Self-healing Clusters: Game of Nodes and the Scaling Throne

on YouTube ↗33m

" summary

Shows a live demo wiring Node Problem Detector, descheduler, and cluster autoscaler so pending pods trigger a new node, the descheduler rebalances by evicting duplicates, and a tainted node forces eviction. Shows NPD creating node taints for memory/pressure, descheduler configured via a ConfigMap with low/high utilization and remove-duplicate policies, and VPA in recommendation mode to surface resource requests. Useful if wanting concrete install and config steps that actually work on GKE.

AI-written summary · about

" chapters (10)

1. 0:00Introduction
2. 1:14Self-healing clusters in Kubernetes
3. 4:09Kubernetes stability challenges
4. 9:46Resilient cluster building blocks
5. 13:39Node Problem Detector
6. 15:30Descheduler policies
7. 18:23Cluster Autoscaler
8. 21:54Live demo: Self-healing tools
9. 30:34Key takeaways
10. 32:20Q&A

" related talks

• 01h 01mKubernetes – going beyond the basicsShahid Iqbal · NDC Sydney 2019
• 50mKubernetes - going beyond the basicsShahid Iqbal · NDC London 2019
• 29mProactive cluster autoscaling in KubernetesChris Nesbitt-Smith · Conf42 DevSecOps 2022
• 59mKubernetes - going beyond the basicsShahid Iqbal · NDC Minnesota 2019
• 28mThe Magic of Kubernetes Self-Healing CapabilitiesSaad Ali · KubeCon + CloudNativeCon Europe 2019
• 21mSelf-Healing KubernetesMax Leonard Inden · ContainerDays Hamburg 2018
• 29mProactive cluster autoscaling in KubernetesDaniele Polencic · Python Web Conf 2023
• 37mAuto-scaling party:VPA, HPA, KEDA, nodes, how do they dance?-Nic Vermande & Oluebube Princess EgbunaOluebube Princess Egbuna, Nicolas Vermande · ContainerDays 2023