Linux-Networking Digest #319, Volume #11 Fri, 28 May 99 17:13:41 EDT
Contents:
Re: NIS (Bjoern Gerhart)
Re: Pinging for behind IP Masq?? ("Aaron Weiker")
Re: YOUNG BEAUTIFUL MODEL AVAILABLE (David Kennedy)
Re: RedHat 6.0 & ipmasqadm ("Jeff Volckaert")
DNS & /var/adm/messages (Ivo Naninck)
Curious anon ftp problem (Fred Riley)
3com 3c900 (Rev. 0) and Redhat 6.0 ("Peter Bekatoros")
pppd fails to establish connection ([EMAIL PROTECTED])
Intercpeting IP packets without kernel intervention (Areef Reza)
DHCP, Pump, RedHat5.2 Solved +question (Francois Magnan)
NIS+ and RedHat 6.0 problems (Brad Zoltick)
Re: Multiple IP Address on single Network Card ("Curt")
Re: Kernel 2.2 TCP incompatible with certain Unices? (Solved) (Reality is a point of
view)
Re: NIS ("Peter M. Nielsen")
Re: mail rerouting ([EMAIL PROTECTED])
ROOT-access and SSH-login ([EMAIL PROTECTED])
Telnet problems ("Hugh Saunders")
SMB Shares in /etc/fstab (Adam C. Emerson)
Re: Linux: ICMP Redirect, IP Source Routing unterdruecken (Detlef Bosau)
Re: Linux: ICMP Redirect, IP Source Routing unterdruecken (Detlef Bosau)
Re: Linux: ICMP Redirect, IP Source Routing unterdruecken (Detlef Bosau)
DSL and Redhat (specifically Sympatico High Speed) ("Glenn Poirier")
----------------------------------------------------------------------------
From: Bjoern Gerhart <[EMAIL PROTECTED]>
Crossposted-To: linux.dev.newbie,linux.redhat.install
Subject: Re: NIS
Date: Fri, 28 May 1999 21:13:16 +0200
Hello
>
> Is it neccesary to have a NIS-server in a network. We have a linux machine
> which is supposed to be faxserver and a internal newsserver. It's running in
> a Windows NT environment
NIS-domains are useful for UNIX-machines (I�d compare a NIS-domain for
UNIX with the NT-domain for Windows). If you just have one
linux machine and the others are WinNT-machines, I�d suggest you to use
the package �samba� for linux.
With version 2 or higher you can simulate a WindowsNT-PDC-server (not
all functions added yet).
With samba you can also share services on the Linux server for your
windows-maschines.
Kind regards
Bjoern
--
Bjoern Gerhart e-mail: [EMAIL PROTECTED]
TFH-Berlin University of Applied Sciences
------------------------------
From: "Aaron Weiker" <[EMAIL PROTECTED]>
Subject: Re: Pinging for behind IP Masq??
Date: Fri, 28 May 1999 15:16:24 -0400
I have set up IP Masquerading to work on my redhat 6 box. I followed the IP
Masquerading HOWTO step by step and got it to work. I did however have to
download and compile IPCHAINS in order to set it up. You will want to make
sure though that you have the IP Masquerading ICMP patch compiled into the
kernel. I think by default redhat has it as a module so all you would have
to do is edit the rc.firewall file and make sure that it loads the
appropriate modules. Again they are in the HOWTO.
Aaron
YES, I was able to ping and do a traceroute from behind the linux box, from
a 95 computer even.
Matt Goebel wrote in message <7iij64$hjr$[EMAIL PROTECTED]>...
>I'm using Redhat 6.0 isn't ICMP masq enabled by default?
>Matt Goebel <[EMAIL PROTECTED]> wrote in message
>news:7ihovc$a9s$[EMAIL PROTECTED]...
>> Is there a way to ping from a client on a masqueraded connection?
>>
>>
>
>
------------------------------
From: [EMAIL PROTECTED] (David Kennedy)
Subject: Re: YOUNG BEAUTIFUL MODEL AVAILABLE
Date: Fri, 28 May 1999 19:21:16 GMT
No, she actually a NT user who is trying to setup a linux firewall.
