Skip to content

Instantly share code, notes, and snippets.

@Jaykul

Jaykul/About autoscaling node pools in AKS.md

Last active July 18, 2023 19:43
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save Jaykul/0fd178825a5d0bcbdb46175ad2dc73ed to your computer and use it in GitHub Desktop.
Save Jaykul/0fd178825a5d0bcbdb46175ad2dc73ed to your computer and use it in GitHub Desktop.
Download ZIP
AKS Autoscaling requires CPU quota +1
Raw
About autoscaling node pools in AKS.md

If you ever see a pod that won't deploy, and it says something like:

Reason: NotTriggerScaleUp
Message: pod didn't trigger scale-up: 1 node(s) had untolerated taint {CriticalAddonsOnly: true}, 1 in backoff after failed scale-up

Don't waste time trying to ask Kubernetes why this failed. Check your scale up settings in the Azure portal, and hit the "Scale node pool" link: image

In this case, (as you can see on the right hand side), the scale up failure is due to lack of quota in Azure for the CPU cores.

Back when we originally deployed AKS, we increased the "Ev5" quota in our 4 main environments to 100, but we recently decided to deploy AKS to a new subscription for dev purposes, and we set our settings so low that it was able to deploy without a problem -- until today, when I tried to deploy a test instance of Elastic Search.

ℹ NOTE: The minimum quota is your "vCPU per node" X your "maximum node count" plus 1 (because we always have to be able to deploy at least one extra node while performing node upgrades). Thus, the default quota of 10 vCPUs only allows a single 4 vCPU node.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment