Linux-Networking Digest #401, Volume #10 Sat, 6 Mar 99 16:13:54 EST
Contents:
LAT-ACK bug? (Paul Farber)
pppd: tcgetattr: Input/output error(5)????? (Mike Thatcher)
Re: Help! How do I build a tulip ethernet driver under REDHAT ? (Stephen Kwan)
Re: 2 NICs in Linux (Colin)
"Stale NFS file handle" problem (2.0.36, RH5.2) (Juergen Nickelsen)
Re: Linux2.2.1 ipchains firewall: help! (need education) (root)
Re: Machine name themes - what do you use? (Juergen Nickelsen)
false error and drop readings from ifconfig? (Steve Maring)
Re: Maximum # of Ethernet Interfaces for Linux? (Bob)
iproute2 and ip tunneling problem (Steve Maring)
Re: can't ping Windows 95 from Linux (David Kirkpatrick)
Re: Problem: Diskless Boot with NFS (Oscar Stiffelman)
Re: Q(nfs): sharing most of /tftpboot/$CLIENT/etc/ between clients (Oscar Stiffelman)
Re: smbmount won't work (Bill Hayles)
Re: ppp doesn't use ISP's Name Server (David Kirkpatrick)
Re: Mirroring an NT-Server with Linux (Jay Thorne)
Cannot connect to shared drives on SAMBA SERVER from WIN98 CLIENT ("Ronald Hovens")
Problem (Norbert Kraiczy)
Re: Linux as a router to replace school NT4 box? (David Kirkpatrick)
----------------------------------------------------------------------------
From: Paul Farber <[EMAIL PROTECTED]>
Subject: LAT-ACK bug?
Date: Sat, 6 Mar 1999 13:14:16 -0500
Hello all..
been getting these lingering TCP connections on my mail server:
tcp 65 25365 mail.f-tech.net:10003 mail.hotmail.com:smtp
LAST_ACK
they all seem to come from hotmail and from what I understand the
conenction is waiting for a packet from the other side to say "yes, close
the connection" but is never getting it.
Why is linix not timing out the connection? I am running RH 5.1 kernel
2.0.34 with no patches.
Paul D. Farber II
Farber Technology
Ph. 570-628-5303
Fax 570-628-5545
[EMAIL PROTECTED]
------------------------------
From: Mike Thatcher <[EMAIL PROTECTED]>
Subject: pppd: tcgetattr: Input/output error(5)?????
Date: Fri, 05 Mar 1999 07:42:53 -0800
Reply-To: [EMAIL PROTECTED]
My RH5.1 system ppp was working but now is reporting this error. Can
anyone provide some insight into what is going on so I can get back
online.
Mike
------------------------------
From: Stephen Kwan <[EMAIL PROTECTED]>
Subject: Re: Help! How do I build a tulip ethernet driver under REDHAT ?
Date: Fri, 05 Mar 1999 10:58:04 -0500
In article <MPG.11490718e05709e989685@news>, lenny wintfeld
<[EMAIL PROTECTED]> wrote:
> Hi-
>
> I've got to build a revised tulip driver for my Netgear FA310tx
> ethernet card. Evidently I need v .89 or higher & I've got v .88 now. The
> instructions that came with the driver are very clear but are for non
> RPM linux's. Can you give me instructions on how to do the build under
> REDHAT's quite different file tree?
Lenny,
Go to RedHat and download the latest kernel. The latest kernel
includes the proper tulip driver. Follow RedHat's instructions and
rebuild the kernel with the tulip driver.
This accomplishes at least 2 things: you'd be running the latest stable
kernel with all the associated bug fixes, and you'll have the proper
tulip driver.
I have the same ethernet card as you do and I'm running 2.0.36 kernel.
The included tulip driver works fine.
------------------------------
From: Colin <[EMAIL PROTECTED]>
Subject: Re: 2 NICs in Linux
Date: Sat, 06 Mar 1999 11:54:58 -0500
Doug Goldstein wrote:
>
> You must also specify both cards here. The syntax for it is.
> append="ether=IRQ,I/O ADDR,eth0 ether=IRQ,I/O ADDR,eth1"
You might even get away with:
append="ether=0,0,eth1"
--
Reply to "cwv [at] idirect (dot) com"
------------------------------
From: Juergen Nickelsen <[EMAIL PROTECTED]>
Subject: "Stale NFS file handle" problem (2.0.36, RH5.2)
Date: 5 Mar 1999 15:55:46 GMT
[I have also posted this to the linux-net mailing list.]
