Linux-Networking Digest #28, Volume #10 Thu, 28 Jan 99 03:14:01 EST
Contents:
Re: DOES LINUX SUCK (MRoman)
Does ipfwadm work with SLIP? ("Fulton B. Gonzalez")
Re: Cable Modem on Linux - use dhcpcd 1.3.17 (Craig J Copi)
Blind Linux Boxes (Tom Ballard)
Re: DOES LINUX SUCK (Jim Richardson)
IP Aliasing/Mailserver (Rizwan Syed)
. (Jan H�hne)
networking with MS-Windows and MS-DOS ("Erik Colson")
Re: Cannot access some sites (David Kirkpatrick)
Improving speed of MSACCESS 2.0 (Win98 client, Linux Rehat 5.0 Server) (Edward John
Barr)
Re: disabling source routing with token ring (Lee Shakespeare)
Re: LINUX PPP on a SPARC10 (James Carlson)
Re: HELP! Linux Compatible Modems? (JaYsOn)
Re: IP Forwarding and Linux 2.2.0 (Malware)
Re: DOES LINUX SUCK (Jim Zubb)
trouble getting modem to dial using uucp (or mgetty ) (Daddy Rabbit)
lpd-0.33 remote printing problem (Steve Beaty)
Re: Admin NT network from a Linux box (Gregory Leblanc)
Adding Routes (William Groombridge)
----------------------------------------------------------------------------
From: MRoman <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.advocacy,linux.redhat.install
Subject: Re: DOES LINUX SUCK
Date: Thu, 28 Jan 1999 00:02:02 -0600
Raymond Doetjes wrote:
> The thing is, that PCMCIA is a lowsy standard and above all a badly documented
> standard. The only people who can get the specs are mostly OS and Hardware
> builders.
>
> I had a lot of problems with PCMCIA and Windoze 95 too so I think it is a bit
> blund to state that Linux might suck because a lowsy PC standard isn't
> implemented for 100%. Besides I had 1 laptop set up with SUSE linux 5.3 without
> any problems, so it also depends on wich PCMCIA cards are in the system.
>
> Lets be short, all PC specs suck. Like powersupple, PCI, ISA, EISA, Intel, VGA.
>
> I think we all should buy IBM AS/400, RS6000 or DIgital ALpha servers. My Alpha
> 533 rules!!!! I can't run Windoze on my ALpha
^^^^^^^^^^^^^^^^^^^^^^^^^ yes u can :) with fx!32 u could
run most NT apps as well.
>
> Raymond.
------------------------------
From: "Fulton B. Gonzalez" <[EMAIL PROTECTED]>
Subject: Does ipfwadm work with SLIP?
Date: Wed, 27 Jan 1999 18:29:31 +0000
I have had a really frustrating time trying to use pppd with my ISP, but I can
connect using SLIP, via a dip script. My ultimate objective is to use my linux box as
an internet gateway for my other boxes, using ipfwadm. Does ipfwadm work with a slip
connection? All the ip forwarding docs seem to assume the the gateway is using a PPP
connection. (I tried using some of the canned ipfwadm scripts, but I couldn't make it
work; I can't tell if it was my SLIP connection or if I hadn't configured ipfwadm
correctly.)
Any help will be appreciated!
--
Fulton B. Gonzalez Email: [EMAIL PROTECTED]
Department of Mathematics Phone: (617) 627-2368
Tufts University Fax: (617) 627-3966
Medford, MA 02155 USA
------------------------------
From: [EMAIL PROTECTED] (Craig J Copi)
Subject: Re: Cable Modem on Linux - use dhcpcd 1.3.17
Date: 27 Jan 1999 18:30:57 GMT
In article <78n6u8$pjc$[EMAIL PROTECTED]>,
Jeffrey T Kowalczyk <[EMAIL PROTECTED]> writes:
> Same here. Newbie RH user who had better get his cablemodem working or is
> stuck using a 14.4 modem. That's some motivation.
>
> The dhcpcd (DHCP client daemon) in RH5.2 is 0.70, and apparently IPv4 broke
> some parts of it, especially the parts cablemodems use. dhcpcd 0.70 Connects
> fine to my NT DHCP server, but it won't initialize the LANCity cablemodem.
