Untitled
unknown
plain_text
3 years ago
1.5 kB
9
Indexable
### Current Architecture Three servers in production (Non Kubernetes) MySQL / Redis (container) / RabbitMQ ### Questions around infra Cloud GKE Kubernetes Version 1.22.12 Service Mesh Istio / Linkerd? -- No Ingress Internet -> Google Load Balancer -> Ingress Controller Pod (HTTPS Lets Encrypt) -> App Service -> App Pod (MySQL / Redis / RabbitMQ) Internet -> Google Load Balancer -> Ingress Controller Pod (HTTP) -> App Service -> App Pod (MySQL / Redis / RabbitMQ) K8s CNI Version kubectl get ns Encryption on? Are we using network policies? ### Questions around application Is the slowness on both http and https? Have we tried http (locally from a k8s pod)? Have we tried to hit from a local pod on the kubernetes local domain (<service_name>.<namespace>.svc.cluster.local)? k exec -it <pod> -- bash stern, ab, curl (for loop) Do we have any tracing enabled on the cluster? Jaeger / Datadog / NewRelic -- Fluentd + Grafana Is there any other error on the nodes? kubeproxy or nodes Does the problem occurs on new nodes as well? Have we tried recycling nodes? No. -- I would recommend terminating the node (and let Google spin up a new one... and test again) Do we have support from Google? No Are other applications on the cluster also slow? How many application do we have in the cluster? I recommend installing a sample app and test that if the app too has some issues. Do we have any load or stress testing tool in place? None
Editor is loading...