Docker Community Forums

Share and learn in the Docker community.

Question about installing new UCP worker

Hi,
I am having some questions about installing a UCP worker and adding it to the UCP manager worker pool.

I have one UCP manager running on one node and want to install a UCP worker on another node in order to install DTR there.

The documentation of how to install a UCP worker node is a bit unclear to me.
As far as I can tel I should just follow the standard UCP installation guide (https://docs.docker.com/ee/ucp/admin/install/), including defining the admin user and password, and then include Step 7: Join worker nodes.

However the standard installation creates a separate swarm and initializes a separate UCP GUI.
I had to leave the swarm with the --force option on the second node and then add the node to the UCP manager.
After I did this I am no longer able to log into the new UCP worker GUI.
There is no mentioning of the leaving of the swarm on the UCP worker in the documentation, but it might be assumed that it is obvious.

Is this the correct procedure to install and add a UCP worker?
Things seem to look ok in the UCP GUI of the UCP manager.

If it is I can move on to install DTR on the worker node.

Thanks in advance.

I am quite confused about what is unclear on the docs…

Also I am not sure why you try to run the command line to start the ucp installer more than once? new replicas for UCP management are added on every management node - this is true for existing manager nodes and for those to join in the future.

You are aware that the steps in step6 and step7 only differ in the setting for “Node Role”?

Though, you can use any of your Swarm manager nodes (that already have UCP installed) to create the commands to be executed on the new node using following commands:
– new manager: docker swarm join-token manager
– new worker: docker swarm join-token worker

Paste the output in the shell of your new node and you are ready to go. UCP will perform the rest of the magic depending on the nodes role.

I realize what my mistake was.
I did not understand that steps 6 and 7 actually installs UCP on the node where you run the generated command.
I though the command was only for joining/adding an already installed UCP to an existing swarm.

The DTR installation documentation did mislead me a bit as it refers to to the standard UCP installation documentation. I had one node running UCP and then looked at the DTR installation documentation which says to install UCP usng the standard documentation. So I started the UCP installation from scratch on the second node, when I should have just installed a worker node using the generated instructions from the UCP “Add Node” GUI.

Now I can proceed with my DTR installation on node two (which is causing some problems due to certificate issues… for another thread…)

Thanks for pointing me in the right direction.

1 Like