Hello. It was a long time ago and I didn't get all the details.

I think it either came down to a broken version of docker at the time that couldn't run properly.

Or it was because of unstability from Docker. All our Erlang apps are meant to take thousands of requests per second and Erlang have particular ideas about how the system/network should behave. It's very hard on Docker and any unstability or incompatibly would make up for outstanding failures.

Either way, we know for sure now that there were many major critical issues with Docker at the time of the Erlang trial. Not adopting it was the right decision, and it still is.

We only use Erlang for critical applications. We really can't afford to push the limits after a few red flags.

I have used Erlang and containers, but not in combination. I know that Erlang (really, beam) has some ideas about what the network should behave like, and highly virtualized/"software defined " networking may confuse it.

What issues are you having? Containers are not really different from non containerised environments. Have you filed an issue?

It runs fine. However, the challenges are the same as any other environment (e.g. AWS): Remember to set a Erlang cookie, and make sure the host name is sane/set and resolvable, otherwise you will run into issues with the portmapper (epmd).

I run several erlang/elixir applications in containers, it works just fine. I'm not sure why someone claimed it doesn't, but I suspect they didn't try very hard if they did have trouble.

Why not use an erlang unikernal instead?