On Thu, 27 May 1999 20:39:17 GMT, Wendell Craig
<[EMAIL PROTECTED]> wrote:
>> Anisha wrote:
>>
>>
>>
>> Hi!
>> My name is Anisha.
>> I am 24 yrs old beautiful model from Paris,
>> I would like to accompany you (discerning gentlemen) for dining,
>> entertainment and pleasure in Europe.
>> I've long black hair and dark eyes,a slim good figure with beautiful
>> face
>> I am well educated, sociable, polite and fun to be with.
>>
>> If you'd like to arrange a liaison, you can contact me at :
>>
>> [Image]
>
>Do you use C++?
>
>--
> Wendell Craig - [EMAIL PROTECTED]
>The box said 'WIN95/98 or better.' so I installed LINUX!
> HEAR MY VOICEOVER DEMO AT http://www.anncrman.com
> OR BY TELEPHONE AT (212)595-0314
------------------------------
From: "Jeff Volckaert" <[EMAIL PROTECTED]>
Subject: Re: RedHat 6.0 & ipmasqadm
Date: Fri, 28 May 1999 08:23:24 -0400
Are you using ipchains now? The old ipfwadm and ipautofw commands probably
don't work now.
http://www.kernelnotes.org/HOWTO/IPCHAINS-HOWTO.html
Jeff Volckaert
Greg Bastian <[EMAIL PROTECTED]> wrote in message
news:7ilaob$139b$[EMAIL PROTECTED]...
> Hi All,
>
> I have set up a machine with RH 6 to act as a masquerading router between
> two networks.
>
> I have the masq'ing working well, but I cannot get ipmasqadm and portfw
> working at all.
>
> I can't find where to turn on debugging. It seems the request from an
> outside client to the external interface (which is being port forwarded)
> doesn't get forwarded on the masq'ing router, even though ip_forwarding is
> enabled.
>
> Any ideas anyone ?
>
> Greg.
>
>
------------------------------
From: Ivo Naninck <[EMAIL PROTECTED]>
Subject: DNS & /var/adm/messages
Date: Fri, 28 May 1999 21:38:34 +0200
Reply-To: [EMAIL PROTECTED]
Hi All,
I have installed bind-8.x as a cacheing DNS.
syslog now says
May 28 08:09:39 bentley named[86]: Cleaned cache of 0 RRs
May 28 08:09:39 bentley named[86]: USAGE 927871779 927749379
CPU=0.11u/0.05s CHILDCPU=0u/0s
May 28 08:09:39 bentley named[86]: NSTATS 927871779 927749379 A=41 PTR=1
May 28 08:09:39 bentley named[86]: XSTATS 927871779 927749379 RR=72
RNXD=8 RFwdR=47 RDupR=2 RFail=0 RFErr=0 RErr=0 RAXFR=0 RLame=0 ROpts=0
SSysQ=17 SAns=12 SFwdQ=33 SDupQ=23 SErr=0 RQ=42 RIQ=0 RFwdQ=0 RDupQ=0
RTCP=0 SFwdR=47 SFail=0 SFErr=0 SNaAns=4 SNXD=6
May 28 09:09:39 bentley named[86]: Cleaned cache of 1 RR
May 28 09:09:39 bentley named[86]: USAGE 927875379 927749379
CPU=0.11u/0.05s CHILDCPU=0u/0s
May 28 09:09:39 bentley named[86]: NSTATS 927875379 927749379 A=41 PTR=1
May 28 09:09:39 bentley named[86]: XSTATS 927875379 927749379 RR=72
RNXD=8 RFwdR=47 RDupR=2 RFail=0 RFErr=0 RErr=0 RAXFR=0 RLame=0 ROpts=0
SSysQ=17 SAns=12 SFwdQ=33 SDupQ=23 SErr=0 RQ=42 RIQ=0 RFwdQ=0 RDupQ=0
RTCP=0 SFwdR=47 SFail=0 SFErr=0 SNaAns=4 SNXD=6
May 28 10:09:39 bentley named[86]: Cleaned cache of 0 RRs
May 28 10:09:39 bentley named[86]: USAGE 927878979 927749379
CPU=0.11u/0.05s CHILDCPU=0u/0s
May 28 10:09:39 bentley named[86]: NSTATS 927878979 927749379 A=41 PTR=1
May 28 10:09:39 bentley named[86]: XSTATS 927878979 927749379 RR=72
RNXD=8 RFwdR=47 RDupR=2 RFail=0 RFErr=0 RErr=0 RAXFR=0 RLame=0 ROpts=0
SSysQ=17 SAns=12 SFwdQ=33 SDupQ=23 SErr=0 RQ=42 RIQ=0 RFwdQ=0 RDupQ=0
RTCP=0 SFwdR=47 SFail=0 SFErr=0 SNaAns=4 SNXD=6
May 28 11:09:39 bentley named[86]: Cleaned cache of 0 RRs
Does this mean bind cleans its cache every our?
