Address already in use 127.0.0.1 port 9293 error while trying to start two instances of Puma on localhost #3142
Description
Puma version: 6.2.2 (ruby 3.1.2-p20)
I tried to start two instances of Puma server (one API and another WebApp) (through hanami gem based application) on my development machine but was unable to start the server for WebApp successfully. API server started normally but WebApp server showed error Address already in use - bind(2) for "127.0.0.1" port 9293.
Please find below the server startup logs.
API App Logs
jignesh@jignesh-Latitude-7290:~/hanami_projects/my_api_app$ bundle exec hanami server --port=2400
13:59:36 - INFO - Using Guardfile at /......./my_api_app/Guardfile.
13:59:36 - INFO - Puma starting on port 2400 in development environment.
13:59:36 - INFO - Guard is now watching at '/......./my_api_app'
[15170] Puma starting in cluster mode...
[15170] * Puma version: 6.2.1 (ruby 3.1.2-p20) ("Speaking of Now")
[15170] * Min threads: 5
[15170] * Max threads: 5
[15170] * Environment: development
[15170] * Master PID: 15170
[15170] * Workers: 2
[15170] * Restarts: (✔) hot (✖) phased
[15170] * Preloading application
[15170] * Listening on http://0.0.0.0:2400
[15170] Use Ctrl-C to stop
[15170] * Starting control server on http://127.0.0.1:9293
[15170] * Starting control server on http://[::1]:9293
[15170] - Worker 0 (PID: 15177) booted in 0.0s, phase: 0
[15170] - Worker 1 (PID: 15179) booted in 0.0s, phase: 0
[15170] ! Terminating timed out worker (worker failed to check in within 60 seconds): 15177
[15170] - Worker 0 (PID: 25100) booted in 0.0s, phase: 0
WebApp Logs
jignesh@jignesh-Latitude-7290:~/hanami_projects/my_web_app$ bundle exec hanami server
15:32:31 - INFO - Using Guardfile at /......./my_web_app/Guardfile.
15:32:32 - INFO - Puma starting on port 2300 in development environment.
15:32:32 - INFO - Guard is now watching at '/......./my_web_app'
[25763] Puma starting in cluster mode...
[25763] * Puma version: 6.2.2 (ruby 3.1.2-p20) ("Speaking of Now")
[25763] * Min threads: 5
[25763] * Max threads: 5
[25763] * Environment: development
[25763] * Master PID: 25763
[25763] * Workers: 2
[25763] * Restarts: (✔) hot (✖) phased
[25763] * Preloading application
[25763] * Listening on http://127.0.0.1:2300
[25763] Use Ctrl-C to stop
/home/jignesh/.rvm/gems/ruby-3.1.2@my_web_app/gems/puma-6.2.2/lib/puma/binder.rb:335:in `initialize': Address already in use - bind(2) for "127.0.0.1" port 9293 (Errno::EADDRINUSE)
from /home/jignesh/.rvm/gems/ruby-3.1.2@my_web_app/gems/puma-6.2.2/lib/puma/binder.rb:335:in `new'
from /home/jignesh/.rvm/gems/ruby-3.1.2@my_web_app/gems/puma-6.2.2/lib/puma/binder.rb:335:in `add_tcp_listener'
from /home/jignesh/.rvm/gems/ruby-3.1.2@my_web_app/gems/puma-6.2.2/lib/puma/binder.rb:329:in `block in add_tcp_listener'
from /home/jignesh/.rvm/gems/ruby-3.1.2@my_web_app/gems/puma-6.2.2/lib/puma/binder.rb:328:in `each'
from /home/jignesh/.rvm/gems/ruby-3.1.2@my_web_app/gems/puma-6.2.2/lib/puma/binder.rb:328:in `add_tcp_listener'
from /home/jignesh/.rvm/gems/ruby-3.1.2@my_web_app/gems/puma-6.2.2/lib/puma/binder.rb:164:in `block in parse'
from /home/jignesh/.rvm/gems/ruby-3.1.2@my_web_app/gems/puma-6.2.2/lib/puma/binder.rb:147:in `each'
from /home/jignesh/.rvm/gems/ruby-3.1.2@my_web_app/gems/puma-6.2.2/lib/puma/binder.rb:147:in `parse'
from /home/jignesh/.rvm/gems/ruby-3.1.2@my_web_app/gems/puma-6.2.2/lib/puma/runner.rb:78:in `start_control'
from /home/jignesh/.rvm/gems/ruby-3.1.2@my_web_app/gems/puma-6.2.2/lib/puma/cluster.rb:410:in `run'
from /home/jignesh/.rvm/gems/ruby-3.1.2@my_web_app/gems/puma-6.2.2/lib/puma/launcher.rb:194:in `run'
from /home/jignesh/.rvm/gems/ruby-3.1.2@my_web_app/gems/puma-6.2.2/lib/puma/cli.rb:75:in `run'
from /home/jignesh/.rvm/gems/ruby-3.1.2@my_web_app/gems/puma-6.2.2/bin/puma:10:in `<top (required)>'
from /home/jignesh/.rvm/gems/ruby-3.1.2@my_web_app/bin/puma:25:in `load'
from /home/jignesh/.rvm/gems/ruby-3.1.2@my_web_app/bin/puma:25:in `<main>'
from /home/jignesh/.rvm/gems/ruby-3.1.2@my_web_app/bin/ruby_executable_hooks:22:in `eval'
from /home/jignesh/.rvm/gems/ruby-3.1.2@my_web_app/bin/ruby_executable_hooks:22:in `<main>'
Following is the Puma config file app/config/puma.rb content. In both the applications the configuration is identical.
