Linux-Networking Digest #138, Volume #11 Thu, 13 May 99 08:13:30 EDT
Contents:
Re: General.NFS.Questions ("Christopher R. Thompson")
Have a problem with the connect() function: PLS help if u can (Gaurav Sehgal)
diald (tarkin)
"Direct Cable Connection" ("Direnzy")
Linux as WEB-, Proxy- Router in Novellnet ("Karl Hempel")
Re: Using two netmasks on one machine? ("Curt")
Re: Toshiba Tecra on EBAY ("David Murray")
Q: Newbie. Can't get client address (Iond Research Srl)
How to fire the "IN.TFTPD" up ?? ([EMAIL PROTECTED])
fdisk /MBR ??? ("Mikael Wetterstrand")
Re: smbmount (Sim)
IP Masquerade question (Skeeter Hartwig)
both isa and pci ne2000 cards? (Damon des Jumeaux)
Re: 3c509 and IRQ conflicts ("Frater M.A.Ch.H. 999")
Re: Reboot linux 5.2 box from telnet connection ([EMAIL PROTECTED])
Re: ifconfig (Max Wheatley)
Re: Another IP Masq... (cont'd) (mist)
----------------------------------------------------------------------------
From: "Christopher R. Thompson" <[EMAIL PROTECTED]>
Crossposted-To:
comp.protocols.nfs,comp.os.linux.help,alt.linux,alt.os.linux,comp.os.linux.misc,comp.os.linux.questions
Subject: Re: General.NFS.Questions
Date: Thu, 13 May 1999 02:45:24 -0700
Christopher wrote:
> In article <[EMAIL PROTECTED]>,
> Christopher <[EMAIL PROTECTED]> wrote:
>
> > I have recently been attempting to build a Linux Beowulf type cluster
> using
> > the Red Hat distibution Linux-Pro 5.4. The Linux kernel version is
> 2.0.33 using
> > rpc.nfsd Universal NFS Server Version 2.2beta16 and rpc.mountd
> Universal NFS
> > Server Version 2.2beta37.
> >
>
> Questions:
>
> 1. Stale NFS File handles:
>
> a) Program "top" displays process status running under Linux. When
> running on a "root-mounted" client "top" displays "Stale NFS File
> handle: /var/run/utmp" and quits. This condition only occurs when
> the /var directory structure physically part of the root directory.
>
> When the host's physical /var directory is mounted to the client...
> "top" runs fine and doesn't complain. Does anyone know why?
>
> b) Does anyone know where /var/run/utmp gets created? Is there any
> process accounting software on linux that shows file I/O
> statistics so I can track how files are created and used?
rc.sysinit does rm -f /etc/mtab~ /fastboot /etc/nologin>/var/run/utmp
>
>
> 2. rpc.mountd. When attempting to initiate rpc.nfsd and rpc.mountd on
> an NFS client rpc.mountd quits with message:
>
> May 9 02:13:32 Shelly mountd[191]: Could not bind name to socket
> 0.0.0.0:635: Address already in use.
>
> Is this becaues my network address and netmask are 0.0.0.0 for
> eth0? I thought I needed this to route internet ip packets to the
> internet server. Is there another way to tell rpc.mountd to use
> ip address 192.168.50.3 for the socket address?
Answer: Yes. You need to specify a network address and a
correspondingnetmask. Your router should handle traffice destined for the
internet.
They still call me Curious George... and I still don't know why?
------------------------------
Date: Sat, 01 May 1999 07:37:16 +0000
From: Gaurav Sehgal <[EMAIL PROTECTED]>
Subject: Have a problem with the connect() function: PLS help if u can
Hi,
I wrote a basic network program in Linux Readhat 5.2 , but when i
compiled the program it gave me an error message for the following line:
connect(sd," parse",10);
the error message said : incompatible type for argument 2 of `connect'
I can't see the problem as in UNIX/Linux one can specify filenames in
the second parameter of the connect command to connect to a file rather
than a remote server.
