Linux-Networking Digest #847, Volume #10 Tue, 13 Apr 99 22:13:43 EDT
Contents:
Re: Just a plain old IP address and dedicated line... (Alan Westhagen)
Re: 3Com 3C905 - Host error, FIFO diagnostic register 2000 ("Druce Vertes")
Re: NFS server crash on all our linux boxes ("Sebastien Boving")
Re: How do i setup forwarding? ("Belgarion")
Re: 3 com 3c905b TX network card problems... ("ryan")
Re: Adding header lines using command line mail program (era eriksson)
Re: how to use IP-adres in firewall script ("Belgarion")
How does daemon take control from inetd? (Stephen Quattlebaum)
BOOTP and NT4 Workstations ("RBS")
MI/X X client question ("John E. Kuslich")
broken interface...backup plan ([EMAIL PROTECTED])
Re: Samba (smbd) wont load (Ingo Buescher)
Re: NFS server crash on all our linux boxes (Craig J Copi)
Re: Can I do this from a laptop? (Marc Hering)
----------------------------------------------------------------------------
From: Alan Westhagen <[EMAIL PROTECTED]>
Subject: Re: Just a plain old IP address and dedicated line...
Date: Tue, 13 Apr 1999 09:20:51 -0700
"David R. Christianson" wrote:
>
> Okay, maybe I'm a nerd, but I just want full control over every aspect
> of my system. I want to administer my own network, my own web site, my
> own mail and news. But no- I have to pay someone else to provide these
> "services," forcing me to use their email, their proxy servers, and
> their web hosting if I want to do things legally. This may be fine for
> ninety-nine percent of computer users, but not me.
>
> Which brings up another question; considering the price of cable modem
> services, why - when leasing a dedicated connection (even a dang 56k)
> and a single IP address - is the cost so <expletive deleted> high in
> comparison? Is there anyone out there that can provide only what is
> necessary to be fully connected and at a fair price, so I can have fun
> and do the rest?
>
> I refuse to believe I am the only one who feels this way. I aplolgize
> for the outburst, but sheesh, already.
>
> --Dave
Depending where you live, a DSL connection may be the answer. It
was for me. My ISP provides me with one static IP address, publishes
the domain names westhagen.com, mail.westhagen.com, www.westhagen.com,
ftp.westhagen.com and thor.westhagen.com to the DNS, and forwards
all traffic destined for those addresses to my static IP. This allows
me to handle my own mail, including any @westhagen.com email addresses
that I want. It gives www.westhagen.com a permanent presence on
the internet, etc.
I use IP masquerading to provide transparent internet access to
a two W95 machines, another linux machine and a solaris workstation,
via my linux gateway server. Each user on this internal network has an
account on the server, which is her email destination, as far
as the outside world is concerned. She fetches her mail from the
server by using either Eudora or Netscape navigator. Qpopper runs
on the server to provide this pop mail service.
I live in Seattle. The ISP is Northwest Nexus and the transport
provider is US WEST.
Alan Westhagen
------------------------------
From: "Druce Vertes" <[EMAIL PROTECTED]>
Subject: Re: 3Com 3C905 - Host error, FIFO diagnostic register 2000
Date: Tue, 13 Apr 1999 13:31:13 -0400
Thanks, Bill,
Sounds like you know whereof you speak. It was probably a question of using
the right initialization options in conf.modules, I tried the default,
options=12 and a few others that seemed to make sense. I'm no longer sure
exactly which options created this particular sequence.
I finally replaced the 3Com with an Intel EtherExpress Pro 100. Had to swap
PCI slots twice, and rebuild the kernel a couple of times, but it seems to
work.
Druce Vertes
(CC 86)
Bill Paul <[EMAIL PROTECTED]> wrote in message
news:7eo4dh$jg$[EMAIL PROTECTED]...
> Daring to challenge the will of the almighty Leviam00se, Druce Vertes
> ([EMAIL PROTECTED]) had the courage to say:
>
> : Thanks, changed to IRQ 10 but no luck yet.
