Linux-Networking Digest #966, Volume #10 Mon, 26 Apr 99 18:13:42 EDT
Contents:
Remote Install?? ("Newbie")
Re: Bridge, router, firewall (John Lewis)
Re: PPP dial in problems (Clifford Kite)
Random Character Swapping ??? Help! ("J.R. Satterfield")
Re: GET REDHAT 6.1 FiNAL! ("Trip6")
Re: linux-2.0.36 loses eth0 route after three weeks... (Ronald Cole)
Re: Remote Install?? (Karl Heinz Kremer)
Re: Network setup tips, please? (Dimitrios Liappis)
----------------------------------------------------------------------------
From: "Newbie" <[EMAIL PROTECTED]>
Subject: Remote Install??
Date: Tue, 27 Apr 1999 14:53:21 -0600
Is it possible to remotely install linux on a new pc (nothing on the hdd at
all) using a boot disk.
log onto a server and then install linux.
Is this possible with any distibutions?
If so how do I go about doing it??
------------------------------
From: John Lewis <[EMAIL PROTECTED]>
Crossposted-To: comp.protocols.smb
Subject: Re: Bridge, router, firewall
Date: Mon, 26 Apr 1999 21:06:12 +0000
Ron:
Also, have you tried tuning a kernel for your application - there are some
optimizations in the config files that tune your systems ip interfaces for
routing. Also, I hear in the grapevine that some folks see significant
performance increases using amd-lance chipset ethernet adapters...
Ron Black wrote:
> Thanks - I will read the docs and give it a try today..
>
> Vidar Andresen wrote:
>
> > In article <[EMAIL PROTECTED]>, Ron Black <[EMAIL PROTECTED]> wrote:
> > >How can this be done ?
> >
> > /usr/doc/HOWTO/mini/bridge*
> >
> > Claims to have some solution. And yes, i have a bridge with up to 3
> > machines connected to it, in my personal privat network, 10Mbps only,
> > no kind of firewall, little security, i have to trust myself that
> > much.. so..
> >
> > I use version bridgex-0.2 bridgex.tar.gz from september 1996. I had
> > to install monkeylinux to compile it, and then transfer it to my
> > redhat 5.1... well.
> >
> > How the 2.2.* kernel does this i dont know.
> >
> > >We are concerned about security and do not completely trust the foreign
> > >network. We know that it has been compromised and want to prevent
> > >unauthorized entry into our systems via the foreign network.
> >
> > I am not into that bit of it.
> >
> > >The foreign network owners have provided us with a router and an ip
> > >address of 10.3.204.254. We have installed a linux box that routes
> > >between our network 192.168.101.x and the foreigners 10.3.204.x. But we
> > >have trouble. The PC's that access both networks use network addresses
> > >in the range of 10.3.204.10 to 10.3.204.100 they use a netmask of
> > >255.255.255.0. They point to gateways 10.3.204.253 (our linux router
> > >box) and 10.3.204.254 (the foreign router).
> >
> > I dont know if it matters, but the netmask on the class A (10.3.204.*)
> > network above is not strictly after the book. My win95 is providing,
> > semi-automatically, a 255.0.0.0 netmask on ip-ranges as the 10.3.204.*
> > above. How dhcp handles this i dont know.
> >
> > Net-3-howto:
> >
> > For administrative reasons some time early in the development of the
> > IP protocol some arbitrary groups of addresses were formed into
> > networks and these networks were grouped into what are called classes.
> > These classes provide a number of standard size networks that could be
> > allocated. The ranges allocated are:
> >
> > ----------------------------------------------------------
> > | Network | Netmask | Network Addresses |
> > | Class | | |
> > ----------------------------------------------------------
> > | A | 255.0.0.0 | 0.0.0.0 - 127.255.255.255 |
> > | B | 255.255.0.0 | 128.0.0.0 - 191.255.255.255 |
> > | C | 255.255.255.0 | 192.0.0.0 - 223.255.255.255 |
> > |Multicast| 240.0.0.0 | 224.0.0.0 - 239.255.255.255 |
> > ----------------------------------------------------------
> >
> > >Periodically we experience difficulties on the network wherein the
> > >connections to the nt server and the unix app servers are busted. These
> > >problems are experienced by machines on both sides of the network. I
> > >suspect the problem has something to do with network noise/traffic.
> > >Maybe as a result of browsing and wins.
> >
> > A 10Mbps (?) net shared withouth any switches between hungry machines..
