Factor out ceph setup in prep for ceph-helm implementation.

Change-Id: I121204e90550ccb6dcfc0b084ace6230a8d6b8f5
Signed-off-by: Bryan Sullivan <bryan.sullivan@att.com>

3 files changed, 244 insertions, 165 deletions

diff --git a/tools/kubernetes/ceph-baremetal.sh b/tools/kubernetes/ceph-baremetal.sh
new file mode 100644
index 0000000..55d2a7f
--- /dev/null
+++ b/tools/kubernetes/ceph-baremetal.sh
@@ -0,0 +1,199 @@
+#!/bin/bash
+# Copyright 2017 AT&T Intellectual Property, Inc
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+#. What this is: script to setup a Ceph-based SDS (Software Defined Storage)
+#. service for a kubernetes cluster, directly on the master and worker nodes.
+#. Prerequisites:
+#. - Ubuntu xenial server for master and agent nodes
+#. - key-based auth setup for ssh/scp between master and agent nodes
+#. - 192.168.0.0/16 should not be used on your server network interface subnets
+#. Usage:
+#  Intended to be called from k8s-cluster.sh in this folder. To run directly:
+#. $ bash ceph-baremetal.sh "<nodes>" <cluster-net> <public-net> [ceph_dev]
+#.     nodes: space-separated list of ceph node IPs
+#.     cluster-net: CIDR of ceph cluster network e.g. 10.0.0.1/24
+#.     public-net: CIDR of public network
+#.     ceph_dev: disk to use for ceph. ***MUST NOT BE USED FOR ANY OTHER PURPOSE***
+#.               if not provided, ceph data will be stored on osd nodes in /ceph
+#.
+#. Status: work in progress, incomplete
+#
+
+function setup_ceph() {
+  node_ips=$1
+  cluster_net=$2
+  public_net=$3
+  ceph_dev=$4
+  echo "${FUNCNAME[0]}: Deploying ceph-mon on localhost $HOSTNAME"
+  echo "${FUNCNAME[0]}: Deploying ceph-osd on nodes $node_ips"
+  echo "${FUNCNAME[0]}: Setting cluster-network=$cluster_net and public-network=$public_net"
+  mon_ip=$(ip route get 8.8.8.8 | awk '{print $NF; exit}')
+  all_nodes="$mon_ip $node_ips"
+  # Also caches the server fingerprints so ceph-deploy does not prompt the user
+  # Note this loop may be partially redundant with the ceph-deploy steps below
+  for node_ip in $all_nodes; do
+    echo "${FUNCNAME[0]}: Install ntp and ceph on $node_ip"
+    ssh -x -o StrictHostKeyChecking=no ubuntu@$node_ip <<EOF
+sudo timedatectl set-ntp no
+wget -q -O- 'https://download.ceph.com/keys/release.asc' | sudo apt-key add -
+echo deb https://download.ceph.com/debian/ $(lsb_release -sc) main | sudo tee /etc/apt/sources.list.d/ceph.list
+sudo apt update
+sudo apt-get install -y ntp ceph ceph-deploy
+EOF
+  done
+
+  # per http://docs.ceph.com/docs/master/start/quick-ceph-deploy/
+  # also https://upcommons.upc.edu/bitstream/handle/2117/101816/Degree_Thesis_Nabil_El_Alami.pdf#vote +1
+  echo "${FUNCNAME[0]}: Create ceph config folder ~/ceph-cluster"
+  mkdir ~/ceph-cluster
+  cd ~/ceph-cluster
+
+  echo "${FUNCNAME[0]}: Create new cluster with $HOSTNAME as initial ceph-mon node"
+  ceph-deploy new --cluster-network $cluster_net --public-network $public_net --no-ssh-copykey $HOSTNAME
+  # Update conf per recommendations of http://docs.ceph.com/docs/jewel/rados/configuration/filesystem-recommendations/
+  cat <<EOF >>ceph.conf