I would be very obliged if anyone could help me with this problem.
thanks
gaurav
email: [EMAIL PROTECTED]
------------------------------
From: tarkin <[EMAIL PROTECTED]>
Subject: diald
Date: Thu, 13 May 1999 01:57:09 -0500
OK, I have diald init scripts in run levels 1, 2, 3, 5. I can ps for diald, and
see it is running (btw, should the ps show some command line options inserted
after the diald command?? I would expect so, but don't see any...). However,
shouldnt there be a route added now? a default route for ppp0 or something like
this? At anyrate, i have edited the config scripts for chat (connect) and for
diald.conf. They are as setup as could be. I am following the tutorial in
"LINUX Network Toolkit" (G. Sery, Paul) by IDG Books. It all seems very
verbose, and I cannot get any sort of response from diald when I try to even
ping outside. Runnung RH5.2, kernel 2.2.5 Also, 2 machine LAN.
newbie, please be gentle....
Brent
------------------------------
From: "Direnzy" <[EMAIL PROTECTED]>
Subject: "Direct Cable Connection"
Date: Sat, 1 May 1999 00:27:02 -0700
Reply-To: "Direnzy" <[EMAIL PROTECTED]>
This is a multi-part message in MIME format.
=======_NextPart_000_000E_01BE9369.55BAB0A0
Content-Type: text/plain;
charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
Does linux have a direct cable connection like win95 (Lpt1 to Lpt1)?
I haven't seen any docs on it so I'm guessing that type of networking
is unavailiable. L8R.
=======_NextPart_000_000E_01BE9369.55BAB0A0
Content-Type: text/html;
charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META content=3D"text/html; charset=3Diso-8859-1" =
http-equiv=3DContent-Type>
<META content=3D"MSHTML 5.00.2014.210" name=3DGENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY bgColor=3D#d8d0c8>
<DIV><FONT face=3D"Times New Roman" size=3D2>Does linux have a direct =
cable=20
connection like win95 (Lpt1 to Lpt1)?</FONT></DIV>
<DIV><FONT face=3D"Times New Roman" size=3D2>I haven't seen any docs on =
it so I'm=20
guessing that type of networking</FONT></DIV>
<DIV><FONT face=3D"Times New Roman" size=3D2>is unavailiable.=20
L8R.</FONT></DIV></BODY></HTML>
=======_NextPart_000_000E_01BE9369.55BAB0A0==
------------------------------
From: "Karl Hempel" <[EMAIL PROTECTED]>
Subject: Linux as WEB-, Proxy- Router in Novellnet
Date: Thu, 13 May 1999 09:08:42 +0200
Reply-To: "Karl Hempel" <[EMAIL PROTECTED]>
Hi,
i had put a Linux-Server (SuSe Linux 6.0) into ower Novell 4.11 Network. It
works as Router, Web-, Proxy-, FTP- and Mailserver. All works very good but
from Novellserver or Workstation (Windows 95) i can not use ftp or ping from
Lan to Internet. The answer is "unknown address". From Linuxserver or with
Browser it works.
Perhaps the Novellserver catch the commands ? On Novellserver run NIS.
What can i do ? Help, please !
Thanks
Karl
PS.: Sorry it is bad english !
------------------------------
Reply-To: "Curt" <[EMAIL PROTECTED]>
From: "Curt" <[EMAIL PROTECTED]>
Subject: Re: Using two netmasks on one machine?
Date: Thu, 13 May 1999 05:48:48 -0500
It should work.
Do you have forwarding on? Take a look at /etc/sysconfig/network
FORWARD_IPV4=yes
<[EMAIL PROTECTED]> wrote in message
news:7he4f4$sak$[EMAIL PROTECTED]...
> I currently have a Red Hat Linux 5.2 (Kernel 2.2.7) installed, but I
> have a problem that I cannot get it to respond to two IP addresses on
> two networks. Well...I don't know how to do it.
>
> There are two networks, and I want my machine to be visible on both of
> them. I have tried IP Aliasing, but that only gives you an option for
> extra IP addresses and not Netmasks. Is there a way to respond to two
> Netmasks?
>
> Thanks,
> Terence.