I can't find in the man page how to enlarge this cacheing period.
Can it be done?
--
Best regards, and don't let the bits byte!
Ivo Naninck.
~
~
:wq!
------------------------------
From: Fred Riley <[EMAIL PROTECTED]>
Subject: Curious anon ftp problem
Date: 28 May 1999 13:50:15 -0500
I've a Linux machine at ftp.cti.hull.ac.uk which I've tried to configure
for anonymous ftp. If you connect to it via a command-line ftp client it
works as expected. If you connect to it using a graphical ftp client,
such as WS_FTP or a web browser, you get an empty directory listing,
although these clients work fine if I log in as a regular system user.
I've had a hunt through the FAQs, HOWTOs, and man pages at the Linux
Documentation Project and have found very little on configuring ftp
servers under Linux, and certainly nothing on this problem. I've
followed the instructions in "Linux Unleashed" about setting up anon
ftp. I've asked our sysadmin, who installed Linux on our PC for us, and
he doesn't seem to know what's happening. Permissions for all
directories in /pub and below are set to world read and execute, and
they're owned by user root and group root (I've experimented with other
ownerships without success).
Sadly, I'm very much a novice when it comes to Unix/Linux admin,
although I'm comfortable with Unix from a user viewpoint, so I'd
appreciate any suggestions as to what might be happening couched in
language that I can understand.
Please copy any replies to my email address (remove NOSPAM). Many thanks
in advance.
Cheers
Fred Riley
Software Development Adviser
Language Institute, University of Hull, UK
------------------------------
From: "Peter Bekatoros" <[EMAIL PROTECTED]>
Subject: 3com 3c900 (Rev. 0) and Redhat 6.0
Date: Fri, 28 May 1999 14:51:01 -0400
On bootup (and then after) I'm receiving the following message:
eth0: transmit timed out, tx_status 00 status e601.
eth0: Interrupt posted but not delivered - IRQ blocked by another device?
I've checked my logs as well as /proc/interrupts and the nic is definitely
receiving an interrupt. Module 3c59x is loading correctly too. I've tried
changing the first available IRQ in the BIOS as well as moving the card
around and still get the same message. The PNP OS bios setting is disabled
also. If I don't ifdown the nic, I eventually get a core dump. The only
other card in the system is an ATI Mach64 VT PCI card.
System: DFI VPM586 Dual Pentium (133x2)
VLSI Chipset
CMD 646 rev. 1 IDE Controller
64M Ram
3.2 Gig EIDE HD
Anybody have any ideas? The card worked fine under Redhat 5.1 (2.0.34)
Thanks,
Pete Bekatoros ([EMAIL PROTECTED])
Fresh Meadow Mechanical Corp.
------------------------------
From: [EMAIL PROTECTED]
Crossposted-To: linux.redhat.ppp
Subject: pppd fails to establish connection
Date: Fri, 28 May 1999 18:34:47 GMT
Hello there.
I have been trying to get pppd to connect to my office LAN from home,
without success.
I have a Toshiba laptop with RedHat Linux 5.2 freshly installed on it.
The PCMCIA works fine, since I am able to connect to the office modem.
I have to use PAP when connecting to the LAN from Win95 (Dial-up
networking works fine on the same machine). I have successfully
connected to ISP's who don't use PAP.
I am at my wits end, and since I don't know how to read debug logs, I
would be grateful for any help.
I have included the debug log (generated when I say "pppd -d
/dev/modem"), the options and the pap-secrets files below.