Hello all,
I seem to have a problem writing large files from a Solaris NFS client
to a Linux (2.0.36) NFS server.
E. g. I have a large file to unzip -- after writing some data gunzip
terminates with the message "Stale NFS file handle" like this:
$ ls -l
total 14570
-rw-r--r-- 1 ni ni 7428958 Mar 4 12:35 foo.gz
$ /bin/pwd
/.automount/dali/root/export/tellique/home/ni/tmp
$ gunzip -c foo.gz > fooo
gunzip: stdout: Stale NFS file handle
$ ls -l
total 51696
-rw-r--r-- 1 ni ni 7428958 Mar 4 12:35 foo.gz
-rw-rw-r-- 1 ni ni 11509760 Mar 5 12:13 fooo
This is quite reproducible. The same happens with cp and with other
applications writing large files.
The NFS client is a Sun UltraSPARC running Solaris 2.6; the server is a
400 MHz Intel P2 machine with Redhat 5.2, kernel 2.0.36, and large, fast
SCSI disks. The directory is mounted via the automounter; the amd.conf
on the sun is
/defaults fs:=${autodir}/${rhost}/root/${rfs};\
opts=nodev,rsize=4096,wsize=4096,hard,intr
* rhost:=${key};type:=host;rfs:=/
(without the backslash and line break), amd is called as
/usr/sbin/amd -a /.automount -l syslog -c 1000 /net /etc/amd.conf
The /etc/eports on the Linux server is
/ *.tellique.de(no_root_squash)
I am at a loss at this point; I have read the NFS-HowTo, searched the
web and the archives of the linux-net mailing list, and found nothing
related.
Can someone help? Have I overlooked some important source of
information? I'd gladly RTFM or a FAQ if I knew which.
Thanks in advance for any help!
--
Juergen Nickelsen <[EMAIL PROTECTED]>
Tellique Kommunikationstechnik GmbH
Gustav-Meyer-Allee 25, 13355 Berlin, Germany
Tel. +49 30 46307-552 / Fax +49 30 46307-579
------------------------------
From: [EMAIL PROTECTED] (root)
Crossposted-To: comp.security.firewalls,utah.linux
Subject: Re: Linux2.2.1 ipchains firewall: help! (need education)
Date: 6 Mar 1999 14:28:48 GMT
KA ([EMAIL PROTECTED]) wrote:
: Question 1: I'm running DNS behind the firewall (I used it to learn
: about DNS, and have been lazy to remove it afterwards.) What is needed
: to allow DNS to pass through the firewall? I can't figure out what I
: need to let in to get it to work. If I open up the firewall (set all
: input policies to accept (no firewall)) everything works fine.
All of those connections are not a sort of "one way communication", but
bidirectional instead. Therefore you need an
output rule: procotol udp, your computer, ports > 1023 to nameserver, port 53
input rule: protocol udp, nameserver, port 53 to your computer, ports > 1023.
clients always use ports > 1023, server < 1023 (ok, not always; x11, ftp and
so on are different, but anyway...)
: Question 2: The only way I can get something to work (e.g. http, ftp,
: telnet) from behind the firewall to the outside is to add an input rule
: allowing traffic to that port. For example, I telnet does not work
: unless I include:
: ipchains -A eth1-in -P TCP -s 0/0 telnet -j ACCEPT
your computer is the client, so it's enough to allow incoming packets to
ports > 1023; you have to use MUCH more narrow rules: not all of the ports,
but only certain ranges; not all ip addresses of the whole internet, only
those ones you really need.
: My question is why do I allow incoming traffic on this port in order to
: work? My understanding is that my machine uses a high-number port to
: connect to the target machine's port (e.g. 23 for telnet), and that
: machine uses a high-number port to return communication. I know this is
^^^^
no, it uses port 23: You are the telnet client, so you use ports > 1023,
and the peer is the server, using port 23. Communication is bidirectional,
so from your computer, >1023 to peer, 23 and from peer, 23, to your computer,
> 1023.
: I assume that I need to shut down the daemons listening on these ports
: (yet to do). I don't want anybody connecting from the outside.
Of course you need, but only as an additional measure.