>
> : I would have thought that if it works under Win 95 with IRQ 9 that would
> : rule out a conflict...
>
> : Also upgraded to 2.2 kernel, that got rid of the errors in the messages
log,
> : but still no link LED on the Ethernet card, same behavior as before.
>
> Hm. Well, since the error doesn't come up anymore you probably won't
> care, but according to my 3Com 3c90X Technical Reference, 0x200 (bit
> 13) set in the Fifo Diagnostic register indicates an RX underrun
> error, whichis described as follows:
>
> "When set, this bit causes a hostError interrupt that requires
> either an RxReset or a GlobalReset command to clear. An rxUnderrun
> occurs when the host reads data out of the receive FIFO faster
> than the network can fill it, or faster than the FIFO can supply
> data to the rxData register, resulting in the host accidentally
> reading invalid data."
>
> : wasn't immediately able to get it working at 10MB either.
>
> : Next step is probably going to be to put in another Ethernet adapter,
but
> : this is a real pain.
>
> Well, before you do that, I noticed the following:
>
> : > > Apr 5 23:07:55 homer kernel: sysctl: ip forwarding off
> : > > Apr 5 23:07:55 homer kernel: 3c59x.c:v0.99H 11/17/98 Donald Becker
> : > > http://cesdis.gsfc.nasa.gov/linux/drivers/vortex.html
> : > > Apr 5 23:07:55 homer kernel: eth0: 3Com 3c905 Boomerang 100baseTx
at
> : > > 0xef00, 00:60:08:8d:64:3a, IRQ 9
> : > > Apr 5 23:07:55 homer kernel: 8K word-wide RAM 3:5 Rx:Tx split,
MII
> : > > interface.
> : > > Apr 5 23:07:55 homer kernel: Media override to transceiver type 4
> : > > (100baseTX).
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>
> It looks like you started tried to force the media type to 100BaseTX.
> However this is the wrong thing to do for the 3c905. You should be using
> the MII type (6) instead. The 3c905 and the 3c905B differ in that the
> 3c905 has an external MII transceiver, while the 3c905B has one built
> in. The 100baseTX (4) transceiver type actually means 'built-in 100Mbps
> tranceiver' which only applies to the 3c905B. It's wrong for the 3c905:
> you should be using MII (6) instead. There are simiular ambiguities
> when you compare, say, the 3c905 to the 3c900. The 3c900 cards are
> 10Mbps only and use the on-board 10Mbps transceiver in the boomerang
> chipset, but the 3c905B, which also supports 10Mbps, uses the external
> MII transciever instead and leaves the internal one idle.
>
> Note that the transceiver on the 3c905B NIC is set up to look like
> an MII compliant transceiver (you control it through the same management
> interface as the 3c905) but it's considered a different animal. This
> is probably because the 3c905B has support for real MII transceivers
> and 3Com wanted to draw a disctinction between that and the built-in
> one.
>
> -Bill
>
> --
>
============================================================================
=
> -Bill Paul (212) 854-6020 | System Manager, Master of Unix-Fu
> Work: [EMAIL PROTECTED] | Center for Telecommunications
Research
> Home: [EMAIL PROTECTED] | Columbia University, New York City
>
============================================================================
=
> "Mulder, toads just fell from the sky!" "I guess their parachutes didn't
open."
>
============================================================================
=
------------------------------
From: "Sebastien Boving" <[EMAIL PROTECTED]>
Subject: Re: NFS server crash on all our linux boxes
Date: Tue, 13 Apr 1999 19:09:10 +0200
Oliver Stahlhut wrote in message <[EMAIL PROTECTED]>...
>Hi there!
>
>This mail won't help you, too ... but i would like to state that all
>that you have written is absolutely correct. Same setup & same
>environment here (Linux, SGI, SUN, Windoze), but with SuSE 5.x and 6. I
>have been describing the problems for a long time in this group (and
>others).
