View source for Elixir/Ports and external process wiring

This deceivingly simple programming adventure veers unexpectedly into piping and signaling between unix processes.

== Context: controlling "rsync" ==
{{Project|source=https://gitlab.com/adamwight/rsync_ex/|status=beta|url=https://hexdocs.pm/rsync/Rsync.html}}

My exploration begins while writing a beta-quality rsync library for Elixir which transfers files in the background and can monitor progress.  I hoped to learn better how to interface with long-lived external processes—and I got more than I wished for.

[[File:Monkey eating.jpg|alt=A Toque macaque (Macaca radiata) Monkey eating peanuts. Pictured in Bangalore, India|right|300x300px]]

Starting rsync should be as easy as calling out to a shell:<syntaxhighlight lang="elixir">
System.shell("rsync -a source target")
</syntaxhighlight>
This has a few shortcomings, such as the static filenames—it feels unsafe to even demonstrate how string interpolation like <code>#{source}</code> could be misused so let's skip straight to the next tool,  <code>System.cmd</code> which doesn't expand its argv:<syntaxhighlight lang="elixir">
System.find_executable(rsync_path)
|> System.cmd([~w(-a), source, target])
</syntaxhighlight>This is safer, but the calling thread loses control and gets no feedback until the transfer is complete.

To run a external process asynchronously we reach for Elixir's lowest-level <code>Port.open</code> which maps directly to ERTS <code>open_port</code><ref>https://www.erlang.org/doc/apps/erts/erlang.html#open_port/2</ref>.  These are tremendously flexible, here we demonstrate turning a few knobs:<syntaxhighlight lang="elixir">
Port.open(
  {:spawn_executable, rsync_path},
  [
    :binary,
    :exit_status,
    :hide,
    :use_stdio,
    :stderr_to_stdout,
    args:
      ~w(-a --info=progress2) ++
        rsync_args ++
        sources ++
        [args[:target]],
    env: env
  ]
)
</syntaxhighlight>

Progress lines come in with a fairly self-explanatory format:
<syntaxhighlight lang="text">
      3,342,336  33%    3.14MB/s    0:00:02
</syntaxhighlight>

{{Aside|text=
rsync has a variety of progress options, we chose overall progress above so the meaning of the percentage is "overall percent complete".

Here is the menu of alternatives:

; <code>--info=progress2</code> : report overall progress

; <code>--progress</code> : report statistics per file

; <code>--itemize-changes</code> : list the operations taken on each file
}}

Each rsync output line is sent to the library's <code>handle_info</code> callback as <code>{:data, line}</code> and after the transfer is finished we receive a conclusive <code>{:exit_status, status_code}</code>.

We extract the percent_done column and strictly reject any other output:
<syntaxhighlight lang="elixir">
with terms when terms != [] <- String.split(line, ~r"\s", trim: true),
         percent_done_text when is_binary(percent_done_text) <- Enum.at(terms, 1),
         {percent_done, "%"} <- Float.parse(percent_done_text) do
      percent_done
    else
      _ ->
        {:unknown, line}
    end
</syntaxhighlight>The <code>trim</code> lets us ignore spacing and newline trickery—or even a leading carriage return as you can see in the rsync source code,
<syntaxhighlight lang="c">
rprintf(FCLIENT, "\r%15s %3d%% %7.2f%s %s%s", ...);
</syntaxhighlight>The carriage return <code>\r</code> deserves a special mention: this "control" character is just a byte in the binary data coming over the pipe from rsync, but it plays a control function because of how the tty interprets it.  On the terminal the effect is to overwrite the current line!

A repeated theme is that data and control are leaky categories.  We come to the more formal control side channels later.
{{Aside|text=
[[File:Chinese typewriter 03.jpg|right|200x200px]]

On the terminal, rsync progress lines are updated in place by emitting a [[w:Carriage return|carriage return]] control character, <code>\r</code>, <code>0x0d</code> sometimes rendered as <code>^M</code>.  The character seems to be named after pushing the physical paper carriage of a typewriter back to the beginning of the line without feeding the roller.

