[SNMP4J] high CPU usage with SNMP4J 1.11 and DefaultTcpTransportMapping
Brice Fines
bfines at sermepa.es
Mon Mar 1 17:01:29 CET 2010
Hi Frank,
I noticed that SNMP4J 1.11 release fixes some leak when using
DefaultTcpTransportMapping([SFJ-22]), so I updated my jar with 1.11.
We use DefaultTcpTransportMapping to send command to our clients.
But while I was trying our app with the new jar, I noticed a high CPU
usage (near 100%) while sending command.
After a few tests, I am sure the high CPU usage matches exactly the TCP
session: it starts when the TCP connection is open to send the SNMP
command, and lasts until the end of the TCP connection (the TCP connection
closes itself by timeout). (even once the SNMP answer to the command is
received, there is still a high CPU usage until the end of the TCP
connection)
I checked that SNMP4J 1.11 was responsible for that behaviour by switching
back to SNMP4J 1.10.2 jar (no changes to the code, same application,
different jar). With version 1.10.2, I have no high CPU usage (nothing,
not even a peak).
I noticed that problem in a Windows environment (sorry, I cannot check it
in an AIX environment).
Please, let me know if you can reproduce it, or if I can provide some
debug trace/code that could help you.
Regards
Brice
More information about the SNMP4J
mailing list