There is no need to allow incoming packets to connect to each of your
trusted ports (< 1023) on packet filtering level. This is not true, if
you want to offer some certain services, but I think you don't want to do so.
: Question 3: (Related to #2) Is this safe? Doesn't this essentially
no, not at all
: negate an input policy of "deny"?
yes
Is it even neccessary
no
: (and can go away)?
yes
Try to find out
1. the peers you really need (you need not each mailserver of the whole world)
2. the services you want to obtain from each single peer
and then formulate rules wich allow nothing else than exact what you just
found out. I. e., in most cases only one service at one peer, not
every service at every machine on the internet.
: ipchains -A eth1-in -p TCP -s $INET -d $IFM 1024: -j ACCEPT
: ipchains -A eth1-in -p UDP -s $INET -d $IFM 1024: -j ACCEPT
UDP from the whole internet? Good god - no. Only from that single
nameserver you use (i. e. one ip address and only one port there)
: Question 4: Is it safe to allow all ICMP traffic in? That's the only way
: I know of how to get stuff to work.
no, but according to the ipchains-HOWTO it is not recommended to block it
totally; well, it depends, if you can do without messages like "no route to
host", "network unreachable", and even without ping, then you can block
it.
Moreover, ICMP doesn't affect the working of, let's say, telnet, ftp, dns
and so on. So you can think of blocking it in the whole.
: ipchains -A eth1-in -p ICMP -s 0/0 -j ACCEPT
: Question 5: What references would you recommend to understand the how
: and why of this stuff?
- ipchains-HOWTO
- Chapman, Zwicky, Installation of internet firewalls
- http://www.cs.purdue.edu/coast/firewalls
bye, juergen.
------------------------------
From: Juergen Nickelsen <[EMAIL PROTECTED]>
Crossposted-To: comp.unix.solaris
Subject: Re: Machine name themes - what do you use?
Date: 5 Mar 1999 15:35:30 GMT
[EMAIL PROTECTED] (Steve Bellenot) writes:
> Some i've seen that haven't been mentioned yet (i think):
I am jumping late into this thread, but perhaps these have not yet
been mentioned: We are using painters (with german/european bias, of
course). Up to now we have
nolde, marc, degas, davinci, escher, picasso, dali, dix, busch,
monet.
I haven't yet been able to come up with an appropriate name for the
printer -- it should perferrably be a graphic artist who worked mostly
in black and white (and escher is already in use).
--
Juergen Nickelsen <[EMAIL PROTECTED]>
Tellique Kommunikationstechnik GmbH
Gustav-Meyer-Allee 25, 13355 Berlin, Germany
Tel. +49 30 46307-552 / Fax +49 30 46307-579
------------------------------
From: Steve Maring <[EMAIL PROTECTED]>
Subject: false error and drop readings from ifconfig?
Date: Fri, 05 Mar 1999 11:56:19 -0500
Reply-To: [EMAIL PROTECTED]
We compiled the 2.2.1 kernel on a bunch of boxes here and everything
seems to be fine. However, when we do an "ifconfig -a" we get this:
lo Link encap:Local Loopback
inet addr:127.0.0.1 Bcast:127.255.255.255 Mask:255.0.0.0
UP LOOPBACK RUNNING MTU:3924 Metric:1
RX packets:6 errors:0 dropped:0 overruns:0
TX packets:0 errors:348 dropped:6 overruns:0
eth0 Link encap:Ethernet HWaddr 00:80:C8:55:66:FE
inet addr:172.17.42.50 Bcast:172.17.42.255
Mask:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:98561 errors:0 dropped:0 overruns:0
TX packets:0 errors:16111418 dropped:78879 overruns:0
Interrupt:11 Base address:0xfc80
Notice the extreme TX errors and drops. We don't, however, notice any
networking problems at the application level.
Anybody know about this?
--
-Steve Maring
[EMAIL PROTECTED]
GTE Enterprise Solutions
Tampa, FL USA
------------------------------
From: Bob <[EMAIL PROTECTED]>
Subject: Re: Maximum # of Ethernet Interfaces for Linux?
Date: Sat, 6 Mar 1999 00:06:56 +0800
How many nic's can you fit into your pc ?