[[w:https://en.wikipedia.org/wiki/Newline#Issues_with_different_newline_formats|Disagreement about carriage return]] vs. newline has caused eye-rolling since the dawn of personal computing.

[[File:Nilgais fighting, Lakeshwari, Gwalior district, India.jpg|left|200x200px]]
}}

== OTP generic server ==
This is where Erlang/OTP really starts to shine: our rsync library wraps the Port calls under a gen_server<ref>https://www.erlang.org/doc/apps/stdlib/gen_server.html</ref> module and this gives us some special properties for free: a dedicated thread which coordinates with rsync independently from anything else, receiving and sending asynchronous messages.  It has an internal state including the latest percent done and this can be probed by calling code, or it can be set up to push updates to a listener.

A gen_server should also be able to run under a [https://adoptingerlang.org/docs/development/supervision_trees/ OTP supervision tree] as well but our module has a major flaw: it can correctly detect and report when rsync crashes or completes, but if our module is stopped by its supervisor it cannot stop its external child process in turn.

== Problem: runaway processes ==
[[File:CargoNet Di 12 Euro 4000 Lønsdal - Bolna.jpg|thumb]]
What this means is that rsync transfers would continue to run in the background even after Elixir had completely shut down, because the BEAM had no way of stopping the process.

To check whether this was something specific to rsync, I tried the same thing with <code>sleep 60</code> and I found that it behaves exactly the same way, hanging until the sleep ends naturally regardless of what happened in Elixir or whether its pipes are still open.

== Bad assumption: pipe-like processes ==
A straightforward use case for external processes would be to run a standard transformation such as compression or decompression.  A program like <code>gzip</code> or <code>cat</code> will stop once it detects that its input has ended, because the main loop usually makes a C system call to <code>read</code> like this:<syntaxhighlight lang="c">
ssize_t n_read = read (input_desc, buf, bufsize);
if (n_read < 0) { error... }
if (n_read == 0) { end of file... }
</syntaxhighlight>The manual for read<ref>https://man.archlinux.org/man/read.2</ref> explains that reading 0 bytes indicates the end of file, and a negative number indicates an error such as the input file descriptor already being closed.  If you think this sounds weird, I would agree: how do we tell the difference between a stream which is stalled and one which has ended?  Does the calling process yield control until input arrives?  How do we know if more than bufsize bytes are available?  If that word salad excites you, read more about <code>O_NONBLOCK</code><ref>https://man.archlinux.org/man/open.2.en#O_NONBLOCK</ref> and unix pipes<ref>https://man.archlinux.org/man/pipe.7.en</ref>.

But here we'll focus on how processes affect each other through pipes.  Surprising answer: they don't affect very much!  Try opening a "cat" in the terminal and then type <control>-d to "send" an end-of-file.  Oh no, you killed it!  You didn't actually send anything, though—the <control>-d is interpreted by bash and it responds by closing its pipe connected to "[[w:Standard streams|standard input]]" of the child process.  This is similar to how <control>-c is not sending a character but is interpreted by the terminal, trapped by the shell and forwarded as an interrupt signal to the child process, completely independently of the data pipe.  My entry point to learning more is this stty webzine<ref>https://wizardzines.com/comics/stty/</ref> by Julia Evans.  Go ahead and try this command, what could go wrong: <code>stty -a</code>

Any special behavior at the other end of a pipe is the result of intentional programming decisions and "end of file" (EOF) is more a convention than a hard reality.  You could even reopen stdin from the application, to the great surprise of your friends and neighbors.  For example, try opening "watch ls" or "sleep 60" and try <control>-d all you want—no effect.  You did close its stdin but nobody cared, it wasn't listening to you anyway.

