tests/run.sh

   1 #!/bin/bash -e
   2 ##############################################################################
   3 # Copyright (c) 2016 NEC Corporation and others.
   4 #
   5 # All rights reserved. This program and the accompanying materials
   6 # are made available under the terms of the Apache License, Version 2.0
   7 # which accompanies this distribution, and is available at
   8 # http://www.apache.org/licenses/LICENSE-2.0
   9 ##############################################################################
  10
  11 [[ "${CI_DEBUG:-true}" == "true" ]] && set -x
  12
  13 IMAGE_URL=https://launchpad.net/cirros/trunk/0.3.0/+download/cirros-0.3.0-x86_64-disk.img
  14 IMAGE_NAME=cirros
  15 IMAGE_FILE="${IMAGE_NAME}.img"
  16 IMAGE_FORMAT=qcow2
  17 VM_NAME=doctor_vm1
  18 VM_FLAVOR=m1.tiny
  19 ALARM_NAME=doctor_alarm1
  20 INSPECTOR_PORT=12345
  21 CONSUMER_PORT=12346
  22 DOCTOR_USER=doctor
  23 DOCTOR_PW=doctor
  24 DOCTOR_PROJECT=doctor
  25 #TODO: change back to `_member_` when JIRA DOCTOR-55 is done
  26 DOCTOR_ROLE=admin
  27
  28 SUPPORTED_INSTALLER_TYPES="apex local"
  29 INSTALLER_TYPE=${INSTALLER_TYPE:-apex}
  30 INSTALLER_IP=${INSTALLER_IP:-none}
  31 COMPUTE_USER=${COMPUTE_USER:-none}
  32 ssh_opts="-o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no"
  33
  34 if [[ ! "$SUPPORTED_INSTALLER_TYPES" =~ "$INSTALLER_TYPE" ]] ; then
  35     echo "ERROR: INSTALLER_TYPE=$INSTALLER_TYPE is not supported."
  36     exit 1
  37 fi
  38
  39 get_compute_host_info() {
  40     (
  41         change_to_doctor_user
  42
  43         # get computer host info which VM boot in
  44         export COMPUTE_HOST=$(openstack server show $VM_NAME | \
  45                 grep "OS-EXT-SRV-ATTR:host" | awk '{ print $4 }' |
  46                 awk -F '.' '{print $1}')
  47     )
  48     if [[ "$INSTALLER_TYPE" == "apex" ]] ; then
  49         COMPUTE_USER=${COMPUTE_USER:-heat-admin}
  50         if [[ "$INSTALLER_IP" == "none" ]] ; then
  51             instack_mac=$(sudo virsh domiflist instack | awk '/default/{print $5}')
  52             INSTALLER_IP=$(/usr/sbin/arp -e | grep ${instack_mac} | awk '{print $1}')
  53         fi
  54         COMPUTE_IP=$(sudo ssh $ssh_opts $INSTALLER_IP \
  55              "source stackrc; \
  56              openstack server show $COMPUTE_HOST \
  57              | awk '/ ctlplane network /{print \$5}'")
  58     elif [[ "$INSTALLER_TYPE" == "local" ]] ; then
  59         COMPUTE_USER=${COMPUTE_USER:-$(whoami)}
  60         COMPUTE_IP=$(getent hosts "$COMPUTE_HOST" | awk '{ print $1 }')
  61         if [[ -z "$COMPUTE_IP" ]]; then
  62             echo "ERROR: Could not resolve $COMPUTE_HOST. Either manually set COMPUTE_IP or enable DNS resolution."
