Various fixes to scripts and conf

.gitignore

@@ -1 +1,7 @@
 .*sw[po]
+cache
+data
+dist
+kubeconfig
+config.yaml
+.dls_adm_token

clusters/base.yaml

@@ -27,7 +27,10 @@ vars:
   netmask: 24
   gateway: 172.17.1.8
   dns:
-  - 172.17.1.1
+    #- 172.17.1.1
+  - 208.67.220.220
+
+  dls_base_url: http://172.17.1.8:7606
 
   etcd:
     image: quay.io/coreos/etcd

hosts.incl/master.yaml

@@ -1,4 +1,4 @@
-#!include hosts/common
+#!include hosts.incl/common
 
 labels:
   "node-role.novit.io/master": "true"

hosts.incl/node.yaml

@@ -1,4 +1,4 @@
-#!include common
+#!include hosts.incl/common
 
 labels:
   "node-role.novit.io/node": "true"

hosts/.placeholder

@@ -1,6 +0,0 @@
-ip: 10.xx.xx.xx
-ips:
-- 127.0.0.1
-- ::1
-cluster: master
-group: master

hosts/m1.yaml

@@ -0,0 +1,4 @@
+#!include hosts.incl/master
+
+ip: 172.17.1.100
+cluster: base

kubeconfig

@@ -0,0 +1,19 @@
+apiVersion: v1
+clusters:
+- cluster:
+    server: https://172.17.1.100:6443
+    certificate-authority-data: LS0tLS1CRUdJTiBDRVJUSUZJQ0FURS0tLS0tCk1JSUNNakNDQVpPZ0F3SUJBZ0lVZFM5bU5PenhscGJBdXVWcWYyQjIydy96MzdVd0NnWUlLb1pJemowRUF3UXcKTXpFUk1BOEdBMVVFQ2hNSWJtOTJhWFF1YVc4eEhqQWNCZ05WQkFNVEZVUnBjbVZyZEdsc0lFeHZZMkZzSUZObApjblpsY2pBZUZ3MHlNekExTVRneE1UQXpNREJhRncweU9EQTFNVFl4TVRBek1EQmFNRE14RVRBUEJnTlZCQW9UCkNHNXZkbWwwTG1sdk1SNHdIQVlEVlFRREV4VkVhWEpsYTNScGJDQk1iMk5oYkNCVFpYSjJaWEl3Z1pzd0VBWUgKS29aSXpqMENBUVlGSzRFRUFDTURnWVlBQkFBOXdac3h0U2l2RUhUQW50aUx4WFFkUmdxR05sempRMGYwelowQwpyc3hmSGF1K0xrQ1JZNDFoN29ZY21XdVArK09FUlNOL3BGUjhNL3lCZ1dRR0JBbDd2Z0JhZEtpZ01xcHMyalhXCmxUTmt1dnJJbzZhMitESWpjTWNITUhyUUpNUFpzSm8vWGRwOWIyaW5yM2U4dnBiUjFEeEwwRDJMcy9WaWFZL2EKWEE1akR0ZHp2YU5DTUVBd0RnWURWUjBQQVFIL0JBUURBZ0VHTUE4R0ExVWRFd0VCL3dRRk1BTUJBZjh3SFFZRApWUjBPQkJZRUZIdUpKVjVwN0EzdEhWWDkrb2dkQ1IxZkVFbFBNQW9HQ0NxR1NNNDlCQU1FQTRHTUFEQ0JpQUpDCkFQZHZlOU9nRjFmTjE5T2VjTXc0WEhSVFBGcmMvVHp6SFA0WmlRQVRKdUlLMS9hVEozK1k4QUd5b3ByNElXbmsKUWZZeWtVVmpFREJIVTFLN1JWU1h3YjdJQWtJQnd5K2FORVd6N0hxZE5QQTRKZXV2M3ZKUGFXdU9vRElERWxoMwora3o0ZjFpUkR5QVFKbXZta2dxRzA1M214RmRZU2VIU0NpQ2hhN242RG5kUTgveDYwODA9Ci0tLS0tRU5EIENFUlRJRklDQVRFLS0tLS0K
+  name: localconfig
+contexts:
+- context:
+    cluster: localconfig
+    namespace: kube-system
+    user: admin
+  name: localconfig
+current-context: localconfig
+kind: Config
+preferences: {}
+users:
+- name: admin
+  user:
+    token: DU3ITMWLQN2TVGVDTWIURJ42S4