>
>
> --== Sent via Deja.com http://www.deja.com/ ==--
> ---Share what you know. Learn what you don't.---
------------------------------
From: "David Murray" <[EMAIL PROTECTED]>
Subject: Re: Toshiba Tecra on EBAY
Crossposted-To: comp.os.linux.help,comp.os.linux.portable,comp.os.liux.questions
Date: Wed, 12 May 1999 17:33:23 GMT
Hmmm.. Does the Toshiba come with Linux pre-installed? Otherwise, how the
heck does this relate?
[EMAIL PROTECTED] wrote in article <[EMAIL PROTECTED]>...
> http://cgi.ebay.com/aw-cgi/eBayISAPI.dll?ViewItem&item=100676035
> Currently (5/12/99) bid at only $575.00!!
>
------------------------------
From: Iond Research Srl <[EMAIL PROTECTED]>
Crossposted-To: comp.unix.programmer,comp.protocols.tcp-ip
Subject: Q: Newbie. Can't get client address
Date: Thu, 13 May 1999 11:14:06 +0200
Hi guys,
I want to develop such an incredible package.
When a client logs on a server and run the package, it writes the client
address on stdout.
I don't know how to get the address, I've been suggested to use
getpeername but I haven't been able to work it out.
Sorry for a probably silly question. I hope it is not an OT, at least
Thanks a lot for your time and help
Luca
------------------------------
From: [EMAIL PROTECTED]
Subject: How to fire the "IN.TFTPD" up ??
Date: Thu, 13 May 1999 10:34:39 GMT
Reply-To: [EMAIL PROTECTED]
Dear All Guru :
How to fire the "/usr/sbin/in.tftpd" up ??
I installed the Slackware with the tcpip1, tcpip2 package.
But how to be sure the "tftpd" runing ??
Any reply will be very thankful .
THANKS
------------------------------
From: "Mikael Wetterstrand" <[EMAIL PROTECTED]>
Crossposted-To:
comp.protocols.nfs,comp.os.linux.help,alt.linux,alt.os.linux,comp.os.linux.misc,comp.os.linux.questions
Subject: fdisk /MBR ???
Date: Thu, 13 May 1999 13:09:27 +0200
Hello
Well, I just wanna know what the command /mbr does?
I know u can use it with fdisk to "get ridd of" Lilo , but what does it
really do?
------------------------------
From: [EMAIL PROTECTED] (Sim)
Subject: Re: smbmount
Date: Thu, 13 May 1999 07:27:28 GMT
On Thu, 13 May 1999 11:30:49 -0400, hullsy <[EMAIL PROTECTED]> wrote:
>Has anyone got smbmount to work my command line is
>
>smbmount //monsta.spawn.net/winhd /mnt/winhd -n
>
> monsta.spawn.net being the win98machine and the share is called
>winhd I have not set a password
> the mount point is obviously /mnt/winhd and the -n is for no
>passwords
I found that I had to add and -I or a -U tag to get it working: see
home.tvd.be/ws35056/linux/samba.html
Simon
------------------------------
From: [EMAIL PROTECTED] (Skeeter Hartwig)
Subject: IP Masquerade question
Reply-To: [EMAIL PROTECTED]
Date: Thu, 13 May 1999 11:08:27 GMT
I've got an app that sends UDP packets to some remote server. The
server receiving my app's UDP packets needs to know the port number of
the original source. This is currently a problem because when my
firewall/masquerade box does the masquerading/forwarding, it uses a
different source port. This new source port is the one detected by
the remote server. For example, my '/sbin/ipfwadm -Ml' output of a
recent trial looks like this:
IP masquerading entries
prot expire source destination ports
udp 01:37.00 mayhem thor.won.net 27015 (61276) -> 27010
The original source port was 27015, but when it got masqueraded and
forwarded, the new source port is 61276, and that's what the remote
server sees. The problem is that, in short, I need the remote server
to see 27015, or the whole exercise is pointless. Is there a solution
to this, or am I going to have to try to get the programmers to
incorporate port info in the data being exchanged, rather than relying
on the packet headers?
For anyone interested, what I'm doing here is running a Half-Life game
server on a machine that is being masqueraded. The server actually
works perfectly using ipportfw on the masquerade machine to forward
packets to the game machine. The problem I'm having is when the game
server is first fired up and it contacts the master server list
machine to have itself be placed into the list (for Gamespy and
friends).