Thanks a ton.
PVS
--debug log begins--
May 28 23:15:52 localhost kernel: IP Protocols: IGMP, ICMP, UDP, TCP
May 28 23:15:52 localhost kernel: Linux IP multicast router 0.07.
May 28 23:15:52 localhost kernel: Linux version 2.0.36
([EMAIL PROTECTED])
May 28 23:15:52 localhost kernel: Serial driver version 4.13 with no
serial options enabled
May 28 23:15:52 localhost kernel: tty00 at 0x03f8 (irq = 4) is a 16550A
May 28 23:15:52 localhost kernel: sysctl: ip forwarding off
May 28 23:15:52 localhost kernel: Swansea University Computer Society
IPX 0.34 for NET3.035
May 28 23:15:52 localhost kernel: IPX Portions Copyright (c) 1995
Caldera, Inc.
May 28 23:15:52 localhost kernel: Appletalk 0.17 for Linux NET3.035
May 28 23:15:52 localhost kernel: Linux PCMCIA Card Services 3.0.9
May 28 23:18:44 localhost login[227]: ROOT LOGIN ON tty3
May 28 23:18:44 localhost PAM_pwdb[227]: (login) session closed for user
root
May 28 23:18:54 localhost kernel: CSLIP: code copyright 1989 Regents of
the University of California
May 28 23:18:54 localhost kernel: PPP: version 2.2.0 (dynamic channel
allocation)
May 28 23:18:54 localhost kernel: PPP Dynamic channel allocation code
copyright 1995 Caldera, Inc.
May 28 23:18:54 localhost kernel: PPP line discipline registered.
May 28 23:18:54 localhost kernel: registered device ppp0
May 28 23:18:55 localhost pppd[289]: pppd 2.3.5 started by root, uid 0
May 28 23:18:55 localhost pppd[289]: Using interface ppp0
May 28 23:18:55 localhost pppd[289]: Connect: ppp0 <--> /dev/modem
May 28 23:18:55 localhost pppd[289]: Warning - secret file
/etc/ppp/pap-secrets has world and/or group access
May 28 23:18:55 localhost pppd[289]: sent [LCP ConfReq id=0x1 <magic
0xdb2bed55> <pcomp> <accomp>]
May 28 23:18:55 localhost pppd[289]: rcvd [LCP ConfReq id=0x1 <mru 1522>
<asyncmap 0xa0000> <auth 0xc027 01 00 00 03 00 00 00 0e> <magic
0x5c7ca7af> <accomp> < 11 04 05 f2> < 12 02> < 13 09 03 00 80 d3 29 38
e0>]
May 28 23:18:55 localhost pppd[289]: sent [LCP ConfRej id=0x1 < 11 04 05
f2> < 12 02> < 13 09 03 00 80 d3 29 38 e0>]
May 28 23:18:55 localhost pppd[289]: rcvd [LCP ConfAck id=0x1 <magic
0xdb2bed55> <pcomp> <accomp>]
May 28 23:18:58 localhost pppd[289]: sent [LCP ConfReq id=0x1 <magic
0xdb2bed55> <pcomp> <accomp>]
May 28 23:19:25 localhost last message repeated 9 times
May 28 23:19:28 localhost pppd[289]: Hangup (SIGHUP)
May 28 23:19:28 localhost pppd[289]: Modem hangup
May 28 23:19:28 localhost pppd[289]: Connection terminated.
May 28 23:19:29 localhost pppd[289]: Exit.
--debug log ends--
--options begins--
lcp-max-configure 30
defaultroute
name my_login_name
--options ends--
--pap-secrets begins--
my_login_name * my_password
--pap-secrets ends--
Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.
------------------------------
From: Areef Reza <[EMAIL PROTECTED]>
Subject: Intercpeting IP packets without kernel intervention
Date: Fri, 28 May 1999 15:14:32 -0400
Hi,
I am trying to intercept IP packets
and dispatches them myself. I don't
want the kernel to do processing on the
packets that I have captured.
By using raw sockets or by accessing the link
layer, I can intercept an IP packet, but how
do I block the kernel from processing it ?
Is it possible to do it from a user level
program, or do I have to write a kernel module ?
