Linux-Networking Digest #941, Volume #10 Thu, 22 Apr 99 14:13:29 EDT
Contents:
3c59x and kernel 2.2.5 and 2.2.6 (Petr Olivka)
Re: Which is the best NIC? (Rod Smith)
IP chains/MASQ Guru's needed ([EMAIL PROTECTED])
Re: Can tcpdump capture a complete UDP datagarm fragmented by IP layer? (Barry
Margolin)
Re: 2 NICs, 2 IPs: how is it determined which one is used? (Pekka Savola)
sharing ISDN connection costs within a LAN??? (Michael Ebert)
Sendmail and firewall problem, Help Please! (Frank Spaniak)
Re: Alright - dhcpcd is ticking me off... (Jerry Shenk)
Re: Alright - dhcpcd is ticking me off... ("ryan")
EzPPP -- reopen modem?? (Rob Kent)
Re: IP chains/MASQ Guru's needed ("Leopold Toetsch")
Re: ipautofw, ipportfw, redir, ipchains, ipfwadm?? ("Leopold Toetsch")
Re: Accessing Inet from company network (Lew Pitcher)
----------------------------------------------------------------------------
Date: Thu, 22 Apr 1999 13:09:14 +0200
From: Petr Olivka <[EMAIL PROTECTED]>
Subject: 3c59x and kernel 2.2.5 and 2.2.6
Hi !
I did compile and use 2.2.6 on my pc, later 2.2.5 and now I did come
back to 2.2.2
my ethernet card 3c590 on PCI after 100-200 MB data transfer did lost
_tx_ - I get message "transmit timeout _tx_ status 0 status e000" or
somethink like this.
I did test second the same card with the same result. With half and full
duplex the same.
I did list patches 2.2.2 to 2.2.6 but in 3c59x nothing changed.
in 2.2.2 this card work fine with half and full duplex too. Kernel
2.2.3 and 2.2.4 not tested.
poli
------------------------------
From: [EMAIL PROTECTED] (Rod Smith)
Subject: Re: Which is the best NIC?
Date: 22 Apr 1999 13:28:39 GMT
Reply-To: [EMAIL PROTECTED]
[Posted and mailed]
In article <[EMAIL PROTECTED]>,
Anthony Ewell <[EMAIL PROTECTED]> writes:
>
> What I do to avoid the hassles you describe is top junk the old network
> card
> and replace it with a "DEC Tulip chip set" base card. They work flawlessly
> in Linux and work very well in Windows computers.
I also like DEC Tulip-based cards. Unfortunately, the various DEC Tulip
chipsets are no longer being made, so "genuine" Tulip cards are now quite
scarce. Fortunately, several companies have stepped in with Tulip clone
chipsets, and most of the inexpensive PCI ethernet boards these days now
come with these. Unfortunately, none of the clone chipsets (AFAIK) is
really 100% compatible with the originals, so they require driver tweaks.
The main problem here is simply in getting updated drivers. As of kernel
2.2.3, the kernel came with driver version 0.89. I'm using 0.90q, and I'm
told that 0.91 is now available. Although this sounds like a small
difference, it's not, at least not when dealing with the clone boards I've
used. The latest Tulip drivers can be obtained from:
http://cesdis.gsfc.nasa.gov/linux/drivers/tulip-devel.html
Anybody who gets these drivers will have to recompile his/her kernel
and/or kernel modules, depending upon how the system is set up.
> My favorite Tulip card is the Kingston KNE100TX (dual 10/100 ethernet).
> They cost around $50.00 and are backed up by Kingston's truly EXTRAORDINARY
> tech support.
I've used two of these. One had the original DEC Tulip 21140 chipset, and
worked well for about a month. Then it started having sporadic problems
when I booted up the machine cold (that is, turned off for several hours).
After warming up, it was OK. I figured the board had some subtle
heat-related defect, so I called Kingston and they replaced it. The
replacement had a 21143 chipset, which gave me problems until I upgraded
to the 0.90q driver. I don't know what they're using now.