On Sat, 20 Feb 1999, Richard Thomas wrote:
> Date: Sat, 20 Feb 1999 06:12:13 GMT
> From: Richard Thomas <[EMAIL PROTECTED]>
> Newsgroups: comp.os.linux.networking
> Subject: Maximum # of Ethernet Interfaces for Linux?
>
> I curious as to how many ethernet interfaces a linux box can have...
> Also, is it a wise decision to make a linux box a combination
> firewall/proxy/router?
>
> Richard Thomas
> Computer Services Coordinator
> William H. Lemmel Middle School
> Baltimore, MD
> [EMAIL PROTECTED]
>
>
>
>
Bob PHILLIPS
Partner/System Administrator
[EMAIL PROTECTED] | ISP to the nor'west of Western Australia
| http://www.norcom.net.au
Yes, I am on the interthingy | If it aint broke, fix it, then it will be
==========================================================================
Pilbara Systems PO Box 2762 SOUTH HEDLAND WA 6722 AUSTRALIA
==========================================================================
------------------------------
From: Steve Maring <[EMAIL PROTECTED]>
Subject: iproute2 and ip tunneling problem
Date: Fri, 05 Mar 1999 12:00:33 -0500
Reply-To: [EMAIL PROTECTED]
Anybody know why I might be getting the following?
# ip tunnel add tunl0 mode ipip remote 172.17.42.50 local 207.223.13.75
ioctl: No buffer space available
--
-Steve Maring
[EMAIL PROTECTED]
GTE Enterprise Solutions
Tampa, FL USA
------------------------------
From: David Kirkpatrick <[EMAIL PROTECTED]>
Subject: Re: can't ping Windows 95 from Linux
Date: Sat, 06 Mar 1999 14:06:00 +0000
Reply-To: [EMAIL PROTECTED]
Anonymous this
Anonymous wrote:
>
> Hi,
>
> I have two machines - one is running Win95. Another can dual boot to
>
> Windows 95 or Linux. Both are connected with a cross cable.
>
> The problem is that I can ping each other while running both in Windows 95
>
> but can't ping while one boot up in Linux and another in Win95.
>
> Both has set the IP already.
>
> IP: 192.168.1.1 IP: 192.168.1.2
>
> Subnet: 255.255.255.0 for both
>
> Would you give me advice how to solve ths?
>
> Regards,
>
> Rory Cham
>
>
>
> ------------------ Posted via SearchLinux ------------------
> http://www.searchlinux.com
--
[EMAIL PROTECTED]
[EMAIL PROTECTED]
[EMAIL PROTECTED]
------------------------------
From: Oscar Stiffelman <[EMAIL PROTECTED]>
Subject: Re: Problem: Diskless Boot with NFS
Date: 6 Mar 1999 19:31:55 GMT
I got a similar message when I tried to boot to a RamDisk as the root
partition. I think the problem may have been an incomplete /lib. Can you
do a full copy of the critical root directories to make sure you havn't
missed anything?
-- Oscar
Frank Jacobi wrote:
> Hello everyone out there in the fast growing linux community!
>
> I have a small problem with my small network. It is the diskless
> machine, a old IBM Thinkpad 340 Laptop with a Realtek parallel port
> network adapter, that does some strange stuff when booting.
> It boots the disk and loads the drivers and starts the negotiation,
> then I get the following:
> Root-NFS: Got BOOTP answer from 192.168.1.1, my address is 192.168.1.2
> Root-NFS: Got filehandle for /tftpboot/192.168.1.2 via RPC
> VFS: Mounted root (nfs filesystem)
>
> that is all I get
>
> there is permanent traffic on the network, but nothing real is
> transmitted, just some link resolving stuff
>
> Thanks in advance!
> Frank Jacobi
>
================== Posted via SearchLinux ==================
http://www.searchlinux.com
------------------------------
From: Oscar Stiffelman <[EMAIL PROTECTED]>
Subject: Re: Q(nfs): sharing most of /tftpboot/$CLIENT/etc/ between clients
Date: 6 Mar 1999 19:31:44 GMT
I would like to hear the response to this message.
We have 70 nodes using NFS and a common /dev, /etc, /lib, /sbin, /bin, /usr
We have been running for many months continuously without any crashes.
-- Oscar
Harald Kirsch wrote:
>
>
> A while ago I came across a patch of part of the NFS-system (don't
> know whether client- or server-side) which allows the following:
>
> Suppose a set of files (not dirs) on the server named e.g.