+osd max object name len = 256
+osd max object namespace len = 64
+EOF
+  cat ceph.conf
+
+  echo "${FUNCNAME[0]}: Deploy ceph packages on other nodes"
+  ceph-deploy install $mon_ip $node_ips
+
+  echo "${FUNCNAME[0]}: Deploy the initial monitor and gather the keys"
+  ceph-deploy mon create-initial
+
+  if [[ "x$ceph_dev" == "x" ]]; then
+    n=1
+    for node_ip in $node_ips; do
+      echo "${FUNCNAME[0]}: Prepare ceph OSD on node $node_ip"
+      echo "$node_ip ceph-osd$n" | sudo tee -a /etc/hosts
+      # Using ceph-osd$n here avoids need for manual acceptance of the new server hash
+      ssh -x -o StrictHostKeyChecking=no ubuntu@ceph-osd$n <<EOF
+echo "$node_ip ceph-osd$n" | sudo tee -a /etc/hosts
+sudo mkdir /ceph && sudo chown -R ceph:ceph /ceph
+EOF
+      ceph-deploy osd prepare ceph-osd$n:/ceph
+      ceph-deploy osd activate ceph-osd$n:/ceph
+      ((n++))
+    done
+  else
+    echo "${FUNCNAME[0]}: Deploy OSDs"
+    for node_ip in $node_ips; do
+      echo "${FUNCNAME[0]}: Create ceph osd on $node_ip using $ceph_dev"
+      ceph-deploy osd create $node_ip:$ceph_dev
+    done
+  fi
+
+  echo "${FUNCNAME[0]}: Copy the config file and admin key to the admin node and OSD nodes"
+  ceph-deploy admin $mon_ip $node_ips
+
+  echo "${FUNCNAME[0]}: Check the cluster health"
+  sudo ceph health
+  sudo ceph -s
+
+  # per https://crondev.com/kubernetes-persistent-storage-ceph/ and https://github.com/kubernetes/kubernetes/issues/38923
+  # rbd  is not included in default kube-controller-manager... use attcomdev version
+  sudo sed -i -- 's~gcr.io/google_containers/kube-controller-manager-amd64:.*~quay.io/attcomdev/kube-controller-manager:v1.7.3~' /etc/kubernetes/manifests/kube-controller-manager.yaml
+  if [[ $(sudo grep -c attcomdev/kube-controller-manager /etc/kubernetes/manifests/kube-controller-manager.yaml) == 0 ]]; then
+    echo "${FUNCNAME[0]}: Problem patching /etc/kubernetes/manifests/kube-controller-manager.yaml... script update needed"
+    exit 1
+  fi
+  mgr=$(kubectl get pods --all-namespaces | grep kube-controller-manager | awk '{print $4}')
+  while [[ "$mgr" != "Running" ]]; do
+    echo "${FUNCNAME[0]}: kube-controller-manager status is $mgr. Waiting 60 seconds for it to be 'Running'"
+    sleep 60
+    mgr=$(kubectl get pods --all-namespaces | grep kube-controller-manager | awk '{print $4}')
+  done
+  echo "${FUNCNAME[0]}: kube-controller-manager status is $mgr"
+
+  echo "${FUNCNAME[0]}: Create Ceph admin secret"
+  admin_key=$(sudo ceph auth get-key client.admin)
+  kubectl create secret generic ceph-secret-admin --from-literal=key="$admin_key" --namespace=kube-system --type=kubernetes.io/rbd
+
+  echo "${FUNCNAME[0]}: Create rdb storageClass 'slow'"
+  cat <<EOF >/tmp/ceph-sc.yaml
+apiVersion: storage.k8s.io/v1
+kind: StorageClass
+metadata:
+   name: slow
+provisioner: kubernetes.io/rbd
+parameters:
+    monitors: $mon_ip:6789
+    adminId: admin
+    adminSecretName: ceph-secret-admin
+    adminSecretNamespace: "kube-system"
+    pool: kube
+    userId: kube
+    userSecretName: ceph-secret-user
+EOF