> >
> > On that (10Mbps ?) i guess a fast 486 with 4 nic's able to take
> > whatever comes could work as a cheap switch. And, as the switch
> > demands promiscous mode on the cards, it will also give you a point
> > where you could listen on 'tcpdump -i eth2' or 'tcpdump -i eth0' or..
> > to see where the problem/noise/traffic are taking place.
> >
> > Things like ntop, iptraf, and so on..
> >
> > (I use 3com509b and a 486DX2-66. It performes well on 10Mbps.)
> >
> > >How can we reconfigure the linux router box to make for easier browsing,
> > >more reliable and faster network performance while keeping unauthorized
> > >foreigners out. I'm sure that we can get a microsoft solution but our
> > >management team has asked for a linux solution. Is it possible for the
> > >linux box to be a bridge, a router and a firewall all at the same time.
> > >If so how ?
> >
> > /usr/doc/HOWTO/mini/bridge*
> >
> > Mvh Vidar Andresen
------------------------------
From: kite@NoSpam.%inetport.com (Clifford Kite)
Subject: Re: PPP dial in problems
Date: 26 Apr 1999 11:50:24 -0500
Hartmann Schaffer ([EMAIL PROTECTED]) wrote:
: I have a bad problem getting ppp to accept an incoming call, regardless
: of whether I use AutoPPP (from mgetty) or a shell script calling pppd.
: According to the log, the connect negotiation goes fine up to the point
: where the caller sends the pap authorization request. Then pppd sits
: there doing nothing until the caller hangs up, upon which event it deems
: it appropriate to send a login [ PAP] acknowledge
<snip>
: Apr 19 21:44:06 inferno kernel: ppp: receive buffer, count = 24
: Apr 19 21:44:06 inferno kernel: 7E FF 7D 23 C0 21 7D 22 ~.}#.!}"
: Apr 19 21:44:06 inferno kernel: 7D 21 7D 20 7D 38 7D 22 }!} }8}"
: Apr 19 21:44:06 inferno kernel: 7D 26 7D 20 7D 20 7D 20 }&} } }
: Apr 19 21:44:06 inferno kernel: ppp: receive buffer, count = 56
: Apr 19 21:44:06 inferno kernel: 7D 20 7D 23 7D 24 C0 23 } }#}$.#
: Apr 19 21:44:06 inferno kernel: 7D 25 7D 26 FF FF 21 7D }%}&..!}
: Apr 19 21:44:06 inferno kernel: 32 7D 27 7D 22 7D 28 7D 2}'}"}(}
: Apr 19 21:44:06 inferno kernel: 22 7D 3B 89 7E
Above is a dump of the LCP ConfAck. Below the first of two Identification
messages begins.
7E C0 21 "};.~~.!
: Apr 19 21:44:06 inferno kernel: 0C 02 00 12 00 00 36 BF ......6.
: Apr 19 21:44:06 inferno kernel: 4D 53 52 41 53 56 34 2E MSRASV4.
: Apr 19 21:44:06 inferno kernel: 30 30 17 BE 7E
The first ID message ends above and the second ID message begins below.
7E C0 21 00..~~.!
: Apr 19 21:44:06 inferno pppd[1675]: rcvd [LCP ConfAck id=0x1 <asyncmap 0x0> <auth
:pap> <magic 0xffff2112> <pcomp> <accomp>]
: Apr 19 21:44:06 inferno pppd[1675]: sent [LCP EchoReq id=0x0 magic=0xffff2112]
: Apr 19 21:44:06 inferno kernel: ppp: receive buffer, count = 54
: Apr 19 21:44:06 inferno kernel: 0C 03 00 1A 00 00 36 BF ......6.
