Guy Templeton
Guy Templeton
FWIW I'm unlikely to pick this work up in the near future given we've moved from using Calico to a different CNI internally so this is no longer a priority...
My general view would be that the current code using the AWS APIs is a significant improvement on the previous setup, and that it provides us with the functionality we...
/area provider/aws
/area provider/aws Hey, Thanks for raising this, my view would be this is a feature request rather than a bug (though maybe our docs should be clearer that currently the...
/area provider/aws Hey @youwalther65, the reason Karpenter is able to handle this nicely whilst the CA doesn't currently is their different operating models, with Karpenter directly creating nodes, whilst the...
/area provider/aws
Hey Shubham, yep, looks like you're correct and this was mistakenly included in the release notes for 1.24 despite not making it into that release branch, will update the release...
There's a number of different factors that could be involved here, including what scaling (both up and down) has happened before this point, and whether you're overriding the default `--scale-down-utilization-threshold`...
The Cluster Autoscaler operates only on the proportion of resources requested by the pods on the node, it doesn't scale based on the raw CPU usage.
Interested!