I've also used a Linksys EtherFast 10/100 board, which has a PNIC
(Lite-On) clone chipset. It's been quite reliable with the 0.90 and 0.90q
drivers. Linksys has Linux drivers on their web site, and they've
reportedly donated equipment for Linux development, so they've got a bit
of a plus for that in my book.
> A Tulip card will save you acres and acres of head scratching and hair
> pulling.
Sadly, at the moment it's likely to be just the opposite because of the
transition from "genuine" Tulip chipsets to the clones, unless you can
find one of the few remaining boards with a real DEC chipset, or if you've
got an existing one sitting around. That will no doubt change before too
long, though, as the marketplace settles again and the newer Tulip drivers
make their way into current kernels.
--
Rod Smith
[EMAIL PROTECTED]
http://www.channel1.com/users/rodsmith
NOTE: Remove the "uce" word from my address to mail me
------------------------------
From: [EMAIL PROTECTED]
Subject: IP chains/MASQ Guru's needed
Date: Thu, 22 Apr 1999 14:28:40 GMT
Hi all
I have a SUSE linux box running 2.2.6 with ipchains/MASQ running. I
have within /etc/rc.config in the Firewall & Masqeurading sections the proper
items (all are not included below)
FW_START="yes"
FW_LOCALNETS="192.168.0.0/24 24.112.x.x "
FW_INT_DEV="eth1"
FW_WORLD_DEV="eth0"
FW_ROUTER=""
FW_FRIENDS="yes"
FW_INOUT="yes"
MSQ_START="yes"
MSQ_NETWORKS="192.168.0.0/24"
MSQ_DEV="eth0"
MSQ_MODULES="ip_masq_cuseeme ip_masq_ftp etc"
******************************************************
and /etc/fw-friends contains the said reserved class C clients:
192.168.0.1
192.168.0.2
192.168.0.3
/sbin/init.d/masquerade status returns:
IP masquerading entries
ipchains -L returns:
Chain input (policy ACCEPT):
Chain forward (policy ACCEPT):
target prot opt source destination ports
user_msq all --- 192.168.0.0/24 anywhere n/a
Chain output (policy ACCEPT):
Chain user_msq (1 references):
target prot opt source destination ports
MASQ all ------ anywhere anywhere n/a
*******************************************************
On internal network with Linux box with two NICs that show via ifconfig:
lo Link encap:Local Loopback
inet addr:127.0.0.1 Bcast:0.0.0.0 Mask:255.0.0.0
UP LOOPBACK RUNNING MTU:3924 Metric:1
RX packets:8040 errors:65 dropped:0 overruns:0
TX packets:0 errors:0 dropped:0 overruns:8040
dummy Link encap:10Mbps Ethernet HWaddr 00:00:00:00:00:00
inet addr:24.112.x.x Bcast:24.112.43.255 Mask:255.255.252.0
UP BROADCAST RUNNING NOARP MTU:1500 Metric:
RX packets:0 errors:0 dropped:0 overruns:0
TX packets:0 errors:0 dropped:0 overruns:0
eth0 Link encap:10Mbps Ethernet HWaddr
00:80:C8:E6:52:59
inet addr:24.112.x.x Bcast:24.112.43.255 Mask:255.255.252.0
00:80:C8:E6:52:59
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:0 errors:22484 dropped:0 overruns:0
TX packets:4 errors:0 dropped:0 overruns:55550
Interrupt:9 Base address:0xd000
eth1 Link encap:10Mbps Ethernet HWaddr 00:80:C8:E6:38:70
inet addr:192.168.0.1 Bcast:192.168.0.255 Mask:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:0 errors:0 dropped:0 overruns:0
TX packets:0 errors:0 dropped:0 overruns:540
Interrupt:3 Base address:0xb800
*******************************************************
Finally route -n returns:
Kernel IP routing table
Destination Gateway Genmask Flags Metric Ref Use Iface
24.112.x.x 0.0.0.0 255.255.255.255 UH 0 0 0 dummy
192.168.0.0 " " 255.255.255.0 U 0 0 0 eth1
24.112.x.0 " " 255.255.252.0 U 0 0 0 eth0
24.112.x.0 " " " " " " U 0 0 0 dummy
24.112.x.0 " " " " " " U 0 0 0 eth0
127.0.0.1 0.0.0.0 255.0.0.0 U 0 0 0 lo
0.0.0.0 24.112.x.1 0.0.0.0 UG 0 0 0 eth0
I cant see why one cannot ping an internal client
(e.g. win95 with TCP IP set as 192.168.0.2/255.255.255.0 and gateway as
192.168.0.0/192.168.0.1 or gateways set singularly), or ping from
client to linux box.
