Merge pull request #1635 from pigletfly/add-faq

docs: add FAQ about reuse kube-apiserver
This commit is contained in:
karmada-bot 2022-04-24 11:57:09 +08:00 committed by GitHub
commit 98ecd8f239
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 18 additions and 0 deletions

View File

@ -20,3 +20,21 @@ to keep a consistent user experience and behavior.
It's worth noting that not all controllers are needed by Karmada, for the recommended controllers please It's worth noting that not all controllers are needed by Karmada, for the recommended controllers please
refer to [Recommended Controllers](./userguide/configure-controllers.md#recommended-controllers). refer to [Recommended Controllers](./userguide/configure-controllers.md#recommended-controllers).
## Can I install Karmada in a Kubernetes cluster and reuse the kube-apiserver as Karmada apiserver?
The quick answer is `yes`. In that case, you can save the effort to deploy
[karmada-apiserver](https://github.com/karmada-io/karmada/blob/master/artifacts/deploy/karmada-apiserver.yaml) and just
share the APIServer between Kubernetes and Karmada. In addition, the high availability capabilities in the origin clusters
can be inherited seamlessly. We do have some users using Karmada in this way.
There are some things you should consider before doing so:
- This approach hasn't been fully tested by the Karmada community and no plan for it yet.
- This approach will increase computation costs for the Karmada system. E.g.
After you apply a `resource template`, take `Deployment` as an example, the `kube-controller` will create `Pods` for the
Deployment and update the status persistently, Karmada system will reconcile these changes too, so there might be
conflicts.
TODO: Link to adoption use case once it gets on board.