Hello, I've been using a nightly build (r1095243) for a while to develop a test plan that runs against Jetty servers. Today I downloaded the 2.6 release, tried the .jmx with 80 concurrent users and the number of processed requests/second was way off. With r1095243 I could get approx 125 requests/seconds, with 2.6 released build approx 30 requests/second. JMeter is run on the same computer (Debian) with the same Java vm (Java(TM) SE Runtime Environment build 1.6.0_26-b03)
A quick diff on bin/jmeter shows this difference o1@vsrv005:~/bin$ diff jmeter/bin/jmeter jakarta-jmeter-r1095243/bin/jmeter 77,78c77,78 < # Increase MaxPermSize if you use a lot of Javascript in your Test Plan : < PERM="-XX:PermSize=64m -XX:MaxPermSize=128m" --- > # PermSize is a scam. Leave it like this: > PERM="-XX:PermSize=64m -XX:MaxPermSize=64m" I will now try to change MaxPermSize to 64m, but I doubt that would cause a 4x difference in reported performance. I am also the only person currently using the test server I am running against, so the load is quite under control. Diffing jmeter.properties show differences, but they seem to be for LAF and other settings that should not affect performance. In both the nightly and the released version, the only customization I did was onload.expandtree=false Other than changing MaxPermSize to 64m, is there any other obvious thing to try? thanks Ivan -- View this message in context: http://jmeter.512774.n5.nabble.com/Performance-difference-between-released-2-6-and-nighly-r1095243-tp5522066p5522066.html Sent from the JMeter - User mailing list archive at Nabble.com. --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
