| apiVersion: v1 | |
| kind: Pod | |
| metadata: | |
| name: netkat | |
| spec: | |
| hostNetwork: true | |
| containers: | |
| - name: netkat | |
| image: aojea/netkat:0.1.1 | |
| command: ["/bin/sh", "-ec", "sleep 10000"] |
References:
Kube-proxy iptables consider affinity per Service:ServicePort, each ServicePort in the same Service has a different affinity.
There is no distributed consistency between nodes to keep track of the Services sessions, when using LoadBalancer type Services,
Conformance test are a subset of the e2e test Kubernetes test.
The e2e tests are based on Ginkgo https://onsi.github.io/ginkgo/ and live in the Kubernetes repository https://github.com/kubernetes/kubernetes/tree/master/test/e2e.
There is no guarantee of compatibility for the e2e binaries, hence the e2e binary version MUST match your kubernetes cluster version.
Benchmark_Key-4 32955 36670 ns/op 109 B/op 0 allocs/op
Benchmark_Value-4 25776 47899 ns/op 140 B/op 0 allocs/op
It seems that by key requires is much better.
Also, it seems that the larger the size of the pod matters, I've tried with larger pods size and the results are worse for Value.
Most of the time is spent in runtime.duffcopy when iterating by value
| # Create test namespaces | |
| sudo ip netns add testNS | |
| # Connect the namespace to the host using a veth pair | |
| sudo ip link add name vethHost type veth peer name vethNS | |
| sudo ip link set netns testNS dev vethNS | |
| # Configure the namespaces network so they can reach each other | |
| sudo ip netns exec testNS ip link set up dev lo | |
| sudo ip netns exec testNS ip link set up dev vethNS |
ServiceType,Services,Endpoint/Service,FilterRules,NATRules,Latency
ClusterIP,1,1,5,11,1.16653ms
ClusterIP,501,1,5,2511,8.228011ms
ClusterIP,1001,1,5,5011,14.535362ms
ClusterIP,1,501,5,1511,3.95692ms
ClusterIP,501,501,5,754011,804.495785ms
ClusterIP,1001,501,5,1506511,1.480063753s
| 00:00:00.000000000 [ovn/openshift-ovn-kubernetes_ovnkube-master-f89lp_nbdb.log] + [[ -f /env/_master ]] | |
| 00:00:00.000000000 [ovn/openshift-ovn-kubernetes_ovnkube-master-f89lp_nbdb.log] + ovn_kubernetes_namespace=openshift-ovn-kubernetes | |
| 00:00:00.000000000 [ovn/openshift-ovn-kubernetes_ovnkube-master-f89lp_nbdb.log] + ovndb_ctl_ssl_opts='-p /ovn-cert/tls.key -c /ovn-cert/tls.crt -C /ovn-ca/ca-bundle.crt' | |
| 00:00:00.000000000 [ovn/openshift-ovn-kubernetes_ovnkube-master-f89lp_nbdb.log] + transport=ssl | |
| 00:00:00.000000000 [ovn/openshift-ovn-kubernetes_ovnkube-master-f89lp_nbdb.log] + ovn_raft_conn_ip_url_suffix= | |
| 00:00:00.000000000 [ovn/openshift-ovn-kubernetes_ovnkube-master-f89lp_nbdb.log] + [[ 10.0.0.4 == *\:* ]] | |
| 00:00:00.000000000 [ovn/openshift-ovn-kubernetes_ovnkube-master-f89lp_nbdb.log] + db=nb | |
| 00:00:00.000000000 [ovn/openshift-ovn-kubernetes_ovnkube-master-f89lp_nbdb.log] + db_port=9641 | |
| 00:00:00.000000000 [ovn/openshift-ovn-kubernetes_ovnkube-master-f89lp_nbdb.log] + ovn_db_file=/etc/ovn/ov |
Follow the Openshift Azure documentation to setup your Azure account and configure all the prerequisites.
You'll need to create a custom release payload with following PRs:
| apiVersion: machine.openshift.io/v1beta1 | |
| kind: Machine | |
| metadata: | |
| annotations: | |
| machine.openshift.io/instance-state: Running | |
| creationTimestamp: "2021-03-18T17:53:14Z" | |
| finalizers: | |
| - machine.machine.openshift.io | |
| generateName: aojeadual-cwdlb-worker-centralus1- | |
| generation: 2 |
| #!/usr/bin/env bash | |
| set -x | |
| convert_kubelet() { | |
| echo "Converting kubelet on nodes $NODES" | |
| for n in $NODES; do | |
| # /var/lib/kubelet/config.yaml | |
| cat <<EOF >> "/var/lib/kubelet/config.yaml" | |
| featureGates: | |
| IPv6DualStack: true |