I was experimenting load balancing our service by having two replicas on a cluster of two nodes.

Then I noticed the following:

Yesterday when I checked the pods, it looked like this:

pod-jq5vr   4/4       Running   0          2m        10.4.1.5    node-vvmb
pod-qbs69   4/4       Running   0          2m        10.4.0.10   node-jskq

This morning:

pod-hvjs8   4/4       Running   0          17h       10.4.1.6   node-vvmb
pod-jq5vr   4/4       Running   0          18h       10.4.1.5   node-vvmb

There must have been node recreation going on between yesterday and this morning, but the pod are on the same node now.

My questions:

1. How can I evenly distribute the pods on both nodes?
2. Does it matter to keep them on the same node? It seems my current configuration does not guarantee they must be on the separate nodes.