Deploying a Phoenix app from Windows to CentOS

Questions / Help
, , ,
1

I am bound by company policy to use Windows for development (only WIndows admins present…) but the productive system schould run on a CentOS.
I tried deploying a simple Phoenix app (converting user input as markdown to html code) using distillery as described in the Using Distillery With Phoenix article but removing the Erlang ERTS by setting set include_erts: false in rel/config.exs and compiling Erlang from source on CentOS.

I tried running the app but nothing happens, when I run the app with the foreground command I get the following error that I don’t comprehend:

bin/markdownhtml foreground
{"init terminating in do_boot",{load_failed, [gen,ets,supervisor,proc_lib,lists,kernel,heart,gen_server,gen_event,file_io_server,file_server,filename,file,error_logger,erl_parse,erl_lint,erl_eval,code_server,code,application_master,application_controller,application,error_handler]}}
init terminating in do_boot ()

Crash dump is being written to: erl_crash.dump...done
Shutting down node cleanly..
Node is not running!

What should this error tell me except that something went wrong?

How else do you suggest should I deploy?

2

Most likely it is a compatibility issue. From Erlang crash dump man page:

Init terminating in do_boot ()
The primitive Erlang boot sequence was terminated, most probably because the boot script has errors or cannot be read. This is usually a configuration error; the system can have been started with a faulty -boot parameter or with a boot script from the wrong OTP version.

Excluding ERTS often result in faulty behavior. For consistency I guess it’s probably best to always bundle it.

Now, since the host system (Windows) is different from the target (CentOS), you have four options:

  1. Build the release inside a CentOS Docker container. This is the most lightweight solution and highly recommended.
  2. Build the release inside a CentOS VM (via Virtualbox, VMWare, etc.).
  3. Build the release on a CentOS build server.
  4. Build the release on the production server.

Option 1 and 2 can be done on your host machine. Option 3 needs you to have an extra server for building releases, and that might not be feasible depending on your company. I don’t think option 4 is recommended though, but it’s possible. Be aware that all options requires your container/VM/machine to have Elixir installed.

1 Like
3

Thanks for the reply, I chose the docker variant, but run the entire application via a docker container based on msaraiva/elixir. The deployment process I used is almost the same as and derived from the one described in this article.

2 Likes