------------------------------
From: Damon des Jumeaux <[EMAIL PROTECTED]>
Subject: both isa and pci ne2000 cards?
Date: 13 May 1999 07:34:42 GMT
hi there. i'm trying to set up a third, pci, ethernet card in this box,
which will eventually replace one of the isa cards and thus free up a
much-needed isa slot. however, i'm having problems getting the ne and
ne2k-pci modules to coexist.
to wit, every time i try to modprobe ne2k-pci, i get:
/lib/modules/2.0.36/net/ne2k-pci.o: init_module: Device or resource busy
8390: Device or resource busy
...so both ne and ne2k-pci are trying to use the `8390' module, and it
doesn't seem to like that.
is there some easy way of letting them share the 8390 module? or
failing that, to create another instance of it for ne2k-pci? (though
i'd prefer not to waste the memory on that if possible...)
please mail be directly at [EMAIL PROTECTED] if possible.
any help is much appreciated.
-damon
--
Damon Harper des Jumeaux _/\_ "Then you'll excuse me...
[EMAIL PROTECTED] __\ /__ I'm in the middle of fifteen
Les Jums: [EMAIL PROTECTED] \ / things, all of them annoying."
http://www.jumeaux.bc.ca/damon/ |/||\| - Commander Susan Ivanova, B5
------------------------------
From: "Frater M.A.Ch.H. 999" <[EMAIL PROTECTED]>
Subject: Re: 3c509 and IRQ conflicts
Date: Thu, 13 May 1999 06:05:10 -0500
What are your results to the following commands?
cat /proc/ioports
cat /proc/interrupts
Also, you are setting the I/O port and interrupt in DOS/Windoze using the
utilities, and NOT trying to do it while under Linux, correct? If so, then
what are your I/O ports and IRQ's in DOS/Windoze?
When I was having annoying trouble getting my card to take off (It was
detected immediately, but would delay initialization), I had nailed it down
to one, all-too-simple issue... when I setup the insmod argument, all I
needed to tell it was:
insmod 3c509
Instead, I had been telling it that, as well as the I/O port and IRQ... but
didn't need to. That's all programmed in the EEPROM when you run the setup
utility in DOS/Windoze. So also make sure that it's not loading due to you
telling it too much. All you should need is: insmod 3c509 --- The Linux
kernel can find out all the rest that it needs directly from the hardware.
Make sure you also power down between changing an I/O port address or IRQ
setting before trying to use it with that new configuration. If you do not,
it will not be using the new information yet.
I don't know if this helps at all, but hopefully it will. Let me know what
the /proc reports say.
AZOTh 999
[EMAIL PROTECTED]
Jebediah <[EMAIL PROTECTED]> wrote in message
news:7hcvi7$ke0$[EMAIL PROTECTED]...
> I just installed the Red Hat 6.0 a few nites ago. I was hoping it would
> fix the problem with my 3Com 3c509 ethernet card, but this was not the
> case. I have narrowed the prob down to an IRQ conflict, mainly due to my
> SoundBlaster AWE 64 card residing on IRQ 5. The ethernet card seems to
> like residing on this IRQ as well. I used the 3c5x9.exe that came on the
> floppies with my card. I changed the IRQ to my empty location of 9 and
> tried it in Linux again. Still no luck. I would try another IRQ but all
> of them are taken and the SoundBlaster is pretty well set on 5 due to
other
> devices. I would also try IRQ 12 where my mouse resides, but I can not
put
> it into USB mode for Linux yet. Finally in desperation I put the ethernet
> on IRQ 10 and IO 0x300... This at least made my card fail a lot faster, so
> it is recognized. My question is how do I make the card work or am I
> screwed up a tree?
>
> ------------------ Posted via SearchLinux ------------------
> http://www.searchlinux.com
------------------------------
From: [EMAIL PROTECTED]
Subject: Re: Reboot linux 5.2 box from telnet connection
Date: Thu, 13 May 1999 06:33:54 GMT
In article <7hdd4c$233m3$[EMAIL PROTECTED]>,
"Ian" <[EMAIL PROTECTED]> wrote:
> Any command MUST be in your path.
Not true. You always have the option of typing the full path to the
executable.
