at:
http://www.debian.org/doc/FAQ/debian-faq-12.html#ss12.5

Change the owner of your script to you and set its setuid bit.

$ chown myusername thefilename
$ chmod u+s thefilename

Be extra careful about access permissions when you do this because
other users may be able to execute this script as you.

> 2.  Is there a way to script the startup of a remote program?  I want
> to have wmmon and wmppp start up every time I log into my machine.
> However they are located on my other linux box.

You can use ssh (or rsh for that matter) to connect and run the file.  This can 
be kind of complicated to set up since you don't want to have to type your 
password.  But basically, you'd be putting a command like 'ssh my.machine.name 
my_command_name' in your .bashrc.  The target computer would need to be running 
sshd.  On the local computer ask ssh to generate a key pair (ssh-keygen), and 
append the public key it generates to the file ~/.ssh/authorized_keys on the 
remote computer.


Re: ppp connection speed

1998-08-06 Thread Marsh Ray
Try putting the 'REPORT CONNECT' at the beginning?

- Marsh

>/usr/sbin/chat -v -r /home/patricko/speed   \
>TIMEOUT 60  \
>ABORT   '\nBUSY\r'  \
>ABORT   '\nNO CARRIER\r'\
>ABORT   '\nNO DIALTONE\r'   \
>ABORT   '\nRINGING\r\n\r\nRINGING\r'\
>''  ATDT$1  \
>CONNECT ''  \
>REPORT CONNECT  \
>ogin:--ogin:$ACCOUNT\
>assword:$PASSWORD



--  
Unsubscribe?  mail -s unsubscribe [EMAIL PROTECTED] < /dev/null


Re: Automatic X startup

1998-08-06 Thread Marsh Ray

On new hamm systems, all I had to do was change the 
line in '/etc/X11/config' from 'no-start-xdm' to 'start-xdm'.

- Marsh

From: Ossama Othman <[EMAIL PROTECTED]>

>> Could anyone tell me how to get X automatically started. Now I have to
>> write 'startx' every time.
>
>Try configuring xdm to start at boot.  You can probably use
>xbase-configure to configure it.  If that doesn't work check /etc/init.d
>for a script called xdm.  If you don't have xdm but have something like
>"xdm-new.dist," than move that file to /etc/init.d/xdm.  The next time you
>boot, xdm should start X for you and provide you with an X login
>prompt/window.



--  
Unsubscribe?  mail -s unsubscribe [EMAIL PROTECTED] < /dev/null


Re: hardware purchase advice: modems

1998-08-03 Thread Marsh Ray

>I would like a 56k but am not sure what to get. I do NOT want to be
>stuck in a winmodem hell. Have the problems with 56k modems been 
>resolved (ie is there a standard yet?). What is good to get
>(ie it must not need any special "windows" driver but...they ALL say
>"made for Windows 95" so..its hard to tell whats best)


I always buy external modems.  You can be sure that it's not a
"winmodem", and my personal belief is that they offer better
lightning/surge protection.

>suggestions? any good brands? any brands to stay away from?
>(alternately...where can I get a good $$ ;) )


Some tests have suggested (www.boardwatch.com) that the
3Com/USRobotics 56K/V90 models have superior 56k performance
in many cases.

- Marsh


--  
Unsubscribe?  mail -s unsubscribe [EMAIL PROTECTED] < /dev/null


Re: Hamm and 3com's 905 PCI.

1998-07-29 Thread Marsh Ray

My experience was that:

3Com is now selling the 3c905B card.  Notice the "B".  Drivers
for this appeared in the 2.0.34 kernel.

The boot-floppies I used had 2.0.33.

- Marsh



-Original Message-
From: Liran Zvibel <[EMAIL PROTECTED]>
To: debian-testing@lists.debian.org ; Debian
Mailing List 
Date: Wednesday, July 29, 1998 12:25
Subject: Hamm and 3com's 905 PCI.


