@pat_nadolny doing a bit of reading, it looks like Prometheus doesn't deal well in autoscaled clusters 🤔 We deploy prometheus 'statically' using Helm, which assumes resources will be stable after deployment. There is however a

prometheus-operator

which is set up to handle autoscaling by reconfiguring Prometheus when cluster conditions change 😅 Found a tutorial on using that and Grafana together on EKS, and would be keen to give it a go. @taylor pointed me to Kubernetes operators a while back, and they are pretty cool!