Linux-Networking Digest #432, Volume #10          Tue, 9 Mar 99 05:13:47 EST

Contents:
  Re: DNS, Mail, News and Internet access (Clemens Heise)
  Re: How do I boot Monitor-less 486 without a login and then... ? ("Peck R. Wood")
  Re: For all you Nicrosoft lovers (Wildman, the Cuberstalker)
  Linux Dial on Demand ([EMAIL PROTECTED])
  Do I install Kernel 2.0 or wait for 2.2 ("Joe")
  Re: Why the Bottleneck? ([EMAIL PROTECTED])
  Re: 3c574 PCMCIA problems (Chuck Slivkoff)
  weird module behavior and pppd rh5.2/2.2.2 (tedd)
  Re: Will Linux work for me? ("Frank Grunenberg")
  Re: IP watching between 2 Hosts (Clemens Heise)
  Re: Linux and Windows NT server? ("Frank Grunenberg")
  is my nic working? ("Gareth Lloyd")
  Re: Do I install Kernel 2.0 or wait for 2.2 (Bob Tennent)
  Strange Networking Problem (Steve Clark)

----------------------------------------------------------------------------

From: [EMAIL PROTECTED] (Clemens Heise)
Subject: Re: DNS, Mail, News and Internet access
Date: Tue, 09 Mar 1999 07:52:12 GMT
Reply-To: [EMAIL PROTECTED]

Cameron !
On Mon, 08 Mar 1999 16:38:13 +0000, Cameron Mulliner <[EMAIL PROTECTED]> wrote:

>I am currently running Redhat 5.2 with the 2.2.2 kernel. This PC
>acts as a DNS server, Mail server and an gateway to a Cisco 1603
>router which connects to my ISP. The way it works at the moment is
>that windaze95 workstations lookup the linux PC and packets that are
>meant for none local recipients are sent to the default gateway which
>is the Cisco router. This ISDN router dials the ISP and forwards all
>packets. Mail is configured in a similar way.
>
>Is this the quickest way to connect to the internet and send mail via
>the
>internet or are their quicker ways such as IP forwarding and
>Masquerading?

The Config seems to be OK. Sure, you can add a bit more security by using NAT and IP 
Forw. But the scheme is the same.
All Packets for non local hosts, were sent to a default gateway. And this decides what 
to do with it. 

>Also are there any benefits with having a local news server as well. If
>there are
>how do I configure one?

You can use your Linux box for setting up a NNTP server. Just use INN in stead of 
CNEWS. If you have questions for
configuration, mail to my personal account.


---
Clemens Heise
eMail [EMAIL PROTECTED] | http://www.conetix.de/~heise

------------------------------

From: "Peck R. Wood" <[EMAIL PROTECTED]>
Subject: Re: How do I boot Monitor-less 486 without a login and then... ?
Date: Tue, 09 Mar 1999 02:47:54 -0500
Reply-To: [EMAIL PROTECTED]

This is a multi-part message in MIME format.
==============41042DD36AC61D9ACD469448
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit



IanP wrote:
> 
> I have a shiny new PII with two hard disks, the first with windows 98
> and the second with Redhat 5.1. networking (soon) with my faithful old
> 486.  I want the 486 to boot without asking for a login and password,
> being monitorless. 

Once you set up the networking to where you boot the 486 and access it
from win98 the way you want, you can unplug the monitor, keyboard and
mouse.  You can set up whatever services you want on the 486, samba,
apache, etc. and they will all run without you having to login.  Once
you can telnet to the box, you can even reboot it remotely and wait for
it to come back up so you can telnet in.  There are some programs that
require the console (monitor, keyboard, etc.) but there are equivalent
programs that don't for most of them.

 The purpose being that I want to telnet (or if there
> is something better...?) from Windows on the PII to a minimal Linux
> installtion on the 486 which will have the PII's Linux partitions
> mounted on it allowing me to use the programs on the PII.