>
> So, I started on the quest to upgrade to dhcpcd 1.3.17
> (http://www.phystech.com), but it will require some shenanigans on our part
> to get it to work on RH5.2. It was easy enough to install, downloaded it to
> /tmp and ran:
>>tar (options I can't remember) dhcpcd-1.3.17.tar.gz
>>cd dhcpcd-1.3.17
>>make install
Or you can just get the rpm at from the rpm repository.
http://rufus.w3.org/linux/RPM/
> But I got "modprobe: can't locate module net-pf-17" on boot up. What was all
> this stuff? My conf.modules was only a couple of lines for my ne2000 isa
> NIC. "modprobe -c" showed that I had net-pf-1 thru -5 aliases set for
> various protocals, although I had nothing to do with that. I am just trying
> to get a basic IP ethernet, plus the cablemodem.
>
> Anyhow, you have to have a kernel recompiled for SOCK_PACKET support for
> dhcpcd, apparently. Nice little task for a newbie linux user... That's
> taking me a few days to get right, make fumbles out with some error about
> this or that missing package. I'm re-adding every relevant package I can fit
> right now, wishing I had enough space for a server class install.
>
> You won't find SOCK_PACKET in "make config", the recommendation I've gotten
> so far is to add the line "CONFIG_PACKET=y" to the generated .config after
> you run make config. Where in the file it goes, I'm only guessing. Then
> again, I haven't gotten through a successful build yet, so I can't confirm
> that that's it.
Maybe you need to turn on EXPERIMENTAL stuff? It should be under
network options.
> I'd like to just go up to the 2.2 kernel, but RH5.2 leaves you with a whole
> bunch of packages you have to upgrade first to be prepared. (see below) I'm
> going to try that next if this build doesn't work out.
>
> So the short recommendation is: Get dhcpcd.1.3.17, recompile a kernel with
> SOCK_PACKET support, and pray that this was all worth it. Send word if you
> have any luck.
When I upgraded to linux 2.2 I had a similar problem. I had compiled
SOCK_PACKET as a module. I had to add
alias net-pf-17 sock_packet
to my /etc/conf.modules file. Then it worked fine. Perhaps it is because
I'm still running an older modutils (2.1.85 I believe).
Craig
------------------------------
From: Tom Ballard <[EMAIL PROTECTED]>
Subject: Blind Linux Boxes
Date: Thu, 28 Jan 1999 01:06:52 -0500
Greetings,
I have two Linux (redhat) boxes setting side by side, both connected to
the same hub, both running on the same class c which is devoted to these
machines. Machine # 1 has an ip of .2 while machine # 2 has an ip of
.50. They are not separated by subnets. Both machines can see the
outside world and machine # 2 can ping and nslookup domains on machine
number 1 while machine # 1 can only do a forward nslookup on the
hostname of machine # 2. Mail sent from machine # 2 to # 1 gets a 553
error (yes, the MX are correctly identified in the DNS records) while an
attempt to send mail from machine # 1 to # 2 will only end up in a hung
process.
I have other machines that are class c - crossed, but on the same
network and both machines can see them.
Again, both machines perform all functions normally except between
themselves. I'm fairly confident they would function properly if I were
to place them on separate subnets, but the data between them (which is
considerable) would then have to go through the router.
Any ideas why these boxes aren't talking?
Thanks in advance.
--
Tom Ballard, Hostmaster
Catchawave Communications
------------------------------
From: [EMAIL PROTECTED] (Jim Richardson)
Crossposted-To: comp.os.linux.advocacy,linux.redhat.install
Subject: Re: DOES LINUX SUCK
Date: 28 Jan 1999 06:45:20 GMT
Reply-To: [EMAIL PROTECTED]
On Wed, 27 Jan 1999 15:52:47 -0700,
Keith Peterson, in the persona of <[EMAIL PROTECTED]>,
brought forth the following words...:
>>Conversely, I recently installed nethack on that very same machine's linux
>>side. I was sitting at home at the time, using a telnet client. All I did
>>was "rpmfind nethack", then confirmed the download, and then used
>>"rpm -i /down/nethack*". That was it.