>
>To draw a conclusion: NFSD-service on Linux doesn't work reliable and I
>would never use it in a production environment. That's what I experience
>and have been told by official Linux-support! ... Multiprocessor
>machines cause the most problems. I use a watchdog to restart the
>nfsservers when they hang.
how did you implement this?
>We have been running a separate little Linux-cluster for our students
>with 7 machines and didn't experience those problems. Well ... seems the
>SUN's or SGI's do the killing job. But we'll get rid of them in the next
>years ;)
someone else suggested Sun's too. We have netgroups for each type of os:
esat.sun has all suns, esat.hp has all hp, ... same for esat.sgi, esat.dec,
esat.linux-i386, ... Then there is one netgroup called esat.unix which
contains all previously mentionned netgroups.
we usually export to esat.unix. I will try on these linux servers to export
to all but esat.sgi and esat.sun. If it doesn't crash anymore, i'll try to
add esat.sgi. This way we will at least know which clients cause the
problem.
have to go.
bye and thanks for the reply,
Sebastien.
>
> Oliver
>--
>/*
> Oliver Stahlhut - Universitaet Hannover
> Institut f�r Theoretische Nachrichtentechnik
> und Informationsverarbeitung (TNT)
>
> mailto:[EMAIL PROTECTED]
> http://www.tnt.uni-hannover.de/~stahlhut/
>*/
--
=============================================
[EMAIL PROTECTED]
http://www.ves.kotnet.org/seb
=============================================
------------------------------
From: "Belgarion" <[EMAIL PROTECTED]>
Subject: Re: How do i setup forwarding?
Date: Tue, 13 Apr 1999 18:32:55 -0700
Enable "IP Forwarding" in the kernel, and go read the Ethernet-HOWTO. :)
If you want to do IP masquerading, which is having the box communicate with
multiple machines over 1 IP, then read the IP-Masquerading HOWTO...
Erik
M. Emmerich wrote in message <7ev63v$[EMAIL PROTECTED]>...
>Hi!
>
>I want to have forwarding in a Linux (SuSE 5.3) box, but i don't know how
to
>set it up. The box has 2 NICs, which are recognized correctly. How do i
tell
>the kernel that i want all packets arriving at eth0 to be forwarded to eth1
>(or vice-versa) ?
>
>thank you in advance
>
>M.E.
>
>
------------------------------
From: "ryan" <[EMAIL PROTECTED]>
Subject: Re: 3 com 3c905b TX network card problems...
Date: Tue, 13 Apr 1999 12:05:41 -0700
i am running 2 linux machine both have 3c905b tx's. they work great. I am
running them at 100. i have sucsessfully run them under the 2.0.36 kernel
and the new 2.2 kernels. I am retively new to linux. but if you have any
more questions i will try to help.
------------------------------
From: era eriksson <[EMAIL PROTECTED]>
Crossposted-To: comp.mail.misc,comp.mail.mime
Subject: Re: Adding header lines using command line mail program
Date: 13 Apr 1999 19:57:46 +0300
Followups (not that I expect any) set to comp.mail.misc, on a whim.
Please set Followup-To: when posting to multiple, only tangentially
related groups.
On Mon, 05 Apr 1999 02:40:43 +0800, Chris Severn
<[EMAIL PROTECTED]> posted to comp.mail.misc, comp.mail.mime,
comp.os.linux.networking:
> I'm interested in running a mail sending program (such as mail,
> fastmail, elm) on the command line, passing the body of the message
> through stdin, but ensuring that the header of the message contains a
> few particular lines - namely "Mime-Version: 1.0", "Content-Type:
> application/octet-stream; name=thefile.gif", and
> "Content-Transfer-Encoding: Base64".
<...>
> account with my ISP to do nearly everthing which I want. The extract
> from my .procmailrc file follows :
> | formail -I "" | mimencode -u >/tmp/convert.tmp ; tifftopnm
> /tmp/convert.tmp | ppmtogif | mimencode | mail -s "Fax Received `date`"
> [EMAIL PROTECTED] ; rm /tmp/convert.tmp
Any particular reason you're not using formail for this as well? Or
even echo, if there's nothing in the original headers you want to
keep. (And, any particular reason to go via a temp file?)
