The production environment seems to have a Bundler or environment configuration mismatch. We are unable to get systemd or capistrano to start delayed_job. We can however start/stop it manually as the role account.
This is working in stage. Added a bunch of debug to the Rails startup files and the bin/delayed_job binstub and found that it is failing when hitting the Bundler.require(*Rails.groups) line in application.rb. This line uses the Gemfile to determine what gems to load.
The journalctl output shows a mysql socket error.
The debug output shows that the Gem paths and Bundler versions are identical regardless of whether we start it manually or via systemd.
Ashley noticed that we had the SSM_ROOT_PATH env variable set for the systemd unit file for puma. Adding it to the one for delayed_job fixed the issue.
The production environment seems to have a Bundler or environment configuration mismatch. We are unable to get systemd or capistrano to start delayed_job. We can however start/stop it manually as the role account.
This is working in stage. Added a bunch of debug to the Rails startup files and the
bin/delayed_job
binstub and found that it is failing when hitting theBundler.require(*Rails.groups)
line inapplication.rb
. This line uses the Gemfile to determine what gems to load.The journalctl output shows a mysql socket error.
The debug output shows that the Gem paths and Bundler versions are identical regardless of whether we start it manually or via systemd.