# frozen_string_literal: true
max_threads_count = ENV.fetch("HANAMI_MAX_THREADS", 5)
min_threads_count = ENV.fetch("HANAMI_MIN_THREADS") { max_threads_count }
threads min_threads_count, max_threads_count
port ENV.fetch("HANAMI_PORT", 2300)
environment ENV.fetch("HANAMI_ENV", "development")
workers ENV.fetch("HANAMI_WEB_CONCURRENCY", 2)
on_worker_boot do
Hanami.shutdown
end
preload_app!
So for my WebApp I tried making following changes to my_web_app/config/puma.rb based on the suggestions in #2113
#port ENV.fetch("HANAMI_PORT", 2300)
bind "tcp://127.0.0.1:2300"
but no luck.
Exploring more I found few more resources at following locations
but couldn't get exactly what I should do in my case.
Just in case it helps following are the contents of my /etc/hosts
127.0.0.1 localhost
127.0.1.1 jignesh-Latitude-7290
127.0.0.1 api.some_api.local
# The following lines are desirable for IPv6 capable hosts
::1 ip6-localhost ip6-loopback
fe00::0 ip6-localnet
ff00::0 ip6-mcastprefix
ff02::1 ip6-allnodes
ff02::2 ip6-allrouters
So I tried seeking help from hanami community and Stackoverflow and I received a response to my Stackoverflow post wherein it was suggested to do following:
bundle exec puma --control-url tcp://127.0.01:9293 --port=2300
bundle exec puma --control-url tcp://127.0.01:9294 --port=2400
And trying that out it solved my problem. But it raised another question in mind that why the --control-url option worked but not the bind configuration option I defined in my puma config file (please refer the code snippets shared above).
As can be seen in the logs shared hanami automatically starts Puma available in my applications Gemfile. So the options defined in puma config should be used by Puma, right? Also the documentation for hanami server command implementation (shown below)
says
The server is just a thin wrapper on top of Rack::Server
So that way also the puma config options should get passed to Puma when Rack::Server internally invokes Puma (as part of invocations referenced below), right?
https://github.com/puma/puma/blob/v6.2.2/lib/rack/handler/puma.rb#L136
https://github.com/hanami/cli/blob/v2.0.3/lib/hanami/cli/commands/app/server.rb#L49
https://github.com/hanami/cli/blob/v2.0.3/lib/hanami/cli/server.rb#L29
https://github.com/rack/rack/blob/v2.2.7/lib/rack/server.rb#L52
https://github.com/rack/rack/blob/v2.2.7/lib/rack/server.rb#L167
https://github.com/rack/rack/blob/v2.2.7/lib/rack/handler.rb#L60
https://github.com/rack/rack/blob/v2.2.7/lib/rack/handler.rb#L40
https://github.com/rack/rack/blob/v2.2.7/lib/rack/handler.rb#L13
Note: Above links refer the version of code my applications uses hanami (2.0.3), hanami-cli (2.0.3), rack (2.2.7).
And bind is documented as a supported config option at following link https://github.com/puma/puma/tree/v6.2.2#binding-tcp--sockets and also in the DSL at https://github.com/puma/puma/blob/master/lib/puma/dsl.rb#L245-L277 (DSL reference found under https://github.com/puma/puma/tree/v6.2.2#configuration-file)
**So the question is: are there differences between bind and control-url options because of which when defined bind option in my puma config file it didn't worked but using the control-url option directly from command line worked? If yes, then in which contexts each of those options should be used?
A subjective thought: I thought it would be easy to just use a different port for starting both the instances but a lot of hours passed to find out the cause for the issue I shared but still I couldn't find a solution and had to seek help from community. I don't think starting multiple instances of server on same machine is a rare scenario and that too esp for development purpose and it's really unfortunate if it can't be achieved in smooth manner.
Thanks.