>
>
>Ah. So simple! I myself used that technique to install a true-type font
>server.
>
>>Now, just for fun, between typing the previous sentence and this sentence,
>>I uninstalled nethack again from that machine. I did this by typing
>>"rpm -e nethack".
>
>
>I removed the font server in the same manner. Simple!
one question, why?
>
>Couldn't launch a terminal window (missing libraries), couldn't launch
>anything under X at all. Exited X to do some checking, couldn't launch X.
>
>Despite certain libraries being needed by my system, RPM took it upon itself
>to remove them anyways. Not only were they required, they were required by
>RPM packages still installed. There is no excuse for this action on the part
>of my software.
>
>>Next, I tried uninstalling the qt libraries, with "rpm -e qt". But rpm told
>>me in no uncertain terms that I shouldn't do that, because a lot of things
>>would break if I did. It listed the things.
>
>
>Query a package: "xxx is not installed."
>Install it: Installation fails because "xxx is already installed."
>
This sounds like rpm < ver 2.3, if so, upgrade to the latest, the not
installed/can't install bug was fixed in (IIRC) 2.4.1 'course, you could
have used --force but that's another issue...
--
Jim Richardson
Anarchist, pagan and proud of it
WWW.eskimo.com/~warlock
Linux, because life's too short for a buggy OS.
------------------------------
From: [EMAIL PROTECTED] (Rizwan Syed)
Subject: IP Aliasing/Mailserver
Date: 27 Jan 1999 18:22:59 GMT
Hello,
I guess this is to whoever can help. I have two domains
registered. I have a machine which is set up and working as a
webserver as well as a mailserver for the first domain. I
decided to set up an alias for the second domain. After I
have my ISP do the necessary DNS entries, what all would I have
to do to receive emails on the second domain?
Example:
First domain: foo1.com www.foo1.com [EMAIL PROTECTED]
Second domain: foo2.com www.foo2.com [EMAIL PROTECTED]
I want both these to work on the same machine. I set up the aliases
and all. The web addresses work for both the sites. All that's
left is the mail. I figured that I owuld need to make changes in
the sendmail.cf and sendmail.cw files?
The sendmail.cf file has a section listed Cwlocalhost. Lets say the
machine name is bar. Under it, it had the the entry for my
original domain:
Cwbar.foo1.com foo1.com bar
So I went ahead and added the following line:
Cwfoo2.com foo2
I also added the above line in the sendmail.cw file.
Will this be enough? Is there something I'm leaving out. I would really
appreciate a quick answer either varifying and/or extra stuff that I
need to change/add.
Thank you
Rizwan
--
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Rizwan Syed
[EMAIL PROTECTED]
------------------------------
From: [EMAIL PROTECTED] (Jan H�hne)
Subject: .
Date: Wed, 27 Jan 1999 19:48:45 +0100
Hello!
I've got an ISDN-Card (Teles 16/3 no PNP)
When I load under SuSE 5.3 with the kernel 2.2.0 when loading failured!
The error message: HiSax can't initialized: Module can't load!!
What must I do?
Ciao Jan
Mail: [EMAIL PROTECTED]
------------------------------
From: "Erik Colson" <[EMAIL PROTECTED]>
Subject: networking with MS-Windows and MS-DOS
Date: Thu, 28 Jan 1999 07:20:14 +0100
Hi !
I would like to know if other people here did install a Linux server (LAN)
to share files, printers and a single mail-adress on a dialup ISP to
MS-WINDOWS and MS-DOS clients ?
Was it difficult ? time consuming ? stable ?
What are encountered or should-know problems ?
Thanx !
Erik
------------------------------
From: David Kirkpatrick <[EMAIL PROTECTED]>
Subject: Re: Cannot access some sites
Date: Wed, 27 Jan 1999 12:57:02 +0000
Reply-To: [EMAIL PROTECTED]
Eitan
How did you hook them up? Is this it?