Not sure what you mean here.  If you can mount the partitions using nfs
(is there nfs for win98?), you still wouldn't be able to use programs
for linux under win98. This would be no problem (not as much of a
problem) when you boot up the redhat partition on the PII.
> 
> Is there a more efficient way to do this?  I would appreciate any
> pointers in the right direction, thanks.
> 
> Ian P.
> 
> [EMAIL PROTECTED]
> [EMAIL PROTECTED]
==============41042DD36AC61D9ACD469448
Content-Type: text/x-vcard; charset=us-ascii;
 name="peckrwood.vcf"
Content-Transfer-Encoding: 7bit
Content-Description: Card for Peck R. Wood
Content-Disposition: attachment;
 filename="peckrwood.vcf"

begin:vcard 
n:R. Wood;Peck R. Wood
x-mozilla-html:FALSE
version:2.1
email;internet:[EMAIL PROTECTED]
x-mozilla-cpt:;0
fn:Peck R. Wood R. Wood
end:vcard

==============41042DD36AC61D9ACD469448==



------------------------------

From: [EMAIL PROTECTED] (Wildman, the Cuberstalker)
Crossposted-To: alt.os.linux,comp.os.linux.hardware,comp.os.linux.setup
Subject: Re: For all you Nicrosoft lovers
Date: 9 Mar 1999 03:13:47 GMT
Reply-To: [EMAIL PROTECTED]

On Mon, 08 Mar 1999 22:29:35 +0000, DanH <[EMAIL PROTECTED]>
wrote:
>Well, it is a good way to break the ice and wake people up about what M$
>is doing and opening that there are "other" operating systems out there.

And posting it to Linux newsgroups is the way to do this? Talk about
preaching to the choir...

-- 
The Wildman     ICQ# 32609427
Fight spam - http://www.spamfree.org
Five is a sufficiently close approximation to infinity.

------------------------------

From: [EMAIL PROTECTED]
Subject: Linux Dial on Demand
Date: Tue, 09 Mar 1999 03:12:18 GMT

I have a small 486, running as an ISDN router for a small subnet in my home
office. I have reserved static IP addresses from my ISP.

The router is set up with kernel 2.2.2, with ppp running in dial-on-demand
mode. Everything works fine on the front.

The problem is, that when I try to access one of the computers on the subnet,
from another using telnet (or whatever), the router dials and makes the
connection to the ISP, *even though I am accessing a local address*. Even
worse, if the link is down for any reason, then I am unbale to telnet to any
other machines on the local network (It connects, and waits indefinately for
the PPP link to come up).

Subnet is xxx.xxx.113.225 through xxx.xxx.113.239, subnet mask 255.255.255.240
Default gateway xxx.xxx.113.225 - (i.e. the router)

Please help, this is costing be a fortune!


Cheers

Richard Turnbull [EMAIL PROTECTED]

Tel +65 252 0287





============= Posted via Deja News, The Discussion Network ============
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

------------------------------

From: "Joe" <[EMAIL PROTECTED]>
Subject: Do I install Kernel 2.0 or wait for 2.2
Date: Mon, 8 Mar 1999 22:14:02 -0500

I am looking to install Linux on a machine.  I figure using Red Hat or
Cardera etc. would be easier than doing it on my own.  Should I wait for ver
2.2?  If so when do you think it will be available by the distributors/

Doing it without the distributor packages sounds interesting.  I have no
experience in Linux.
Thank for your advice.

--
Joseph Norcott
Business Technologies Of New England, Inc.
[EMAIL PROTECTED]
www.btne.com



------------------------------

From: [EMAIL PROTECTED]
Subject: Re: Why the Bottleneck?
Date: Tue, 09 Mar 1999 01:35:19 GMT

In article <[EMAIL PROTECTED]>,
  [EMAIL PROTECTED] (Jerry Gardner) wrote:

> I have a machine that dual-boots Linux 2.0.36 and FreeBSD 3.1. This
> machine is connected to another machine running WinNT 4.0 on an
> isolated ethernet. Both machines have 3Com 3C905B NICs.

> When I transfer large (~100MB) files between the FreeBSD box and the
> NT box, the throughput averages 1000-1100 KB/sec. When I reboot into
> Linux and FTP the same size files, the throughput drops to 600-700
> KB/sec. This is on an isolated 10BaseT ethernet with *no* other
> traffic.

