-2

I am running a network server application in CentOS 6.5, having 200 forks with 100,000 maximum client per fork. Normally the traffic is 200K requests per minute. I have seen that CPU load gets high up to 200% and then comes back to normal around 2% within minutes. It is happening around every 30 minutes or so.

The same application running in RHEL 5.8 have no such issues.

I want to know that is there major difference in performance of CentOS with RHEL.

I have clean the code for any memory leaks using valgrind, but having same issue. These things come in my mind that CentOS is taking load for creating forks after they die. CentOS kernel is not using fast timers. CentOS memory manager is not efficient as RHEL. Should I try CentOS 6.3, either it is more stable than new release CentOS 6.5.

Kamran Abdul Jabbar
  • 21
  • 3

1 Answers1

1

After many days, we have found that there is a bug in our code that is doing segmentation fault in RHEL and due to multi-process application new process born after that and RHEL efficiently release the memory of that process that is caused the segmentation fault.

However when we try the same application in CentOS, the bug do not trigger the segmentation fault but make the process in sleep mode with extremely high ram consumption nearly 25%. In this way the process stay in the system and holding the memory which causes high load and spikes.

Conclusion is that one need to be very good in memory leak check in coding while running in CentOS as unlike RHEL CentOS is not smart enough to heal itself.

Kamran Abdul Jabbar
  • 21
  • 3