Habitat plans for chef-server


#1

Hello,

I have an interest in running our chef-server in K8s. I see that all of the individual components of the chef-server have been habitatized. Has this been done by anyone and if so is there any literature or examples floating around to get me started?

Thanks,
Nathan


#2

I’ve actually been interested in this myself but still dealing with some things so haven’t got around to it. But I’d like to know more.


#3

We’ve definitely got examples of Habitat and Kubernetes working together, but I don’t recall anyone spinning up chef-server in it specifically. Maybe take a look at the Habitat Operator to get started:


#4

Hey @nmcginnis @y-me-y I can definitely talk to you about this - we’ve got those habitat plans to the point where they can wire together in containers in a simplistic way (1 of each, with binds) and that could be ported to run in k8s without a ton of trouble. But I’m pretty sure that today the bind/template logic will break in horrible ways if there’s more than 1 of any member in the service group.

Also relevant and some lessons learned is this 3-part series: https://medium.com/@irvingpop where I played with scaling up the sister stack from Chef Automate.


#5

@irvingpop @garrett - I think the idea is what is the long term strategy for deploying chef infrastructure with habitat or automate for applications. Is it feasible to use multi node layout with ha for core chef services. I.E. can I deploy a new supermarket easily with habitat yet?


#6

@y-me-y Not yet, but things are afoot. Caveat: I’m not a Chef product manager nor do I get to make strategic decisions about this, but this topic is where two of my personal interests intersect ( building highly scalable and available Chef services, and habitizing difficult things) :slight_smile:


#7

@irvingpop, I am cool with that. Just it seems to me if you want someone to buy into using your tech to deploy the apps - you should also buy into that and make an offering in habitat :man_shrugging: