Chef service restart_command not running on AWS opsworks instnace

Question

We are facing a strange issue with a restart_command on Chef service definition not being executed. We have a task in AWS OpsWorks that is being executed with the following service definition :

service "celery-worker-1" do
  action [ :nothing ]
  supports :restart=>true, :status=>true
  retries 3
  restart_command 'sv force-stop celery-worker-1 ; sv start celery-worker-1'
  if node[:opsworks][:instance][:layers].include?('celery-worker')
    subscribes :restart, "deploy_revision[testapp]", :delayed
  end
end

Then this is called at the end of the file from notifies

  elsif node[:opsworks][:instance][:layers].include?('celery-worker')
    notifies :restart, resources(:service => "celery-worker-0", :service => "celery-worker-1")

When this task is executed from OpsWorks the logs show no errors or issues :

[2018-02-09T08:33:34+00:00] INFO: Processing service[celery-worker-0] action nothing (testapp::configure line 17)
[2018-02-09T08:33:34+00:00] INFO: Processing service[celery-worker-1] action nothing (testapp::configure line 27)

But when we check on the server itself the celery-workers were not restarted. Executing manually the command from restart_command on the server, works without any issues. So, it seems Chef is not executing this restart_command for some reason :

'sv force-stop celery-worker-1 ; sv start celery-worker-1'

Thanks in advance for the help.

Answer 1

That would mean that either node[:opsworks][:instance][:layers].include?('celery-worker') is false or deploy_revision[testapp] is not updating. For the latter you can see in the output, if you get something like deploy_revision[testapp] (up-to-date) then it isn't updating so no notification trigger. For the layers data, you would have to check that manually.