heymp / containers-on-demand Goto Github PK

If we knew for sure what endpoint to ping inside of all of these containers then we could add automatic health checks. Need to figure out a way to do it constantly.

For example, I am currently just assuming that I can run:
curl --fail http://127.0.0.1:${Number(options.port)}

The problem is that I can assume the host.

Path currently only works on the web component side. If the user selects the &redirect option and specifies a &path option, the server should redirect the user to the specified host AND path.

We have to explicitly allow whitelisted docker images or this thing won't be secure at all. We need them to live in environment variables so the use can easily add their own.

We should also add regex patterns.

Per my conversation with Alex from OpenFaas, by leveraging only the control-plane from openfaas we could use the open faas cli to spin up our containers and label them with the correct traefik labels to wire up the urls correctly. 👍

Currently, the consumer can specify the host url of the new containers that are created by containers-on-demand. We make it so you can specify either a fallback host url or restrict the host urls to a regex pattern.

https://github.com/alexellis/jaas

Issues

• Needs to support adding deploy labels
• Example container timeout on startup

We need to impliment some form of user session so that when a user hands containers on demand a specific id, it will know if the container already exists for that user.

Trying to run through your README instructions, I got a signing issue trying to build server/Dockerfile, fixed by:

RUN apt-get install --allow-unauthenticated -y apache2-utils