Whatever the case maybe, could somebody please
provide me with some pointers, where I can find code
and examples ?
I am using FreeBSD 2.2.8 and Linux (RedHat 5.1).
thanx in advance,
-- Areef.
------------------------------
Subject: DHCP, Pump, RedHat5.2 Solved +question
From: [EMAIL PROTECTED] (Francois Magnan)
Date: Fri, 28 May 1999 19:02:07 GMT
Hi,
I am now able to use DHCPCD in RedHat5.2 with the new kernel
of RedHat6.0: 2.2.5. I followed
http://www.pdc.kth.se/~jas/dhcpcd-on-redhat-with-linux-2.2.html
The only thing I am missing now is that the ifdhcpc-done script
doesn't get executed after I call DHCP. I need this badly because my
firewall script must get reexecuted! What can I do?
Francois
--
______________________________________________________
Francois Magnan
Departement de Mathematique & Statistiques
Universite de Montreal
email: [EMAIL PROTECTED] (MIME, NeXTMail Ok!)
------------------------------
From: Brad Zoltick <[EMAIL PROTECTED]>
Subject: NIS+ and RedHat 6.0 problems
Date: 28 May 1999 19:08:36 GMT
Re: Has anyone gotten NIS+ to work correctly with RedHat 6.0.
I downloaded and installed the following packages:
pam_keylogin-1.2
autofs-3.1.3
nis-utils-1.0
My /etc/nsswitch.conf is identical to the one specified by Thorsten Kukuk.
I was able to generate a NIS_COLD_START client file.
What I don't understand is how the two libc's (libc-5.4 and glibc-2.1.1)
coexist and how programs compiled with one can work with programs
compiled with the other while using a common /etc/nsswitch.conf
file. I compiled and installed the above packages using
./configure; make; make install
and all seemed to go well. I am uncertain if I am linking
with glibc-2.1.1 or the standard libc-5.4 shared C library.
[I find this rather confusing.]
My network consists of 8 Solaris 2.x systems. The NIS+ root
server is a Solaris system.
After installing the above packages, I can list all the tables
from my RedHat Linux hosts
niscat hosts.org_dir
niscat cred.org_dir
.
but if I type "finger login_name", the system responds
with "no such user".If I use Thorsten Kukuk's libc5
/etc/nsswicth.conf file, finger returns the correct
information.
What is most troublesome is if I try to start keyserv or nis_cachemgr
on my Linux system, networking both stops and my rpc.nisd process on
my NIS+ root server begins to use up all the CPU time.
Has anyone implemented NIS+ properly? If so, can you
send me some additional notes or point me to some
better documentation.
Thanks.
Brad Zoltick
National Institutes of Health
Bethesda, Maryland, USA
Email: [EMAIL PROTECTED]
------------------------------
Reply-To: "Curt" <[EMAIL PROTECTED]>
From: "Curt" <[EMAIL PROTECTED]>
Subject: Re: Multiple IP Address on single Network Card
Date: Fri, 28 May 1999 14:36:28 -0500
See http://metalab.unc.edu/LDP/HOWTO/mini/IP-Alias.html
Dan <[EMAIL PROTECTED]> wrote in message news:7imfu1$1nv$[EMAIL PROTECTED]...
> I would like to setup multiple IP's on a single netword card in Linux. The
> kernel that I am using is 2.0.34. Can any one point me to the
documentation
> on how to do this. I currently have a firewall set up which uses port
> forwarding to forward port 80 to my webserver behind the firewall. I now
> want to set up another web server behind the firewall which will be
> publically registered under a new IP. For this reason I need my network
card
> which is on the outside of the firewall to assume two IP address's and
> forward port 80 for each IP. I understand that it is possible to have a
> network card assume two ip address's but am not sure how. Can anyone
> help?????
>
> Thanks
> Dan
>
>
------------------------------
From: [EMAIL PROTECTED] (Reality is a point of view)
Subject: Re: Kernel 2.2 TCP incompatible with certain Unices? (Solved)
Date: Fri, 28 May 1999 19:50:30 GMT
+---- [EMAIL PROTECTED] wrote (Fri, 28 May 1999 14:05:59 GMT):
| A quick peek at a packet dump made me think it was a problem
+----
A longer peek makes me think the echo > tcp_* workaround works.