>Hi.
>After using Debian 2.0 at home (and enjoying it very much) I decided to
>bring it to work, install it on some computers and show the world that
>Debian is better then both Win.NT and RedHat5.1.
>Well, I didn't bring the CD with me since I knew that I can have it here
>through NFS.
>The problem is that the installation floppies (the drv1440, I mean)
>doesn't have 3com 90x PCI drivers, and I can't work. (I'm writing this
>from my other computer that has RedHat (that will hopefully become
>Debian as soon as I can get it to work with the network card)).





--  
Unsubscribe?  mail -s unsubscribe [EMAIL PROTECTED] < /dev/null


Re: Graphical Linux Logo on boot

1998-07-23 Thread Marsh Ray
From: Curt E. Spann <[EMAIL PROTECTED]>

> I am wondering if there is any Linux version of the boot screen like in Win95.
> So as linux is booting instead of seeing all that informtion it will be
> a graphical image of Linux or something. If there isn't one I would like
> to make one.  Is it even possibly to make it?  Will I have to modify the
> kernel or what?

There has been plenty of discussion about this over on the linux kernel
development lists.  In general, it's easy to do for any specific
hardware configuration, (Intel with VGA is all Win95 supports) but
no way to do it across platforms. Maybe if/when the KGI/GGI stuff makes
it into the kernel, this capability will be added.

If this is important for your application, someone has probably written
a kernel patch.

- Marsh


--  
Unsubscribe?  mail -s unsubscribe [EMAIL PROTECTED] < /dev/null


Re: unsubscribe "helper line"

1998-07-09 Thread Marsh Ray

On Thu, Jul 09, 1998 at 01:50:13AM -0400, Daniel Martin at cush wrote:
>
>>> Unsubscribe?  mail -s unsubscribe [EMAIL PROTECTED] <
/dev/null
>
>> Does this message strike anyone as significantly less helpful than the
>> former "mail a message with the subject "unsubscribe" to
>> [EMAIL PROTECTED]" message?  I'll grant you that
>> people screwed that up too, but this seems to be asking for people to
>> get confused.


Let's keep in mind that many folks won't be reading this
list on a Debian (or even Linux) system.  However, we must not
judge them for this, we can only pity them, for the wheels of
destiny have inflicted many a misfortune upon otherwise noble souls.

However, if the unsubscription instructions aren't "cross-platform",
people are sure to start posting in frustration to the list itself.

- Marsh




--  
Unsubscribe?  mail -s unsubscribe [EMAIL PROTECTED] < /dev/null


Re: What's a good video card?

1998-07-08 Thread Marsh Ray

>*-Shaleh ( 7 Jul)
>|
>| price range.  As a general comment I say avoid Diamond.  For a long time
>| they have not supported any form of driver for their cards (even their
>| Windows support is not all that great).

From: [EMAIL PROTECTED] <[EMAIL PROTECTED]>
>Didn't this occur around 3 or 4 years ago?

Yes, but . . .
To this day we still don't have working Win95 drivers
for their "Speedstar Pro VL" product.  Our company received
several of these in new computers.  If they're going to abandon
their own products, my company will not be purchasing any more
of them.

Story goes they got into a finger-pointing contest with
Microsoft and Cirrus Logic over who should write the drivers.

- Marsh



--  
Unsubscribe?  mail -s unsubscribe [EMAIL PROTECTED] < /dev/null


Re: Getting notebook internal modem to work.

1998-07-07 Thread Marsh Ray
From: Geoff Brimhall <[EMAIL PROTECTED]>


> I recently got a rather brilliant pc notebook, and have
> been able to utilize just about everthing it has through
> debian and recompiling the kernel.

> Except for one thing - the internal modem.

> setserial does not detect the modem's COM port,
> nor the irq (it does register the regular
> external serial port though). 

> In fact, something rather strange is going on
> with the modem because even Windoze did not
> register the modem's COM port until the driver
> was properly loaded for it. 

> What Windoze registered, before the driver, was
> the DMA and IRQ of the device - but really had
> no idea what to do with it until the driver was
> loaded. Then suddenly the COM port was present.

> Any ideas on how to get it to work ? I've tried
> hard-coding setserial to use the IRQ and COM ports
> for the modem, but it just does not register
> anything. Somehow linux needs to link more
> tightly the DMA to the IRQ ?