+  # TODO: find out where in the above ~/.kube folders became owned by root
+  sudo chown -R ubuntu:ubuntu ~/.kube/*
+  kubectl create -f /tmp/ceph-sc.yaml
+
+  echo "${FUNCNAME[0]}: Create storage pool 'kube'"
+  # https://github.com/kubernetes/examples/blob/master/staging/persistent-volume-provisioning/README.md method
+  sudo ceph osd pool create kube 32 32
+
+  echo "${FUNCNAME[0]}: Authorize client 'kube' access to pool 'kube'"
+  sudo ceph auth get-or-create client.kube mon 'allow r' osd 'allow rwx pool=kube'
+
+  echo "${FUNCNAME[0]}: Create ceph-secret-user secret in namespace 'default'"
+  kube_key=$(sudo ceph auth get-key client.kube)
+  kubectl create secret generic ceph-secret-user --from-literal=key="$kube_key" --namespace=default --type=kubernetes.io/rbd
+  # A similar secret must be created in other namespaces that intend to access the ceph pool
+
+  # Per https://github.com/kubernetes/examples/blob/master/staging/persistent-volume-provisioning/README.md
+
+  echo "${FUNCNAME[0]}: Create andtest a persistentVolumeClaim"
+  cat <<EOF >/tmp/ceph-pvc.yaml
+{
+  "kind": "PersistentVolumeClaim",
+  "apiVersion": "v1",
+  "metadata": {
+    "name": "claim1",
+    "annotations": {
+        "volume.beta.kubernetes.io/storage-class": "slow"
+    }
+  },
+  "spec": {
+    "accessModes": [
+      "ReadWriteOnce"
+    ],
+    "resources": {
+      "requests": {
+        "storage": "3Gi"
+      }
+    }
+  }
+}
+EOF
+  kubectl create -f /tmp/ceph-pvc.yaml
+  while [[ "x$(kubectl get pvc -o jsonpath='{.status.phase}' claim1)" != "xBound" ]]; do
+    echo "${FUNCNAME[0]}: Waiting for pvc claim1 to be 'Bound'"
+    kubectl describe pvc
+    sleep 10
+  done
+  echo "${FUNCNAME[0]}: pvc claim1 successfully bound to $(kubectl get pvc -o jsonpath='{.spec.volumeName}' claim1)"
+  kubectl get pvc
+  kubectl delete pvc claim1
+  kubectl describe pods
+}
+
+if [[ "$1" .ne "" ]]; then
+  setup_ceph "$1" $2 $3 $4
+else
+  grep '#. ' $0
+fi
diff --git a/tools/kubernetes/ceph-helm.sh b/tools/kubernetes/ceph-helm.sh
new file mode 100644
index 0000000..96a310c
--- /dev/null
+++ b/tools/kubernetes/ceph-helm.sh
@@ -0,0 +1,32 @@
+#!/bin/bash
+# Copyright 2017 AT&T Intellectual Property, Inc
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+#. What this is: script to setup a Ceph-based SDS (Software Defined Storage)
+#. service for a kubernetes cluster, using Helm as deployment tool.
+#. Prerequisites:
+#. - Ubuntu xenial server for master and agent nodes
+#. - key-based auth setup for ssh/scp between master and agent nodes
+#. - 192.168.0.0/16 should not be used on your server network interface subnets
+#. Usage:
+#  Intended to be called from k8s-cluster.sh in this folder. To run directly:
+#. $ bash ceph-helm.sh "<nodes>" <cluster-net> <public-net> [ceph_dev]
+#.     nodes: space-separated list of ceph node IPs
+#.     cluster-net: CIDR of ceph cluster network e.g. 10.0.0.1/24
+#.     public-net: CIDR of public network
+#.     ceph_dev: disk to use for ceph. ***MUST NOT BE USED FOR ANY OTHER PURPOSE***
+#.               if not provided, ceph data will be stored on osd nodes in /ceph
+#.
