Simulation finished with the error = null

21 views
Skip to first unread message

Philipp Giesen

unread,
Jan 22, 2014, 9:37:55 AM1/22/14
to peerfac...@googlegroups.com
Hi,

I tried to run the Simulator with my monitoring application and a rather large amount of 10k nodes. After a long run, my simulation finishes and log file reads 

WARN  [SimulationThread - 86]# java.lang.UNIXProcess@1bd15f8c: 2014-01-16 09:25:22,691 ERROR [Simulator - 216]# Simulation finished with the error = null
WARN  [SimulationThread - 86]# java.lang.UNIXProcess@1bd15f8c: 2014-01-16 09:25:22,692 INFO  [Simulator - 221]# Realtime Duration of experiment (m:s) 1291:12
WARN  [SimulationThread - 86]# java.lang.UNIXProcess@1bd15f8c: 2014-01-16 09:25:22,692 INFO  [Simulator - 223]# Simulated time is 3:24:37:136 (H:m:s:ms)

Simulated time should be 04:30h, so the run ended prematurely. The error however states null, which is not very helpful. I already modified the Simulator.java and changed the block

try {

scheduler.start();
finishedWithoutError = true;

} catch (RuntimeException e) {
 

to

try {

scheduler.start();
finishedWithoutError = true;

} catch (Exception e) {
 

to maybe catch an Exception which not belongs to the RuntimeException class. So far no success. However, reducing the runtime or node count drastically (1000 nodes or just 60mins runtime) seems to work.

My application sends a lot of messages through the network and therefore generates a lot of operations. Is there a limitation towards the amount of generated operations? Or can the reason for simulation failure be something else? Did you ever experience something similar during long runs of the simulator?

Kind regards,
Philipp

Philipp Giesen

unread,
Jan 22, 2014, 9:41:42 AM1/22/14
to peerfac...@googlegroups.com
to maybe catch an Exception which not belongs to the RuntimeException class. So far no success.

To be clear: The error still is null. Even though the try block seems to throw an Exception. I don't understand how this can be.
 
 However, reducing the runtime or node count drastically (1000 nodes or just 60mins runtime) seems to work.

The simulation then finishes withour any error. I am still interested in the results for 10000 nodes and a longer runtime though.

Regards,
Philipp
Reply all
Reply to author
Forward
0 new messages