Does the Java 6 concurrent garbage collector perform some daily activity?

Question

JVM version: Oracle Java 6 update 45

GC analysis seems to show that the concurrent garbage collector is taking some action on a 24-hour cycle that causes strange behavior, where there are more and more New Space GCs when new space seems to have a lot of free space -- does anybody know if this garbage collector has some daily activity that would cause a pause every 24 hrs? Perhaps some in-depth daily sweep that totally locks down the heap like a serial garbage collector? Perhaps the garbage collector is tuning itself?

Why do you ask? Do you observe something like that? Did you try [`-Xloggc`](http://stackoverflow.com/questions/1161647/how-to-redirect-verbose-garbage-collection-output-to-a-file)? — Joachim Sauer, Oct 03 '13 at 14:49
Yes, the output of -Xloggc was analyzed to find this behavior. — Greg Mattes, Oct 03 '13 at 14:55

score 1 · Answer 1 · answered Oct 03 '13 at 23:09

I have not heard of any automatic 24h cleanup cycles built into the Hotspot's Parallel GC, I would be surprised if there was any such feature.

To rule out the automatic tuning, try -XX:-UseAdaptiveSizePolicy this will disable automatic GC ergonomy changes.

Then, I would suggest detail GC logging and class histogram printing in those intervals where you observe the 24-hour hiccup. Also, taking thread dumps during the hiccup could help you to rule out the possibility that it is introduced by your application.

Does the Java 6 concurrent garbage collector perform some daily activity?

1 Answers1