workflow-manager needs an update to reflect that we're not called Deis anymore

[kingdonb]

2018/06/03 17:15:12 Starting periodic jobs at interval 12h0m0s
2018/06/03 17:15:12 Serving on :8080
2018/06/03 17:15:12 unable to decorate cluster data with available updates data: &errors.errorString{s:"latest version not available for deis-builder"}
2018/06/03 17:15:12 unable to decorate cluster data with available updates data: &errors.errorString{s:"latest version not available for deis-builder"}

I have a workflow-manager-api server up and another running at http://versions-staging.teamhephy.info and https://versions.teamhephy.info, with pointers to the old Deis version numbers (but it says "hephy") and neither can provide service to a v2.18.0 Deis Workflow cluster because the versions I posted are updated as hephy

Normally it is not necessary to post versions manually into the workflow-manager-api because it is part of the Jenkins release workflow, but since we've never generated a release through Jenkins, I had to copy the Deis release info, update it to say "hephy" and then POST it manually into a versions API backed with an empty database.

This worked! I was able to then export the psql structure and rows, and I have a seed file that I can port over to new wfm-api services.

But none of the Deis services in v2.18.0 know that they are in a hephy chart.

(You can install the -1th release of Hephy today, and point it at the new wfm-api in your values.yaml by changing the setting at the bottom like this):

workflow-manager:
  versions_api_url: https://versions-staging.teamhephy.info
  doctor_api_url: https://doctor-staging.teamhephy.info

[kingdonb]

The error at the top of report just indicates that workflow-manager is checking in with the API server, looking for the latest version of "deis-builder" and Hephy wfm-api does not know anything about deis-builder, only answers for hephy-*

Each new cluster that checks in with workflow-manager-api increments the count, but the update checks will fail because workflow-manager needs to look for eg. hephy-builder

$ curl https://versions.teamhephy.info/v3/clusters/count
3

yebyen:~$ curl https://versions-staging.teamhephy.info/v3/versions/stable/hephy-registry
{"data":[{"component":{"name":"hephy-registry"},"version":{"data":{"description":"final release of Deis registry","fixes":"EOL March 2018"},"released":"2017-05-31T21:40:58Z","train":"stable","version":"2.4.2"}}]}
yebyen:~$ curl https://versions-staging.teamhephy.info/v3/versions/stable/hephy-builder
{"data":[{"component":{"name":"hephy-builder"},"version":{"data":{"description":"final release of Deis Builder","fixes":"EOL March 2018"},"released":"2017-07-06T16:59:24Z","train":"stable","version":"2.12.0"}}]}

vs deis-builder:

yebyen:~$ curl https://versions-staging.teamhephy.info/v3/versions/stable/deis-builder
{"data":[]}

This report simply means that workflow-manager needs an update to look in the right place for newer versions of Deis/hephy components.

I don't know how much this API service ever enabled seamless upgrades of Deis workflow. If I understand correctly, the workflow-manager is supposed to phone home, and then decorate the cluster components every 12 or 24 hours when there is an upgrade available, or I guess even when they are the latest available version so that (some kind of homebrew, user-managed) automated inventory services can monitor and let the cluster operator know if everything's ok/action is needed to upgrade.