This on on baseT ethernet and hub/cables work under NT.
Any recommendations and/or ideas as to what is causing this problem would be
greatly appreciated.
Thanx for your time
Mac
============= Posted via Deja News, The Discussion Network ============
http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own
------------------------------
From: Barry Margolin <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux,comp.protocols.tcp-ip
Subject: Re: Can tcpdump capture a complete UDP datagarm fragmented by IP layer?
Date: Thu, 22 Apr 1999 15:16:03 GMT
In article <7fm4o3$1et$[EMAIL PROTECTED]>, Witman Peng <[EMAIL PROTECTED]> wrote:
>I use tcpdump or snoop to capture the UDP datagram on a specific port. But
>if the datagram is fragmented by IP layer, can I get all these fragments or
>just the first fragment containing the UDP header? Thanks in advance.
The port number only appears in the first fragment, so if you specify this
in your tcpdump filter expression you won't get the other fragments. If
you just specify the IP addresses and UDP protocol, you'll get all the
fragments, but you must put them together yourself. Tcpdump shows the ID
field and fragment offset information, which is what you need to correlate
them.
--
Barry Margolin, [EMAIL PROTECTED]
GTE Internetworking, Powered by BBN, Burlington, MA
*** DON'T SEND TECHNICAL QUESTIONS DIRECTLY TO ME, post them to newsgroups.
Please DON'T copy followups to me -- I'll assume it wasn't posted to the group.
------------------------------
From: [EMAIL PROTECTED] (Pekka Savola)
Subject: Re: 2 NICs, 2 IPs: how is it determined which one is used?
Date: Thu, 22 Apr 1999 15:17:57 GMT
>Your question doesnt make much sense.. but the answer is (in some form) "In
>your routing table"
Doesn't help much, because I can't find anything at all wrong with my
routing and/or ifconfig.
As you can see, the outbound nic is eth0, and default gw is set for
eth0 and eth0 alone.
Just got a freaky idea.. what if I set up a reject route to 0.0.0.0
for my internal NIC? Have you experimented with this kind of stuff?
Here they are:
=======
eth0 Link encap:Ethernet HWaddr 00:60:97:81:21:F6
inet addr:130.233.25.176 Bcast:130.233.31.255
Mask:255.255.240.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:3274159 errors:1 dropped:0 overruns:0 frame:2
TX packets:724425 errors:0 dropped:0 overruns:0 carrier:3
collisions:17650 txqueuelen:100
Interrupt:9 Base address:0x6100
eth1 Link encap:Ethernet HWaddr 00:A0:D2:1A:01:3C
inet addr:192.168.1.2 Bcast:192.168.1.255
Mask:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:1140523 errors:0 dropped:0 overruns:0 frame:0
TX packets:714676 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:100
Interrupt:11 Base address:0x6200
lo Link encap:Local Loopback
inet addr:127.0.0.1 Mask:255.0.0.0
UP LOOPBACK RUNNING MTU:3924 Metric:1
RX packets:1908 errors:0 dropped:0 overruns:0 frame:0
TX packets:1908 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
======
192.168.1.0 0.0.0.0 255.255.255.0 U 0 0
0 eth1
130.233.16.0 0.0.0.0 255.255.240.0 U 0 0
0 eth0
0.0.0.0 130.233.31.254 0.0.0.0 UG 0 0
0 eth0
----
Pekka Savola pekkas at netcore dot fi
---
Across the nations the stories spread like spiderweb laid upon spiderweb,
and men and women planned the future, believing they knew truth. They
planned, and the Pattern absorbed their plans, weaving toward the future
foretold. -- Robert Jordan: The Path of Daggers
------------------------------
From: Michael Ebert <[EMAIL PROTECTED]>
Subject: sharing ISDN connection costs within a LAN???
