k8s required antiAffinity

[wedobetter]

redpanda/src/go/k8s/pkg/resources/statefulset.go

Lines 440 to 445 in d172f26

	RequiredDuringSchedulingIgnoredDuringExecution: []corev1.PodAffinityTerm{
	{
	LabelSelector: clusterLabels.AsAPISelector(),
	Namespaces: []string{r.pandaCluster.Namespace},
	TopologyKey: corev1.LabelHostname},
	},

Is this really a hard requirement?
I have just installed and am testing redpanda in a 2 node k8s cluster, it won't let me run more than 2 replicas.
Why is it advised against running more than one pod per node?

Thanks

[emaxerrno]

@wedobetter i think this is a mistake. When we built our cloud this made sense for our deployments, but I think it makes sense to make this configurable for other users that are not our cloud.

[emaxerrno]

more color here: redpanda can basically take actually advantage of the mull machine. and it already is a thread-per-core design so it was designed from scratch to scale vertically