|
I used PMAT to analyze verbosegc output from SUN JVM. It really messed up. Here is the summary
"
File name : E:\dlc\dlc-bprap1-deftServer1-native_stdout.log.2009-04-08
Number of verboseGC cycles : 16
Number of all Garbage Collections : 1,853
Number of Full Garbage Collections : 0
Number of Minor Garbage Collections : 1,853
First Garbage Collection : Thu Feb 5 15:34:02 1970 (There's no startup timestamp information. The timestamp at View->Option->Start time is used.)
Last Garbage Collection : Mon Apr 26 17:01:29 1971 (There's no startup timestamp information. The timestamp at View->Option->Start time is used to calculate last garbage collection time.)
Maximum size of New generation : 4,471,128,064 bytes
Maximum size of Tenured generation : 5,905,580,032 bytes
Maximum size of Permanent generation : 6,174,015,488 bytes
Overall Garbage Collection overhead : 0%
Total time spent in non Garbage Collection : 38,449,016 seconds
Maximum Garbage Collection overhead : 105% (Sat Sep 12 19:56:30 1970)
Number of 100% overhead : 5
Total Garbage Collection pause : 630 seconds
Maximum Permanent Generation usage : 139,624,448 bytes ( 2% Sat Mar 14 10:07:55 1970)
Average Permanent Generation usage : 135,350,661 bytes
Maximum Tenured Generation usage : 799,904,768 bytes ( 13% Fri Feb 6 01:47:58 1970 )
Average Tenured Generation usage : 612,638,443 bytes
Maximum New Generation usage : 652,740,608 bytes ( 14% Sun Apr 25 11:54:33 1971 )
Average New Generation usage : 118,235,757 bytes
Java Heap Activity Analysis and Recommendations report
"
As you can see , the New generation and Tenured generation are running into 4gb -5gb etc., This is incorrect as the max heap is 1GB and browsing output shows the heap lower than this.
I used PMAT from ISA as well as I downloaded the copy from developerworks, they both have same result. Does anyone else have similar issues? I see that its supposed to be able to handle output from HotSpot JVM.
|