Installing Custom Node to Remote Server

Core Development
1

Hello!

I am currently working on using the Admin HTTP API to install nodes to Node-RED servers. Installing to a local server is simple enough, and has no issues, but I am curious about installing a custom node to a remote Node-RED server.

The documentation shows that the POST method takes either a path to the node on your machine or the name of a module in the npm repository. Seeing as passing a local path to a directory would not work on a remote server, it seems that the only way to put a node onto a remote server is to publish the custom node to npm and install it that way. Is this thinking correct, or is there another way to do this that I am missing? In addition to this, if my thinking is correct, are there plans to add support for uploading custom nodes directly to a remote server through the API to install them rather than having to publish to npm?

Thank you!

2

I have never done it, but presumably you could copy the node to the remote server then use the api to install it from that directory.

3

This could be a good solution but, at the same time, is there a good way to generalize this to cover any remote server? I feel like there would be issues that arise around access credentials and differences in OS structure that could prevent a good general-purpose method of accomplishing it this way. But I could be overthinking it

4

If you need to publish to multiple servers but don't want to publish to npm then I believe there are various ways of having a private npm repository. Google will show you some no doubt. Alternatively you could just put them in a git repository on github or similar (which can be private) and pick them up from there.

5

one way is to use npm pack to package your node locally - then you can distribute the tgz file it creates as you wish - then run npm install yournodename.tgz in the .node-red directory as usual.

6

Does the old method of using the nodes folder still work? If it does, you could copy the nodes though you would still need to restart NR.

7

To address the bit of your question that hasn't been answered - there are no plans for this because it has never come up as a requirement.

When I created the api in the first place, I did recognise it might be something we could add - but at the time there was no need for it, so nothing more happened.

I'm sure its something we could put on the backlog to explore at some point - without committing either way on it.

1 Like
8

Thank you everybody for the responses. These were all helpful!