On Fri, Jul 26, 2019 at 4:34 PM Terence Monteiro <
[email protected]> wrote:

> Hi Fineracters,
>
> Normally about 1.5 is the minimum recommended RAM for Fineract v1 and some
> docs suggest 4Gb is good. I have run PHP, Ruby and Perl applications on VPS
> with 1Gb RAM (though some PHP applications can even require 10Gb). I was
> reading this thread
> https://www.quora.com/Is-it-true-that-Node-js-helps-in-reducing-server-costs-compared-to-PHP-and-Java-servers
>  saying
> that NodeJS can handle many more requests with the same resources.
>

That thread to me seems an oversimplification of the topic... ;-)

I think that if we can find a way to run Fineract on a Raspberry Pi, it
> would be very cool to demonstrate at the next ApacheCon
>

It would be cool demo/show case. I think practical application would be
much more limited though.

FYI Java on the RPi is not a great story. It's because of the ARM vs. Intel
thing - nobody really invested properly into optimizing the JVM for ARM;
OpenJDK doesn't perform as well on ARM as on x86. There are other JVM impls
than OpenJDK, notably Azul and J9 - it's possible that they are performing
better, but it would need more testing and all.

and also will help address some of the memory issues that cause Fineract to
> crash requests with OutOfMemoryException. Per the above thread, Java Web
> servers can be designed with multiple threads per process but it is much
> easier to write them with single thread per process, so would an
> alternative multi-thread per process Java Web Server help make it possible
> to run Fineract on a Pi 3b+ or we should just wait till a future Raspberry
> Pi ships with more RAM? Thoughts please.
>

If I can be blunt, IMHO that's the wrong approach... whatever memory issues
that cause Fineract to crash requests with OutOfMemoryException should be
analysed and fixed. Is there a JIRA for this?

Instead of single thread per process, what you typically do, for a Java
server application such as Fineract, is the classical multi-thread per
process Java Web Server, up to a point; if you need more, you would scale
out by running several of them - Kubernetes! ;-)


> --
> Best Regards,
> Terence Monteiro,
> Mob: +91 96633 13728
> www.sanjosesolutions.in
> <http://www.sanjosesolutions.in>
> "Terroy", 67, 10th Cross,,
> Lingarajpuram, Bangalore - 84.
>

Reply via email to