Compute sever and app server issues

Hi, we are developing commercial applications based on the latest rhino. Compute and app server. At present, we are doing the test locally. I have a few questions:

The first problem is that when rhino.compute starts, there is a parent port of 5000. The program already supports multiple child processes of compute.geometry. It can also be turned off automatically when not in use. This function is very good. The problem is that when I start with app server, I can only connect to one URL and specify the port of one of the compute.geometry processes, such as 6001. Once there is no request for a long time and the child process is shut down, a new child process cannot be generated. I want to ask if I can make the app server connect to the parent interface, that is, 5000, when it starts, so that I can not rely on the interface of any child process, that is, 6001. Is it because I don’t use it right?

Second, can app server be deployed on other cloud services besides heroku?

Third, when will the relevant documents be released? Will it be much more convenient to use?

Fourth, is there a separate charge for each compute. Geometry subprocess?

Thank you very much, looking forward to your reply!

You want to always connect to the port of the parent rhino.compute process (in this case 5000). It is rhino.computes job to manage and forward jobs to the child compute.geometry processes. Do not directly call the compute.geometry processes (port 6001,…).

AppServer can be deployed anywhere that you can run a node.js web server. Heroku is not required. We used Heroku as an example since it is easy to set up and publish to.

Our current documents are available at

Are there topics that you would like us to expand on?

No, core hour billing is solely based on the number of cores on your server and does not take into account how many compute.geometry processes are running. For a given server, the charge is the same if you have 10 copies of compute.geometry running or one copy running.