socket timeout, process not responding.

[Marvin Tapessur_1]

Hi Guru's,

I have a very weird problem. We had JDEdwards running on 11.11 and migrated to 11.23 3 weeks ago. 2 days ago we've been seeing some weird application issues where users weren't able to log in anymore.

Looking more closely it seems that the network process of JDE jdenet_n wasn't responding to connections on port TCP 6009 anymore. I could see the process being active and in netstat I could see that the system would LISTEN on port 6009 but a telnet on port 6009 timed out.
This is not a firewall issue, I did a telnet locally on the system which also didnt work and we don't have a firewall running locally.

I used tusc to monitor the parent jdenet_n process and I can see that the connect syscall timed out when connecting the socket. See attachment for some tusc output on the parent process. Any ideas as what is happening here? To solve the problem we have to restart the application.

[Steven Schweda]

> [...] This is not a firewall issue [...]

Ok. How about routing? Does the server know
how to get a response back to the client?
"netstat -rn" on both ends could be
interesting, or even something like Telnet
from the server back to the client.

[Marvin Tapessur_1]

Steven, thanks for your reply.

It's not a routing issue either. I'm on the application server doing a "telnet localhost 6009" and it doesn't connect even though the in netstat I can see the "*:6009 LISTEN" entry.

That's why a I "tusc" the PID responsible for listening on 6009. See attachment of my previous post. The syscall connect() timed out.

You should know that it runs fine again when we restart the application but after 2, 3 days JDE stops responding on port 6009.

My feeling is a system limit that's being reached. But which one?