For CDK vagrant-service-manager should print OpenShift setup information during `vagrant up`

[LalatenduMohanty]

Vagrant-service-manager (1.0.0) should print OpenShift setup information during vagrant up for CDK.

For CDK, vagrant-service-manager should print the below OpenShift information as it is the default in CDK.

==> default: 
==> default: Successfully started and provisioned VM with 2 cores and 3072 MB of memory.
==> default: To modify the number of cores and/or available memory set the environment variables
==> default: VM_CPU respectively VM_MEMORY.
==> default: 
==> default: You can now access the OpenShift console on: https://10.1.2.2:8443/console
==> default: 
==> default: To use OpenShift CLI, run:
==> default: $ vagrant ssh
==> default: $ oc login 10.1.2.2:8443
==> default: 
==> default: Configured users are (<username>/<password>):
==> default: openshift-dev/devel
==> default: admin/admin
==> default: 
==> default: If you have the oc client library on your host, you can also login from your host.

[LalatenduMohanty]

Looks like lib/vagrant-service-manager/services/open_shift.rb needs to print the stdout

module Vagrant
  module ServiceManager
    class OpenShift
      def initialize(machine, ui)
        @machine = machine
        @ui = ui
        @extra_cmd = build_extra_command
      end

      def execute
        full_cmd = "#{@extra_cmd} sccli openshift"

        @machine.communicate.sudo(full_cmd) do |type, data|
          if type == :stderr
            @ui.error(data)
            exit 126
          end
        end
      end

      private

      def build_extra_command
        cmd = ''
        CONFIG_KEYS.select {|e| e[/^openshift_/] }.each do |key|
          unless @machine.config.servicemanager.send(key).nil?
            env_name = key.to_s.gsub(/openshift_/,'').upcase
            cmd += "#{env_name}='#{@machine.config.servicemanager.send(key)}' "
          end
        end
        cmd.chop
      end
    end
  end
end

[hferentschik]

Now, whether the service-manager does print this setup information or not is debatable, but how is this going to solve the issue? There is nothing wrong with the current Vagrantfile afaict. It should be possible to add the script provisioning at this step. We need to find the underlying reason why Vagrant exits prematurely.

[LalatenduMohanty]

@hferentschik Agree.

It does not solve https://github.com/projectatomic/adb-atomic-developer-bundle/issues/337.

But IMO it should print required openshift info. It is a step towards having a minimal Vagrantfile.

Ideall sccli openshift should print the info in stdout and vsm should pick them and display to user.

[navidshaikh]

Ideall sccli openshift should print the info in stdout and vsm should pick them and display to user.

+1 , this should also help handling services by one single component than having multiple components (sccli, vsm, Vagrantfile) configuring a single service. This helps configuring a service completely by one component, ensuring the rest part of configuration is not left over by another components and also not having the dependency on other components to do the rest of configuration.

[coolbrg]

@LalatenduMohanty @navidshaikh : Moreover, we found that vagrant service-manager plugin is not executing extra provisioners present in Vagrantfile (https://github.com/projectatomic/vagrant-service-manager/issues/166).

[navidshaikh]

Moreover, we found that vagrant service-manager plugin is not executing extra provisioners present in Vagrantfile (#166)

Because proper exit codes were not returned from the opensihft configuration commands from sccli

[hferentschik]

It does not solve projectatomic/adb-atomic-developer-bundle#337.

Ok, as long as we agree on that, I am fine :-)

But IMO it should print required openshift info.

Sounds reasonable. We just have to make sure that 'vagrant up' completes successfully and that potential additional provisioning steps get executed as expected.

It is a step towards having a minimal Vagrantfile.

+1

[LalatenduMohanty]

Corresponding issue in adb-utils https://github.com/projectatomic/adb-utils/issues/55

[navidshaikh]

and for removing the lines from Vagrantfile https://github.com/projectatomic/adb-atomic-developer-bundle/issues/341

[hferentschik]

BTW, the Vagrantfile is not such a bad location for this final access information, especially if we can get it into the box'es Vagrantfile. Just wondering whether it is really the responsibility of sccli or the plugin really. Also there is already vagrant service-manager env openshift.

[LalatenduMohanty]

@hferentschik We should not put the OpenShift access information in to the Vagrantbox's Vagrantfile. Because that would mean always getting OpenShift information even if we just want docker , k8s or mesos. Though we can put complex ruby code in the Vagrantfile too to handle. But IMO code outside of the Vagrantfile (i.e. VSM) is better.

I think VSM/SCCLI should have this info as these are interfaces to CDK/ADB.

[hferentschik]

We should not put the OpenShift access information in to the Vagrantbox's Vagrantfile.

true, true

So why not call service-manager env openshift and displaying its content instead of adding more logging elsewhere?

[LalatenduMohanty]

So why not call service-manager env openshift and displaying its content instead of adding more logging elsewhere?

I am not opposed to the idea. The end result is vagrant up should have the required information.

[hferentschik]

I am not opposed to the idea

:-)

[hferentschik]

There is a drawback though, unless we have an explicit command which we can call at the very end, we might not get the desired output at the end of the provisioning. In this case the information might get lost among other messages.

[coolbrg]

@LalatenduMohanty @navidshaikh @hferentschik Since, now the idea is to auto-generate Vagrantfiles, this can be implemented along with https://github.com/projectatomic/vagrant-service-manager/issues/163

[hferentschik]

Since, now the idea is to auto-generate Vagrantfiles, this can be implemented along with #163

not quite sure I see the connection here.

[coolbrg]

@hferentschik , if it is only moving print message part then it is different than #163.

Then, it means that the shell provisioning part will be migrated to vagrant-service-manager, right?

[hferentschik]

Then, it means that the shell provisioning part will be migrated to vagrant-service-manager, right?

I guess that was the intend.

[navidshaikh]

Even if we add feature for auto-generating Vagrantfile, having the openshift information message print part should be moved to vagrant-service-manager from Vagrantfile. As Hardy said, there is no connection.

[coolbrg]

@navidshaikh @hferentschik : If we go this way of printing the message, then we need to remove the shell provisioning lines from provided Vagrantfiles since vagrant up process will print duplicate messages.

What is your opinions?

[navidshaikh]

If we go this way of printing the message, then we need to remove the shell provisioning lines from provided Vagrantfiles since vagrant up process will print duplicate messages.

We remove the one which is extra! We remove the shell provisioning part from Vagrantfile.