+#. Status: work in progress, incomplete
+#
diff --git a/tools/kubernetes/k8s-cluster.sh b/tools/kubernetes/k8s-cluster.sh
index 3d896bf..015e217 100644
--- a/tools/kubernetes/k8s-cluster.sh
+++ b/tools/kubernetes/k8s-cluster.sh
@@ -24,17 +24,18 @@
 #. $ bash k8s-cluster.sh master
 #. $ bash k8s-cluster.sh agents "<nodes>"
 #.     nodes: space-separated list of ceph node IPs
-#. $ bash k8s-cluster.sh ceph "<nodes>" <cluster-net> <public-net> [ceph_dev]
+#. $ bash k8s-cluster.sh ceph "<nodes>" <cluster-net> <public-net> <ceph-mode> [ceph_dev]
 #.     nodes: space-separated list of ceph node IPs
 #.     cluster-net: CIDR of ceph cluster network e.g. 10.0.0.1/24
 #.     public-net: CIDR of public network
+#.     ceph-mode: "helm" or "baremetal"
 #.     ceph_dev: disk to use for ceph. ***MUST NOT BE USED FOR ANY OTHER PURPOSE***
 #.               if not provided, ceph data will be stored on osd nodes in /ceph
 #. $ bash k8s-cluster.sh helm
 #.     Setup helm as app kubernetes orchestration tool
 #. $ bash k8s-cluster.sh demo
 #.     Install helm charts for mediawiki and dokuwiki
-#. $ bash k8s-cluster.sh all "<nodes>" <cluster-net> <public-net> [ceph_dev]
+#. $ bash k8s-cluster.sh all "<nodes>" <cluster-net> <public-net> <ceph-mode> [ceph_dev]
 #.     Runs all the steps above
 #.
 #. Status: work in progress, incomplete
@@ -122,167 +123,6 @@ function setup_k8s_agents() {
   echo "${FUNCNAME[0]}: Cluster is ready when all nodes in the output of 'kubectl get nodes' show as 'Ready'."
 }
 
-function setup_ceph() {
-  node_ips=$1
-  cluster_net=$2
-  public_net=$3
-  ceph_dev=$4
-  echo "${FUNCNAME[0]}: Deploying ceph-mon on localhost $HOSTNAME"
-  echo "${FUNCNAME[0]}: Deploying ceph-osd on nodes $node_ips"
-  echo "${FUNCNAME[0]}: Setting cluster-network=$cluster_net and public-network=$public_net"
-  mon_ip=$(ip route get 8.8.8.8 | awk '{print $NF; exit}')
-  all_nodes="$mon_ip $node_ips"
-  # Also caches the server fingerprints so ceph-deploy does not prompt the user
-  # Note this loop may be partially redundant with the ceph-deploy steps below
-  for node_ip in $all_nodes; do
-    echo "${FUNCNAME[0]}: Install ntp and ceph on $node_ip"
-    ssh -x -o StrictHostKeyChecking=no ubuntu@$node_ip <<EOF
-sudo timedatectl set-ntp no
-wget -q -O- 'https://download.ceph.com/keys/release.asc' | sudo apt-key add -
-echo deb https://download.ceph.com/debian/ $(lsb_release -sc) main | sudo tee /etc/apt/sources.list.d/ceph.list
-sudo apt update
-sudo apt-get install -y ntp ceph ceph-deploy
-EOF
-  done
-
-  # per http://docs.ceph.com/docs/master/start/quick-ceph-deploy/
-  # also https://upcommons.upc.edu/bitstream/handle/2117/101816/Degree_Thesis_Nabil_El_Alami.pdf#vote +1
-  echo "${FUNCNAME[0]}: Create ceph config folder ~/ceph-cluster"
-  mkdir ~/ceph-cluster
-  cd ~/ceph-cluster
-
-  echo "${FUNCNAME[0]}: Create new cluster with $HOSTNAME as initial ceph-mon node"
-  ceph-deploy new --cluster-network $cluster_net --public-network $public_net --no-ssh-copykey $HOSTNAME
