1

my app is running util at some point the JVM recursively does minor garbage collections over and over again and never stops. my java command line :

java -Xmx4G -Xms4G -XX:+PrintGCDetails -XX:+PrintGCTimeStamps -Xloggc:/home/root/gc.log -jar myApp.jar

The gc output:

 4.238: [Full GC [PSYoungGen: 21968K->9997K(420864K)] [ParOldGen: 74537K->83814K(150144K)] 96505K->93811K(571008K) [PSPermGen: 10761K->10757K(21824K)], 0.3216090 secs] [Times: user=0.99 sys=0.01, real=0.32 secs]
    8.660: [GC [PSYoungGen: 400333K->23552K(517568K)] 484147K->117422K(667712K), 0.0554510 secs] [Times: user=0.22 sys=0.00, real=0.06 secs]
    8.979: [GC [PSYoungGen: 517504K->23488K(632384K)] 611374K->117358K(782528K), 0.0263860 secs] [Times: user=0.09 sys=0.01, real=0.03 secs]
    9.333: [GC [PSYoungGen: 622208K->32K(637888K)] 716078K->117418K(788032K), 0.0249140 secs] [Times: user=0.09 sys=0.01, real=0.02 secs]
    9.656: [GC [PSYoungGen: 603040K->32K(603072K)] 720426K->117418K(753216K), 0.0059430 secs] [Times: user=0.02 sys=0.00, real=0.00 secs]
    9.960: [GC [PSYoungGen: 603040K->32K(638464K)] 720426K->117418K(788608K), 0.0051460 secs] [Times: user=0.01 sys=0.00, real=0.01 secs]
    10.263: [GC [PSYoungGen: 602464K->32K(602496K)] 719850K->117418K(752640K), 0.0037990 secs] [Times: user=0.01 sys=0.00, real=0.01 secs]
    10.575: [GC [PSYoungGen: 602464K->32K(638656K)] 719850K->117418K(788800K), 0.0037570 secs] [Times: user=0.02 sys=0.00, real=0.00 secs]
    10.876: [GC [PSYoungGen: 603168K->32K(637952K)] 720554K->117418K(788096K), 0.0039310 secs] [Times: user=0.02 sys=0.00, real=0.00 secs]
    11.180: [GC [PSYoungGen: 603168K->32K(638528K)] 720554K->117418K(788672K), 0.0039980 secs] [Times: user=0.01 sys=0.00, real=0.01 secs]
    11.480: [GC [PSYoungGen: 605216K->64K(637952K)] 722602K->117450K(788096K), 0.0036720 secs] [Times: user=0.01 sys=0.00, real=0.01 secs]
    11.786: [GC [PSYoungGen: 605248K->64K(640384K)] 722634K->117450K(790528K), 0.0049470 secs] [Times: user=0.02 sys=0.00, real=0.00 secs]
    12.091: [GC [PSYoungGen: 609664K->32K(639872K)] 727050K->117418K(790016K), 0.0051130 secs] [Times: user=0.02 sys=0.00, real=0.01 secs]
    12.395: [GC [PSYoungGen: 609632K->32K(642752K)] 727018K->117418K(792896K), 0.0042050 secs] [Times: user=0.01 sys=0.00, real=0.00 secs]
    12.704: [GC [PSYoungGen: 614816K->32K(642304K)] 732202K->117418K(792448K), 0.0036980 secs] [Times: user=0.01 sys=0.01, real=0.00 secs]
    13.012: [GC [PSYoungGen: 614816K->32K(645504K)] 732202K->117418K(795648K), 0.0040370 secs] [Times: user=0.01 sys=0.00, real=0.00 secs]
        Heap
         PSYoungGen      total 1397696K, used 1209524K [0x00000007aaab0000, 0x0000000800000000, 0x0000000800000000)
          eden space 1397312K, 86% used [0x00000007aaab0000,0x00000007f47cd078,0x00000007fff40000)
          from space 384K, 16% used [0x00000007fffa0000,0x00000007fffb0000,0x0000000800000000)
          to   space 384K, 0% used [0x00000007fff40000,0x00000007fff40000,0x00000007fffa0000)
         ParOldGen       total 2796224K, used 98577K [0x0000000700000000, 0x00000007aaab0000, 0x00000007aaab0000)
          object space 2796224K, 3% used [0x0000000700000000,0x0000000706044600,0x00000007aaab0000)
         PSPermGen       total 21248K, used 11122K [0x00000006fae00000, 0x00000006fc2c0000, 0x0000000700000000)
          object space 21248K, 52% used [0x00000006fae00000,0x00000006fb8dca60,0x00000006fc2c0000)

