We are having issues where a process isn't always closing the handles and will eventually run Integrator out of memory.
I can see the total number of handles in use but I don't see a way to determine how many handles each active job is using.
Is there a way to determine which job is eating up the handles?
How do you determine how many handles an active job is using with Lansa Integrator?
Re: How do you determine how many handles an active job is using with Lansa Integrator?
I found that java jobs and traces are hard to follow.
While you can enable trace and get more information on the job, like amount of time consume by the execution in threads, the trace is heavy for a production environment, it is meant only for debugging/tracing to solve an issue.
If this problem is making java to not finalize normally and I need to know what process are starting and ending correctly, I will create a simple log file, when you start your java process, you insert a record with the basic information, like program name, date, time, jobnbr, status and when the programs end normally, the status will be updated to finalize (and maybe the date/time/ ended/duration), or the record deleted.
That way a simple inspection to that file would tell you what process are failing to complete normally and most likely will be the same ones.
you could even have a flag set in your system, if active, you start this recording, if not you just ignore it, so you can enable it system wide when needed and make it part of your standard jsm jobs.
While you can enable trace and get more information on the job, like amount of time consume by the execution in threads, the trace is heavy for a production environment, it is meant only for debugging/tracing to solve an issue.
If this problem is making java to not finalize normally and I need to know what process are starting and ending correctly, I will create a simple log file, when you start your java process, you insert a record with the basic information, like program name, date, time, jobnbr, status and when the programs end normally, the status will be updated to finalize (and maybe the date/time/ ended/duration), or the record deleted.
That way a simple inspection to that file would tell you what process are failing to complete normally and most likely will be the same ones.
you could even have a flag set in your system, if active, you start this recording, if not you just ignore it, so you can enable it system wide when needed and make it part of your standard jsm jobs.