Commcarehq Performance Issue after updating commcare-cloud

HI

i’ve observed a performance issue on web application after updating commcare-cloud repo to the latest. I’ve seen the change removes a jython_memory setting from environments/echis/app-processes.yml

my previous setting was jython_memory: "12288m", the setting has been removed on the latest commcare-cloud repo.

can anyone tell me where to look for a better performance web app?

Demisew,

I don’t know what could be causing the performance issue you describe. The jython_memory setting was related to an old CommCare service called “touchforms” which is no longer in use.

Perhaps someone else on this forum has suggestions for where to look.

  • Ethan

Powering the Web Apps feature there is a java service called “formplayer” that runs on the server. If that service is sharing a machine with other application and database services, then there’s no one rule for how to optimize formplayer performance—there are many possible causes of slowness, including general high resource (e.g. CPU/RAM) utilization on the machine, overconsumption of resources by a particular process, or under-allocation of resources (probably RAM) to formplayer. The problem can also be unrelated to the formplayer service/machine itself, for example if a database or other service it depends on is slow.

If you have reason to believe that increasing the memory allocated to formplayer would improve the situation, you can do so by editing the formplayer_memory variable in app-processes.yml (e.g. "3584m" for 3584 MB) and running commcare-cloud <env> update-config followed by commcare-cloud <env> service formplayer restart. Restarting formplayer will cause a brief interruption to formplayer, approximately 15 seconds.

Hi danny,
thank you for the response. I’ve previously increased RAM usage to 12GB that is:
formplayer_memory: "12288m" so that memory won’t be an issue for formplayer.

And the machine is not using too much resources(CPU/RAM) after i moved celery and pillow tasks to separate machine.