Jstack output when jvm is doing forever minor gcs:

"Service Thread" daemon prio=6 tid=0x0000000010e51000 nid=0x3c8c runnable [0x0000000000000000]
   java.lang.Thread.State: RUNNABLE

"C2 CompilerThread1" daemon prio=10 tid=0x0000000010e50000 nid=0x449c waiting on condition [0x0000000000000000]
   java.lang.Thread.State: RUNNABLE

"C2 CompilerThread0" daemon prio=10 tid=0x0000000010e3d800 nid=0x46b8 waiting on condition [0x0000000000000000]
   java.lang.Thread.State: RUNNABLE

"Attach Listener" daemon prio=10 tid=0x0000000010e3c800 nid=0x37c8 waiting on condition [0x0000000000000000]
   java.lang.Thread.State: RUNNABLE

"Signal Dispatcher" daemon prio=10 tid=0x0000000010e35000 nid=0x4088 runnable [0x0000000000000000]
   java.lang.Thread.State: RUNNABLE

"Finalizer" daemon prio=8 tid=0x0000000002186000 nid=0x3df8 in Object.wait() [0x000000001216f000]
   java.lang.Thread.State: WAITING (on object monitor)
        at java.lang.Object.wait(Native Method)
        - waiting on <0x0000000001fcd4d0> (a java.lang.ref.ReferenceQueue$Lock)
        at java.lang.ref.ReferenceQueue.remove(ReferenceQueue.java:135)
        - locked <0x0000000001fcd4d0> (a java.lang.ref.ReferenceQueue$Lock)
        at java.lang.ref.ReferenceQueue.remove(ReferenceQueue.java:151)
        at java.lang.ref.Finalizer$FinalizerThread.run(Finalizer.java:177)

"Reference Handler" daemon prio=10 tid=0x000000000217f800 nid=0x1cfc in Object.wait() [0x0000000011f5f000]
   java.lang.Thread.State: WAITING (on object monitor)
        at java.lang.Object.wait(Native Method)
        - waiting on <0x0000000001fcf510> (a java.lang.ref.Reference$Lock)
        at java.lang.Object.wait(Object.java:503)
        at java.lang.ref.Reference$ReferenceHandler.run(Reference.java:133)
        - locked <0x0000000001fcf510> (a java.lang.ref.Reference$Lock)

"main" prio=6 tid=0x000000000208d000 nid=0x4704 runnable [0x000000000244e000]
   java.lang.Thread.State: RUNNABLE
        at java.util.regex.Pattern.atom(Pattern.java:2199)
        at java.util.regex.Pattern.sequence(Pattern.java:2097)
        at java.util.regex.Pattern.expr(Pattern.java:1964)
        at java.util.regex.Pattern.compile(Pattern.java:1665)
        at java.util.regex.Pattern.<init>(Pattern.java:1337)
        at java.util.regex.Pattern.compile(Pattern.java:1022)
        at java.util.regex.Pattern.matches(Pattern.java:1128)
        at com.app.Search.search(Search.java:100)


"VM Thread" prio=10 tid=0x0000000010d92800 nid=0x4208 runnable

"GC task thread#0 (ParallelGC)" prio=6 tid=0x00000000020d4800 nid=0x4538 runnable

"GC task thread#1 (ParallelGC)" prio=6 tid=0x00000000020d6000 nid=0x4390 runnable

"GC task thread#2 (ParallelGC)" prio=6 tid=0x00000000020d7800 nid=0x4874 runnable

"GC task thread#3 (ParallelGC)" prio=6 tid=0x00000000020d9800 nid=0x4558 runnable

"VM Periodic Task Thread" prio=10 tid=0x0000000010e62000 nid=0x455c waiting on condition

JNI global references: 189

The Search.search() is searching strings that match the given regex, the code

Pattern pattern = Pattern.compile(regex);
        final Set<String> result = new HashSet<>();
        for (final String word : words) {


            if (pattern.matcher(word).matches()) {

                result.add(word);

            }
        }

Thank you.

Kevin
  • 187
  • 2
  • 12

2 Answers2

6

This seems like perfectly healthy behaviour of a JVM. Seems your code got into a loop and keeps creating temporary objects which have to be GCed.

