Node 14.17.1 support for new Meteor version 2.3

[liitfr]

Hello @abernix !

New version of meteor 2.3 has been just released and it uses node 14.
I've forked your package in order to try to make MUP support this new Meteor version
Looks like it works fine.

Hope you will consider this update.

PS: I had to replace occurences of abernix by liitfr since docker tags depends on each others.
If you accept to check this PR I'll put again abernix everywhere, and then close my fork since it won't be useful anymore.

For people who need a workaround, you can use my copy of abernix/meteord.
Example in your MUP config :

    docker: {
      image: 'liitfr/meteord:node-14-base',
    },

Thanks for your consideration.
Mathias

[xpressabhi]

Thanks, I am gonna give it a try.

[abernix]

I'm interested in merging this (or at least the first commit? Happy to give you contribution credit for sure if you want it though!)

Have you been able to run the tests successfully? e.g.,:

cd tests/
NODE_VERSION=14.17.3 bash ./run_tests.sh

This repo isn't wired up with CI at the moment, so this is manual. (Honestly, I'm not sure if they've all passed recently or not as it's been so long since I've published a version!)

[abernix]

If the answer is "the tests seem to fail but the image seems to work", I'm somewhat accepting of that as the status quo and would be game to publish it still. Would like to get them working, but that's likely an investment in debugging time that I don't have for this project right now unfortunately!

[abernix]

Btw, I was incorrect. There is CI setup on this repository (it just doesn't run on forks) and tests are failing.

I'll need to investigate if you can't — I can imagine that it's something to do with the Meteor 1 -> 2 bump that's not considered by the tests, but probably worth understanding.

The first thing that jumped out to me was that it seems that Mongo is required now and that leaving out the MONGO_DB environment variable – even for applications which don't necessitate Mongo — will result in the server failing to start. That seems easily surmountable, but a quick attempt still didn't get me over the line.