You should also describe a few things about the usage pattern. Is there an increasing number of unconsumed messages over time? Do you use a non-trivial number of scheduled messages?
And for the heap usage itself, do you see a low staeady state but a sudden spike at the time that the OOM occurs, or is there a slow growth that eventually hits the max heap size? You said you're continually monitoring the free heap (after a full GC, presumably, since heap usage at any other time is largely irrelevant) and it's at 1.2 GB free (2.8 used), which sounds fine, but is that how it was just before the crash? Also, which GC strategy are you using? And 8u60 is pretty old and there were a number of improvements to the G1GC strategy early in the Java 8 development cycle, so if you're using G1, is a JVM upgrade an option to rule out a G1 bug? The error message indicates that it created a heap dump. Have you analyzed it? Tim On Thu, Dec 19, 2019, 11:09 PM brusdev <[email protected]> wrote: > Hi, > > for memory usage, could be useful taking some screenshot of objects in the > heap dump with VisualVM [1]: > - screenshot/s of objects in the heap dump sorted by count; > - screenshot/s of objects in the heap dump sorted by size; > > Thanks, > Domenico > > [1] https://visualvm.github.io/ > > > > -- > Sent from: > http://activemq.2283324.n4.nabble.com/ActiveMQ-User-f2341805.html >
