After being connected either through rlogin or telnet (secure packages aren't installed because it's behind a firewall and telnet packages are latest packages available), the computer will simply disconnect and give the popular "Connection reset by peer" reasoning. Then, when you try to reconnect from any computer.. not just the one you were disconnected from, you get the "Connection refused." message. Also, when it does this, I try to get on the computer and try to telnet out to another computer and it won't do that either. Any ideas on what could be causing this?

Try this:

1.go to dir /etc/xinetd.d/
2.open file 'telnet' for telnet and 'rlogin'
for rlogin
3. replace 'disable = yes' by 'disable = no'
4. restart the machine!!

I did the same thing and it worked for me!

regards
Sushil

NEVER reboot the mahcine unless you need to!! that's a windows user solution! use

/etc/init.d/xinetd restart

instead.

But then i'd guess that thet's not the problem, as you say you can telnet in the first place, but reconnections aren't working.... no?

Two things wrong that solution:

A. I am allowed to connect I just get disconnected so that setting is already set correctly.

B. This is RedHat 6.2 this time so it does not have xinetd. Sorry, I should have clarified that. My fault.

More help?

Okay you say you can connect. I just have dumb questions for right now.

Can you ping anything outside your network?
Is this machine internal to your network the one your trying to connect to?
Is your machine that your using a slave/client on the network(type)your using?
Is your machine setup as a server (Whatever kind of server)?
Is your machine the network(type)master?
Was this machine recently rebuilt/Just loaded new Linux version?
When you say you can connect how long is it before you get disconnected? And what are you connecting to? Do you have a proxy server also? Finally after you get disconnected or prior to getting disconnected do you see any error statements, besides the one you mentioned. Also are there any processes running on your machine that would cause telnet to disconnect.?

If you answer these I might be able to help you better. The more idea I have of what's wrong the more I can know about your problem

Allow me to give you a complete description. This is an internal network. First of all, we have the firewall machine. It's an intel machine running RedHat version 7.1. Next, we have an alpha machine running RedHat version 7.1 as well with kernel version 2.2.19. Then we have an alpha "cluster." The main node is the only node that you can reach but from the main node you can reach the other nodes. The main node is also an alpha machine but it is running RedHat version 6.2 with kernel version 2.2.18smp (obviously a dual processor alpha). Next, we have another intel machine which is our soon-to-be RAID server. It, however, does not have the same problem as the two alpha computers have. I can be connected to it for hours with absolutely no difficulty. The firewall is set up as the gateway on all of the internal computers. So you don't simply go straight from one alpha to the other... you go from one alpha, to the firewall, and back to the other alpha. So here are the tests I've tried:

1. Connected from firewall to each individual computer. Disconnected.

2. Connected from firewall to one alpha and then to another alpha. Connection from one alpha to the other died and then, sometime later, connection from firewall to alpha died.

I've tried all of these using ssh, rlogin, and telnet and it happens completely at random. About 2 weeks ago it started doing this and then it stopped for no reason... but now it has just started doing it again and we would like to fix
this because this needs to be stable very badly. One solution that we are currently looking into is a kernel upgrade. We need a kernel upgrade anyways so what could it hurt but if there's a simpler solution for now or what if that doesn't even fix it? Does anyone have a clue? Thanks for your help.

Okay I don't know if the disconnection happens when use are telneting or whatever from the nodes to the alpha but to me since you said it was doing this for 2 weeks then suddenly stopped to me it sounds like a hardware problem with the network. Maybe somebody's ehternet card is screwing up or one of the TCP/IP cables is bad....Like a connection is going on and off.....I mean you say you get disconnected at random maybe one of these cables is barely plugged in. It definately sounds like hardware to me. If the disconnection is random it sounds more like hardware. If it was software the disconnection probably wouldn't be random and more than likely somebody would see a software problem because your RedHat systems would be complaining and you'd be seeing network errors in your console or some error complaining about something. So I'm going with hardware (Now remember I didn't say it couldn't be software) It might be but I highly doubt it. I'd start at checking the firewall machine...I'd also run have your nodes all telnet to some IP and then run diagnostics on the firewall machine. First I'd check the firewall machines TCP/IP cable. Second do the connection process with the nodes and run diagnostics on the firewall machine. Third try replacing the ethernet card in the firewall machine. It definately sounds like the firewall is where the problem is as in each case you were connected to the firewall. Before a Kernel upgrade(I'd do that anyway just cuz I'm picky and would want them all to have the same Kernel)try hardware troubleshooting. If you have to go through the firewall to connect to the other machines (different ones) each single machine doesn't have a problem with it (I hope not) So that leaves the firewall. Also you have a router sitting somewhere....Make sure to check it....I'm guessing all your nodes and your alpha are connected into a central router and then sent off to the firewall. So It might not even be the firewall....It could be the damn router. It could be the connection from the router to the firewall. Well time to go digging (I just hope you don't have subfloor)(I hate digging under subfloor)

Well hope I helped -
Dave
Airadigm Communications
Network Systems

We actually have a 100 Mb switch but you have the basic idea. We thought that could be the problem so we borrowed another switch and gave it a shot and the same problem came up. We've checked all the wiring. I don't think it's the firewall or the switch thought because those two computers do it but there's the other Intel RAID server that does not do it and it's connected to the firewall through the exact same switch. We've had ethernet card problems out of one of the alpha machines and we replaced that card a month or two ago. I would think that it was the network card if the other computer wasn't doing the exact same thing. It's very weird. Any more ideas?

After I get disconnected from one machine, I've often done a couple of other things.. I've pinged the machine from the firewall and ping the firewall from the machine. In both cases, I got a ping response but I could not log into the alpha machine from the firewall but I could SSH from the alpha machine to the firewall or to the other alpha. Strange?!?!?

I'm sure you've done this before but find out what port one of the nodes that gets disconnected is on in the switch and which port the Intel Raid Serve is on and switch em around and then try to telnet...Just to double check your switch. Just because one server can stay connected doesn't mean that all the ports are good. You have one port on that switch you know that works so give it a whirl. Okay another idea. Is the intel RAID server also running redhat. Finally, if that fails slap in a kernel upgrade and see what happens then. It could be a communications conflict.....Maybe something as dumb as one of the kernels is familiarized with IPv4 and IPv6 but the oldest kernel only knows IPv4....That might cause some kind of problem between the kernels. It's a definite possibility. I never tried messing around with network configuration conflicts involving IPv4 machines and IPv4/IPv6 machines. You never know though. Also just out of curiosity do you guys have like a performance meter running so you can check out the network packets/collisions and machine interrupt levels.

No network performance meter. I know it's not the switch because we tried another switch from another group that we KNOW works and it had the same problem. The RAID server is also running RedHat. Guess we're gonna have to shoot for the kernel upgrade.. thanks anyways.

and I checked... all the machines are IPv6.