Linux1
+-----+-------HUB--------------192.168.1.0,
Box1 Box2 192.114.159.17
----+---------HUB----------
|
192.114.159.25
What is the route output from these machines?
--
[EMAIL PROTECTED]
[EMAIL PROTECTED]
[EMAIL PROTECTED]
------------------------------
From: Edward John Barr <[EMAIL PROTECTED]>
Subject: Improving speed of MSACCESS 2.0 (Win98 client, Linux Rehat 5.0 Server)
Date: Wed, 27 Jan 1999 05:58:19 +0000
In article <[EMAIL PROTECTED]>,
James Youngman <[EMAIL PROTECTED]> writes
><[EMAIL PROTECTED]> writes:
>
>> Newbies to linux. Have Samba working ok. Running MSACCESS 2.0 residing on
>> Rehat 5.0 server from Win 98 client. Problem: MSACCess runs too slow. Have
>> tried the speed fix (Socket options= TCP_NoDelay) found in Samba
>> documentation to no avail. Suggestions?
>
>If the file is not accessed locally on the server as well, turn on
>oplocks. This will speed up (e.g.) the case where the number of
>concurrent users of the file is low.
>
I'm new to Linux but have a lot of experience in the database world and
Access has reputation of bogging down when used on a network. In the non
Linux world I tend to use either Sybase SQL anywhere or Advantage, both
of which being non Billyware, ate fast and reliable.
[EMAIL PROTECTED]
Edward J Barr
------------------------------
From: Lee Shakespeare <[EMAIL PROTECTED]>
Subject: Re: disabling source routing with token ring
Date: Wed, 27 Jan 1999 18:55:24 +0000
Matt Kressel wrote:
>
> Lee Shakespeare wrote:
> >
> > Hi Folks.
> >
> > I'm trying to connect a linux box (2.0.35) to a rather large token ring
> > network and am encountering a few problems. The network has a number of
> > 3com token ring hubs, which are joined by a 3com 7000 ATM switch. I can
> > see all the machines on the local token ring, but none on the other
> > rings. An NT box which sits on the same ring can see the whole network.
> >
> > I've been told by the people in the know, that I need to "disable source
> > route routing" for the token ring card. A quick usenet search has
> > turned up little, source routing hadn't used to work, but does now. The
> > source code doesn't make an explicit references to it, and I'm no kernel
> > hacker.
> >
> > Does anyone know how I can disable source routeing for the token ring
> > card?
> >
>
> This is actually a kernel option. See the Kernel compiling howto to
> compile a new kernel. Its actually not that hard, just three commands,
> once you do it a few times. There is an option in the kernel compile
> configuration called "Drop source routed frames" you would say yes
> here. I don't think you can configure this with a live kernel, so
> correct me if I'm wrong but I think recompiling is your only option in
> this case.
>
I'm not sure that it's simple. If my understanding is correct, the
source routeing I'm talking about occurs at the token ring protocol
level, not at the TCP/IP protocol level. I'm not a token ring expert,
so I'm afraid I'm getting this all second hand. I'm told source route
bridging (a more accurate term I'm told) is an IBM extension which
allows different rings to talk to one another, regardless network
protocol. Unfortunately this confuses the big 3Com ATM switch at the
centre of the network and it ignores all the token ring packets from my
Linux box.
I hope this is a little clearer, thanks anyway for the suggestion.
Lee.
--
[EMAIL PROTECTED] : Lee Shakespeare
------------------------------
From: James Carlson <[EMAIL PROTECTED]>
Crossposted-To:
comp.os.linux.setup,comp.os.linux.misc,comp.os.linux.hardware,alt.os.linux,comp.protocols.ppp,linux.redhat.install
Subject: Re: LINUX PPP on a SPARC10
Date: 27 Jan 1999 13:56:00 -0500
"Chenard, Sylvain [CAR:VS13:EXCH]" <[EMAIL PROTECTED]> writes:
> Jan 18 21:29:16 oliver6 pppd[1479]: Receive serial link is not 8-bit
> clean:
> Jan 18 21:29:16 oliver6 pppd[1479]: Problem: all had bit 7 set to 0
> Jan 18 21:30:15 oliver6 pppd[1479]: Exit.