If you just want the body, use a B flag rather than an empty formail.
On the other hand, you can fix the Subject: header while you're at it.
As another gratuitous change, I've removed the call to date(1) [you
can get the date from the From_ pseudoheader instead]:
:0fhwi # Replace the headers with your own (tweak for a neater date)
* ^From [^ ]+[ ]+\/[^ ]+
| echo "Subject: Fax Received $MATCH\
\
Content-Type: application/octet-stream; name=thefile.gif\
\
Content-Transfer-Encoding: base64\
\
Mime-Version: 1.0\
\
"
# Yes, the backslashitis is ugly and counter-intuitive.
:0fbw # Transform the body by pushing it thru the following pipeline
| mimencode -u | tifftopnm | ppmtofif | mimencode
:0 # Now send off the results
! [EMAIL PROTECTED]
# Or perhaps you wouldn't even need to replace all headers, or resend?
The date grabbing could admittedly be improved a lot if you want to
enhance the human-readableness of the date stamp in the Subject. (Alan
Stebbens' Procmail library contains a nice date parser you could use.)
If you have printf(1) or an echo that understands options like -e and
so forth, you could probably clean up the backslash madness a lot. Or
you could put the headers in a variable (or a file, like somebody
already suggested, but that's kind of ugly as well IMHO) as outlined
in the FAQ.
(Yes, iiit's PLUG TIME: <http://www.iki.fi/era/procmail/mini-faq.html>)
Hope this helps,
/* era */
Or did I misunderstand your question? Anyway, echo works just fine
from the command line, too :-)
--
.obBotBait: It shouldn't even matter whether <http://www.iki.fi/era/>
I am a resident of the state of Washington. <http://members.xoom.com/procmail/>
* Sign the European spam petition! <http://www.politik-digital.de/spam/en/> *
------------------------------
From: "Belgarion" <[EMAIL PROTECTED]>
Subject: Re: how to use IP-adres in firewall script
Date: Tue, 13 Apr 1999 18:37:59 -0700
not sure about ipchains, but don't you have to have a /32 at the end of the
IP address to signify that it's a full address? I could easily be wrong
here, but that's how it is with ipfwadm and other firewall tools.
As for the cut script, looks kosher to me, but you should probably just echo
that to your shell and see what output you get to make sure the IP is being
cut correctly.
Erik
razoon wrote in message <[EMAIL PROTECTED]>...
>I want to run a firewallscript from commandline.
>
>IPADDR=`/sbin/ifconfig ipp p0 | /bin/grep 'inet addr' | /usr/bin/cut
>-f >2 -d: | /usr/bin/cut -f 1 -d' '`
>
>ipchains -A input -p udp -j DENY -d $IPADDR 137:139 gives an error.
>
>How can i generate my ip-addres?
>What is the right syntax?
>
------------------------------
Date: Mon, 12 Apr 1999 21:38:16 -0500
From: Stephen Quattlebaum <[EMAIL PROTECTED]>
Subject: How does daemon take control from inetd?
I have a fairly sparse background in networking so please bear with me in
the following question. I'm just learning networking (concentrating on
how Linux does it as well).
I'm trying to write a simple proof-of-concept Linux daemon to implement a
simple protocl called NMP (Nearest-Mirror-Protocol) I'm designing, and I
think I have everything I need to know figured out except for one thing -
when inetd detects a connection to a port and spawns a daemon, how does
the daemon connect to the proper client? The answer to this question
would appear to be simple: the daemon connects to the proper port and the
client and daemon are then connected. But I know that more than one
client can connect to a given port and be served at a time, so how does
the one-to-one correspondence come about? How does inetd get an IP
address, or whatever else it sends, to the daemon so that the daemon can
connect to the proper port and IP address?
