How to get a Rails App Deployed with Puma and Capistrano to start on reboot

Question

I've managed to successfully deploy a Rails 4, Puma, Nginx App using Capistrano. When I deploy cap production deploy everything works great. My problem is if the server reboots for whatever reason or if it crashes, it doesn't restart.

I'm using a Debian 8 on DigitalOcean. It seems Debian 8 uses systemd , so I've followed the instructions from Puma but it didn't work. After some research I've found a couple more scripts and the one that seemed most sensible was this:

[Unit]
Description=Rails-Puma Webserver

[Service]
Type=simple
User=myuser
WorkingDirectory=/home/myuser/apps/myapp
ExecStart=/home/myuser/.rvm/rubies/ruby-2.2.2/bin/systemd_rails server -e production
TimeoutSec=15
Restart=always

[Install]
WantedBy=multi-user.target

I've saved the above file in /etc/systemd/system/rails-puma.service then I've enabled it: sudo systemctl enable rails.service and finally started it: sudo systemctl start rails-puma.service

Which unfortunately didn't work. This is the result of sudo systemctl status rails-puma.service :

    ● rails-puma.service - Rails-Puma Webserver
   Loaded: loaded (/etc/systemd/system/rails-puma.service; enabled)
   Active: failed (Result: start-limit) since Thu 2016-07-07 12:11:58 EDT; 4s ago
  Process: 4373 ExecStart=/home/myuser/.rvm/rubies/ruby-2.2.2/bin/systemd_rails server -e production (code=exited, status=203/EXEC)
 Main PID: 4373 (code=exited, status=203/EXEC)

Jul 07 12:11:58 mrcProd systemd[1]: rails-puma.service: main process exited, code=exited, status=203/EXEC
Jul 07 12:11:58 mrcProd systemd[1]: Unit rails-puma.service entered failed state.
Jul 07 12:11:58 mrcProd systemd[1]: rails-puma.service start request repeated too quickly, refusing to start.
Jul 07 12:11:58 mrcProd systemd[1]: Failed to start Rails-Puma Webserver.
Jul 07 12:11:58 mrcProd systemd[1]: Unit rails-puma.service entered failed state.

What am I doing wrong here?

Answer 1

I have a similar service but I declared slightly different /etc/systemd/system/puma.service

[Unit]
Description=Puma Control
After=network.target auditd.service

[Service]
Type=oneshot
RemainAfterExit=yes
ExecStart=/etc/puma/puma-start
ExecStop=/etc/puma/puma-stop

[Install]
WantedBy=multi-user.target

then in /etc/puma/puma-start

#!/bin/bash -

cd /home/changeuser/apps/changeapp/current && ( export RACK_ENV="production" ; /home/changeuser/.rvm/bin/rvm default do bundle exec puma -C /home/changeuser/apps/changeapp/shared/puma.rb --daemon )

and in /etc/puma/puma-stop

#!/bin/bash -

cd /home/changeuser/apps/changeapp/current && ( export RACK_ENV="production" ; /home/changeuser/.rvm/bin/rvm default do bundle exec pumactl -S /home/changeuser/apps/changeapp/shared/tmp/pids/puma.state stop )

remember after setting up to execute

chmod +x /etc/puma/puma-start
chmod +x /etc/puma/puma-stop
systemctl enable puma

and then to test

systemctl start puma
systemctl stop puma
systemctl restart puma
systemctl status puma