This is a seriously frequently asked question. See any of the
archives. If you see this, then the other side is most likely not
running PPP.
> Even if I put the debug option in "options" file, I got nothing more...
That means that your syslogd is misconfigured. You'll need to send
*.debug to the file.
> send ATQ0V1E1X4\r
> wait OK 5
> dial T*70,,*67,,555-1212 #As an example
> sleep 1
> send \r
> send \r
> wait Userid: 15
> send e(myunixID)\r
> wait Password? 15
[...]
Most ISPs do NOT want you to do that if you're using PPP. Instead,
they want you to just dial in and start using PPP, since that's what
Windoze does. If you start chatting a user name at them, then they'll
usually assume you don't speak PPP.
The problem here is most likely in the chat portion, not in PPP. Run
chat with the "-v" flag to get debug messages in syslog.
> The Sparc uses both different UART and irq than PC. The following shows
> you what I got:
These are completely different architectures. You can't compare them.
--
James Carlson, Consulting S/W Engineer <[EMAIL PROTECTED]>
IronBridge Networks / 55 Hayden Avenue 71.246W Vox: +1 781 372 8132
Lexington MA 02421-7996 / USA 42.423N Fax: +1 781 372 8090
"PPP Design and Debugging" --- http://people.ne.mediaone.net/carlson/ppp
------------------------------
Date: Fri, 29 Jan 1999 02:57:50 +0800
From: JaYsOn <[EMAIL PROTECTED]>
Subject: Re: HELP! Linux Compatible Modems?
Ramanlal Mistry wrote:
> Does anyone have a list (or at least know a few) of internal 56K modems that
> are compatible with Linux 5.0?
Linux5.0?.....or do u mean AnyDistro 5.0
> Or, does anyone know of a Linux driver to
> support a U.S. Robotics 56K PCI Winmodem? Please give very specific product
> information.
>
Usually u can go to the website of your linux distro and look for hardware
compatibility list....
>
> I appreciate your help.
>
> Raj
------------------------------
From: Malware <[EMAIL PROTECTED]>
Subject: Re: IP Forwarding and Linux 2.2.0
Date: Wed, 27 Jan 1999 17:07:31 +0100
Hi Darren,
you wrote:
> Since upgrading to 2.2.0 kernel IP forwarding is broken.
>
> The server can see the internet and the clients can see the server (and
> have default routes to it) but no forwarding seems to be happening...
RTFM - /usr/src/linux/Documentation/Changes
Malware
------------------------------
From: Jim Zubb <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.advocacy,linux.redhat.install
Subject: Re: DOES LINUX SUCK
Date: Wed, 27 Jan 1999 23:11:39 -0800
Keith Peterson wrote:
>
> >Conversely, I recently installed nethack on that very same machine's linux
> >side. I was sitting at home at the time, using a telnet client. All I did
> >was "rpmfind nethack", then confirmed the download, and then used
> >"rpm -i /down/nethack*". That was it.
>
> Ah. So simple! I myself used that technique to install a true-type font
> server.
>
> >Now, just for fun, between typing the previous sentence and this sentence,
> >I uninstalled nethack again from that machine. I did this by typing
> >"rpm -e nethack".
>
> I removed the font server in the same manner. Simple!
>
> Couldn't launch a terminal window (missing libraries), couldn't launch
> anything under X at all. Exited X to do some checking, couldn't launch X.
>
> Despite certain libraries being needed by my system, RPM took it upon itself
> to remove them anyways. Not only were they required, they were required by
> RPM packages still installed. There is no excuse for this action on the part
> of my software.
Please tell the package maintainer it did this so that it
can be fixed, thank you. Actually probably best to post here
first about the package so others can verify it wasn't user
error.
I much prefer this to the MS way of doing things,
arbitrarily installing dlls into the system directory,
happily downgrading unless the installer is smart enough to
actually check for the versions (unlikely).
NT is a little bit nicer, because you can usually reinstall
service packs to fix major problems. Of course it is the
suttle problems that drive you batty.
