Open tdiaque opened 3 months ago
Hi @tdiaque - this is something we've heard a few times, so it's on our radar! There are some complexities around the wider impact to our architecture, and we have a lot of other exciting features coming up (like better control of the permissions you give your pipelines!) so it's not something we're looking at right now. We will definitely let you know if and when we get to this though!
Hi @catmo-syntasso, Thank you for the information. Has this feature/request been accepted for future work or is it still under consideration?
Hey @tdiaque, Cat is on holiday for the week so wanted to get back to you.
That is a good question and can confirm it is in our backlog for the future. We will be leaving this open for visibility of the issue for others as well.
There are a few key considerations we need to design and cover before we can pull in the work. The biggest is that given resources are not scheduled until after their workflow completes, we would need to run a resource workflow before knowing if the Promise workflow had been run successfully for its future destination. This can be orchestrated and managed if the workflows only depend on declarative outputs, but as soon as an imperative action is required it can create challenges that we want to make sure we design for.
With this in mind, there are a few workarounds that we can share with you that depend on some assumptions. Would you be interested in talking these through and getting a plan together to get you unblocked?
Hi @abangser, Thank you for the information. Our team would be happy to meet and discuss. I can send an email to feedback@syntasso.io if that is the email I should utilize.
Hello again @tdiaque! Here's a scheduling link where you can setup time with us: https://meetings-eu1.hubspot.com/cat-morris/kratix-working-session
Hello,
My current environment is set up so that our Kubernetes clusters (these are what Kratix considers worker clusters) are fully owned by the tenants. Whereas I understand more traditional setups are focused around having Kubernetes clusters (worker clusters) and tenants separated by namespaces within the clusters.
With how Kratix is currently handling dependencies upon a Promise being applied; that is, pushing any dependencies to the Kubernetes clusters means that every single of my Kubernetes clusters is going to have the dependencies pushed even if there is no need for the service at the time.
Is there a way to hold dependencies from being pushed to the clusters until the resource request is applied? Currently I have both Backstage and Kratix integrated with each other, so the resource request would be submitted from Backstage.
My goal is to give tenants full control of their own Kubernetes cluster and have the ability to request a service (Jenkins, Postgres, etc.) as needed. However, I do not want a Jenkins operator deployed to the Kubernetes cluster without being needed until the resource request is received.