If this question is answered simply by just looking at the linux sockets
interface, then please forgive me. I thought I'd go ahead and fire the
question off *before* trying to write the code (though I am reading up on
the sockets interface right now) to avoid major re-writes. I have looked
all through my available docs (info, man, /usr/doc, and the RedHat Linux
Library), but all the info I've found on inetd has been on how to
configure it, which was useful to me but didn't help any with the
programming side.
Thanks.
****
Stephen Quattlebaum
[EMAIL PROTECTED]
Death multitasks. Someday, your timeslice will be up.
The opinions expressed by me are not necessarily the opinions of the
University of Alabama or it's faculty/administration.
------------------------------
From: "RBS" <[EMAIL PROTECTED]>
Subject: BOOTP and NT4 Workstations
Date: Tue, 13 Apr 1999 23:24:03 +0200
Hello,
I have a Linux server which has a BOOTP server.
When I look what is happening on the network I see that my Windows 95
station ask for a new IP address every 1 hour. But when I look for the NT4
Workstation, I see they ask for a new address every second.
Is there a their special parameters to set on the NT4 WKS ?
Thank you
------------------------------
From: "John E. Kuslich" <[EMAIL PROTECTED]>
Reply-To: [EMAIL PROTECTED]
Subject: MI/X X client question
Date: Tue, 13 Apr 1999 10:42:35 -0700
I downloaded the free X client from the microimages web site for
installation on my Win 95 machine.
I can't get the damn thing to install.
I get the error message "Unable to update TNTPROC.INI".
I read the FAQ and followed all the advice about not putting a space in
a directory name.
Can any one help or is this software just crap.
JK
--
CRAK Software (Password Recovery Software)
Http://www.crak.com
[EMAIL PROTECTED]
602 863 9274 or 1 800 505 2725 In the USA
------------------------------
From: [EMAIL PROTECTED]
Crossposted-To: alt.os.linux
Subject: broken interface...backup plan
Date: Tue, 13 Apr 1999 17:23:10 GMT
lets say i have a lease line connection via eth0. this interface is the
default route. now, something happens to the line and packets can no longer
use this route. so, i want to bring up a backup connection (ppp) as a
temporary fix.
1) how do i determine that the line is down? ....so that i know to bring up
the ppp connection.
2) and vice-versa...line gets fixed, so i switch back
any ideas and/or solutions appreciated....
thanks
============= Posted via Deja News, The Discussion Network ============
http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own
------------------------------
From: Ingo Buescher <[EMAIL PROTECTED]>
Subject: Re: Samba (smbd) wont load
Date: 11 Apr 1999 09:38:00 GMT
Pedro Garrett <[EMAIL PROTECTED]> wrote:
> I'm trying to set up Samba on my Redhat 5.1 Linux (i386) box and the nmbd
> daemon loads fine (and the Win98 machine on my network can find the linux
> box ), but the smbd daemon errors out with the following message:
> bind failed on port 139 socket_addr=0.0.0.0 (Address already in use)
I had the same problem when setting up samba. This happened because
my inetd already listened to the port. Look in /etc/inetd.conf and
change it if necessary.
IB
________________________________________________________________
Ingo Buescher <[EMAIL PROTECTED]>
<[EMAIL PROTECTED]>
Fingerprint: D9C3 0899 78BF 15DF 0C99 B6B2 EFFC 8DF1 BAB7 703F
G-Code: [Sph:s FHu Oeu WTn Sxm ~1200/~30 Sz+ CS-12 PT-12/13 M++
Tc+ R--- H0 | A--- W-- EF++ S-- C G* CR | RLBM/RLMA a
cl++ e+ h+ i+ j-- o-]
------------------------------
From: [EMAIL PROTECTED] (Craig J Copi)
Subject: Re: NFS server crash on all our linux boxes
Date: 13 Apr 1999 18:28:41 GMT
In article <[EMAIL PROTECTED]>,
Oliver Stahlhut <[EMAIL PROTECTED]> writes:
> Sebastien Boving wrote:
>>
>> Hi,
>>
>> I'd like to report a problem i've been trying to solve since a long time
>> now. I almost never had reactions to my posts except people having the same
>> problem as i described. Also RH support doesn't seem to know the answer.
