Hi all,
Occassionally, when we have a client that appears to be not responding, we issue a "kill -10 PID" to get a stacktrace. It helps us determine what the client is actually doing. Can this method be used on an appserver agent? We appear to hit some deadlock situations, and would like to find the cause.
I know I could use 4GL logging for this, but as it only happens in production, I'd like to have a less invasive option.
- Martin
> I'd like to have a less invasive option.
Article: proGetStack / kill -SIGUSR1 causes memory leak, eventually crash in AppServer
knowledgebase.progress.com/.../000035325
"Workaround: Avoid sending -SIGUSR1 signals when possible."
proGetStack is the command you're looking for knowledgebase.progress.com/.../P134912 and the KB confirms it works on AppServer agents.
Besides proGetStack, you can also use "kill -10 PID" to get stack trace for appserver agents.
> I'd like to have a less invasive option.
Article: proGetStack / kill -SIGUSR1 causes memory leak, eventually crash in AppServer
knowledgebase.progress.com/.../000035325
"Workaround: Avoid sending -SIGUSR1 signals when possible."
Thanks guys, great help!
Good to know about the memory leak. However, in this case we resolve the deadlock by disconnecting or killing the agent anyway. So that memory leak may not be a problem.