Date: Thu, 22 Apr 1999 17:03:58 +0200
I have the following problem:
In the appartement where I live we have a little home network. The server
runs with linux and provides an internet connection via ISDN. The problem
is to assign the connection costs to the users (means to the ip number in=
=20
the local network). If an user asks for an internet address the
connection is automatically established and I can log the connection time.=
=20
In order to match the connection time with the user (ip address of the
host in the local network) I need to log the ip packets. How can I do
this efficiently? Have anybody had that problem? What for solutions do you
know?
Consider that for example several network users can use the internet line
at the same time.
I thought on two possibilities:
1. logging of the packets with a firewall. Problem of the package
"ipfwadm" is, I get the packets but cannot delete them after reading. That
means I have running a cronjob and get the packets from somewhere from the
/proc filesystem, but they have an expire time. So when I ask 1 min later
for the last packets I get also the old ones.
2. I use a packet sniffer like "sniffit". It doesn=B4t work good for ppp
connections. If there is no connection (no ippp device) it exits with
segemntation fault. I can log the ethernet device of the server as well
but then I get a lot of other traffic that I don=B4t want.
You see, I don=B4t know a good solution for that problem. Am I the only on=
e
who has to think about that?????=20
I appreciate any help!
=20
cu,
michael
--=20
=======================================
[EMAIL PROTECTED]
========================================
------------------------------
From: Frank Spaniak <[EMAIL PROTECTED]>
Subject: Sendmail and firewall problem, Help Please!
Date: Thu, 22 Apr 1999 11:18:47 -0500
I'm running SDSL through a Linux RH5.2 firewall Masq'ing 5 boxes (2 servers, 3
workstations behind firewall) but I cant get sendmail 8.9.3 to send out past the
firewall. I Can receive mail just fine. but I have to disable the firewall to
get it to send, then it works.
I like Linux, Straight forward to Install and configure, few gotchas (although I
made the seemingly obligatory mistake of forgetting to update lilo mistake when
compiling and installing the kernel, I should know how to do the rescue disk
anyways :-) ). It runs alot leaner than any other *nix i've seen recently. IP
Masq and squid proxy work great.
I'm lost as to how to fix this.
I cant tell if this is a sendmail, ident or DNS problem.
My ISP handles DNS although it does the same thing even if I use a caching
nameserver.
Mailq reports
(deferred: Connection timed out with other.domain.com.)
I get this on any email I want to send.
===========
In the /var/log/messages I get messages like
Apr 22 07:01:00 mysys kernel: IP fw-in deny eth0 TCP aaaa.bbbb.ccc.ddd:43278
204.233.xxx.yyy:79 L=44 S=0x00 I=29054 F=0x0040 T=421
This is the finger port, I don't get this at all. Possibly unrelated?
===============
The fire wall rules for the authorization port (113) are
# AUTH server (113)
# -----------------
# Reject, rather than deny, the incoming auth port. (NET-3-HOWTO)
ipfwadm -I -a reject -P tcp -W $EXTERNAL_INTERFACE \
-S $ANYWHERE -D $IPADDR 113
# AUTH client (113)
# -----------------
ipfwadm -I -a accept -P tcp -k -W $EXTERNAL_INTERFACE \
-S $ANYWHERE 113 \
-D $IPADDR $UNPRIVPORTS
ipfwadm -O -a accept -P tcp -W $EXTERNAL_INTERFACE \
-S $IPADDR $UNPRIVPORTS \
-D $ANYWHERE 113
As I understand it, sendmail will identify the sending machine on an
unpriv port it seems to be random port. According to the sendmail site this
deffered message is a symptom of an ident problem.