> Why is Linux network performance so much less (30-40%) than FreeBSD
> performance in this case? Where is the bottleneck? Is there some
> tuning I need to do to fix this?

The "tuning" you'd need to do is to throw away the Linux networking code and
replace it with 4.4BSD-Lite(2) or some variant.  Sorry, but Linux simply
wasn't designed to handle high amounts of network traffic.  I've run into the
same problem more times than I can count, and the problem you describe is
*precisely* the type of problem that forces large internet sites like Yahoo!
and XOOM.com to use FreeBSD rather than Linux.

The problem you describe is most likely caused by poor network buffering on
the part of the Linux box.  FreeBSD buffers network I/O in structures called
"mbuf clusters", which are essentially just chunks of pre-malloc`ed memory
that are chained together to form buffers.  Linux uses what are called
"skbuff" structures, which are variable length linear buffers that tend to
perform better under most circumstances but which waste vast amounts of
memory under high network load.  Unless I'm mistaken, the skbuffs also need
to be malloc`ed every single time one is needed, which adds considerable
overhead (relative speaking, of course).  The mbuf clusters are never
deallocated, which can be a bad thing in some cases (like when you want that
memory _back_, damnit! :) but which speeds up network buffering like you
wouldn't believe.

I'm actually suprised to find that there's a "reasonable" situation in which
FreeBSD outperforms Linux, since I'd only ever seen these types of
differences crop up when machines were pushing on the order of 10-15Mb/sec
(that's bit, not byte!).  In fact, the numbers you quote make me think that
something *else* is causing at least part of the problem, and that you might
be able to improve performace slightly by modifying some of the associated
network structures (look for references to "skbuff" in kernel configuration
files).

Of course, since you're obviously able to use both FreeBSD and Linux, this
question is probably more academic than practical, in which case I hope I've
provided at least a little insight :).

-Bill Clark

============= Posted via Deja News, The Discussion Network ============
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

------------------------------

From: Chuck Slivkoff <[EMAIL PROTECTED]>
Subject: Re: 3c574 PCMCIA problems
Date: 9 Mar 1999 03:23:47 GMT

Tom Roper wrote:
> 
> I have installed RH 5.2 (2.0.36) on my Thinkpad 760c.  I have a 3Com
> 3c574TX PCMCIA NIC.  I can ping the address I give to the card but
> nothing else.  When I boot, the card and the hub indicate connection,
> however pinging any other address fails.  When I run the route
> command, after a few minutes the default gateway will show up.
> Ifconfig shows IRQ 10  I/O 300h.  I tried changing the IRQ, nothing
> has worked.  Has anyone else had this problem?  I have checked the
> PCMCIA pages, portable pages, and FAQs.  Nothing has helped.  Any
> idea??

Get the 3.0.9 pcmcia package, and roll the 3c574_cs.c file back to the
3.0.4 pcmcia version.  I have a 3CCFEM556 combo-card which uses the same
3c574 driver. The driver is still considered "buggy", though.  Keep an
eye on the message forums here for updates:

  http://hyper.stanford.edu/HyperNews/get/pcmcia/home.html

-chuck

------------------------------

From: tedd <[EMAIL PROTECTED]>
Subject: weird module behavior and pppd rh5.2/2.2.2
Date: Tue, 09 Mar 1999 08:47:04 +0000


Hi.

I'm using pppd version 2.3 patch level 5 installed as part of a stock redhat 5.2
x86 system.