>>
>> Problem affects all our RH5.0+ systems. Recently i have installed kernel
>> 2.2.3 and hoped for a change but the same problem still occurs.
>>
>> The problem is that the NFS server on linux seems to stop answering requests
>> after a while. I suspect it is the mountd part which stop doing its work, or
>> maybe the portmapper. Usually there is no crash, it simply still runs but
>> does not what it is supposed to do. Clients hang while trying to mount,
>> sometimes severely (no way to kill the proces), sometimes just for a few
> .
> .
> .
>
> Hi there!
>
> This mail won't help you, too ... but i would like to state that all
> that you have written is absolutely correct. Same setup & same
> environment here (Linux, SGI, SUN, Windoze), but with SuSE 5.x and 6. I
> have been describing the problems for a long time in this group (and
> others).
>
> To draw a conclusion: NFSD-service on Linux doesn't work reliable and I
> would never use it in a production environment. That's what I experience
> and have been told by official Linux-support! ... Multiprocessor
> machines cause the most problems. I use a watchdog to restart the
> nfsservers when they hang.
>
> We have been running a separate little Linux-cluster for our students
> with 7 machines and didn't experience those problems. Well ... seems the
> SUN's or SGI's do the killing job. But we'll get rid of them in the next
> years ;)
>
> Oliver
If you are upgrading a linux-2.0 distribution up to linux-2.2 by yourself and
aren't keeping track of the problems/fixes on your own then what exactly were
you expecting to happen? It is a well known problem that linux-2.2 nfs does
not play with with other OS's. HJ Lu has released a new version of knfs that
contains a patch against 2.2.5 to fix most (all?) of these problems.
Your best bet is to wait for the distributions to release 2.2 based versions.
If they don't work after that then you can complain to them about it.
--
Craig J Copi | [EMAIL PROTECTED]
Case Western Reserve University | http://erebus.phys.cwru.edu/~copi/
Department of Physics | (216) 368-8831
------------------------------
From: Marc Hering <[EMAIL PROTECTED]>
Subject: Re: Can I do this from a laptop?
Date: Wed, 14 Apr 1999 01:49:39 GMT
Not to mention that win95 will beat the laptop to death with only
8mb,,,(then again,,,,win95 would beat the laptop to death with ANY
amount of memory ) ;)
Belgarion wrote:
>
> Samba would be your best bet for file sharing, but a null modem just won't
> do unless you are just sharing small files and don't care about speed.
>
> X will beat the desktop machine to death with 8 megs of ram, even if it is
> just running as a server. But yes, you can do it, with X server software
> (look for Xwin32) and the 'xhost' command on your linux box.
>
> Erik
>
> Ng, Choon Hooi wrote in message
> <[EMAIL PROTECTED]>...
> >I am trying to network my laptop, but not sure if I can do this. Okay,
> >this is what I want to do.
> >
> >Desktop: 486 8Meg, running RH 5.1, with 3Com 10BT NIC installed
> >Laptop, Pentium120 8Meg, running Win95, with no NIC.
> >
> >I want to be able to share files stored in the desktop from the laptop,
> >perhaps something like the Samba. I did try connect the desktop from my
> >laptop thro a null modem cable. Works great as far as using it as a text
> >terminal. But what I really want to do is, the laptop is to run Win95,
> >but able to get/share some files located in the destop running Linux,
> >and maybe run some x-windows applications. Yes, I know I can just buy a
> >pcmcia nic and stick it in, but I am considering other alternatives at
> >the moment.
> >
> >Any idea how can I do this? Any help is appreciated.
> >
> >rgds,
> > ch
> >
> >
> >
------------------------------
** 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
******************************