This looks like the firewall kicking in causing identd to fail.
Am I the only one that has had this problem?
Thanks for any help.
------------------------------
Date: Thu, 22 Apr 1999 12:47:59 -0400
From: Jerry Shenk <[EMAIL PROTECTED]>
Subject: Re: Alright - dhcpcd is ticking me off...
Win9x's DHCP client is about half broken. Run these two lines as root.
If that fixes things, all them into an rc file someplace. Some of the
HOWTOs detail the reasons for this.
----manually ran two lines for Win95/98 support:
======route add =net 10.1.1.0 netmask 255.255.255.0 dev
eth0
======route add =host 255.255.255.255 dev eth0255.0 dev eth0
Chris Milkosky wrote:
>
> Hi all!
>
> I've got a little problem here that's driving me nuts. My desktop computers
> (running Win98 and Win95) connect to my cable ISP through the cable modem
> provided, but my laptop running Redhat Linux 5.2 won't do the same.
>
> I've gone through the newsgroups and FAQs, MINI HOWTOs, etc... And so far
> none of the suggestions are helping me. I have two PCMCIA adapters on the
> laptop that seem to be working properly. I did an ifconfig eth0 and the
> same for eth1, and it looks like they see packets out there (used tcpdump to
> see that). But, it seems that every time I use dhcpcd to connect to a DHCP
> server, I get the "no DHCPOFFER messages" message.
>
> What's the deal with that? I've tried passing the hostname to the server
> with the -h option, and that did nothing....
>
> Anyone have any ideas? Again I have Redhat 5.2 running on a IBM Thinkpad
> 701C (486) with 16 Mb RAM. I tried it on two separate PCMCIA cards - a 3com
> 589 and a Xircom, and had the same results.
>
> Thanks in advance for any help.
>
> Chris
------------------------------
From: "ryan" <[EMAIL PROTECTED]>
Subject: Re: Alright - dhcpcd is ticking me off...
Date: Thu, 22 Apr 1999 17:05:25 GMT
Be careful with the newest version as it will not work with 2.0 kernels
Chris Milkosky wrote in message ...
I don't think I'm using the new kernel though - I'm using 2.0.36...
and the dhcpcd rpm on redhat's site is 0.70-2, which is what I've got.
Still think I should grab the sources and make the new version?
Thanks,
Chris
Bob <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]...
Chris Milkosky wrote:
Hi all!
"no DHCPOFFER messages" message.
What's the deal with that?
It means your dhcpcd is not the newest version.
-Bob
--
Never attribute to malice that which can be adequately
explained by stupidity --Robert Heinlein, 'Logic of Empire'
News -- companies, sectors
iNews -- internet companies
------------------------------
From: Rob Kent <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.help
Subject: EzPPP -- reopen modem??
Date: Thu, 22 Apr 1999 15:42:54 GMT
I've just finished installing EzPPP, which is actually a nice little front
end to the pppd, basically eliminates messing around with config files.
You essentially set up the configuration graphically, dial, and then it
spawns a pppd with all the right options.
As root, it's working fine. I also want other users to have access,
though, and so i've changed some permissions around so that any user SHOULD
be able to use it specifically, I changed the following devices
lrwxrwxrwx 1 root tty 5 Apr 17 18:52 /dev/modem -> ttyS2
crw-rw-rw- 1 root tty 4, 66 Apr 21 14:15 /dev/ttyS2
to be globally read-writable, and turned the ppp daemon's SUID on, as well
as making it globally executable:
lrwxrwxrwx 1 root root 8 Apr 17 14:36 /usr/sbin/pppd -> pppd-2.3*
-r-sr-sr-x 1 root root 117032 Mar 27 21:36 /usr/sbin/pppd-2.3*
The thing is, the pppd keeps dying immediately after any normal (ie, non
root) user dials up and logs in (all of which works correctly), and the
following line shows up in the syslog:
Apr 21 14:03:29 pangea pppd[348]: Failed to reopen /dev/modem: Permission
denied
Which I don't understand, since everyone can read and write /dev/modem, and
the pppd runs as root anyway! Any ideas?