>
> /anydir/theFile
> /anydir/theFile%theHost1
> /anydir/theFile%theHost2
>
> Further suppose a mount on clients `theHost1', `theHost2' and
> `otherHost' like:
>
> mount -t nfs server:/anydir /anydir
>
> The patch I am talking about makes sure that all three clients will
> have a file named
>
> /anydir/theFile
>
> however the nfs-system maps one of the three files listed above
> onto that name such that
>
> `theHost1' in fact accesses server:/anydir/theFile%theHost1,
> `theHost2' in fact accesses server:/anydir/theFile%theHost2,
>
> while
>
> `otherHost' accesses the default file server:/anydir/theFile
>
>
> This mechanism is pretty nice to let serveral diskless clients share a
> mostly common /etc, /lib or /bin directory on a server.
>
>
> PLEASE: Don't tell me that I can get a similar behaviour with
> softlinks. At least pam does not seem to like /etc/password to be a
> softlink, and /etc/password is one of the few files the clients shall
> not share.
>
>
> Where is this patch?
>
>
> Regards,
> Harald Kirsch
> --
> ---------------------+------------------+--------------------------
> Harald Kirsch (@home)| | Now I rebooted.
> [EMAIL PROTECTED] | | --- Jerry Pournelle, BYTE
> gegen Punktfilitis hilft nur `chmod u-w ~'
================== Posted via SearchLinux ==================
http://www.searchlinux.com
------------------------------
From: Bill Hayles <[EMAIL PROTECTED]>
Crossposted-To: comp.protocols.smb,linux.samba
Subject: Re: smbmount won't work
Date: Sat, 06 Mar 1999 18:55:53 +0000
ozric wrote:
>
> No I rpmed the Samba 2.0.3 and it did not have smbmount in it ? So you
> are saying that I have to compile it. Ok I will remove the smbfs rpm and
> compile samba I just hate that I have to --prefix the path to put all
> the stuff where I want it.
>
You have to explicitly include the smbmount option when you run
/configure :-
/configure --with-smbmount
And if it makes you feel any better, that had to be pointed out to me,
too!
--
Who needs a life when there's linux?
Bill Hayles
[EMAIL PROTECTED]
------------------------------
From: David Kirkpatrick <[EMAIL PROTECTED]>
Subject: Re: ppp doesn't use ISP's Name Server
Date: Sat, 06 Mar 1999 14:47:09 +0000
Reply-To: [EMAIL PROTECTED]
Setup /etc/resolv.conf. cd /usr/doc/HOWTO, less PPP-HOWTO and
search on resolv.conf.
Mike Niemann wrote:
>
> I'm connected just fine to my ISP, and ppp appears working fine. I can
> get to www sites (via Netscape) by number... but not by name. Tried
> nslookup, but it never returns either.
>
> Does IP packet forwarding have anything to do with this?
>
> It doesn't look like named is running. Is there an easy way to display
> all processes... even daemons? I'm not running a local name server.
>
> Config: RH5.2, KDE1.1, kppp, local peer-to-peer lan, ISP=Netcom.
> Host=pro200nx, Domain=netcom.com, IP=10.1.1.4.
>
> The only "odd" entry I see my Windoze networking is that "use IP
> compression" is enabled. Haven't seen a Linux anlog to that.
>
> Regards, Mike
>
> PS - a simple setup for small lan (W98,NT4,no server), ISP dialout
> which includes which services/daemons/host/domain/ whatever names
> would be much appreciated.
--
[EMAIL PROTECTED]
[EMAIL PROTECTED]
[EMAIL PROTECTED]
------------------------------
From: Jay Thorne <[EMAIL PROTECTED]>
Subject: Re: Mirroring an NT-Server with Linux
Date: Sat, 06 Mar 1999 19:56:32 GMT
Wolfgang Falk wrote:
>
> Hello,
>
> has anyone an idea how to mirror an nt4.0 server on a linux-box? i need it
> to prepare the replacement of this nt-server. this mirroring is just for two
> weeks (in a huge multi-user area) to have two redundant servers, because of
> several reboots on each machine. both should detect the reboot of the other
> machine and take over the fileserver-tasks with the latest version of data
> .. i know, it s not an easy job .. but if anyone can give me some hints i d
> be very glad ;-)
>
> greetings,
> hawk
mount the NT disks on the linux box with smbmount. (comes with samba)
cp -aR them to the linux fileshare areas.
