I have a client with several PC's connecting to their new RH 7.3 Server. Just this evening they started to lose their connection. In checking the server the telnet process is still running for each of the dropped connections. The telnet client has been used for several years to connect to SCO box that is still on the same network. I'm baffled at what could be causing these connections to drop. Any idea??

PS this is a secure network so we opted not to use ssh.

I have more information regarding this issue.

This network has a SCO box, a RH 7.3 an NT4, several routers and a bunch of PC's.

If I grab a hub and plug in just some of the PC's and the RH box everything works. I have not had time yet to selectively readd all of the devices to the network one by one to see if one causes the problem.

Further if I telnet to the SCO box and from the SCO box telnet to the RH box everything seems to work correctly.

I have tried to both tighten down and completely relax ipchains to see if it was somehow spurattically rejecting traffic causing telnet to die.

I would appreciate ***any*** help with this problem.

What can signal a RH 7.3 host to terminate a telnet session? I'm thinking that some other IP traffic traffic on the network is causing the host to close this connection.

what are you using before 7.3 ?
from the 7.0 or 7.1 the telnet server is running via the xinetd, this means that you must start the xinetd to allow the telnet for client connection

bela

you;d be supprised at what you find in the man for telnetd
check it out it might help

Well we didn't upgrade from a previous version of RH but rather we installed a new server running RH. The software previously ran on a SCO 5 server that was accessed from a telnet client.

I have poured through all of the man pages on the xinetd and in.telnetd.

In testing with this very same server in my office connected to my LAN worked correctly. At the client site some traffic or condition on their network seems to be related to this problem.

My desired plan is:

Client PC -->(telnet)------> RH 7.3 server
(This is unstable and drops the telnet sessions in a time spanning 10 seconds to 30 minutes)

My current work around is:

Client PC -->(telnet)---->SCO 5 server --->(telnet)---> RH server
(This is stable but of course I now need to turn off the SCO server)

To further test this I have taken the network switch and disconnected all network devices except for the RH server and client PC and it stayed up for ever. This lead me to think that some other traffic on the LAN causes the RH server to drop the connection when talking to the PC. Could it be traffic with another LAN protocol or something probing a port perhaps. Further it doesn't matter which of the Win9x clients on the network are being used or which telnet package on Win9x client.

I've been having a similiar issue. It doesn't seem to be dependent on the kernel version since it happens with 2.4.2, 2.4.14, 2.4.18, 2.4.19, and 2.4.20. It does not occur when sshing from the internal network to the dmz or from the dmz to the internal network. It only happens when a connection from outside the firewall connects to anything on the dmz, internal network or even the firewall.
At first I thought it was the ruleset for ipchains, which is rather extensive, but now I am of the mind that there is a lot of network congestion going through one access point (the firewall). I've been delving into traffic shaping and queueing, and learning Cisco IOS for a Cisco 2900 switch to replace the PC firewall. But, for now, it seems to be network congestion.. time to pull out etherboy again. Heh.

Now I'm frightened...

I'm mostly an applications kinda guy with enought knowledge of the whole DMZ zone to know that it's an area between North and South Korea... LOL

This network I'm dealing with has an NT 4 server on it and I'm betting that it is part of the problem.... LOL

So I take it that no one else has encountered this yet? I have had an open Ticket with Red Hat for 7 days thus far and I now know that it has been assigned to a Tech but that alone took 5 days. I'm so glad I found a temporary work around or I and my Red Hat recommendation would have been thrown out to the curb.

Is Red Hat support normally this awful or is just me?