Hawthorn.2 native installation: heavy RAM consumption overcome?

[Andreas Gelever]

Hi!

I've installed the newest release of Open edX Hawthorn.2 on my developer PC running Ubuntu 16.04 LTE for testing purposes.

The problem is that the whole bunch of installed services and systems consumes all my available RAM of 8GB as well as swap of 3.7 GB.

Can I safely turn off some of them to release some memory space?

Sorry, I am not an expert in Open edX architecture (I come from Java community) with no idea why it needs ruby, rabbitMQ, memcached even Java and other tools (of course I am aware it is written in Python))). But particularly I noticed about 14 heavy python processes under www-data user in System monitor consuming 200MB each! And 8 heavy gunicorn processes under the same user consuming approx. the same amount of memory each. Screenshots are attached.

As this is not a production installation, and my objective is only to test some basic features like Studio and LTI interoperability,

I would like to disable unnecessary functions, just to release a bit of RAM on my PC for browser and my Java IDE.

If this is possible, please provide some instructions how to disable unnecessary services safely with the possibility to revert them back as needed.

Any suggestions and links to a bit deeper information than is placed in on-line documentation would be greatly appreciated!

Thank you.

[Nguyễn Văn Tú]

For testing purposes, you can reduce the number of workers, which is safe and greatly reduce the amount of ram usage. The number of workers is set in /edx/app/edxapp/lms_gunicorn.py and /edx/app/edxapp/cms_gunicorn.py.

--
You received this message because you are subscribed to the Google Groups "General Open edX discussion" group.
To view this discussion on the web visit https://groups.google.com/d/msgid/edx-code/37335784-95b9-4561-b32a-9efcdd42969c%40googlegroups.com.

[Andrey Popov]

Maybe there is also a way to reduce the number of celery workers too?

[Stan V]

https://stackoverflow.com/questions/3874422/developing-with-djangocelery-without-running-celeryd

[Andreas Gelever]

Great thanks. I've made it done with 1 process per installation!