--
Gary Johnson [EMAIL PROTECTED]
Privacy on the net is still illegal.
------------------------------
From: "Peter M. Nielsen" <[EMAIL PROTECTED]>
Crossposted-To: linux.dev.newbie,linux.redhat.install
Subject: Re: NIS
Date: Fri, 28 May 1999 22:43:40 +0200
Do I have to reinstall Red Hat, or only install samba, or how do I do it.
What if I f.x. want to connect 12 linux machines via telnet to a NT server,
what is the proper sollution to this....
Best regards, Peter
Bjoern Gerhart skrev i meddelelsen <[EMAIL PROTECTED]>...
Windows NT environment
>
>NIS-domains are useful for UNIX-machines (I�d compare a NIS-domain for
>UNIX with the NT-domain for Windows). If you just have one
>linux machine and the others are WinNT-machines, I�d suggest you to use
>the package �samba� for linux.
>With version 2 or higher you can simulate a WindowsNT-PDC-server (not
>all functions added yet).
>With samba you can also share services on the Linux server for your
>windows-maschines.
>
>Kind regards
> Bjoern
>
>--
> Bjoern Gerhart e-mail: [EMAIL PROTECTED]
> TFH-Berlin University of Applied Sciences
------------------------------
From: [EMAIL PROTECTED]
Subject: Re: mail rerouting
Date: Fri, 28 May 1999 20:39:21 GMT
This may or may not work.
Set up some sort of script to telnet into your providers mail server
(port 25) and then have the script send the command "ETRN" (no quotes)
Should dump mail for your domain if that's how it's set up.
Read about the ETRN command, in any event or call your ISP.
In article <[EMAIL PROTECTED]>,
Xavier Garcia <[EMAIL PROTECTED]> wrote:
> Hi
>
> I have a mail server in my local network (pop/smtp) but not online
24/24
> so what i would like is that for exemple every 2 hours it checks the
> mail on my provider's mail server and download them. For sending mail
> too, store the pending mail into my mail server and when it is online,
> sent it to my provider's mail server.
> There's no problem to set the connection up every 2hours (diald) but I
> just don't know how to upload and download the mails to/from my
> provider's mail server.
>
> Thanx in advance
> xg
>
>
Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.
------------------------------
From: [EMAIL PROTECTED]
Subject: ROOT-access and SSH-login
Date: Fri, 28 May 1999 20:51:49 GMT
Hello!
I have two questions. I were very happy, if somebody could help me!
RedHat Linux 6.0
Current situation:
*Normally root has no login-access (except console)
How to setup these things?:
*root access only for one machine (cs2, eth2, 192.168.2.1) in the
local network and no other!
*e-mail client (on cs2, and only on cs2) should have access to root's
mail over POP3 (other accounts access are ok, ipop3d is running).
Current situation:
*SSH (Secure Shell) connection is not possible...
How can I enable:
*SSH for all users
I tried various things... but could not solve these two problems.
Thank you in advance!
Sincerely
Erkan
PS: If you reply, please write a copy to my e-mail-address, too...
------------------------------
From: "Hugh Saunders" <[EMAIL PROTECTED]>
Subject: Telnet problems
Date: Fri, 28 May 1999 21:09:25 +0100
Hi folks.
I have a problem getting Telnet to work, so here is my set-up and problem;
please help!
I have a windows PC that I want to connect to my new Linux server. The Linux
box is running RedHat Linux 6, with an ethernet NIC. TCP/IP is working on
both computers, since each can ping the other, and I can FTP to the Linux
box from the WinPC.
However, I cannot telnet on the Linux box from Windows. Telnetd is running,
since under Linux I can telnet to localhost and log in. If I telnet from the
Windows PC, I get the System message:
"RedHat Linux 6
on an i586 etc."
But no log in prompt. From the books that I have got, I think that I need to
set up pseudo-terminals (pty s), but none of the docs (HOWTOs etc.) seem to
mention how I should do this. The telnetd man page mentions a pty man page
that doesn't seem to exist.
HELP! I've been trying to solve this for two days!
