build: Use KinD for syncing helm chart resources

scripts/download-binaries.sh

@@ -78,3 +78,5 @@ echo -n "#   kubebuilder:    "; "${dest_dir}/kubebuilder" version
 echo -n "#   helm:           "; "${dest_dir}/helm" version --client --short
 echo -n "#   golangci-lint:  "; "${dest_dir}/golangci-lint" --version
 echo -n "#   go-bindata:     "; "${dest_dir}/go-bindata" -version
+
+"${root_dir}/scripts/download-e2e-binaries.sh"

scripts/sync-up-helm-chart.sh

@@ -22,10 +22,13 @@ source "$(dirname "${BASH_SOURCE}")/util.sh"
 
 ROOT_DIR="$(cd "$(dirname "$0")/.." ; pwd)"
 WORKDIR=$(mktemp -d)
+trap-add 'rm -rf "${WORKDIR}"' EXIT
 NS="${KUBEFED_NAMESPACE:-kube-federation-system}"
 CHART_FEDERATED_PROPAGATION_DIR="${CHART_FEDERATED_PROPAGATION_DIR:-charts/kubefed}"
 TEMP_CRDS_YAML="/tmp/kubefed-crds.yaml"
 
+export PATH=${ROOT_DIR}/bin:${PATH}
+
 OS=`uname`
 SED=sed
 if [ "${OS}" == "Darwin" ];then
@@ -36,13 +39,6 @@ if [ "${OS}" == "Darwin" ];then
   SED=gsed
 fi
 
-# Check for existence of kube-apiserver and etcd binaries in bin directory
-if [[ ! -f ${ROOT_DIR}/bin/etcd || ! -f ${ROOT_DIR}/bin/kube-apiserver ]];
-then
-  echo "Missing 'etcd' and/or 'kube-apiserver' binaries in bin directory. Call './scripts/download-binaries.sh' to download them first"
-  exit 1
-fi
-
 # Remove existing generated crds to ensure that stale content doesn't linger.
 rm -f ./config/crds/*.yaml
 
@@ -59,38 +55,17 @@ done
 
 mv ${TEMP_CRDS_YAML} ./charts/kubefed/charts/controllermanager/crds/crds.yaml
 
-# Generate kubeconfig to access kube-apiserver. It is cleaned when script is done.
-cat <<EOF > ${WORKDIR}/kubeconfig
-apiVersion: v1
-clusters:
-- cluster:
-    server: 127.0.0.1:8080
-  name: development
-contexts:
-- context:
-    cluster: development
-    user: ""
-  name: kubefed
-current-context: ""
-kind: Config
-preferences: {}
-users: []
-EOF
-
-# Start kube-apiserver to generate CRDs
-${ROOT_DIR}/bin/etcd --data-dir ${WORKDIR} --log-output stdout > ${WORKDIR}/etcd.log 2>&1 &
-util::wait-for-condition 'etcd' "curl http://127.0.0.1:2379/version &> /dev/null" 30
-
-${ROOT_DIR}/bin/kube-apiserver --etcd-servers=http://127.0.0.1:2379 --service-cluster-ip-range=10.0.0.0/16 --cert-dir=${WORKDIR} 2> ${WORKDIR}/kube-apiserver.log &
-util::wait-for-condition 'kube-apiserver' "kubectl --kubeconfig ${WORKDIR}/kubeconfig --context kubefed get --raw=/healthz &> /dev/null" 60
+declare -rx KUBECONFIG="${WORKDIR}/kubeconfig"
+kind create cluster --name=kubefed-dev
+trap-add 'kind delete cluster --name=kubefed-dev' EXIT
 
 # Generate YAML templates to enable resource propagation for helm chart.
 echo -n > ${CHART_FEDERATED_PROPAGATION_DIR}/templates/federatedtypeconfig.yaml
 echo -n > ${CHART_FEDERATED_PROPAGATION_DIR}/crds/crds.yaml
 for filename in ./config/enabletypedirectives/*.yaml; do
   full_name=${CHART_FEDERATED_PROPAGATION_DIR}/templates/$(basename $filename)
 
-  ./bin/kubefedctl --kubeconfig ${WORKDIR}/kubeconfig enable -f "${filename}" --kubefed-namespace="${NS}" --host-cluster-context kubefed -o yaml > ${full_name}
+  ./bin/kubefedctl --kubeconfig ${WORKDIR}/kubeconfig enable -f "${filename}" --kubefed-namespace="${NS}" -o yaml > ${full_name}
   $SED -n '/^---/,/^---/p' ${full_name} >> ${CHART_FEDERATED_PROPAGATION_DIR}/templates/federatedtypeconfig.yaml
   $SED -i '$d' ${CHART_FEDERATED_PROPAGATION_DIR}/templates/federatedtypeconfig.yaml
 
@@ -101,7 +76,4 @@ for filename in ./config/enabletypedirectives/*.yaml; do
 done
 
 # Clean kube-apiserver daemons and temporary files
-kill %1 # etcd
-kill %2 # kube-apiserver
-rm -fr ${WORKDIR}
 echo "Helm chart synced successfully"

scripts/util.sh

@@ -119,3 +119,10 @@ function check-command-installed() {
     exit 1
   }
 }
+
+trap-add() {
+  local sig="${2:?Signal required}"
+  hdls="$(trap -p "${sig}" | cut -f2 -d \')"
+  # shellcheck disable=SC2064 # Quotes are required here to properly expand when adding the new trap.
+  trap "${hdls}${hdls:+;}${1:?Handler required}" "${sig}"
+}