I have a question.
Let's imagine a situation.
For example: We have three node within the cluster. And we have replication-controller where we have spec: replicas: 3
I would like to have not more one pod per node. Because if one node will have more one pod while node die, i'll lose two or more pods.
How can i do that?
The Kubernetes scheduler already prioritizes spreading pods from the same replication controller out across your nodes.
However, if you want to be 100% sure that no two pods will end up on the same node, you can set the container's
HostPort
field. No two containers with the sameHostPort
can ever run on the same node.If you want to get even fancier, you could write your own scheduler plugin :)
In general, though, the idea of Kubernetes is that you shouldn't have to think about your nodes except in special circumstances; you can instead trust the system to keep your applications running.