Thanks,
Hugh Saunders
------------------------------
From: Adam C. Emerson <[EMAIL PROTECTED]>
Subject: SMB Shares in /etc/fstab
Date: Fri, 28 May 1999 20:09:40 GMT
Greetings,
Is there any way to have a SMB filesystem mounted automatically
from /etc/fstab?
I've tried:
//carcity/company /company smbfs defaults,umask=000 0 0
But all I get is:
smb_read_super: wrong data argument
mount: wrong fs type, bad option, bad superblock on //carcity/company,
or too many mounted file systems
Thanks.
--
Adam C. Emerson [EMAIL PROTECTED]
http://www.calvin.edu/~aemers19/
Preach from it unto the Righteous, that they may renounce their
ways and repent. -- Honest Book of Truth
------------------------------
Date: 28 May 1999 22:23:00 +0200
From: [EMAIL PROTECTED] (Detlef Bosau)
Crossposted-To: de.comm.internet.routing
Subject: Re: Linux: ICMP Redirect, IP Source Routing unterdruecken
[EMAIL PROTECTED] meinte am 27.05.99
zum Thema "Re: Linux: ICMP Redirect, IP Source Routing unterdruecken":
> Sehe ich eigentlich nicht so - im Gegenteil: das erste Paket, was
> an den falschen Gateway geschickt wird, generiert ein (1)
> zusaetzliches Paket, und alle weiteren Pakete muessen dann nur noch
> einmal uebers LAN. Das ist gar nicht so bloed.
Relativ prominente Systeme wie z.B. Solaris 2.5 und gated 3.5.6,
wenn ich die Versionen richtig im Kopf habe, sind bei icmp
redirect nur noch abgepfiffen. Bums aus, schon stand der Laden.
Und die Redirects dann rauszukriegen, war ein grauenvolles Theater.
Auch habe ich legacy systems gesehen, die sich ueber Redirect
Routen eingefangen haben, die schlicht ein komplettes Warenwirtschaftssystem
flachgelegt haben und niemand kannte das System.
Redirect ist ISP Denke. Die ist durch zwei Aspekte gekennzeichnet.
1. Kapazitaet: Brumm Brumm Quietsch Quietsch, Prof. August-Wilhelm
Scheer sagt ja auch immer, es muss bummern.
2. SLA: Null. Wobei man mit den Mathematikern noch diskutiert,
ob es dem Betrage nach kleine Zahlen als die Null gibt.
Ich habe allein im letzten Jahr mehrere Nachtschichten durch
redirects gehabt.
Und wenn ich etwas nicht brauche, wenn mir etwas keinen Nutzen bringt,
und ich nur etwas, was ich nicht brauche, debuggen muss, habe ich
gegen die absolut wesentlichste aller Designgrundregeln verstossen:
By all means: KEEP IT SIMPLE!
Wenn Rechner, denen ich nicht explizit, z.B. durch gated,
_befohlen_ habe, etwas zu lernen, anfangen, ein lernbegieriges
Eigenleben zu fuehren, werde ich grundsaetzlich nervoes.
Gruss
Detlef
--
Detlef Bosau [EMAIL PROTECTED]
Bienroder Weg 79 Tel.: +49 531 303383
D2: +49 172 6819937
38106 Braunschweig, Germany Fax: +49 531 303364
>>>> PGP Public Key als Empfangsbestaetigung <<<<
## CrossPoint v3.1 R ##
------------------------------
Date: 28 May 1999 22:16:00 +0200
From: [EMAIL PROTECTED] (Detlef Bosau)
Crossposted-To: de.comm.internet.routing
Subject: Re: Linux: ICMP Redirect, IP Source Routing unterdruecken
[EMAIL PROTECTED] meinte am 26.05.99
zum Thema "Re: Linux: ICMP Redirect, IP Source Routing unterdruecken":
> In comp.os.linux.networking Helge Oldach
> > <[EMAIL PROTECTED]> wrote: Klar. Aber kein Geld: Das
> > verdreifacht gegenueber einer serioesen Konfiguration die Anzahl
> > der Pakete und verlangt ergo nach Fast Ethernet. Ohne Not,
> > wohlgemerkt.