You could try geting some stacktraces via jstack and try to find out where the program is looping.

piet.t
  • 11,718
  • 21
  • 43
  • 52
6

Based on your one sample (more samples would be helpful) it appears that you are creating lots of garbage here.

    at java.util.regex.Pattern.<init>(Pattern.java:1337)
    at java.util.regex.Pattern.compile(Pattern.java:1022)
    at java.util.regex.Pattern.matches(Pattern.java:1128)
    at com.app.Search.search(Search.java:100)

Compiling a pattern creates lot of objects and I suggest doing this as rarely as possible (if at all)

For a more informated view, you really need to use a memory profiler. Try VisualVM, it is built-in to the JDK, or Java Mission Control if you have Java 7 update 40+

JVM recursively does minor garbage collections over and over again and never stops.

I suspect you meant repeatedly, not recursively.

That is what the GC is designed to to do. You are creating a large number of short live objects and the GC is cleaning the up pretty quickly.

What is suspicious is how cleanly the GC is running. Is this a simple program which just produces garbage? If anything real applications have much uglier logs.

You are producing about 2 GB per second of garbage. This is an upper limit for some older systems and newer systems can produce about 8 GB per second.

I suspect if you used a memory profiler you could eliminate most of the garbage you are creating and your program would run quite a bit faster.

jvm is doing forever minor gcs:

It is not, you can just add up the time spent in the GC in a 3 second period.

10.263: [GC [PSYoungGen: 602464K->32K(602496K)] 719850K->117418K(752640K), 0.0037990 secs] [Times: user=0.01 sys=0.00, real=0.01 secs]
10.575: [GC [PSYoungGen: 602464K->32K(638656K)] 719850K->117418K(788800K), 0.0037570 secs] [Times: user=0.02 sys=0.00, real=0.00 secs]
10.876: [GC [PSYoungGen: 603168K->32K(637952K)] 720554K->117418K(788096K), 0.0039310 secs] [Times: user=0.02 sys=0.00, real=0.00 secs]
11.180: [GC [PSYoungGen: 603168K->32K(638528K)] 720554K->117418K(788672K), 0.0039980 secs] [Times: user=0.01 sys=0.00, real=0.01 secs]
11.480: [GC [PSYoungGen: 605216K->64K(637952K)] 722602K->117450K(788096K), 0.0036720 secs] [Times: user=0.01 sys=0.00, real=0.01 secs]
11.786: [GC [PSYoungGen: 605248K->64K(640384K)] 722634K->117450K(790528K), 0.0049470 secs] [Times: user=0.02 sys=0.00, real=0.00 secs]
12.091: [GC [PSYoungGen: 609664K->32K(639872K)] 727050K->117418K(790016K), 0.0051130 secs] [Times: user=0.02 sys=0.00, real=0.01 secs]
12.395: [GC [PSYoungGen: 609632K->32K(642752K)] 727018K->117418K(792896K), 0.0042050 secs] [Times: user=0.01 sys=0.00, real=0.00 secs]
12.704: [GC [PSYoungGen: 614816K->32K(642304K)] 732202K->117418K(792448K), 0.0036980 secs] [Times: user=0.01 sys=0.01, real=0.00 secs]
13.012: [GC [PSYoungGen: 614816K->32K(645504K)] 732202K->117418K(795648K), 0.0040370 secs] [Times: user=0.01 sys=0.00, real=0.00 secs]

The total time spent in GC in 2.749 seconds is 0.037 seconds which means it is only in GC 1.4% of the time, so 98.6% of the time it is not GCing.

Peter Lawrey
  • 525,659
  • 79
  • 751
  • 1,130
  • 1
    The only thing to add is that there is no real indication of *recursion* here. – Stephen C Sep 20 '13 at 12:54
  • You are right Peter, I should have meant "repeatly". When GCs are running in such a high frequency, JProfiler showed java.util.regex.Matcher was using 500M+ memory. The reason was It was searching words from 6345 distinct words against a regex. – Kevin Sep 20 '13 at 13:34
  • The Matcher is not used very well there. You should use the Matcher.reset method so it would avoid that much of objects , see more in http://stackoverflow.com/questions/11391337/java-pattern-matcher-create-new-or-reset – lcestari Sep 20 '13 at 13:39
  • Thanks so much Peter! Its much better now – Kevin Sep 20 '13 at 13:53