: Apr 19 21:44:06 inferno kernel: 4D 53 52 41 53 2D 31 2D MSRAS-1-
: Apr 19 21:44:06 inferno kernel: 56 45 52 53 41 2D 48 4F VERSA-HO
: Apr 19 21:44:06 inferno kernel: 4D 45 D1 09 7E 7E C0 23 ME..~~.#
: Apr 19 21:44:06 inferno kernel: 01 24 00 13 07 72 70 61 .$...xxx
: Apr 19 21:44:06 inferno kernel: 72 6B 65 72 06 52 6F 62 xxxx.xxx
: Apr 19 21:44:06 inferno kernel: 65 72 74 27 58 7E xxx'X~
: Apr 19 21:44:06 inferno pppd[1675]: rcvd [LCP code=0xc id=0x3 00 00 36 bf 4d 53 52
:41 53 2d 31 2d 56 45 52 53 41 2d 48 4f 4d 45]
: Apr 19 21:44:06 inferno pppd[1675]: sent [LCP CodeRej id=0x2 0c 03 00 1a 00 00 36 bf
:4d 53 52 41 53 2d 31 2d 56 45 52 53 41 2d 48 4f 4d 45]
: Apr 19 21:44:06 inferno pppd[1675]: rcvd [PAP AuthReq id=0x24
: user="xxxxxxx" password="xxxxxx"]
My *guess* would be a pppd bug. The code=0xc is the LCP Identification
code which contains arbitrary text strings to allow a ppp implementation
to identify itself. There are two 0xc codes back-to-back; pppd swallows
the first one without protest but CodeRej's the second which likely
means that it's really confused.
Identification codes are rare, and CodeRej is rare and so not much tested
by the troops in the field. Pppd acts as if it is in a state where it
can't respond to the PAP AuthReq until it gets an LCP request - in this
case the LCP TermReq that the peer eventually sent.
If you want/need to pursue this further then you should consider joining
the Linux ppp mailing list and asking the question there. The pppd
maintainer has been active on the list lately.
Send a message with the line
subscribe linux-ppp
contained in the body to [EMAIL PROTECTED]
To leave the mail list, send 'unsubscribe linux-ppp' to the same
mail address.
--
Clifford Kite <kite@inet%port.com> Not a guru. (tm)
/* Microsoft is a great marketing organization.
* It _has_ to be */
------------------------------
From: "J.R. Satterfield" <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.misc,comp.os.linux.setup,linux.redhat.misc
Subject: Random Character Swapping ??? Help!
Date: 26 Apr 1999 17:15:26 GMT
Well, this may or may not be Linux related... but it's my best guess.
Background:
Pentium 90 running RH5.1 is my firewall masq. machine forone other win98
machine, using 10BT. Linux is connected using ADSL. Linux box has no
monitor and is for the most part administered remotely from the win98
machine or in some occasions from outside.
Problem:
In accessing from the win98 machine and also from outside via telnet, I
am seeing somewhat random (unpredictable, but repeatable) character
swapping. For example, read carefully the results of two consecutive
ping commands:
[root@hoth log]# ping yahoo.com
PING yahoo.com (204.71.177.35): 56 data bytes
64 bytes from 204.71.177.35: icmp_seq=0 ttl=242 time=196.9 ms
64 bytes from 204.71.177.35: icmp_seq=1 ttl=242 time=195.9 ms
64 bytes from 204.71.177.35: icmp_seq=2 ttl=242 time=194.8 ms
64 bytes from 204.71.177.35: icmp_seq=3 ttl=242 time=194.4 ms
64 bytes from 204.71.177.35: icmp_seq=4 ttl=242 time=1:2.5 ms
64 bytes from 204.71.177.35: icmp_seq=5 ttl=242 time=194.8 ms
64 bytes from 204.71.177.35: icmp_seq=6 ttl=242 time=194.4 ms
64 bytes from 204.71.177.35: icmp_seq=7 ttl=242 time=196.3 ms
64 bytes from 204.71.177.35: icmp_seq=8 ttl=242 time=1:3.8 ms
r
--- yahoo.com ping statistics ---
9 packets transmitted, 9 packets recei~ed, 0% packet loss
round-trip min/avg/max = 114.4/117.0/123.8 ms
[root@hoth log]# !!
ping yahoo.com
PING yahoo.com (204.71.177.35): 56 data bytes
64 bytes from 204.71.177.35: icmp_seq=0 ttl=242 time=195.5 ms
64 bytes from 204.71.177.35: icmp_seq=1 ttl=242 time=193.6 ms
64 bytes from 204.71.177.35: icmp_seq=2 ttl=242 time=193.2 ms
64 bytes from 204.71.177.35: icmp_seq=3 ttl=242 time=114.3 ms
64 bytes from 204.71.177.35: icmp_seq=4 ttl=242 time=196.3 ms
64 bytes from 204.71.177.35: icmp_seq=5 ttl=242 time=198.0 ms
64 bytes from 204.71.177.35: icmp_seq=6 ttl=242 time=198.1 ms
64 bytes from 204.71.177.35: icmp_seq=7 ttl=242 time=1:0.3 ms
64 bytes from 204.71.177.35: icmp_seq=8 ttl=242 time=116.8 ms
64 bytes from 204.71.177.35: icmp_seq=9 ttl=242 time=199.2 ms
r
--- yahoo.com ping statistics ---
10 packets transmitted, 10 packets recei~ed, 0% packet loss
round-trip min/avg/max = 113.2/116.5/120.3 ms
[root@hoth log]#
See the ':' where a number should be? and the word received has
a '~' instead of a v... both times...
