This Bugzilla instance is a read-only archive of historic NetBeans bug reports. To report a bug in NetBeans please follow the project's instructions for reporting issues.
Hi Guys, I am using the NB 6 Dev profiler milestone 6, on linux, to profile an app running under Tomcat. When I hit the "trash can" (Garbage collect) icon, I observe the "used Heap" to fall to a small value. What confuses me is that when I run the Unix 'top' command, I don't see any proportionate drop in memory usage. I understand this is probably not a bug, but I was hoping you could nevertheless help me to understand why. The RES (resident memory) reported by top continues to grow, and is virtually unaffected by huge garbage collections that return about 500MB of used heap. Basically, the application shows all the symptoms of a memory leak, but the VM telemetry graph doesn't show that at all. From the VM telemetry it looks like we have a perfectly normal garbage collection going on. If I leave the app long enough, tomcat will eventually throw OutOfMemoryExecption. Any tips or comments would be excellent. Attached is a screenshot showing the drop in used Heap when I hit "GC", the Unix 'top' screen showed only a 0.1 GB drop in RES memory, when the VM telemetry showed 500MB being freed. Granted, the entire heap is allocated, so maybe it should all show up in RES, but why does the RES grow steadily over time when the "available heap" is a flatline?
Created attachment 37851 [details] screenshot of used heap and Unix 'top'
This looks to more like a question about JVM and/or linux kernel. It has nothing to do with profiler. I believe you can observe the same results without profiler. In the future please use users@profiler.netbeans.org mailing list.