-  # Update conf per recommendations of http://docs.ceph.com/docs/jewel/rados/configuration/filesystem-recommendations/
-  cat <<EOF >>ceph.conf
-osd max object name len = 256
-osd max object namespace len = 64
-EOF
-  cat ceph.conf
-
-  echo "${FUNCNAME[0]}: Deploy ceph packages on other nodes"
-  ceph-deploy install $mon_ip $node_ips
-
-  echo "${FUNCNAME[0]}: Deploy the initial monitor and gather the keys"
-  ceph-deploy mon create-initial
-
-  if [[ "x$ceph_dev" == "x" ]]; then
-    n=1
-    for node_ip in $node_ips; do
-      echo "${FUNCNAME[0]}: Prepare ceph OSD on node $node_ip"
-      echo "$node_ip ceph-osd$n" | sudo tee -a /etc/hosts
-      # Using ceph-osd$n here avoids need for manual acceptance of the new server hash
-      ssh -x -o StrictHostKeyChecking=no ubuntu@ceph-osd$n <<EOF
-echo "$node_ip ceph-osd$n" | sudo tee -a /etc/hosts
-sudo mkdir /ceph && sudo chown -R ceph:ceph /ceph
-EOF
-      ceph-deploy osd prepare ceph-osd$n:/ceph
-      ceph-deploy osd activate ceph-osd$n:/ceph
-      ((n++))
-    done
-  else
-    echo "${FUNCNAME[0]}: Deploy OSDs"
-    for node_ip in $node_ips; do
-      echo "${FUNCNAME[0]}: Create ceph osd on $node_ip using $ceph_dev"
-      ceph-deploy osd create $node_ip:$ceph_dev
-    done
-  fi
-
-  echo "${FUNCNAME[0]}: Copy the config file and admin key to the admin node and OSD nodes"
-  ceph-deploy admin $mon_ip $node_ips
-
-  echo "${FUNCNAME[0]}: Check the cluster health"
-  sudo ceph health
-  sudo ceph -s
-
-  # per https://crondev.com/kubernetes-persistent-storage-ceph/ and https://github.com/kubernetes/kubernetes/issues/38923
-  # rbd  is not included in default kube-controller-manager... use attcomdev version
-  sudo sed -i -- 's~gcr.io/google_containers/kube-controller-manager-amd64:.*~quay.io/attcomdev/kube-controller-manager:v1.7.3~' /etc/kubernetes/manifests/kube-controller-manager.yaml
-  if [[ $(sudo grep -c attcomdev/kube-controller-manager /etc/kubernetes/manifests/kube-controller-manager.yaml) == 0 ]]; then
-    echo "${FUNCNAME[0]}: Problem patching /etc/kubernetes/manifests/kube-controller-manager.yaml... script update needed"
-    exit 1
-  fi
-  mgr=$(kubectl get pods --all-namespaces | grep kube-controller-manager | awk '{print $4}')
-  while [[ "$mgr" != "Running" ]]; do
-    echo "${FUNCNAME[0]}: kube-controller-manager status is $mgr. Waiting 60 seconds for it to be 'Running'"
-    sleep 60
-    mgr=$(kubectl get pods --all-namespaces | grep kube-controller-manager | awk '{print $4}')
-  done
-  echo "${FUNCNAME[0]}: kube-controller-manager status is $mgr"
-
-  echo "${FUNCNAME[0]}: Create Ceph admin secret"
-  admin_key=$(sudo ceph auth get-key client.admin)
-  kubectl create secret generic ceph-secret-admin --from-literal=key="$admin_key" --namespace=kube-system --type=kubernetes.io/rbd
-
-  echo "${FUNCNAME[0]}: Create rdb storageClass 'slow'"
-  cat <<EOF >/tmp/ceph-sc.yaml
-apiVersion: storage.k8s.io/v1
-kind: StorageClass
-metadata:
