Best installation? halyard vs vs helm charts



I’d like to get an idea from the community about the pros and cons of different methods of getting a production-ready spinnaker deployed. From the docs, spinnaker looks like the right solution for my organization’s CD needs, but I’d like to evaluate it in a form as close to what our final production system will look like.

Another desire is to deploy spinnaker in a manner as close to possible as our actual application deployment (AWS, kubernetes, helm charts) to reduce the amount of cognitive overhead necessary for our SRE team.

After a very small amount of investigation, I see three approaches to this: halyard, armory dot io and the helm charts in the github repo kubernetes/charts, at /stable/spinnaker

My sense of the pros/cons of these are:

Halyard (kubernetes V1/V2 deployment)


  • supported by the official spinnaker project
  • super easy to get started


  • doesn’t work out of the box as well as I hoped
  • a bit of a black box
  • requires a bit of extra config through the hal cli

armory dot io


  • enterprise support
  • packaged product, hopefully easy to deploy


  • unclear on pricing model (have contacted them, waiting for a response)
  • worried about vendor lock-in
  • worried about lack of configurability/flexibility

Helm charts


  • fits in well with our organisational expertise
  • more easy to configure/customise (hopefully)


  • doesn’t work out of the box as well as I had hoped
  • worried that charts won’t be supported/updated as spinnaker project progresses.

I’ve only been investigating this for a few days, so I may have missed or misunderstood many details. Any insight on how Spinnaker is being deployed in production would be very welcome.

Thank you!


We’re working to update the helm charts to rely on Halyard, rather than hard-code configuration. I think when that’s done the choice between Helm & Halyard shouldn’t be so stark.


Hi lwander

Thanks for the response! What form would that take, having the helm charts rely on Halyard? Does that mean you would deploy halyard using a helm chart in your cluster, and then allow it to spin everything up from there?


Yes – that’s the plan.


Thanks for the info. I’m hoping to build something similar following the instructions in this blog post: Best installation? halyard vs vs helm charts.

I’ll report back about how well I get on.


Does this work live anywhere public right now? I’m super interested in this, as I was planning on rolling something like this by hand.


Hey @eac!

I’ll probably fire up a branch in the next week or so. If you’d like I can ping you via Github to give it a look and contribute.


@blaffoy I posted a similar question in the DevOps subreddit here: You hit the nail on the head. I’ve tried all three, and came to similar conclusions.

The PR that @viglesias mentioned is here: I couldn’t get the installation working personally; see the PR above for my issue. Looking forward to the development of that installation method since it seems to be the best of both worlds: automated install using Helm, but in an official way with Halyard.


I use Helm charts, did some updates and added fiat and kayenta myself, pretty good for me at least.


For what it’s worth, the newest Helm chart defers to Halyard for installation:


Can’t trust helm , better stay with halyard , though its a pain initially but once you get the ~/.hal/config ready and up , its just couple of changes that need to appended accordingly , and by the way halyard offically recommended .