Troubleshooting resources

apiVersion: v1
kind: Pod
metadata:
  name: netshoot-unprivileged-meshed
  namespace: cloud-platform
  labels:
    app: netshoot-unprivileged-meshed
    istio.io/rev: default
spec:
  containers:
  - name: netshoot-unprivileged
    resources:
      limits:
        cpu: 50m
        memory: 50Mi
      requests:
        cpu: 50m
        memory: 50Mi
    image: eu.gcr.io/ems-gap-images/netshoot-unprivileged:latest
    command: ["/bin/sleep", "3650d"]
    imagePullPolicy: IfNotPresent
    securityContext:
      runAsUser: 1000
  restartPolicy: Always

• take a look at node resource usage (including bandwidth)
  • overall node resource trend
  • podwise resource trend
• calico-node runs and not throttling
  • check logs for obvoius errors
• netd runs and not throttling
  • check logs for obvoius errors
• kube-dns runs and not throttling
  • check logs for obvoius errors
• check basic communication
  • from node (one-shot pod or another running workload) (telnet/curl/nc -vz)
  • to node (one-shot pod or another running workload) (telnet/curl/nc -vz)
• symptoms
• connection reset
  • indicates that the other side is cutting an existing connection
  • if sporadic rule out keepalive timeout
  • if reproducible
• connection refused
  • indicates that the other side is not even accepting the connection
• no route to host
  • likely you are doing something wrong check source/dst ips if they are for active resources
• timeout
  • indicates dropped packet on the firewall level or bad routing
  • rule out network policies