#computing #cloud # [[Epistemic status]] #godel-uncertain # Observability #to-digest ## All in one ### Linkerd managed - https://buoyant.cloud/langame/overview ## Low level observability Observability is mainly about metrics, logs, traces, and maybe some more. ### Metrics 1. Expose metrics 2. Use [Prometheus Operator](https://github.com/prometheus-community/helm-charts/tree/main/charts/kube-prometheus-stack) and put a servicemonitor thing with your Helm chart like this: https://github.com/langa-me/helm-charts/blob/main/charts/parlai/templates/servicemonitor.yml ### Logs https://vector.dev/docs/setup/installation/platforms/kubernetes/ or https://github.com/banzaicloud/logging-operator ### Traces https://github.com/jaegertracing/jaeger-operator ## Scaling with custom hacks k8s client Stuff I've done when I was student on my week ends https://github.com/louis030195/niwrad/blob/master/nakama/niwrad/internal/match_deployment.go ## Decision making Questions: - Need persistence, long loading code (AI models, datasets, indexes, Java :D)? -> Kubernetes, compute with no cold start is expensive # External links