Schedulix- Timeout Exception during calculation of next execution.

[Lingareddy Varala]

  Hi everyone,

we have a schedulix batch that runs for every 20 mins and we have encountered the following timeout error - "Timeout Exception during calculation of next execution". when it times out , it was in a failed state and not initiating any new runs. NO failure emails sent since timeout exception occurred at the schedulix scheduler level.

any help is appreciated on how to resolve this timeout error. 

Thank you.

[Ronald Jeninga]

Hi,

it's at least 15 years ago that I've seen that error message.

I'm very interested in finding the cause, but it'll require some deeper investigations.

Can you send me the server log file with the error message as a start?

You can send it to my personal e-mail address; no need to publish it here.

Would you be available for a meeting? I think that would speed up the analysis a lot.

Please send me a date and time (and timezone) along with the server log.

Thank you,

Ronald

PS: Happy New Year!

[Lingareddy Varala]

Thanks Ronald for the quick reply.

Can you send me your personal email address to send the server log.  Yes, I will be available for the meeting. I will send the details once I have your email address. Thank you.

[Ronald Jeninga]

Hi all,

after investigating the issue it turned out that the scheduling server was low on memory.

What happens is that in such a case Java desperately and aggressively tries to clean up its memory.

This includes a kind of "stop the wold" because the Java VM needs exclusive access to the memory.

If the garbage collection is (halfway) successful, the server will continue to run, but it won't take long before it runs out of memory again.

The effect is that the performance of the system drops significantly.

Transactions that usually take several milliseconds all of a sudden take seconds to complete.

The time scheduling thread aborts a calculation if it needs too much time to determine the next starting time of a master.

It will then disable the schedule that caused the problem; it doesn't make sense to spend, say, 15 seconds every 60 seconds for  a calculation that doesn't complete.

Naturally, in this case the root cause for the lengthy calculation wasn't the complexity of the definition but the memory low situation, but the time scheduling thread was unaware of that.

The parameter that configures the available memory can be found in the java.conf file.

It is called BICSUITEMEM and defines the number of MB available to the Java VM.

Best regards,

Ronald