Ram Overhead of Packages for Docker Deployment

[mgraeser]

Hi Everyone,

last year I created Pluto Notebooks for explaining Fourier Transformation and Digitalization for my university course. For demonstration I searched for ways to host this and found this awesome package. I got everything running like a charm. Thank you for that!

However, I now try to get in in the web, and as I have not VServer myself i tried to host it on free Docker hosters. This fails mostly due to the overhead in RAM usage that Julia and its Packages are creating. My notebook has RAM usage of around 150 MB, with potential to decrease it further. However the local docker file running it needs nearly 1,5 GB in RAM. Around 500 MB of this is already allocated when the slider server is running, another 600 MB due to the notebook (probably due to other packages).

Is there any way to reduce the overhead of Julia, PlutoSliderServer or other packages (mainly FFTW, Plots.jl, PlutoUtils.jl).

I found the precompile option and tested it using the static-export-1 branch. It seem to work (after disabling my costum sliders), however the Website does not respond to changing the sliders. I am not sure where the problem lies, as no error is reported.

I am very grateful for any suggestions how to get it running.

[mgraeser]

Hi Everyone,
I was able to get the precompile running. I found that it need to run on a actual webserver (running it in a windows webserver service does not seem to work, probably as some scripts need to be enabled, which are not enabled by default). After I uploaded it to my webspace it worked.

So I mark this as solved.