>
> Verdreifacht? Wie kommst du da drauf. Ich denke das sind deutlich
> weniger Pakete als bei RIP (je nach Struktur der Subnetze). Die
> ICMP Redirects werden ja gelernt.
Also ich werde jetzt Pruegel beziehen.
Aber ob ich Hot Potatoe route (wobei ich schon gar nicht mehr genau
weiss, wie das geht, vielleicht kann es mir nochmal jemand
_richtig_ erlaeutern ;-) jedenfalls dieses totale Zufallsgeroute,
irgendwann wird es ankommen), icmp redirect oder RIP, ist
fuer mich irgendwann das gleiche.
Ich habe Unix-Rechner mit ueber 200 Routen gesehen, weil die mit icmp
redirect zugepflastert wurden. Und wenn ich Netze zu klein
fragmentiere, und dann hoffe, "Struktur" zu kriegen, habe ich dann
nur noch ein Netz, dass unter redirects absaeuft, weil hinten und
vorn Netzmasken nicht stimmen.
(In Zeiten von Switches ist die Netzgroesse eh dadurch bestimmt,
wie gross ich bei Schrott-NT die IP Broadcast Domain machen muss.
Mit Unix koennte man mit ordentlichen Switches ruhig 1000 Rechner
in ein single large vlan packen, das wuerde niemanden mehr stoeren.
Aber da gibt es ja noch was von Gates.)
Aber RIP kann es irgendwie auch nicht sein.
*Schuettel*
Detlef
--
Detlef Bosau [EMAIL PROTECTED]
Bienroder Weg 79 Tel.: +49 531 303383
D2: +49 172 6819937
38106 Braunschweig, Germany Fax: +49 531 303364
>>>> PGP Public Key als Empfangsbestaetigung <<<<
## CrossPoint v3.1 R ##
------------------------------
Date: 28 May 1999 22:26:00 +0200
From: [EMAIL PROTECTED] (Detlef Bosau)
Crossposted-To: de.comm.internet.routing
Subject: Re: Linux: ICMP Redirect, IP Source Routing unterdruecken
[EMAIL PROTECTED] meinte am 27.05.99
zum Thema "Re: Linux: ICMP Redirect, IP Source Routing unterdruecken":
>
> Hmm, wie kommst Du auf Verdreifachung? Ich komme nur auf zwei
> zusaetzliche Pakete pro Kommunikationsbeziehung, die das LAN ueber
> einen anderen als den Defaultrouter verlaesst. Das ist meiner
> Meinung nach billiger als ein Routing-Protokoll, das alle Hosts
> staendig ueber alle Routen auf dem laufenden haelt.
Dem stimme ich zu. Aber ein Routingprotokoll haelt Dich nicht nur
ueber die Routen auf dem laufenden, es sei denn, Du wertest
irgendwelche Distance-Vector Historien als Routingprotokoll.
Ein Link State Protokoll hat naemlich auch etwas mit praeventiver
Netzwerkueberwachung zu tun, darum macht man das naemlich.
Und dass man da bei heutiger Prozessorkapazitaet auch mal gerad
rasch nen SPF-Algorithmus drueberjagen kann, ist eigentlich
ein sehr nuetzlicher Nebeneffekt ;-)
(Davon ganz abgesehen ist gerade bei OSPF der Traffic sehr durchgestyled.
Da wird naemlich _nicht_ wild mit Paketen um sich geschmissen.)
Detlef
--
Detlef Bosau [EMAIL PROTECTED]
Bienroder Weg 79 Tel.: +49 531 303383
D2: +49 172 6819937
38106 Braunschweig, Germany Fax: +49 531 303364
>>>> PGP Public Key als Empfangsbestaetigung <<<<
## CrossPoint v3.1 R ##
------------------------------
From: "Glenn Poirier" <[EMAIL PROTECTED]>
Subject: DSL and Redhat (specifically Sympatico High Speed)
Date: Fri, 28 May 1999 20:13:49 GMT
Anyone have any pointers or know of any resources that will help me with my
setup? I've got my setup running in windows, but I'm not exactly sure what
I'm doing in linux.
Glenn
------------------------------
** 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
******************************
