Hanging process

[Nick Anzevino]

I am working on an HPUX10.20. We have been experiencing a spinning process once every couple of weeks which consumes our CPU. This process gets runs a 1000 times a day. Then all of the sudden it will spin. I perform a kill -3 on the process to generate a core. When I look at the core file, it seems to try an access shared memory. The code does a semaphore lock and then accesses shared memory and the does a semaphore unlock. The debugger points to the line where it access shared memory. The xdb points to line 367 of the attatched code. Which by the way has the semaphore locked so no other process can access shared memory, which in turns creates more CPU problems. Can anyone give me any clues or where else to look.

[Chris Wilshaw]

Does the problem show up as soon as that instance of the process is run, or does it have to run for some time to start spinning?

If it's after running for a while, it could be down to a memory leak.

How much memory is installed in the server, and how much is free when the problem appears?

To check how much memory the process is using, use the following, replacing with the relevant process ID

UNIX95= ps -p -o vsz,user,args

The vsz column shows how much memory the process is using (in KB). The user and args info just confirms the ID that the process is running under, and the command (with arguments if any were supplied to it).

[Nick Anzevino]

It happens when it first starts up and whenit is looking to exit. Both hanging on shared memory calls of different shared memory segments Here is the output from the top command which shows the size of the process.

CPU TTY PID USERNAME PRI NI SIZE RES STATE TIME %WCPU %CPU COMMAND
0 ? 22446 j0f 241 20 2524K 528K run 1519:49 99.89 99.72 net_slave