News: Troubleshooting an application suffering from 130x performance degradation

  1. We also faced at the very same time when the throughput falls the application heap usage has reached close to the maximum heap size allowed by the -Xmx512m option given at the startup script. And from the same moment the variance in heap usage starts to diminish.

    Third symptom was derived from CPU monitoring scripts where all cores become completely utilized right about the same time when the throughput falls and memory usage first reaches close to the maximum available memory.

    To see what we did to find the cause and verify our gut feelings, read further from our blog post.

  2. I assume by "130x worse", you mean "1/130'th of the original speed/throughput", right?


    Nice writeup nonetheless.  GC/heap monitoring/adjusting has always been a black art to me.

  3. Yes, you got the point correctly - the throughput was reduced to 1/130th of the original