Back to the problem at hand, "rsync" is in this latter category of "daemon-like" programs which will carry on even after standard input is closed.  This makes sense enough, since rsync isn't interactive and any output is just a side effect of its main purpose.

== Shimming can kill ==
It's possible to write a small adapter which is sensitive to stdin closing, then converts this into a stronger signal like SIGTERM which it forwards to its own child.  This is the idea behind a suggested shell script<ref>https://hexdocs.pm/elixir/1.19.0/Port.html#module-orphan-operating-system-processes</ref> for Elixir and the erlexec<ref>[https://hexdocs.pm/erlexec/readme.html https://hexdocs.pm/erlexec/]</ref> library.  The opposite adapter is also found in the [[w:nohup|nohup]] shell command and the grimsby<ref>https://github.com/shortishly/grimsby</ref> library: these will keep standard in or out open for the child process even after the parent exits.

I took this approach with my rsync library and included a small C program<ref>https://gitlab.com/adamwight/rsync_ex/-/blob/main/src/main.c?ref_type=heads</ref> which wraps rsync and makes it sensitive to the BEAM port_close.  It's featherweight, leaving pipes unchanged as it passes control to rsync—its only real effect is to convert SIGHUP to SIGKILL (see the sidebar discussion of different signals below).

== Reliable clean up ==
{{Project|status=in review|url=https://erlangforums.com/t/open-port-and-zombie-processes|source=https://github.com/erlang/otp/pull/9453}}
It's always a pleasure to ask questions in the BEAM communities, they have earned their reputation for being friendly and open.  The first big tip was to look at the third-party library [https://hexdocs.pm/erlexec/ erlexec], which demonstrates best practices that can be backported into the language itself.  Everyone speaking on the problem has generally agreed that the fragile clean up of external processes is a bug, and supported the idea that some flavor of "terminate" signal should be sent to spawned programs.

I would be lying to hide my disappointment that the required core changes are mostly to a C program and not actually in Erlang, but it was fascinating to open such an elegant black box and find the technological equivalent of a steam engine inside.  All of the futuristic, high-level features we've come to know actually map closely to a few scraps of wizardry with ordinary pipes, using stdlib read, write, and select<ref>https://man.archlinux.org/man/select.2.en</ref>.

Port drivers<ref>https://www.erlang.org/doc/system/ports.html</ref> are fundamental to ERTS and external processes are launched through several levels of wiring: the spawn driver starts a forker driver which sends a control message to <code>erl_child_setup</code> to execute your external command.  Each BEAM has a single erl_child_setup process to watch over all children.

Letting a child process outlive the one that spawned leaves it in a state called an "orphaned process" in POSIX, and the standard recommends that when this happens the process should be adopted by the top-level system process "init" if it exists.  This can be seen as undesirable because unix itself has a paradigm similar to OTP's Supervisors, in which each parent is responsible for its children.  Without supervision, a process could potentially run forever or do naughty things.  The system <code>init</code> process starts and tracks its own children, and can restart them in response to service commands.  But init will know nothing about adopted, orphan processes or how to monitor and restart them.

The patch [https://github.com/erlang/otp/pull/9453 PR#9453] adapting port_close to SIGTERM is waiting for review and responses look generally positive so far.

{{Aside|text='''Which signal?'''

Which signal to use is still an open question:

; <code>HUP</code> : the softest "Goodbye!" that a program is free to interpret as it wishes

; <code>TERM</code> : has a clear intention of "kill this thing" but still possible to trap at the target and handle in a customized way

; <code>KILL</code> : bursting with destructive potential, this signal cannot be stopped and you may not clean up

There is a refreshing diversity of opinion, so it could be worthwhile to make the signal configurable for each port.
}}

Discussion threads also included some notable grumbling about the Port API in general, it seems this part of ERTS is overdue for a larger redesign.  There's a good opportunity to unify the different platform implementations: Windows lacks the erl_child_setup layer entirely.

== References ==