[SNMP4J] Spurious 1000 millisecond delays when sending traps

[Parhar, Ajeet S]

Hi,

I've been doing some testing with snmp4j recently and getting some
anomalous results which I'm hoping someone can enlighten me on.

My java application basically reads in a message from a Java JMS
messaging bus and uses snmp4j to send out a trap. The application works
more-or-less as expected. However, when I put in a few
System.out.println()s of System.currentTimeMillis() timing information I
notice that it usually takes a few to few tens of millisends for
messages to transit the application. The anomaly is that every now and
again it appears to take an extra 1000 milliseconds, which appears to be
gettings caught in the sending process somehow.

I've experienced this with Java 1.5 on XP and Java 1.6 on Linux.

I believe I'm setting target.setTimeout() to non-1000 values.
I have noticed that the rate at which the delays appear is inversely
proportional to that loading of the hardware that's running the JVM (ie.
The more heavily loaded is the machine, the less frequently the anomaly
appears).
After much digging I discovered this link which seems to me relevant,
but I'm not sure:
http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=4667046
Has anyone else experienced this problem? If so does anyone know why it
may be happening? If not, any ideas on how I can focus in on its source?
	Regards,
	- Ajeet.