Replies: 3 comments 6 replies
-
Option 1: hardcode Keepalived support, let OpenShift fork the project. As an OpenShift developer, I'd hate this. |
Beta Was this translation helpful? Give feedback.
-
Option 2: some sort of CR-level plugins? Imagine, I create an Ironic this way: ---
apiVersion: metal3.io/v1alpha1
kind: Keepalived
metadata:
name: ironic-sample
spec: {}
---
apiVersion: metal3.io/v1alpha1
kind: Ironic
metadata:
name: ironic-sample
spec:
networking:
ipAddress: 172.0.2.2
ipManagerRef:
apiGroup: metal3.io/v1alpha1
kind: Keepalived
name: ironic-sample Then I expect the Keepalived controller to... modify the Deployment? Somehow in a way that the Ironic controller don't override? This is not a well thought solution, but I'd be curious to hear feedback. |
Beta Was this translation helpful? Give feedback.
-
Option 3: add explicit support for Keepalived, use #75 (which I think is useful regardless) for downstreams. |
Beta Was this translation helpful? Give feedback.
-
Hi folks,
I've been thinking about connecting keepalived support to IrSO (without it, metal3-dev-env won't work). Similarly to IronicDatabase, we could just create a Keepalived resource. Not similarly to IronicDatabase, keepalived has to be deployed into the same Deployment as Ironic (but not on HA, I guess?).
On top of that, I'm thinking how to enable downstream modifications. For example, OpenShift does not use Keepalived, it injects something else into the pod.
Beta Was this translation helpful? Give feedback.
All reactions