I am having a problem here. We are having several problems in regards to hung process's on unix (HPUX box), caused by my RF equipment (Mobile data capture units). these contact the host via a simply telnet session and locks the system?
Is it a timeout problem as the timeout is disabled on the host.
Sorry no knowlegde of RF.
Try tusc/truss on the system the connection is set up to.
tusc -p <pid>
Will show you what the process is doing. It is not always already installed on Unix, you can install the package if you like. It's a very handy tool !!
Regs David
Wow, how about a little information here?
What version of HP-UX?
TCP/IP packets will arrive via a network interface card. Does that card actually have an antenna and nothing more? Describe your actual network topology. The fact that the packets travel partially over a rf link is probably rather trivial.
What is the actual symptom? You give two theories: "hung process" and "locks the system". What is actually happening? Do you get an unkillable process that requires a reboot to get rid of it? Does the entire system become unresponsive? Or does your mobile box simply not finish the conversation?
And... "the timeout is disabled on the host".... what is that susposed to mean? HP-UX has many timeouts. What exactly did you tune?
If we are to provide you with any help, you're going to need to describe things with much more precision.
our stations stay active even if you logout untill you reboot the rf device.
give that a shot and see what happens.
In our case, our RF usually result in hung process (zombie) in the unix box due to:
-
the user RESET the RF equipment when the application is still active. -- usually the application (without proper termination) may lock the application database but not the os.
-
the RF equipment lost connection to the UNIX box due to atenna or coverage problem, or sometime the RF controller is down. -- this seemingly caused something like the timeout problem.
Regards
that is really strange that it hangs the DB. the zombie i can understand (we get those too sometimes) but never hangs the DB, and we use an ancient form of informix ids.
i bet there is some switches in the db to kill connections that are idle for x amount of time. this might help from stuff hanging in case of outages.
wish i could be of more help but i am not a db guy.