Toby

A WIP terminal-based observer for the Erlang VM. Just an early prototype here so far.

It's being developed in parallel with ex_termbox and Ratatouille, which respectively provide termbox bindings and a terminal UI kit for Elixir / Erlang.

Running It

First, clone the repo and run mix deps.get. Then to run the application:

mix run --no-halt

Currently, you can only see information about the local node (i.e., the one you just started), but soon you'll also be able to connect to other nodes like you can with the observer GUI.

Building a Release

It's also possible to create a distributable, self-contained executable via Distillery. I'd like to provide these for download in the future, but for now you can build one yourself using the provided config.

Releases built on a given architecture can generally be run on machines of the same architecture.

MIX_ENV=prod mix release --executable --transient

See the output for the location of the built executable (most likely at _build/prod/rel/toby/bin/toby.run).

This is a Distillery release that bundles the Erlang runtime and the toby application. Start it in the foreground:

_build/prod/rel/toby/bin/toby.run foreground

You can also move this executable somewhere else (e.g., to a directory in your $PATH). A current caveat is that it must be able to unpack itself, as Distillery executables are self-extracting archives.

Termbox NIFs & Comparison to Other Approaches

How does this compare to other tools like observer_cli and etop?

These are nice tools written in pure Erlang, but I was looking for a user experience closer to htop or terminal vim/emacs. Pure-Erlang tools cannot access raw terminal events; they can only get line-buffered input (i.e., some characters and then a return).

Because toby uses the termbox library (an alternative to ncurses) under the hood, it can support interactions like scrolling with the keyboard or mouse, clicking, or pressing a single key to perform an action. It also doesn't flicker on updates, and it can respond to window resizes by automatically performing a relayout of the content. These things unfortunately aren't possible in pure Erlang, as it requires putting the terminal into "raw" mode. Toby relies on NIFs from ex_termbox in order to do this.

Isn't it dangerous to use NIFs?

Yes, it certainly is, but the idea isn't to run this in the same VM as your production code.

Rather, it's to run toby as a separate, hidden node and simply connect it to your node or cluster via Erlang distribution protocol. This is similar to how a C node works. By running toby in a separate VM, even if one of the NIFs causes a segfault in the observer node, it should not affect the node(s) being observed.