For example:
/sbin/reboot
will work, regardless of your path.
> Most people have the current directory in
> their path (.)
I don't have hard figures on this, but somehow I doubt the validity of
your claim. (.) is left out of the path by default, and I would say the
number of people savvy enough to know how to change their default path
yet not savvy enough to understand why this is a bad idea is very small.
> but as root this can be a bad thing. Imagine cd'ing to a
> directory that has an executable file in it called cd that does
> nothing. How do you leave the directory?
Well, to answer your question, you'd type "cd .."
[You picked a bad example -- "cd" is a built-in shell command, not an
executable.. the shell never even looks for a file named "cd"]
A better example is a file called "ls-l" [or some other common typo].
Since it is ulikely that another executable exists in your path with
that name, it will be executed if (.) is in your path. Simply having a
file called "ls" in your cwd isn't necessarily a danger, unless you
happen to have (.) *first* in your path. Anyone who would put (.) first
in their path, however, should be shot and/or castrated [depending on
gender].
> > BTW I cannot log onto
> >the linux box as root but must su to root.
> This is a security feature. Stops people from getting straight to root
> from outside.
No, it discourages root passwords being transmitted unencrypted over the
wire. Technically, you're not supposed to `su` except over a secure
connection, either.
With the advent of ssh, the old method of telnetting in as a regular
user and then su'ing should be discouraged, because an unencrypted
password is still sent over the wire. The second most secure way to
become root on a machine is to ssh straight in as root. The most secure
way is to ssh in as root, but using RSARhostsAuthentication rather than
a password [to eliminate the danger of somebody changing the ssh binary
on the client machine].
--
-Bill Clark
Systems Architect
ISP Channel
http://locale.ispchannel.com/
--== Sent via Deja.com http://www.deja.com/ ==--
---Share what you know. Learn what you don't.---
------------------------------
From: Max Wheatley <[EMAIL PROTECTED]>
Subject: Re: ifconfig
Date: Wed, 12 May 1999 19:45:46 +1200
ANTHONY wrote:
>
> Any body can tell me what's the reason whenever I issue the ifconfig
> command (ifconfig eth0 ipaddress) it hangs ?
All to do with resolving names from memory .....
Isn't there on option on ifconfig to stop it resolving names ????
check out the man pages, and things like /etc/networks /etc/resolve.conf
Sorry can't be more use
--
[EMAIL PROTECTED]
[EMAIL PROTECTED]
------------------------------
From: mist <[EMAIL PROTECTED]>
Subject: Re: Another IP Masq... (cont'd)
Date: Thu, 13 May 1999 12:05:53 +0100
Reply-To: mist <new$[EMAIL PROTECTED]>
Rock <[EMAIL PROTECTED]> scribed to us that -
>First, thanks for the replies to the first version of this message. As
>usual, I post, get a couple replies and some ideas, then it takes me a week
>to get over to try things out, and I get a step further.
>
>It was suggested more than once to check routing setup and netstat. First
>off, here's the output of ifconfig when the linux box is connected to the
>internet (w/o the lo):
>
<snip>
>
>I noticed something here about the MSS settings -- it is different for the
>local lan (1500) and the ppp0 connection (1024). I've read that mru=mru for
>things to be happy, and for ppp0, they were both 552. I noticed the lan uses
>1500, so tried to set mtu=mru=1500 in /etc/ppp/options, but what came out is
>the above -- set at 1024.
>
This is correct behaviour. The MTU for the PPP link is less than the
MTU for an ethernet frame.
>I am still in the same situation. The lan (win9x, tcp/ip) pings each other
>fine, with their gateway set as the linux box at 192.168.1.1 but when the
>box connects to internet, the other machines still can't get out.
>
So I take it that you have enabled IP masquerading on the Linux box, set
up rules, etc.? And done the echo 1 > /proc/sys/net/ipv4/ip_forward
(Or some path similar to that?)
>At this point, I'm suspecting the problem is the mismatched mss settings.
>How can I force ppp0 to get to 1500 instead of 1024? Or should I somehow try
>to change the lan settings to 552?
>
No. Don't mess with those at all. The defaults should be fine for
general usage.
--
Mist.
------------------------------
** 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
******************************
