Do not close systemd activated socket on pumactl restart

[fsateler]

Describe the bug

When using systemd socket activation, issuing pumactl restart causes puma to close the systemd-activated socket, leaving the server inaccessible (but still running).

Puma config:

systemd-socket-activate --listen=0.0.0.0:3001 bundle exec puma -C config/puma.rb hello.ru

# Puma can serve each request in a thread from an internal thread pool.
# The `threads` method setting takes two numbers: a minimum and maximum.
# Any libraries that use thread pools should be configured to match
# the maximum value specified for Puma. Default is set to 5 threads for minimum
# and maximum; this matches the default thread size of Active Record.
#
max_threads_count = ENV.fetch("RAILS_MAX_THREADS") { 5 }
min_threads_count = ENV.fetch("RAILS_MIN_THREADS") { max_threads_count }
threads min_threads_count, max_threads_count
set_default_host '0.0.0.0'
# Specifies the `port` that Puma will listen on to receive requests; default is 3000.
#
port        ENV.fetch("PORT") { 3001 }

# Specifies the `environment` that Puma will run in.
#
environment ENV.fetch("RAILS_ENV") { "development" }

dir = File.absolute_path(__dir__ + '/../tmp/sockets')
FileUtils.mkdir_p dir
activate_control_app "unix://#{dir}/pumactl.sock", no_token: true
# Specifies the `pidfile` that Puma will use.
pidfile ENV.fetch("PIDFILE") { "tmp/pids/server.pid" }

# Specifies the number of `workers` to boot in clustered mode.
# Workers are forked web server processes. If using threads and workers together
# the concurrency of the application would be max `threads` * `workers`.
# Workers do not work on JRuby or Windows (both of which do not support
# processes).
#
# workers ENV.fetch("WEB_CONCURRENCY") { 2 }

# Use the `preload_app!` method when specifying a `workers` number.
# This directive tells Puma to first boot the application and load code
# before forking the application. This takes advantage of Copy On Write
# process behavior so workers use less memory.
#
# preload_app!

# Allow puma to be restarted by `rails restart` command.
plugin :tmp_restart

To Reproduce

Use the template hello.ru and the above config (generated by a rails new, plus the control socket):

run lambda { |env| [200, {"Content-Type" => "text/plain"}, ["Hello World"]] }

Run it with:

systemd-socket-activate --listen=0.0.0.0:3001 bundle exec puma -C config/puma.rb hello.ru

Then issue a restart:

bin/pumactl restart

And observe that the inherited socket is closed:

* Restarting...
Puma starting in single mode...
* Puma version: 5.1.0 (ruby 2.6.6-p146) ("At Your Service")
*  Min threads: 5
*  Max threads: 5
*  Environment: development
*          PID: 68425
* Inherited tcp://0.0.0.0:3001
* Closing unused activated socket: tcp:0.0.0.0:3001
* Starting control server on unix:///home/felipe/src/satelabs/buk/test-app2/tmp/sockets/pumactl.sock
Use Ctrl-C to stop

And now the 3001 port is not accessible:

% curl localhost:3001
curl: (7) Failed to connect to localhost port 3001: Connection refused

Expected behavior

I expected the inherited socket to continue being used after a restart, just as it works without socket activation.

Desktop (please complete the following information):

• OS: Linux
• Puma Version: 4.3.6, 5.0.4, 5.1.0

[MSP-Greg]

@fsateler

Thank you for the report, and sorry for the delay. I have a fix, but I've got one question.

I can repo this, sort of. I need to start with:

systemd-socket-activate ** bundle exec --keep-file-descriptors **

rather than:

systemd-socket-activate ** bundle exec **

So, were you using --keep-file-descriptors?

[fsateler]

No, I was not using that flag. AFAICT, adding that does not change anything

[MSP-Greg]

@fsateler

Sorry I wasn't clear last night. --keep-file-descriptors doesn't have anything to do with the bug/issue, but Puma won't start unless I use it. I'm using Windows WSL2/Ubuntu 20.04 locally.

I believe I've got a fix, working on a test...