Redeploy stack exec

[longnameislong]

What you expected to happen?

I require that my stacks are redeployed each run when a change is detected. I am using a file resource to deploy my stack file and when that deploys a change I and using a subscribe on my stack class to run the new exec I have added below.

How to reproduce it?

Deploy a stack and then change the stack file, run puppet with changes and the changes will not be added as the unless option exists

Anything else we need to know?

FYI - I am very new to GIT and puppet that is why I am raising this as an issue instead of a pull, apologies

Could I please ask for this to be added to the module as I am currently running a fork of the project that includes this change.

Code: -extract of stacks.pp

if $ensure == 'present'{
      $docker_stack_flags = docker_stack_flags ({
      stack_name => $stack_name,
      bundle_file => $bundle_file,
      compose_file => $compose_file,
      prune => $prune,
      with_registry_auth => $with_registry_auth,
      resolve_image => $resolve_image,
      })

      $exec_stack = "${docker_command} deploy ${docker_stack_flags} ${stack_name}"
      $unless_stack = "${docker_command} ls | grep ${stack_name}"

      # If you need to redeploy a stack, you can notify Docker::Stack[$stack_name]
      exec { "docker stack refresh ${stack_name}":
        refreshonly => true,
        command     => $exec_stack,
        path        => ['/bin', '/usr/bin'],
        before      => Exec["docker stack create ${stack_name}"],
      }

      exec { "docker stack create ${stack_name}":
        command => $exec_stack,
        unless  => $unless_stack,
        path    => ['/bin', '/usr/bin'],
    }
  }

[davejrt]

I've created a ticket and added this to our internal backlog

[PierreR]

I have the same problem. When my docker-compose file changes, there is a notify to trigger a redeploy but nothing happens even if the logs seems to say otherwise with :

Info: Computing checksum on file /etc/swarm/docker-compose-monitoring.yml
Info: /Stage[main]/Application::Docker::Swarm_manager/Application::Docker::Stack[monitoring]/File[/etc/swarm/docker-compose-monitoring.yml]: Filebucketed /etc/swarm/docker-compose-monitoring.yml to puppet with sum 42b3ef3f34ee00271bad335e561a8f88
Notice: /Stage[main]/Application::Docker::Swarm_manager/Application::Docker::Stack[monitoring]/File[/etc/swarm/docker-compose-monitoring.yml]/content: content changed '{md5}42b3ef3f34ee00271bad335e561a8f88' to '{md5}9fdfab8603f25eb29dd7ad1f7795750c'
Info: /Stage[main]/Application::Docker::Swarm_manager/Application::Docker::Stack[monitoring]/File[/etc/swarm/docker-compose-monitoring.yml]: Scheduling refresh of Docker::Stack[monitoring]
Info: Docker::Stack[monitoring]: Scheduling refresh of Exec[docker stack create monitoring]

This was quite unexpected knowing that running docker stack deploy -c /etc/swarm/docker-compose-monitoring monitoring would update the stack without any issue.

[mattstrain]

So we were having the same issue as this but if I update to the new version and use this new type I get a notice message on every puppet run?
Notice: /Stage[main]/xxx_docker::Service::Stack[v52]/Docker_stack[v52]/ensure: created

[davejrt]

@mattstrain can you please provide a manifest and a more complete log file?

[kwek]

Are you sure this is fixed? Triggering a refresh on docker::stack does not refresh it. The same applies for docker_stack.

I applied the changes in #657 . But not sure how this fix this for docker_stack. Any hints?