scripts/.common

@@ -0,0 +1,128 @@
+#!/bin/bash
+
+## Vars
+
+tknfile=".dls_adm_token"
+
+## Helper funcs
+
+
+pinfo() {
+  echo -e "\e[32m$@\e[39m"
+}
+
+perror() {
+  echo -e "\e[31m$@\e[39m"
+  exit 1
+}
+
+dls() {
+  path=$1
+  shift
+  if [ -z "$DLS_ADM_TOKEN" ]; then
+    test -f $tknfile && DLS_ADM_TOKEN=$(cat $tknfile)
+  fi
+  curl -s -H 'Content-Type: application/json' -H 'Authorization: '$DLS_ADM_TOKEN http://127.0.0.1:7606${path} "$@"
+}
+
+kctl() {
+  kubectl --kubeconfig $ctxdir/kubeconfig "$@"
+}
+
+extract_var() {
+  where=$1
+  what=$2
+  grep -rh " $what: " $ctxdir/$where/*.yaml|awk '{print $2}'
+}
+
+## Run funcs
+
+check_root() {
+  [ $UID != 0 ] && perror This program needs to be run as root. Aborting...
+}
+
+prereqs() {
+  set -ex
+  cd "$(dirname $0)/.."
+  ctxdir="$PWD"
+}
+
+check_conf() {
+  all_clusters=$(ls $ctxdir/clusters|wc -l)
+  if [ "$all_clusters" != "1" ]; then
+    perror "Those helper scripts are not capable of running several clusters at once, check your configuration. Aborting..."
+  fi
+}
+
+fresh_start() {
+  rm -rf $ctxdir/secrets
+}
+
+unlock_store() {
+  # Unlock DLS store after checking if online
+  #
+  if [ -f $tknfile ]; then
+    DLS_ADM_TOKEN="$(cat $tknfile)"
+  fi
+  store_state=$(curl -H 'Content-Type: application/json' -H "Authorization: $DLS_ADM_TOKEN" -sw %{http_code} localhost:7606/hosts -o /dev/null)
+  if [ "$store_state" == "000" ] ; then
+    perror "Direktil Local Server seems not up, please fix. Aborting."
+  elif [ "$store_state" == "200" ] ; then 
+    pinfo "Direktil Local Server store already unlocked"
+  else
+    pinfo "Unlocking the DLS store ..."
+    DLS_ADM_TOKEN=$(dls /public/unlock-store -d "\"${DLS_UNLOCK_TOKEN}\""|tr -d \")
+    pinfo "Admin access token is $DLS_ADM_TOKEN"
+    echo $DLS_ADM_TOKEN > $tknfile
+    chmod 444 $tknfile
+  fi
+}
+
+
+get_hosts() {
+  hosts_files=$(basename $ctxdir/hosts/*.yaml|sed 's/.yaml//')
+  for h in ${hosts_files}; do
+    ip=$(grep ip: $ctxdir/hosts/${h}.yaml|awk '{print $2}')
+    hosts[$h]="$ip"
+  done
+}
+
+get_parts() {
+  for host in ${!hosts[*]}; do
+    mkdir -p $ctxdir/data/$host
+    for part in kernel initrd-v2
+    do
+      partfile=$ctxdir/data/$host/$part
+      test -f $partfile || dls /hosts/$host/$part -o $partfile
+    done
+    diskfile=$ctxdir/data/$host/disk
+    test -f $diskfile || truncate -s ${QEMU_DISK_SIZE:-30G} $diskfile
+  done
+}
+
+
+create_kubeconfig() {
+  if test -f $ctxdir/kubeconfig; then
+    pinfo "kubeconfig file detected in config dir, won't overwrite... remove it for an update."
+    return
+  fi
+  adm_token=$(dls /clusters/base/tokens/admin)
+  ca_cert=$(dls /clusters/base/CAs/cluster/certificate|base64 -w0)
+  vip=$(extract_var clusters public_vip)
+  vip_api_port=$(extract_var clusters api_port)
+  pinfo "Writing new kubeconfig conf in $ctxdir directory, you may want to move it to \~/.kube/ directory for usability"
+  sed -e "s/__VIP_IP__/$vip/" \
+      -e "s/__VIP_API_PORT__/$vip_api_port/" \
+      -e "s/__CA_CERT__/$ca_cert/" \
+      -e "s/__ADM_TOKEN__/$adm_token/" \
+      scripts/.template.kubeconfig > $ctxdir/kubeconfig
+  chmod 444 $ctxdir/kubeconfig
+}
+
+clean() {
+  set +e
+  sudo iptables -t nat -D POSTROUTING -j MASQUERADE -s $QEMU_BR_IP \! -o $QEMU_BR_NAME &>/dev/null
+  sudo iptables -D FORWARD -o $QEMU_BR_NAME -m conntrack --ctstate RELATED,ESTABLISHED -j ACCEPT &>/dev/null
+  sudo iptables -D FORWARD -j ACCEPT -i $QEMU_BR_NAME &>/dev/null
+}
+