After I installed kernel 2.2.2 (which I build myself, not a rpm), I start a ppp
connection and am able to do everything I was under 2.0.36, but I find that pppd
causes the following output in syslog:


Mar  8 22:01:07 localhost chat[426]: Password:
Mar  8 22:01:07 localhost chat[426]:  -- got it 
Mar  8 22:01:07 localhost chat[426]: send (XXXXXXXXX^M)
Mar  8 22:01:08 localhost pppd[424]: Serial connection established.
Mar  8 22:01:09 localhost pppd[424]: Using interface ppp0
Mar  8 22:01:09 localhost pppd[424]: Connect: ppp0 <--> /dev/modem
Mar  8 22:01:11 localhost modprobe: can't locate module ppp-compress-21
Mar  8 22:01:11 localhost modprobe: can't locate module ppp-compress-26
Mar  8 22:01:11 localhost modprobe: can't locate module ppp-compress-24
Mar  8 22:01:11 localhost pppd[424]: local  IP address 209.110.237.95
Mar  8 22:01:11 localhost pppd[424]: remote IP address 165.236.178.55
Mar  8 22:01:11 localhost modprobe: can't locate module ppp-compress-21
Mar  8 22:01:11 localhost modprobe: can't locate module ppp-compress-26
Mar  8 22:01:11 localhost modprobe: can't locate module ppp-compress-24
Mar  8 22:01:41 localhost pppd[424]: CCP: timeout sending Config-Requests


What are these weird ppp-compress-2? modules and what is the last line about? 
Everythings seems to work fine but I haven't been able to figure 
out what's going on here.

I think I've got everything set up nicely, mirroring the way the 2.0.36 kernel is
set up by default using the /boot partition.

I chose to use the kernel thread to load modules and it's set up to find modprobe
correctly.

I'm using modutils 2.1.121.


-- 

 .:.
  :                             [EMAIL PROTECTED]

------------------------------

From: "Frank Grunenberg" <[EMAIL PROTECTED]>
Subject: Re: Will Linux work for me?
Date: Tue, 09 Mar 1999 08:53:39 GMT


>We have in the office:
>   12 nt workstations primarily running autocad.
>    1 accounting computer
>    1 computer for a print server
>    1 computer for the fileserver and internet connection.
>
>The fileserver;
>      PII-300
>      3-com pci network card
>      US robotics 56k modem
>      HP7200i re-writable cd-rom
>
>
>We do daily backup onto the cdrom and have modem sharing
>software so 3 of the computers can access the internet at the
>same time.
>I personally have limited experience on hp-ux workstations.
>Any feedback would be appreciated.
>
>Thanks in advance.
>Walter


This may sound kinda silly, but what are you trying to accomplish?

If your goal is to have net access for all 12 sure no problem.
If your goal is for file/printer sharing no problem...
As far as the backing up I wouldn't mess with that (moving the CD R) until
you are familiar with the Administration of  U*nix

Each flavor is slightly different, but can accomplish the same things...
Linux is one of the better U*nix varients for drivers. where as some other
might cause you more greaf..  Try determining the Hardware compatibility of
your system go to http://www.linux.org and determine you systems potential
"problem" hardware.  If you find the important pieces are missing, then
wait, howevr if you have the necessary hardware then go for it.



------------------------------

From: [EMAIL PROTECTED] (Clemens Heise)
Crossposted-To: 
comp.os.linux.admin,comp.protocols.tcp-ip.domains,de.comm.protocols.tcp-ip
Subject: Re: IP watching between 2 Hosts
Date: Tue, 09 Mar 1999 07:37:24 GMT
Reply-To: [EMAIL PROTECTED]

Barry !
On Mon, 08 Mar 1999 17:16:50 GMT, Barry Margolin <[EMAIL PROTECTED]> wrote:
[...]
>The -p option specifies that it shouldn't go into Promiscuous mode, so it
>will only see traffic to or from itself.  Don't use that option if you want
>to see traffic of other hosts.

Thank you for assistance but it doesnt work. Let me explain:
tcpdump ist installed on Host A. I want to listen on IP between Host X and Y all in 
the same (sub)net. I tried the
Host A> tcpdump host X and host Y
and 
Host A> tcpdump -qni eth0 host X and host Y
But tcpdump takes no effect. When I try the 
Host A> tcpdump -qni eth0 host A and host Y
everything works fine. 

Any ideas ?

>Also, if you have a switching hub, it won't normally send traffic to your
>port if it's addressed to the MAC address on a different port.  Some
>switches allow you to put a port into "watch" mode so it sees the traffic
>for other ports.