-   name: slow
-provisioner: kubernetes.io/rbd
-parameters:
-    monitors: $mon_ip:6789
-    adminId: admin
-    adminSecretName: ceph-secret-admin
-    adminSecretNamespace: "kube-system"
-    pool: kube
-    userId: kube
-    userSecretName: ceph-secret-user
-EOF
-  # TODO: find out where in the above ~/.kube folders became owned by root
-  sudo chown -R ubuntu:ubuntu ~/.kube/*
-  kubectl create -f /tmp/ceph-sc.yaml
-
-  echo "${FUNCNAME[0]}: Create storage pool 'kube'"
-  # https://github.com/kubernetes/examples/blob/master/staging/persistent-volume-provisioning/README.md method
-  sudo ceph osd pool create kube 32 32
-
-  echo "${FUNCNAME[0]}: Authorize client 'kube' access to pool 'kube'"
-  sudo ceph auth get-or-create client.kube mon 'allow r' osd 'allow rwx pool=kube'
-
-  echo "${FUNCNAME[0]}: Create ceph-secret-user secret in namespace 'default'"
-  kube_key=$(sudo ceph auth get-key client.kube)
-  kubectl create secret generic ceph-secret-user --from-literal=key="$kube_key" --namespace=default --type=kubernetes.io/rbd
-  # A similar secret must be created in other namespaces that intend to access the ceph pool
-
-  # Per https://github.com/kubernetes/examples/blob/master/staging/persistent-volume-provisioning/README.md
-
-  echo "${FUNCNAME[0]}: Create andtest a persistentVolumeClaim"
-  cat <<EOF >/tmp/ceph-pvc.yaml
-{
-  "kind": "PersistentVolumeClaim",
-  "apiVersion": "v1",
-  "metadata": {
-    "name": "claim1",
-    "annotations": {
-        "volume.beta.kubernetes.io/storage-class": "slow"
-    }
-  },
-  "spec": {
-    "accessModes": [
-      "ReadWriteOnce"
-    ],
-    "resources": {
-      "requests": {
-        "storage": "3Gi"
-      }
-    }
-  }
-}
-EOF
-  kubectl create -f /tmp/ceph-pvc.yaml
-  while [[ "x$(kubectl get pvc -o jsonpath='{.status.phase}' claim1)" != "xBound" ]]; do
-    echo "${FUNCNAME[0]}: Waiting for pvc claim1 to be 'Bound'"
-    kubectl describe pvc
-    sleep 10
-  done
-  echo "${FUNCNAME[0]}: pvc claim1 successfully bound to $(kubectl get pvc -o jsonpath='{.spec.volumeName}' claim1)"
-  kubectl get pvc
-  kubectl delete pvc claim1
-  kubectl describe pods
-}
-
 function wait_for_service() {
   echo "${FUNCNAME[0]}: Waiting for service $1 to be available"
   pod=$(kubectl get pods --namespace default | awk "/$1/ { print \$1 }")
@@ -406,6 +246,14 @@ function setup_helm() {
   # e.g. helm install stable/dokuwiki
 }
 
+function setup_ceph() {
+  if [[ "$4" .eq "helm" ]]; then
+    source ./ceph-helm.sh "$1" $2 $3 $4
+  else
+    source ./ceph-baremetal.sh "$1" $2 $3 $4
+  fi
+}
+
 export WORK_DIR=$(pwd)
 case "$1" in
   master)
@@ -415,7 +263,7 @@ case "$1" in
     setup_k8s_agents "$2"
     ;;
   ceph)
-    setup_ceph "$2" $3 $4 $5
+    setup_ceph "$2" $3 $4 $5 $6
     ;;
   helm)
     setup_helm
@@ -426,7 +274,7 @@ case "$1" in
   all)
     setup_k8s_master
     setup_k8s_agents "$2"
-    setup_ceph "$2" $3 $4 $5
+    setup_ceph "$2" $3 $4 $5 $6
     setup_helm
     demo_chart dokuwiki
     ;;