thanks,
______________________________________________________________________
rob kent | email address in header | http://cc.kzoo.edu/~k96rk01/
============= Posted via Deja News, The Discussion Network ============
http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own
------------------------------
From: "Leopold Toetsch" <[EMAIL PROTECTED]>
Subject: Re: IP chains/MASQ Guru's needed
Date: Thu, 22 Apr 1999 18:42:07 +0200
[EMAIL PROTECTED] wrote in message
<7fnbmh$gae$[EMAIL PROTECTED]>...
>Hi all
>
>I have a SUSE linux box running 2.2.6 with ipchains/MASQ running. I
>have within /etc/rc.config in the Firewall & Masqeurading sections the
proper
>items (all are not included below)
==SNIP
>FW_FRIENDS="yes"
==SNIP
>and /etc/fw-friends contains the said reserved class C clients:
>192.168.0.1
>192.168.0.2
>192.168.0.3
Don't put inside addresses in here.
FW_FRIENDS is for IPs from the outside, allowed to access your internal net.
Additionally you would open your net to spoofing if you keep these lines.
look at /etc/rc.d/firewall for details.
leo
------------------------------
From: "Leopold Toetsch" <[EMAIL PROTECTED]>
Subject: Re: ipautofw, ipportfw, redir, ipchains, ipfwadm??
Date: Thu, 22 Apr 1999 18:48:51 +0200
Martin NG wrote in message <7fm07f$9vb$[EMAIL PROTECTED]>...
>Hi,
> I was really confused by the packages above. What I want to do is to
>allow internal Web/Ftp server be able to be accessed by computers inside
the
>internal network. What should I have?
>
Nothing, but web & ftp servers.
For the subject read:
Linux IPCHAINS-HOWTO:
It has one chapter titled:
"I'm confused! Routing, masquerading, portforwarding, ipautofw... "
leo
------------------------------
From: [EMAIL PROTECTED] (Lew Pitcher)
Subject: Re: Accessing Inet from company network
Reply-To: [EMAIL PROTECTED]
Date: Thu, 22 Apr 1999 17:36:52 GMT
On Thu, 22 Apr 1999 13:21:27 -0400, "Art" <[EMAIL PROTECTED]> wrote:
>I am a newbie to LINUX, so please go easy on me. I've successfully installed
>LINUX on a spare Compaq 5100 workstation. Since RH 5.2 doesn't have drivers
>for the built in ethernet card, I bought a 3com that is supported. I can see
>the local machines, can do a loopback successfully, and have all the
>DNS/gateway settings the same as Win98, which is installed on another drive
>in this computer, using the same 3com card and enet line.
>
>It must be a configuration problem - is there something special I have to do
>to access the internet through the companies WAN from my LAN? Through
>Netscape, I get "No DNS server found" If I switch to Proxy settings (which
>our network supports in addition to TCP/IP), I get "no connection to the
>Proxy could be made"
>
>Here's the ifconfig output
>
>lo link encap: local loopback
>inet addr:127.0.0.1 Bcast:127.255.255.255 mask: 255.0.0.0
>up broadcast loopback running mtu: 3584 Metric:1
>all packets, errors, collisions = 0
>
>I've been pulling my hair for two weeks over this. Any help would be
>great!!!!
>
>Please respond via email - I'm not able to get to newsgroups from our
>network, so I have to dialup with a 144 (no fun)
>From the ifconfig output above, you have neglected to tell linux
networking to add your ethernet card to the configuration (ifconfig
eth0).
Find the startup script that performs the ifconfig of your NICs, and
update it to include eth0.
You may also need to load the driver for your ethernet card; take a
look at your module load script.
And, for goodness sake, Read The Fine Manuals. That's what they're
there for.
Start with the Net3 howto, and the Ethernet howto.
Lew Pitcher
System Consultant, Development Services
Toronto Dominion Bank
(Opinions expressed are my own, not 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.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
******************************