scripts/.template.kubeconfig

@@ -0,0 +1,19 @@
+apiVersion: v1
+clusters:
+- cluster:
+    server: https://__VIP_IP__:__VIP_API_PORT__
+    certificate-authority-data: __CA_CERT__ 
+  name: localconfig
+contexts:
+- context:
+    cluster: localconfig
+    namespace: kube-system
+    user: admin
+  name: localconfig
+current-context: localconfig
+kind: Config
+preferences: {}
+users:
+- name: admin
+  user:
+    token: __ADM_TOKEN__

scripts/0.start_dls.sh

@@ -0,0 +1,49 @@
+#!/bin/bash
+#
+D2C_CTR_NAME=dir2config
+DLS_CTR_NAME=dls
+# Admin token to unlock the DLS store
+DLS_UNLOCK_TOKEN=changeme
+#
+#
+D2C_IMG=novit.tech/direktil/local-server:b6fa941
+DLS_IMG=novit.tech/direktil/local-server:b6fa941
+
+prereqs_dls() {
+  command -v docker 1>/dev/null || perror "Docker is needed, please install it and run again."
+  systemctl is-active docker &>/dev/null || systemctl start docker
+  docker pull $DLS_IMG
+}
+
+dir2config() {
+  pinfo "Generating config.yaml from Direktil configuration"
+  docker run --rm --name $D2C_CTR_NAME \
+    -v .:/var/lib/direktil -w /var/lib/direktil \
+    --entrypoint=/bin/dkl-dir2config \
+    $D2C_IMG
+}
+
+start_store() {
+  if docker ps|grep " $DLS_CTR_NAME$" ; then
+    pinfo "Container $DLS_CTR_NAME seems already running"
+    return
+  fi
+  docker run --rm --name $DLS_CTR_NAME -p 7606:7606 \
+      -e http_proxy=$http_proxy \
+      -e https_proxy=$https_proxy \
+      -e HTTP_PROXY=$HTTP_PROXY \
+      -e HTTPS_PROXY=$HTTPS_PROXY \
+      -v .:/var/lib/direktil \
+      $DLS_IMG &
+#      -auto-unlock 'N0v!T'
+  sleep 2
+}
+
+source $(dirname $0)/.common
+check_root
+prereqs_dls
+prereqs
+dir2config
+start_store
+unlock_store
+

scripts/1.qemu.sh

@@ -0,0 +1,76 @@
+#! /bin/sh
+#
+# This collection of scripts aims to install a NOVIT cluster easily, with help of QEMU
+#
+#
+# QEMU local bridge name. If you specificy a custom name, you may have to configure qemu bridge helper to allow it
+QEMU_BR_NAME=virbr0
+#
+# QEMU VM default disk size
+QEMU_DISK_SIZE=30G
+# Allocated CPUs to QEMU VMs
+QEMU_VM_CPU=4
+# Allocated Memory to QEMU VMs
+QEMU_VM_MEM=8096
+################
+
+
+## QEMU functions
+#
+prereqs_qemu() {
+  for com in qemu-system-x86_64 truncate docker iptables ; do
+    command -v $com 1>/dev/null || error Command $com not found, please install it. Aborting...
+  done
+}
+setup_network_qemu() {
+  if ! ip li show $QEMU_BR_NAME ; then
+    ip li add name $QEMU_BR_NAME type bridge
+    ip li set $QEMU_BR_NAME up
+  fi
+  QEMU_BR_IP=$(extract_var clusters gateway)
+  QEMU_BR_MASK=$(extract_var clusters netmask)
+  if [ $(echo $QEMU_BR_IP | wc -w) -gt 1 ]; then
+    perror "More than one cluster is configured, not compatible with our quick-start setup, exiting"
+  fi
+  pinfo "Using detected gateway IP $QEMU_BR_IP for bridge $QEMU_BR_NAME"
+  if ! ip a show dev $QEMU_BR_NAME | grep $QEMU_BR_IP ; then
+    ip a add $QEMU_BR_IP/$QEMU_BR_MASK dev $QEMU_BR_NAME
+    sudo iptables -t nat -I POSTROUTING -j MASQUERADE -s $QEMU_BR_IP/$QEMU_BR_MASK \! -o $QEMU_BR_NAME
+    sudo iptables -I FORWARD -o $QEMU_BR_NAME -m conntrack --ctstate RELATED,ESTABLISHED -j ACCEPT
+    sudo iptables -I FORWARD -j ACCEPT -i $QEMU_BR_NAME
+  fi
+}
+run_qemu() {
+  id=1
+  for host in ${!hosts[*]}; do
+    ip route show |grep "${hosts[$host]} dev $QEMU_BR_NAME" ||\
+        ip route add ${hosts[$host]} dev $QEMU_BR_NAME
+    if test -f $ctxdir/data/$host/pid ; then
+      pinfo "Detected a pid file, killing process in case VM was already started"
+      kill $(cat $ctxdir/data/$host/pid) && sleep 1
+    fi
+    pinfo "Starting host $host with ip ${hosts[$host]}"
+    qemu-system-x86_64 -enable-kvm -smp $QEMU_VM_CPU -m $QEMU_VM_MEM \
+      -nic bridge,br=$QEMU_BR_NAME,mac=42:42:42:42:42:0${id} \
+      -kernel $ctxdir/data/$host/kernel -initrd $ctxdir/data/$host/initrd-v2 -vga qxl \
+      -drive format=raw,file=$ctxdir/data/$host/disk &
+    echo $! >$ctxdir/data/$host/pid
+    ((++id))
+  done
+  pinfo "$(ls $ctxdir/data/*/pid|wc -w) host(s) have been started"
+}
+# # # # # # # #
+
+source $(dirname $0)/.common
+check_root
+prereqs
+check_conf
+#fresh_start
+trap clean SIGINT SIGTERM SIGKILL
+declare -A hosts
+setup_network_qemu
+get_hosts
+get_parts
+run_qemu
+#clean
+