  63             exit 1
  64         fi
  65     fi
  66
  67     # verify connectivity to target compute host
  68     ping -c 1 "$COMPUTE_IP"
  69     if [[ $? -ne 0 ]] ; then
  70         echo "ERROR: can not ping to computer host"
  71         exit 1
  72     fi
  73 }
  74
  75 prepare_compute_ssh() {
  76     ssh_opts_cpu="$ssh_opts"
  77
  78     # get ssh key from installer node
  79     if [[ "$INSTALLER_TYPE" == "apex" ]] ; then
  80         sudo scp $ssh_opts root@"$INSTALLER_IP":/home/stack/.ssh/id_rsa instack_key
  81         sudo chown $(whoami):$(whoami) instack_key
  82         chmod 400 instack_key
  83         ssh_opts_cpu+=" -i instack_key"
  84     elif [[ "$INSTALLER_TYPE" == "local" ]] ; then
  85         echo "INSTALLER_TYPE set to 'local'. Assuming SSH keys already exchanged with $COMPUTE_HOST"
  86     fi
  87
  88     # verify ssh to target compute host
  89     ssh $ssh_opts_cpu "$COMPUTE_USER@$COMPUTE_IP" 'exit'
  90     if [[ $? -ne 0 ]] ; then
  91         echo "ERROR: can not ssh to computer host"
  92         exit 1
  93     fi
  94 }
  95
  96 download_image() {
  97     [ -e "$IMAGE_FILE" ] && return 0
  98     wget "$IMAGE_URL" -o "$IMAGE_FILE"
  99 }
 100
 101 register_image() {
 102     openstack image list | grep -q " $IMAGE_NAME " && return 0
 103     openstack image create "$IMAGE_NAME" \
 104                            --public \
 105                            --disk-format "$IMAGE_FORMAT" \
 106                            --container-format bare \
 107                            --file "$IMAGE_FILE"
 108 }
 109
 110 create_test_user() {
 111     openstack user list | grep -q " $DOCTOR_USER " || {
 112         openstack user create "$DOCTOR_USER" --password "$DOCTOR_PW"
 113     }
 114     openstack project list | grep -q " $DOCTOR_PROJECT " || {
 115         openstack project create "$DOCTOR_PROJECT"
 116     }
 117     openstack user role list "$DOCTOR_USER" --project "$DOCTOR_PROJECT" \
 118     | grep -q " $DOCTOR_ROLE " || {
 119         openstack role add "$DOCTOR_ROLE" --user "$DOCTOR_USER" \
 120                            --project "$DOCTOR_PROJECT"
 121     }
 122 }
 123
 124 change_to_doctor_user() {
 125     export OS_USERNAME="$DOCTOR_USER"
 126     export OS_PASSWORD="$DOCTOR_PW"
 127     export OS_PROJECT_NAME="$DOCTOR_PROJECT"
 128     export OS_TENANT_NAME="$DOCTOR_PROJECT"
 129 }
 130
 131 boot_vm() {
 132     (
 133         # test VM done with test user, so can test non-admin
 134         change_to_doctor_user
 135         openstack server list | grep -q " $VM_NAME " && return 0
 136         openstack server create --flavor "$VM_FLAVOR" \
 137                                 --image "$IMAGE_NAME" \
 138                                 "$VM_NAME"
 139         sleep 1
 140     )
 141
 142 }
 143
 144 create_alarm() {
 145     (
 146         # get vm_id as test user
 147         change_to_doctor_user
 148         ceilometer alarm-list | grep -q " $ALARM_NAME " && return 0
 149         vm_id=$(openstack server list | grep " $VM_NAME " | awk '{print $2}')
 150         ceilometer alarm-event-create --name "$ALARM_NAME" \
 151             --alarm-action "http://localhost:$CONSUMER_PORT/failure" \
 152             --description "VM failure" \
 153             --enabled True \
 154             --repeat-actions False \
 155             --severity "moderate" \
 156             --event-type compute.instance.update \
 157             -q "traits.state=string::error; traits.instance_id=string::$vm_id"
 158     )
 159 }
 160
 161
 162 start_monitor() {
 163     pgrep -f "python monitor.py" && return 0
 164     sudo python monitor.py "$COMPUTE_HOST" "$COMPUTE_IP" \
 165         "http://127.0.0.1:$INSPECTOR_PORT/events" > monitor.log 2>&1 &
 166 }
 167
 168 stop_monitor() {
 169     pgrep -f "python monitor.py" || return 0
 170     sudo kill $(pgrep -f "python monitor.py")
 171     cat monitor.log
 172 }
 173
 174 start_inspector() {
 175     pgrep -f "python inspector.py" && return 0
 176     python inspector.py "$INSPECTOR_PORT" > inspector.log 2>&1 &
 177 }
 178
 179 stop_inspector() {
 180     pgrep -f "python inspector.py" || return 0
 181     kill $(pgrep -f "python inspector.py")
 182     cat inspector.log
 183 }
 184
 185 start_consumer() {
 186     pgrep -f "python consumer.py" && return 0
 187     python consumer.py "$CONSUMER_PORT" > consumer.log 2>&1 &
 188 }
 189
 190 stop_consumer() {
 191     pgrep -f "python consumer.py" || return 0
 192     kill $(pgrep -f "python consumer.py")
 193     cat consumer.log
 194 }
 195
 196 wait_for_vm_launch() {
 197     echo "waiting for vm launch..."
 198
 199     (
 200         # get VM state as test user
 201         change_to_doctor_user
 202
 203         count=0
 204         while [[ ${count} -lt 60 ]]
 205         do
 206             state=$(openstack server list | grep " $VM_NAME " | awk '{print $6}')
 207             [[ "$state" == "ACTIVE" ]] && return 0
 208             [[ "$state" == "ERROR" ]] && echo "vm state is ERROR" && exit 1
 209             count=$(($count+1))
 210             sleep 1
 211         done
 212         echo "ERROR: time out while waiting for vm launch"
 213         exit 1
 214     )
 215 }
 216
 217 inject_failure() {
 218     echo "disabling network of compute host [$COMPUTE_HOST] for 3 mins..."
