jjb/packaging/openstack-k8s-create-with-template.sh

   1 #!/bin/bash -l
   2 # SPDX-License-Identifier: EPL-1.0
   3 ##############################################################################
   4 # Copyright (c) 2021 The Linux Foundation and others.
   5 #
   6 # All rights reserved. This program and the accompanying materials
   7 # are made available under the terms of the Eclipse Public License v1.0
   8 # which accompanies this distribution, and is available at
   9 # http://www.eclipse.org/legal/epl-v10.html
  10 ##############################################################################
  11 # shellcheck disable=SC2153,SC2034
  12 echo "---> Create K8S cluster with pre-existing template"
  13
  14 set -eux -o pipefail
  15 # shellcheck disable=SC1090
  16 . ~/lf-env.sh
  17 lf-activate-venv python-openstackclient python-magnumclient
  18
  19 OS_TIMEOUT=20       # Wait time in minutes for OpenStack cluster to come up.
  20 CLUSTER_RETRIES=3   # Number of times to retry creating a cluster.
  21 CLUSTER_SUCCESSFUL=false
  22
  23 mkdir -p "$WORKSPACE/archives"
  24
  25 boot_volume_size="${BOOT_VOLUME_SIZE}"
  26 cluster_name="${CLUSTER_NAME}"
  27 cluster_settle_time="${CLUSTER_SETTLE_TIME:-1m}"
  28 cluster_template_name="${CLUSTER_TEMPLATE_NAME}"
  29 cloud_provider_tag="${CLOUD_PROVIDER_TAG}"
  30 container_infra_prefix="${CONTAINER_INFRA_PREFIX}"
  31 etcd_volume_size="${ETCD_VOLUME_SIZE}"
  32 k8s_version="${K8S_VERSION}"
  33 keypair="${KEYPAIR}"
  34 kube_tag="${KUBE_TAG}"
  35 helm_client_url="${HELM_CLIENT_URL}"
  36 helm_sha256="${HELM_SHA256}"
  37 helm_version="${HELM_VERSION}"
  38 master_count="${MASTER_COUNT:-1}"
  39 master_flavor="${MASTER_FLAVOR}"
  40 master_lb_floating_ip_enabled="${MASTER_LB_FLOATING_IP_ENABLED:-false}"
  41 node_count="${NODE_COUNT:-2}"
  42 node_flavor="${NODE_FLAVOR}"
  43 os_cloud="${OS_CLOUD:-vex}"
  44
  45 echo "INFO: Create a Cluster:${CLUSTER_NAME} for attempts:${CLUSTER_RETRIES}."
  46 for try in $(seq $CLUSTER_RETRIES); do
  47     # shellcheck disable=SC1083
  48
  49     # Create the cluster using pre-defined template. Returns the status which includes the $cluster_uuid
  50     cluster_status=$(openstack --os-cloud "${os_cloud}" coe cluster create "${cluster_name}" \
  51         --cluster-template "${cluster_template_name}" \
  52         --keypair "${keypair}" \
  53         --master-count "${master_count}" \
  54         --node-count "${node_count}" \
  55         --master-flavor "${master_flavor}" \
  56         --flavor "${node_flavor}" \
  57         --labels \
  58 boot_volume_size="${boot_volume_size}",\
  59 container_infra_prefix="${container_infra_prefix}",\
  60 cloud_provider_tag="${cloud_provider_tag}",\
  61 helm_client_sha256="${helm_sha256}",\
  62 helm_client_tag="${helm_version}",\
  63 etcd_volume_size="${etcd_volume_size}",\
  64 kube_tag="${kube_tag}",\
  65 master_lb_floating_ip_enabled=false,\
  66 helm_client_url="${helm_client_url}" \
  67         --floating-ip-disabled)
  68
  69     # Check return status and extract the $cluster_uuid from return status
  70     if [[ -z "$cluster_status" ]]; then
  71         echo "ERROR: Failed to create coe cluster ${cluster_name}"
  72         exit 1
  73     elif [[ "${cluster_status}" =~ .*accepted.* ]]; then
  74         cluster_uuid=$(echo "${cluster_status}" | awk -F' ' '{print $5}')
  75     fi
  76
  77     echo "INFO $try: Wait until ${OS_TIMEOUT} (in minutes) to rollout ${cluster_name}."
  78     for i in $(seq $OS_TIMEOUT); do
  79         sleep 90
  80
  81         CLUSTER_STATUS=$(openstack --os-cloud "$os_cloud" coe cluster show "$cluster_uuid" -c status -f value)
  82         echo "$i: ${CLUSTER_STATUS}"
  83
  84         case "${CLUSTER_STATUS}" in
  85             CREATE_COMPLETE)
  86                 echo "INFO: Cluster ${cluster_name} initialized on infrastructure successful."
  87                 CLUSTER_SUCCESSFUL=true
  88                 break
  89             ;;
  90             CREATE_FAILED)
  91                 reason=$(openstack coe cluster show "${cluster_name}" -f value -c health_status_reason)
  92                 echo "ERROR: Failed to initialize infrastructure. Reason: ${reason}"
  93                 openstack ceo cluster show "${cluster_name}"
  94
  95                 echo "INFO: Deleting cluster and re-try to create the cluster again ..."
  96                 openstack coe cluster delete "${cluster_name}"
  97
  98                 # Post delete, poll for 5m to learn if cluster is fully removed
  99                 for j in $(seq 20); do
 100                     sleep 30
 101                     delete_status=$(openstack coe cluster show "${cluster_name}" -f value -c status)
 102                     echo "$j: ${delete_status}"
 103                     if [[ ${delete_status} == "DELETE_FAILED" ]]; then
 104                         reason=$(openstack coe cluster show "${cluster_name}" -f value -c health_status_reason)
 105                         echo "ERROR: Failed to delete ${cluster_name}. Reason: ${reason}"
 106
 107                         echo "INFO: Deleting failed cluster again: ${cluster_name}"
 108                         openstack coe cluster delete "${cluster_name}"
 109                     fi
 110
 111                     if ! openstack coe cluster show "${cluster_name}" -f value -c status; then
 112                         echo "INFO: Cluster show on ${cluster_name} came back empty. Assuming successful delete"
 113                         break
 114                     fi
 115                 done
 116
 117                 # If we still see $CLUSTER_NAME in `openstack coe cluster show` this infers the delete hasn't fully
 118                 # worked and we can exit forcefully
 119                 if openstack coe cluster show "${cluster_name}" -f value -c stack_status; then
 120                     echo "ERROR: Cluster ${cluster_name} still in cloud output after polling. Quitting!"
 121                     exit 1
 122                 fi
 123                 break
 124             ;;
 125             CREATE_IN_PROGRESS)
 126                 echo "INFO: Waiting to initialize cluster infrastructure ..."
 127                 continue
 128             ;;
 129             *)
 130                 echo "ERROR: Unexpected status: ${OS_STATUS}"
 131                 # DO NOT exit on unexpected status. Openstack cluster sometimes returns unexpected status
 132                 # before returning an expected status. Just print the message and loop until we have
 133                 # a confirmed state or timeout.
 134                 # exit 1
 135             ;;
 136         esac
 137     done
 138     if $CLUSTER_SUCCESSFUL; then
 139         break
 140     fi
 141 done