Merge pull request #1466 from manuelbuil/coreDNS_nodeLocal

Provide support for nodelocal DNSCache

Dockerfile

@@ -97,7 +97,7 @@ RUN CHART_VERSION="1.9.808"                   CHART_FILE=/charts/rke2-cilium.yam
 RUN CHART_VERSION="v3.19.1-build2021061107"   CHART_FILE=/charts/rke2-canal.yaml          CHART_BOOTSTRAP=true   /charts/build-chart.sh
 RUN CHART_VERSION="v3.1907"                   CHART_FILE=/charts/rke2-calico.yaml         CHART_BOOTSTRAP=true   /charts/build-chart.sh
 RUN CHART_VERSION="v1.0.007"                  CHART_FILE=/charts/rke2-calico-crd.yaml     CHART_BOOTSTRAP=true   /charts/build-chart.sh
-RUN CHART_VERSION="1.16.201-build2021072301"  CHART_FILE=/charts/rke2-coredns.yaml        CHART_BOOTSTRAP=true   /charts/build-chart.sh
+RUN CHART_VERSION="1.16.201-build2021072303"  CHART_FILE=/charts/rke2-coredns.yaml        CHART_BOOTSTRAP=true   /charts/build-chart.sh
 RUN CHART_VERSION="3.34.002"                  CHART_FILE=/charts/rke2-ingress-nginx.yaml  CHART_BOOTSTRAP=false  /charts/build-chart.sh
 RUN CHART_VERSION="v1.21.3-rke2r1-build2021072101" \
     CHART_PACKAGE="rke2-kube-proxy-1.21"      CHART_FILE=/charts/rke2-kube-proxy.yaml     CHART_BOOTSTRAP=true   /charts/build-chart.sh

bundle/bin/rke2-killall.sh

@@ -73,5 +73,15 @@ ip link delete flannel.1
 ip link delete vxlan.calico
 ip link delete cilium_vxlan
 ip link delete cilium_net
+
+#Delete the nodeLocal created objects
+if [ -d /sys/class/net/nodelocaldns ]; then
+  for i in $(ip address show nodelocaldns | grep inet | awk '{print $2}');
+  do
+    iptables-save | grep -v $i | iptables-restore
+  done
+  ip link delete nodelocaldns
+fi
+
 rm -rf /var/lib/cni/
 iptables-save | grep -v KUBE- | grep -v CNI- | grep -v cali- | grep -v cali: | grep -v CILIUM_ | iptables-restore

docs/networking.md

@@ -18,6 +18,41 @@ If you don't install CoreDNS, you will need to install a cluster DNS provider yo
 
 CoreDNS is deployed with the [autoscaler](https://github.com/kubernetes-incubator/cluster-proportional-autoscaler) by default. To disable it or change its config, use the [HelmChartConfig](https://docs.rke2.io/helm/#customizing-packaged-components-with-helmchartconfig) resource.
 
+### NodeLocal DNSCache
+
+[NodeLocal DNSCache](https://kubernetes.io/docs/tasks/administer-cluster/nodelocaldns/) improves the performance by running a dns caching agent on each node. To activate this feature, apply the following HelmChartConfig:
+
+```yaml
+---
+apiVersion: helm.cattle.io/v1
+kind: HelmChartConfig
+metadata:
+  name: rke2-coredns
+  namespace: kube-system
+spec:
+  valuesContent: |-
+    nodelocal:
+      enabled: true
+```
+The helm controller will redeploy coredns with the new config. Please be aware that nodelocal modifies the iptables of the node to intercept DNS traffic. Therefore, activating and then deactivating this feature without redeploying, will cause the DNS service to stop working.
+
+Note that NodeLocal DNSCache must be deployed in ipvs mode if kube-proxy is using that mode. To deploy it in this mode, apply the following HelmChartConfig:
+
+```yaml
+---
+apiVersion: helm.cattle.io/v1
+kind: HelmChartConfig
+metadata:
+  name: rke2-coredns
+  namespace: kube-system
+spec:
+  valuesContent: |-
+    nodelocal:
+      enabled: true
+      ipvs: true
+```
+
+
 ## Nginx Ingress Controller
 
 [nginx-ingress](https://github.com/kubernetes/ingress-nginx) is an Ingress controller powered by NGINX that uses a ConfigMap to store the NGINX configuration.

scripts/airgap/dnsNodeCache-test.yaml

@@ -0,0 +1,9 @@
+apiVersion: helm.cattle.io/v1
+kind: HelmChartConfig
+metadata:
+  name: rke2-coredns
+  namespace: kube-system
+spec:
+  valuesContent: |-
+    nodelocal:
+      enabled: true

scripts/build-images

@@ -15,6 +15,8 @@ xargs -n1 -t docker image pull --quiet << EOF >> build/images-core.txt
     ${REGISTRY}/rancher/hardened-kubernetes:${KUBERNETES_IMAGE_TAG}
     ${REGISTRY}/rancher/hardened-coredns:v1.8.3-build20210720
     ${REGISTRY}/rancher/mirrored-cluster-proportional-autoscaler:1.8.3
+    ${REGISTRY}/rancher/mirrored-k8s-dns-node-cache:1.15.13
+    ${REGISTRY}/rancher/library-busybox:1.32.1
     ${REGISTRY}/rancher/hardened-etcd:${ETCD_VERSION}-${IMAGE_BUILD_VERSION}
     ${REGISTRY}/rancher/hardened-k8s-metrics-server:v0.3.6-${IMAGE_BUILD_VERSION}
     ${REGISTRY}/rancher/klipper-helm:v0.6.1-build20210616

scripts/test-helpers

@@ -68,8 +68,7 @@ pod-ready() {
 }
 export -f pod-ready
 
-# ---
-
+# -- wait-for-services() accepts container names as input and checks if they are ready. It does not check Kubernetes services!
 wait-for-services() {
     for service in $@; do
         while [[ "$(pod-ready $service | sort -u)" != 'true' ]]; do

scripts/test-run-basics

@@ -19,6 +19,7 @@ export WAIT_SERVICES="${all_services[@]}"
 
 start-test() {
     #docker exec $(cat $TEST_DIR/servers/1/metadata/name) check-config || true
+    use-nodelocal-DNSCache
     verify-valid-versions $(cat $TEST_DIR/servers/1/metadata/name)
     verify-airgap-images $(cat $TEST_DIR/{servers,agents}/*/metadata/name)
 }
@@ -44,5 +45,14 @@ verify-airgap-images() {
 }
 export -f verify-airgap-images
 
+
+# -- Enable the nodelocal DNSCache so that its images are used
+use-nodelocal-DNSCache() {
+    local DNS_nodeCache_manifest='scripts/airgap/dnsNodeCache-test.yaml'
+    kubectl apply -f $DNS_nodeCache_manifest
+    wait-for-services node-cache
+}
+export -f use-nodelocal-DNSCache
+
 # --- create a basic cluster and check for valid versions
 LABEL=BASICS run-test