DevOps in K8s — PodTopologySpread

DevOps in K8s bootcamp series

Note, full “DevOps in K8s” mind map is available at: “DevOps in K8s Mind Map”

What is PodTopologySpread?

PodTopologySpread (formerly known as "EvenPodsSpread") is a scheduling feature in K8s that aims to achieve a balanced distribution of pods across different topologies. Common examples are nodes, racks, zones, or even custom-defined domains that can be labeled within a cluster.

The purpose of PodTopologySpread is to ensure that the workload is not concentrated in a few nodes or zones but instead is evenly distributed.

An example of PodTopologySpread definition looks like:

apiVersion: v1
kind: Pod
metadata:
  name: mypod
spec:
  topologySpreadConstraints:
  - maxSkew: 1
    topologyKey: zone
    whenUnsatisfiable: DoNotSchedule
    labelSelector:
      matchLabels:
        app: myapp

You can see that anew topologySpreadConstraints field has been added to the Pod's Spec specification for configuring topology distribution constraints. Since this new field is added at the Pod spec level, higher-level controllers (Deployment, DaemonSet, StatefulSet) can also make use of the PodTopologySpread feature.

PodTopologySpread vs…