Flyte enables production-grade orchestration for machine learning workflows and data processing created to accelerate local workflows to production.

Flyte

Hey all, I put together this repo with a bunch of tooling to get the multi-cluster deployment operational on EKS.

<https://github.com/alexifm/flyte-eks-deployment>

I actually haven’t tested it much so it’s probably riddled with bugs.  It’s a cleanup and refactoring of what we’ve had success with so I expect flaws to be minor and not major issues with the logic.

no prob.  hopefully it helps people wade through it a little easier.

Wow! Thanks for putting together a guide, Alex! Really appreciate it. <@U01DYLVUNJE>, shall we add a link to this in our deployment guide somewhere?

I’m commenting on this PR now: <https://github.com/flyteorg/flyte/pull/3363>

I think there’s a lot of disparate sources that have shades of correctness or were correct at some point in time.  For example, I found the below lines to be incorrect (searched the `values.yaml` and the referenced part didn’t exist).  Also, if you see the `values-auth.yaml` file I constructed, this part of the PR appears inadequate for getting Auth working.

 <https://github.com/flyteorg/flyte/pull/3363/files#diff-ee77fbe074a2e541ba69e70ac154aa8fcffcdc3699cb4f0abc8379e4ee2a3b9bR191-R195>

Oh okay. Please leave a comment on the PR. We'll make sure to incorporate your suggestion!

Also, if you see in my repo, I will need to submit a PR to the main flyte repo because the helm chart was insufficient (maybe it’s been updated, haven’t checked in the last week or two)

<https://github.com/alexifm/flyte-eks-deployment/commit/69e327734acf0bca67cd89dc62e7c26e8ca7a9c9>

The extra lines of config you added to your repo aren't present in the flyte helm chart. <@UNR3C6Y4T>, is this something we need to add to the flyte helm chart?
<@U04P6HHMCG0>, please feel free to create a PR meanwhile.

Right.  That commit is a quick hack I put in because I was finding the the `additionalVolumes` and `additionalVolumeMounts` described in the `values-override.yaml` file in <https://docs.flyte.org/en/latest/deployment/deployment/multicluster.html#user-and-control-plane-deployment|this section> weren’t working out of the box.  I was getting errors on some of the containers/init-containers for `flyteadmin` and `clusterresourcesync`and it was due to the secrets not being mounted in enough places.

this is great <@U04P6HHMCG0> thank you. I will use it and report back any findings