Why does this happen?
Also, possibly related, 'w' doesn't always
return complete results... sometimes not telling me about sessions
that I know to be running (like the one originating the 'w' command!)
For example:
[root@hoth log]# w
1:05pm up 34 days, 11:51, 4 users, load average: 1.02, 1.05, 1.00
USER TTY FROM ( LOGIN@ IDLE JCPU PCPU WHAT
jsatter ttyp3 :0.0 Thu 9pm 0.00s 47:40m 47:40m bash
[root@hoth log]# w
1:06pm up 34 days, 11:51, 4 users, load average: 1.02, 1.05, 1.00
USER TTY FROM ( LOGIN@ IDLE JCPU PCPU WHAT
jsatter ttyp2 :0.0 Thu 9pm 5days 0.10s 0.10s
bash
jsatter ttyp0 acy1sun01.mea.co 9:17am 0.00s 2.00s ? -
Oh, and this is weird... I just tried 'w' two more times, once resulting
in a seg. fault, and the second time giving me this:
[root@hoth log]# w
Error: /proc must be mounted
To mount /proc at boot you need an /etc/fstab line like:
/proc /proc proc defaults
In the meantime, mount /proc /proc -t proc
1:07pm up 0 min, 4 users, load average: 1.07, 1.07, 1.01
USER TTY FROM LOGIN@ IDLE JCPU PCPU WHAT
[root@hoth log]#
What the heck is wrong with my machine!? Or is it a problem with my
(small) network? (0 packet loss suggested to me otherwise...)
Thanks
------------------------------
From: "Trip6" <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.setup,linux.redhat.install
Subject: Re: GET REDHAT 6.1 FiNAL!
Date: Mon, 26 Apr 1999 23:30:43 +0200
In advertising, we call this 'high stopping power' ;-)
Xarj wrote in message ...
http://teen.sexhound.net/xxxpic/ 100% FREE WEBSITE WITH 200 PICS OF HOT SEXY
TEENS UPDATED WEEKLY! PLEASE COME AND HAVE A LOOK IF YOU LIKE TRUE TEEN
PICTURES... IT'S ALL IN YOUR INTEREST! http://teen.sexhound.net/xxxpic/
------------------------------
From: Ronald Cole <[EMAIL PROTECTED]>
Subject: Re: linux-2.0.36 loses eth0 route after three weeks...
Date: 26 Apr 1999 14:35:09 -0700
Ronald Cole <[EMAIL PROTECTED]> writes:
> The machine running 2.0.36 is my masquerading firewall (slackware-3.4
> based with security fixes). It is connected to my internal network
> (192.168.1.x). After being up for quite a while, the kernel appears
> to lose the route to the internal network. I can go to the console
> and re-add the route and everything is peachy-keen again.
>
> Has anyone seen this before?
This sniffs more and more like a tulip driver problem... Can't get
packets in or out when the Kingston NIC hangs. Taking the interface
down, bringing it back up and re-adding the route tends to restore
functionality.
Anyone concur with my driver assessment? Should I grab the latest v0.90
and use it?
--
Forte International, P.O. Box 1412, Ridgecrest, CA 93556-1412
Ronald Cole <[EMAIL PROTECTED]> Phone: (760) 499-9142
President, CEO Fax: (760) 499-9152
My PGP fingerprint: 15 6E C7 91 5F AF 17 C4 24 93 CB 6B EB 38 B5 E5
------------------------------
From: [EMAIL PROTECTED] (Karl Heinz Kremer)
Subject: Re: Remote Install??
Date: Mon, 26 Apr 1999 21:29:27 GMT
In article <7g2k3q$ip5$[EMAIL PROTECTED]>,
"Newbie" <[EMAIL PROTECTED]> writes:
> Is it possible to remotely install linux on a new pc (nothing on the hdd at
> all) using a boot disk.
> log onto a server and then install linux.
> Is this possible with any distibutions?
> If so how do I go about doing it??
I don't know about "any" distribution, but it's possible with
SuSE Linux 5.3 and 6.0 (and I would guess 6.1 does the same). You
can either export a NFS drive to the client, or you can install
via ftp. SuSE also allows you to make an installation across the
Internet by using their ftp server as source for the installation.
It's pretty easy: Just use the boot disk and boot your
machine. Then choose the source for the installation. For any
network installation you have to specify an IP address for the client and
for the server and a netmask (and a gateway address if you need to
go across a router).
The rest of the installation is exactly the same as with a local CDRom.
Karl Heinz
--
Karl Heinz Kremer [EMAIL PROTECTED]
http://www.khk.net
------------------------------
From: Dimitrios Liappis <[EMAIL PROTECTED]>
Subject: Re: Network setup tips, please?
Date: 26 Apr 1999 23:11:37 +0300
"Kalle Svensson" <ksvensson@bigfoo***NOSPAM***t.com> writes:
> OK, this is the situation. I'm a network newbie.
> I and a couple of friends of mine want to set up a network consisting of a
> router/gateway/firewall/whatever-you-call-it (constantly connected to the
> internet via ISDN), two servers (Pri./Sec. DNS, WWW, mail, FTP) and an
> undefined amount of workstations that need internet access.
> Several questions pop up in my mind:
>
> * How many "real" IP addresses do I need to apply for? My guess here would
> be three, one for each server and one for the router.
Since real IP addresses cost real money you 'd better think a bit
more regarding this part of your setup. It is not uncommon for a server
to play the role of the router/gateway therefore you might want to convert
the single router thingy to a server/router.
For instance, you might want to run DNS on the first server (and perhaps
sendmail and apache). You may route the second server (running other things such as
dhcpd etc.) via the first one, being the default gateway.
Using ipchains you can carefully setup your primary server/router carefully
disabling all useless ports (Ingoing & Outgoing) and making it more secure.
> * All the machines in the network use IPs in the 192.168.1.0 subnet
> (surprise!). How do I make the servers work with real IPs?
> This question shows just how clueless I am, but what will happen to the
> network if I give them their real IP addresses, say 132.75.4.9 and
> 234.76.193.27 instead of the old 192.168.1.2 and 192.168.1.3?
> I was thinking about something with IP-aliasing the router to all the IPs
> and then port-forwarding to the servers, but this sounds a bit far-fetched.
> Is there an easier way?
I don't follow you completely here. Perhaps you might want to rephrase that.
(this sounded like Data speaking in StarTrek !)
Anyway, you give the real ip/netmask/resolv settings on your primary server.
Then either you use a second NIC on your primary server, or as you said using ip
aliasing, you configure ipchains to route correctly and block all unnecessary ports.
The 2 NIC solution is the simpler and easier. You will use a bogus ip for the second
NIC of course.
Using the second server (don't put everything on the primary) you may want
to do some masquerading (for the rest of the workstations offering ip's like 192.168.
x.x) using ipchains again. This time you will route the packets using the primary
server (infact, through it's second NIC or the aliased IP) as the default gateway.
>
> * Should I set up the router to IP masq. the unnamed workstations. I think
> this seems natural.
Not really. You 'd better (for security reasons mainly) use masquerading
on a server that's behind the server/router (firewall). It will obviously route
the packets coming from the workstations through the router.
>
> * Approx. what kind of box do I need for a router of this kind? 486,
> Pentium, PII or 256 PII CPU cluster (uh, I hope not)?
A 486DX2/66 works fine with me but please use branded, non-shit NE2000 clone
NIC, not because I like to sponsor big companies but because NE2000 use memory
mapped IO and it's a CPU hog. Using a couple of NE2000 clones on a 486DX2 could
easily bring it to its knees on high traffic occasions (and you won't reach more
than approximately 500Kb/sec at best).
> I suppose these aren't all the important questions, but I'm tired (it's late
> here in Sweden, and I've been drinking too much coffee earlier today) so
> I'll end here and see what you say.
How's the weather up there ?
>
> Oh, yeah, almost forgot to mention, we'll be using Debian 2.1.
Good choice
>
> Thanks in advance.
>
> /Kalle Svensson
Don't hesitate to ask/mail for any further questions
(I hope I helped and not messed things up for you !)
Dimitris
==============================
[EMAIL PROTECTED]
remove kill-all-spam. to reply !!
------------------------------
** 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
******************************