Sounds like you have either a "winmodem" (which
will probably never work under Linux) or a PCMCIA
modem (which requrires kernel support and utilities.)

What model notebook?  Does the modem have a model?

- Marsh



--  
Unsubscribe?  mail -s unsubscribe [EMAIL PROTECTED] < /dev/null


Re: Driver bug? 56k USR doesn't transmit properly?

1998-07-06 Thread Marsh Ray
You might try "AT&F&B1" in place of "atz".  I have found
that most USR Sportsters (contrary to documentation) require
&B1 to lock DTE rate (serial speed) upon connection.
The &F ensures you're starting from the factory defaults
every time.

- Marsh


>>ABORTBUSY
>>ABORT"NO CARRIER"
>>ABORTVOICE
>>ABORT"NO DIALTONE"
>>""   atz
>>OK   ATW1M0DT2823500
>>CONNECT  \c
>>ogin i421441
>>word \q
>>""   \q



--  
Unsubscribe?  mail -s unsubscribe [EMAIL PROTECTED] < /dev/null


Re: ppp analog bonding w/ eql

1998-06-26 Thread Marsh Ray

From: Hamish Moffatt <[EMAIL PROTECTED]>

>On Thu, Jun 25, 1998 at 10:26:48PM -0700, George Bonser wrote:
>> they both need to be the SAME IP Address for EQL to work.
>
>Hmmm. I got the impression that it should work on any link. They use
>plip as an example in the doco, I think. I did try it with the same IP
>on both links (a bit weird on ethernet but anyway) and didn't seem to 
>get anywhere further. I'll try again soon and check out the kernel list.


The thing with eql is you need a linux box on both ends of the
connections.  Our company put a linux box out at our ISP's site
so we can use eql.  They're a small ISP and easy to work with.

Other ISPs might support "multilink PPP" but probably not eql.

- Marsh


--  
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]


Re: Once burned...

1998-06-25 Thread Marsh Ray
I just set up a machine sort of like that one . . .

From: [EMAIL PROTECTED] <[EMAIL PROTECTED]>
>I have a Logitech TrackMan Marble (same connector as the standard PS/2 port
>mouse),

use /dev/psaux, I think you need to load a kernel module.

>an HP DeskJet 820 CSE on LPT1,
ATI Video Expression+ (Direct
>Draw) video card,

Should work fine for X, think of it as an "ATI Mach 64"

>and a Sound III 336SP Sound Card

Probably SB16 compatible

>and 333 modem combination (OEM stuff).

Watch out . . this might be one of those "WinModem" type
things which only works under Microsoft Windows!

>The other Linux either didn't recognize any of
>them or I couldn't figure out how to make it; actually I never had the
>chance to try the modem.
>
>The other Linux did recognize both my hard drives and my CD.
>
>Sorry to be so wordy but I'd rather not set it all up again only to find I
>can't use it. Anyway, is there a way to find out specifically if Debian 2
>will be able to support my hardware?


see http://sunsite.unc.edu/LDP/HOWTO/Hardware-HOWTO.html

Usually device support is a kernel rather than a distribution
issue.  Debian 2.0 beta comes with a fairly recent kernel,
but you may still want to build your own kernel.  I needed to update to 2.0.34
in order to support my 3c905b.

- Marsh



--  
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]


Date Disparities

1998-06-23 Thread Marsh Ray
I'm having an interesting experience with one of my
Debian machines.  I've got a few, all synchronized
with ntp.  One of them always reports the time as
20 seconds slow.

As an example, I do this:
---

radha# date ; date +%s
Tue Jun 23 09:41:48 EDT 1998
898609308

--- and at the same time I do this

krishna# date ; date +%s
Tue Jun 23 09:41:28 EDT 1998
898609308

--

Radha returns the correct time, Krishna is 20 seconds
late. Please note that I am getting 2 different time
strings for the same time integer.

I'm sure somebody knows what's going on.  They're
both running bo with libc5.4.38-1 and
shellutils 1.16-2.

Maybe Krishna doesn't know about leap-seconds?

TIA

- Marsh


--  
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]