--
Jim Zubb
[EMAIL PROTECTED]
------------------------------
From: [EMAIL PROTECTED] (Daddy Rabbit)
Subject: trouble getting modem to dial using uucp (or mgetty )
Date: Wed, 27 Jan 1999 19:37:35 GMT
Reply-To: [EMAIL PROTECTED]
uucp accepts the job and places it in the queue however it will not
access the modem. I have been told that before I can dial out I need
to hold the port open while dialing. ( Using RedHat 5.2 ). I enter the
following before attempting to dial out:
tail -f </dev/ttyS1 &
stty opost orcnl</dev/ttyS1
Then, supposedly, I should be able to dial directly from the prompt.
Like so.
echo atdt1234567 > /dev/ttyS1
Problem is, if it works at all it may be 2-3 minutes before the modem
is accessed and the number actually dialed. Then again, it may not
work at all.
Does anyone have any suggestions?
Jim
------------------------------
From: Steve Beaty <[EMAIL PROTECTED]>
Subject: lpd-0.33 remote printing problem
Date: 28 Jan 1999 06:11:10 GMT
i was having difficulty printing from a Solaris machine logged
in as "usera" to a Linux box where an account for "usera" does
not exist, using the lpd-0.33 package. in the printcap man page
it states that having an "rs" field requires the remote user to
have a login on the local machine to print. this is reflected
in the code for the function printit(file) in printfile.c:
360 case 'P':
361 strncpy(logname, line+1, sizeof(logname) - 1);
362 logname[sizeof(logname) - 1] = '\0';
363 if (RS) { /* restricted */
364 if (getpwnam(logname) == NULL) {
365 bombed = NOACCT;
366 sendmail(line+1, bombed);
367 goto pass2;
368 }
369 }
370 continue;
i had no "rs" entry so expected to be able to print. however,
in the function print in printfile.c the following code
appears:
524 /* lets make sure we know what user to run the filters as */
525 if (getpwnam(logname) == (struct passwd *)0 || !*logname)
526 return(ERROR);
checking to see if there is a login for the user. this is so
any filter will be run setuid that user, probably a reasonable
security measure. this seems to be in direct conflict with
the printcap "rs" entry however. the same feature exists in
the sendit(file) function later in printfile.c
in the end, i know how to get around this, and it's probably
not a bug in the code per se, although probably one in the
documentation. just thought i should tell someone about my
findings.
--
Steve Beaty Assistant Professor
Metro State College of Denver [EMAIL PROTECTED]
VOX: (303) 556-5321 Science Building 225A
FAX: (303) 556-5381 http://clem.mscd.edu/~beatys/
------------------------------
From: [EMAIL PROTECTED] (Gregory Leblanc)
Subject: Re: Admin NT network from a Linux box
Date: Thu, 28 Jan 1999 06:49:31 GMT
I don't know what you mean by monitoring traffic, but as for the admin
functions you can telnet to an NT box and use the command line tools.
I haven't heard of found any GUI tools to do anything related to NT,
yet.
Greg
On Mon, 25 Jan 1999 16:51:37 -0500, "Jon Horner" <[EMAIL PROTECTED]>
wrote:
>Hi,
> I have a linux box set up in an NT network. It is the only Linux box,
>and I would rather keep it as my workstation. How can I do my
>administrative tasks from the linux box, such as monitor network traffic,
>reset users, etc?
>
>Thanks,
>Jon.
>
>
Greg Leblanc
Network Admin
Concordia University Portland
gleblanc at cu-portland.edu
------------------------------
From: William Groombridge <[EMAIL PROTECTED]>
Subject: Adding Routes
Date: Wed, 27 Jan 1999 15:17:53 -0500
I have a few Linux machines with 2 NICs, and have had to add a couple of
static routes like
route add -net 10.0.0.0 netmask 255.0.0.0 gw 10.10.10.1
for example... anyway.. whenever I reboot the machine, the routes are
gone, and was wondering how to make them semi-permanent???
Should I just add the route add command to my rc.local file or is there
a proper way to do it.. thanks muchly,
Bill
------------------------------
** 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
******************************