No this net is not switched, just simple Hups between the nodes.
Thanks in advance for your assistance


---
Clemens Heise
eMail [EMAIL PROTECTED] | http://www.conetix.de/~heise

------------------------------

From: "Frank Grunenberg" <[EMAIL PROTECTED]>
Subject: Re: Linux and Windows NT server?
Date: Tue, 09 Mar 1999 09:05:21 GMT


agm wrote in message <[EMAIL PROTECTED]>...
>I've set up a home network with one computer running Windows NT4
>server (PDC) and the other one Linux with the 2.2.0 kernel and Samba
>2.0.0. The problem is that it's only possible to connect to the Linux
>computer from Win NT. When I try to connect to Win NT with smbclient
>-L <hostname>, I get error message "access denied" after typing the
>password.
>I've edited the registry of NT to enable plain text password.
>
>I was able to make a good network connection from Linux to Win 98 in
>the past with samba, what am I doing wrong this time?


As silly as this sounds make sure the user account exists that you are using
to log into NT with...
Then make sure the user account is active...
Then make sure the unix side is set-up correctly...  like the other
gentelmen suggested.
Just a few suggestion...  I know this from 98 networking with NT...  Forget
to type in an actual user...
Just for 2 machines makes no sence when both are in 98 to have passwords and
such...
Heck the machines are right next to one another...
Just a few suggestions...



------------------------------

From: "Gareth Lloyd" <[EMAIL PROTECTED]>
Subject: is my nic working?
Date: Mon, 8 Mar 1999 22:51:50 -0500

whats the easiest way to tell if my RH 5.2 machine has configured my ISA
linksys ether16 card correctly? if i do a ping on 127.0.0.1 it does the
loopback test correctly, i'm just not sure that my nic is working. is there
a utility that will 'probe' my nic and tell me what it thinks i have?

thanks
Gareth Lloyd



------------------------------

From: r d t@c s.q u e e n s u.c a (Bob Tennent)
Subject: Re: Do I install Kernel 2.0 or wait for 2.2
Date: 9 Mar 1999 03:42:53 GMT

On Mon, 8 Mar 1999 22:14:02 -0500, Joe wrote:
 >I am looking to install Linux on a machine.  I figure using Red Hat or
 >Cardera etc. would be easier than doing it on my own.  Should I wait for ver
 >2.2?  If so when do you think it will be available by the distributors/
 >
 >Doing it without the distributor packages sounds interesting.  I have no
 >experience in Linux.
 >
Unless you have a specific requirement that only the 2.2 kernel
will satisfy, there's no need to rush into 2.2.  I suggest you
get a good modern distribution and get used to Linux.  If you
want, you can follow the instructions at the Redhat site and
compile yourself a 2.2 kernel.  Not a big deal, you'll have
to upgrade a bunch of packages, but they're all readily available.
Again, unless you really need a new functionality, you're not likely
to notice any difference with a new kernel.

Bob T.

------------------------------

From: Steve Clark <[EMAIL PROTECTED]>
Subject: Strange Networking Problem
Date: Tue, 9 Mar 1999 08:45:49 -0000

I am trying to get a Linux PC (Dell Optiplex running Red Hat 5.1) to
talk to a Windows NT Server over a UTP Ethernet network but am having
some major problems.

The Linux PC has an onboard 3Com 3c905B (Cyclone) NIC. The kernel module
for this (3c59x.o) loads fine and displays all of the relevant
information.

The interface starts up OK (running 'ifup eth0').

However, I cannot ping the NT Server (which can, by the way communicate
with several other Unixware, Windows NT and Windows 95 PC's).

After attempting to ping the NT box (by IP Address), the arp table shows
the MAC address as 'incomplete'!

If I try the ping from the NT server to the Linux PC, this also fails.
The arp tables on the NT box, however, correctly show the correct MAC
address for the Linux server.

Can anybody help ?

Thanks in advance.

Steve Clark
[EMAIL PROTECTED]





------------------------------


** 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.networking) 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-Networking Digest
******************************

Reply via email to