How Kubernetes PodDisruptionBudgets Cause Rollout Stalls
from DevOps Daily with Fexingo: CI/CD, Kubernetes, and Modern Software Operations Hosted by Fexingo
In this episode of DevOps Daily, Lucas and Luna dive into a subtle Kubernetes scheduling trap: how PodDisruptionBudgets, meant to ensure high availability during voluntary disruptions, can inadvertently cause rollout stalls in production. Using a real-world example from an e-commerce platform handling Black Friday traffic, they explain the interplay between PDBs, replica counts, and node drains. Lucas walks through the scenario where a PDB with 'maxUnavailable: 1' combined with a two-replica deployment and a rolling update leads to a complete freeze. They discuss the root cause, how the kube-scheduler and kube-controller-manager interact with PDBs, and practical mitigations like using 'minAvailable' over 'maxUnavailable', setting appropriate replica counts, and monitoring with custom alerts. Listeners walk away with a concrete debugging approach and a better understanding of why their rollouts sometimes stall for no apparent reason. #Kubernetes #PodDisruptionBudget #DevOps #Scheduling #RolloutStall #HighAvailability #KubeScheduler #NodeDrain #Ecommerce #BlackFriday #MaxUnavailable #MinAvailable #RollingUpdate #ReplicaCount #KubernetesTroubleshooting #CloudNative #SRE #FexingoBusiness Keep every episode free: buymeacoffee.com/fexingo
Embed this episode
No transcript yet.