scripts/2.first_start_k8s.sh

@@ -0,0 +1,88 @@
+#!/bin/bash
+#
+# Starting control plane
+
+
+prereqs_control_plane() {
+  for com in kubectl ; do
+    command -v $com 1>/dev/null || perror "Command $com is not installed, aborting..."
+  done
+}
+prereqs_addons() {
+  for com in curl kubectl ; do
+    command -v $com 1>/dev/null || perror "Command $com is not installed, aborting..."
+  done
+  cluster=$(basename $ctxdir/clusters/*.yaml|sed 's/.yaml//')
+}
+
+checkup() {
+  for host in ${!hosts[*]}; do
+    tries=3
+    while :
+    do
+      pinfo "Checking availability of node $host..."
+      ssh root@${hosts[$host]} <<< true &>/dev/null
+      if [ $? == 0 ]; then
+        pinfo "VM $host is up!"
+        break
+      else
+        ((tries--))
+        if [ $tries < 1 ]; then
+          pinfo "Timeout waiting for node detection, please investigate why node $host is not up by now"
+          break
+        fi
+        sleep 30
+      fi
+    done
+  done
+}
+
+start_control_plane() {
+  for host in ${!hosts[*]}; do
+    ssh root@${hosts[$host]} << EOF
+      if ls /etc/kubernetes/manifests.static/* &>/dev/null ; then
+        mv /etc/kubernetes/manifests.static/* /var/lib/kubelet/manifests/
+      fi
+EOF
+  done
+}
+
+install_addons() {
+  body='{"Kind":"cluster","Name":"'$cluster'","Assets":["addons"]}'
+  download_id=$(dls /authorize-download -d "$body"|tr -d \")
+  dls /public/downloads/${download_id}/addons |\
+      kctl apply -f -
+}
+
+approve_kubelet_certificates() {
+  tries=5
+  nodes_num=$(kctl get node -oname|wc -l)
+  while [ "$nodes_num" != "${#hosts[*]}" ] ; do
+    pinfo "Waiting for certificates requests to be created by Kubelet when it's ready... ($tries/5)"
+    sleep 60s
+    csrs="$(kctl get csr|awk '/Pending/ {print $1}')"
+    if [ "$csrs" != "" ]; then
+      kctl certificate approve $csrs
+    fi
+    ((tries--))
+    if [ "$tries" < 1 ]; then
+      pinfo "Timeout waiting for kubelet certificates creation, please investigate why all nodes are not up by now"
+      break
+    fi
+  done
+  pinfo "All kubelets ($nodes_num) are up, enjoy !"
+}
+
+source $(dirname $0)/.common
+prereqs
+prereqs_control_plane
+prereqs_addons
+check_conf
+unlock_store
+declare -A hosts && get_hosts
+checkup
+start_control_plane
+create_kubeconfig
+install_addons
+approve_kubelet_certificates # clients and serving certs
+

static-pods/master/etcd.yaml

@@ -40,7 +40,7 @@ spec:
           - name:  ETCD_INITIAL_CLUSTER
             value: {{ range $i, $host := hosts_by_group "master" }}{{ if gt $i 0 }},{{end}}{{$host.name}}=https://{{$host.ip}}:2380{{end}}
           - name:  ETCD_INITIAL_CLUSTER_STATE
-            value: {{.etcd.cluster_state}}
+            value: {{ .vars.etcd.cluster_state }}
           - name:  ETCD_INITIAL_CLUSTER_TOKEN
             value: '{{ token "etcd-initial-cluster" }}'
           - name:  ETCDCTL_ENDPOINTS