--
Jay Thorne The Net Result System Services [EMAIL PROTECTED]
Http://net.result.com
Zoom 505 Effect page http://net.result.com/~jay
Zoom 5xx series Patch Database: http://net.result.com/~jay/db.html
------------------------------
From: "Ronald Hovens" <[EMAIL PROTECTED]>
Crossposted-To: comp.protocols.smb
Subject: Cannot connect to shared drives on SAMBA SERVER from WIN98 CLIENT
Date: Sun, 7 Mar 1999 20:53:41 +0100
I am running a Linux samba server (1.9.18p10) and I want to connect to it
with a Laptop that runs on Win98.
TCP networking is running fine: I can telnet and ping from one to another.
Furthermore, I can see the shared devices on the laptop when I run
smbclient -L hostnameoflaptop> on my linux samba server.
My problem is that I cannot connect to (the shared devices on) my samba
server: everytime I enter data in the win98 logon dialog, I get the message
"the domain password you supplied is not correct, or access to logon server
has been denied".
========================
On my laptop I have set:
primary logon: Client for micorsoft network
workgroup in Identification: MYGROUP
properties for Client for microsoft networks: Log on to windows NT domain
MYGROUP
wins enabled for the network connection (wins adress is the linux samba
server IP adress)
========================
Some relevant(?) settings on my samba server (in /etc/smb.conf):
workgroup = MYGROUP
encrypt passwords = yes (I have read that win98 sends encrypted passwords)
smb passwd file = /etc/smbpasswd
os level = 34
domain master = yes
preferred master = yes
domain logons = yes
wins support = yes
I assume that something is wrong with password encryption. What can I do?
Please help.
R. Hovens
------------------------------
From: Norbert Kraiczy <[EMAIL PROTECTED]>
Subject: Problem
Date: Sat, 06 Mar 1999 20:32:10 +0100
Hi,
When I'm run fdisk then I see following:
[Wenn ich fdisk starte und p (print) waehle sehe ich folgendes:]
/temp/hda1/ .....................
/tem/hda2/ ........................
/temp/hda3/ ................
etc.
[Was ist passiert und wie komme ich zurueck zu:]
What's happend ? What is the best way to following (true) parameter:
/dev/hda1/ .....................
/dev/hda2/ ........................
/dev/hda3/ ................
etc.
Norbert
------------------------------
From: David Kirkpatrick <[EMAIL PROTECTED]>
Subject: Re: Linux as a router to replace school NT4 box?
Date: Sat, 06 Mar 1999 15:47:30 +0000
Reply-To: [EMAIL PROTECTED]
Andy,
In general yes it will work and is not a big deal to setup.
There is a big range of speeds of 386/486 machines but a decent
486 will work.
d
Andy wrote:
>
> Hi,
>
> We have just configured a NT4 box to route between two subnets on our school
> LAN, however as the current setup with the NT4 box is tying up a machine we
> are wondering if we could run linux on a old 486 or 386 todo the routing.
>
> We have copies of Red hat Linux 5.1 & S.u.S.E Linux 5.2.
>
> Currently our NT4 box has two ethernet cards assigned ip 192.168.2.31 &
> 192.168.3.31 respectively, both with a netmask of 255.255.255.0.
>
> Should configuring a Linux box to do this be fairly straightforward and
> would a 386 /486 cope with network traffic (about 60 pc's across both
> subnet's). I have setup linux a few years ago as a web server (apache), but
> have never tried routing with it.
>
> The alternative is for the school to buy a new NT box to replace the one
> that is currently handling the routing.
>
> Regards,
>
> Andy
--
[EMAIL PROTECTED]
[EMAIL PROTECTED]
[EMAIL PROTECTED]
------------------------------
** FOR YOUR REFERENCE **
The service address, to which questions about the list itself and requests
to be added to or deleted from it should be directed, is:
Internet: [EMAIL PROTECTED]
You can send mail to the entire list (and comp.os.linux.networking) via:
Internet: [EMAIL PROTECTED]
Linux may be obtained via one of these FTP sites:
ftp.funet.fi pub/Linux
tsx-11.mit.edu pub/linux
sunsite.unc.edu pub/Linux
End of Linux-Networking Digest
******************************