 219     cat > disable_network.sh << 'END_TXT'
 220 #!/bin/bash -x
 221 dev=$(sudo ip route | awk '/^default/{print $5}')
 222 sleep 1
 223 sudo ip link set $dev down
 224 sleep 180
 225 sudo ip link set $dev up
 226 sleep 1
 227 END_TXT
 228     chmod +x disable_network.sh
 229     scp $ssh_opts_cpu disable_network.sh "$COMPUTE_USER@$COMPUTE_IP:"
 230     ssh $ssh_opts_cpu "$COMPUTE_USER@$COMPUTE_IP" 'nohup ./disable_network.sh > disable_network.log 2>&1 &'
 231 }
 232
 233 calculate_notification_time() {
 234     detected=$(grep "doctor monitor detected at" monitor.log | awk '{print $5}')
 235     notified=$(grep "doctor consumer notified at" consumer.log | awk '{print $5}')
 236     echo "$notified $detected" | \
 237         awk '{d = $1 - $2; if (d < 1 && d > 0) print d " OK"; else print d " NG"}'
 238 }
 239
 240 check_host_status() {
 241     expect_state=$1
 242     (
 243         change_to_doctor_user
 244
 245         host_status_line=$(openstack --os-compute-api-version 2.16 server show $VM_NAME | grep "host_status")
 246         if [[ $? -ne 0 ]] ; then
 247             echo "ERROR: host_status not configured for owner in Nova policy.json"
 248             exit 1
 249         fi
 250
 251         host_status=$(echo $host_status_line | awk '{print $4}')
 252         if [ -z "$host_status" ] ; then
 253             echo "ERROR: host_status not reported by: nova show $VM_NAME"
 254             exit 1
 255         elif [[ "$host_status" != "$expect_state" ]] ; then
 256             echo "ERROR: host_status:$host_status not equal to expect_state: $expect_state"
 257             exit 1
 258         else
 259             echo "$VM_NAME showing host_status: $host_status"
 260         fi
 261     )
 262 }
 263
 264 cleanup() {
 265     set +e
 266     echo "cleanup..."
 267     stop_monitor
 268     stop_inspector
 269     stop_consumer
 270
 271     echo "waiting disabled compute host back to be enabled..."
 272     python ./nova_force_down.py "$COMPUTE_HOST" --unset
 273     sleep 180
 274     check_host_status "UP"
 275     ssh $ssh_opts_cpu "$COMPUTE_USER@$COMPUTE_IP" \
 276         "[ -e disable_network.log ] && cat disable_network.log"
 277     sleep 1
 278
 279     (
 280         change_to_doctor_user
 281         openstack server list | grep -q " $VM_NAME " && openstack server delete "$VM_NAME"
 282         sleep 1
 283         alarm_id=$(ceilometer alarm-list | grep " $ALARM_NAME " | awk '{print $2}')
 284         sleep 1
 285         [ -n "$alarm_id" ] && ceilometer alarm-delete "$alarm_id"
 286         sleep 1
 287     )
 288     image_id=$(openstack image list | grep " $IMAGE_NAME " | awk '{print $2}')
 289     sleep 1
 290     [ -n "$image_id" ] && openstack image delete "$image_id"
 291     openstack role remove "$DOCTOR_ROLE" --user "$DOCTOR_USER" \
 292                               --project "$DOCTOR_PROJECT"
 293     openstack project delete "$DOCTOR_PROJECT"
 294     openstack user delete "$DOCTOR_USER"
 295 }
 296
 297
 298 echo "Note: doctor/tests/run.sh has been executed."
 299
 300 trap cleanup EXIT
 301
 302 echo "preparing VM image..."
 303 download_image
 304 register_image
 305
 306 echo "creating test user..."
 307 create_test_user
 308
 309 echo "creating VM and alarm..."
 310 boot_vm
 311 wait_for_vm_launch
 312 create_alarm
 313
 314 echo "get computer host info and prepare to ssh..."
 315 get_compute_host_info
 316 prepare_compute_ssh
 317
 318 echo "starting doctor sample components..."
 319 start_monitor
 320 start_inspector
 321 start_consumer
 322
 323 sleep 60
 324 echo "injecting host failure..."
 325 inject_failure
 326 sleep 10
 327
 328 check_host_status "DOWN"
 329 calculate_notification_time
 330
 331 echo "done"