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     elif [[ "$INSTALLER_TYPE" == "local" ]] ; then
  82         echo "INSTALLER_TYPE set to 'local'. Assuming SSH keys already exchanged with $COMPUTE_HOST"
  83     fi
  84
  85     sudo chown $(whoami):$(whoami) instack_key
  86     chmod 400 instack_key
  87     ssh_opts_cpu+=" -i instack_key"
  88
  89     # verify ssh to target compute host
  90     ssh $ssh_opts_cpu "$COMPUTE_USER@$COMPUTE_IP" 'exit'
  91     if [[ $? -ne 0 ]] ; then
  92         echo "ERROR: can not ssh to computer host"
  93         exit 1
  94     fi
  95 }
  96
  97 download_image() {
  98     [ -e "$IMAGE_FILE" ] && return 0
  99     wget "$IMAGE_URL" -o "$IMAGE_FILE"
 100 }
 101
 102 register_image() {
 103     openstack image list | grep -q " $IMAGE_NAME " && return 0
 104     openstack image create "$IMAGE_NAME" \
 105                            --public \
 106                            --disk-format "$IMAGE_FORMAT" \
 107                            --container-format bare \
 108                            --file "$IMAGE_FILE"
 109 }
 110
 111 create_test_user() {
 112     openstack user list | grep -q " $DOCTOR_USER " || {
 113         openstack user create "$DOCTOR_USER" --password "$DOCTOR_PW"
 114     }
 115     openstack project list | grep -q " $DOCTOR_PROJECT " || {
 116         openstack project create "$DOCTOR_PROJECT"
 117     }
 118     openstack user role list "$DOCTOR_USER" --project "$DOCTOR_PROJECT" \
 119     | grep -q " $DOCTOR_ROLE " || {
 120         openstack role add "$DOCTOR_ROLE" --user "$DOCTOR_USER" \
 121                            --project "$DOCTOR_PROJECT"
 122     }
 123 }
 124
 125 change_to_doctor_user() {
 126     export OS_USERNAME="$DOCTOR_USER"
 127     export OS_PASSWORD="$DOCTOR_PW"
 128     export OS_PROJECT_NAME="$DOCTOR_PROJECT"
 129     export OS_TENANT_NAME="$DOCTOR_PROJECT"
 130 }
 131
 132 boot_vm() {
 133     (
 134         # test VM done with test user, so can test non-admin
 135         change_to_doctor_user
 136         openstack server list | grep -q " $VM_NAME " && return 0
 137         openstack server create --flavor "$VM_FLAVOR" \
 138                                 --image "$IMAGE_NAME" \
 139                                 "$VM_NAME"
 140         sleep 1
 141     )
 142
 143 }
 144
 145 create_alarm() {
 146     (
 147         # get vm_id as test user
 148         change_to_doctor_user
 149         ceilometer alarm-list | grep -q " $ALARM_NAME " && return 0
 150         vm_id=$(openstack server list | grep " $VM_NAME " | awk '{print $2}')
 151         ceilometer alarm-event-create --name "$ALARM_NAME" \
 152             --alarm-action "http://localhost:$CONSUMER_PORT/failure" \
 153             --description "VM failure" \
 154             --enabled True \
 155             --repeat-actions False \
 156             --severity "moderate" \
 157             --event-type compute.instance.update \
 158             -q "traits.state=string::error; traits.instance_id=string::$vm_id"
 159     )
 160 }
 161
 162
 163 start_monitor() {
 164     pgrep -f "python monitor.py" && return 0
 165     sudo python monitor.py "$COMPUTE_HOST" "$COMPUTE_IP" \
 166         "http://127.0.0.1:$INSPECTOR_PORT/events" > monitor.log 2>&1 &
 167 }
 168
 169 stop_monitor() {
 170     pgrep -f "python monitor.py" || return 0
 171     sudo kill $(pgrep -f "python monitor.py")
 172     cat monitor.log
 173 }
 174
 175 start_inspector() {
 176     pgrep -f "python inspector.py" && return 0
 177     python inspector.py "$INSPECTOR_PORT" > inspector.log 2>&1 &
 178 }
 179
 180 stop_inspector() {
 181     pgrep -f "python inspector.py" || return 0
 182     kill $(pgrep -f "python inspector.py")
 183     cat inspector.log
 184 }
 185
 186 start_consumer() {
 187     pgrep -f "python consumer.py" && return 0
 188     python consumer.py "$CONSUMER_PORT" > consumer.log 2>&1 &
 189 }
 190
 191 stop_consumer() {
 192     pgrep -f "python consumer.py" || return 0
 193     kill $(pgrep -f "python consumer.py")
 194     cat consumer.log
 195 }
 196
 197 wait_for_vm_launch() {
 198     echo "waiting for vm launch..."
 199
 200     (
 201         # get VM state as test user
 202         change_to_doctor_user
 203
 204         count=0
 205         while [[ ${count} -lt 60 ]]
 206         do
 207             state=$(openstack server list | grep " $VM_NAME " | awk '{print $6}')
 208             [[ "$state" == "ACTIVE" ]] && return 0
 209             [[ "$state" == "ERROR" ]] && echo "vm state is ERROR" && exit 1
 210             count=$(($count+1))
 211             sleep 1
 212         done
 213         echo "ERROR: time out while waiting for vm launch"
 214         exit 1
 215     )
 216 }
 217
 218 inject_failure() {
 219     echo "disabling network of compute host [$COMPUTE_HOST] for 3 mins..."
 220     cat > disable_network.sh << 'END_TXT'
 221 #!/bin/bash -x
 222 dev=$(sudo ip route | awk '/^default/{print $5}')
 223 sleep 1
 224 sudo ip link set $dev down
 225 sleep 180
 226 sudo ip link set $dev up
 227 sleep 1
 228 END_TXT
 229     chmod +x disable_network.sh
 230     scp $ssh_opts_cpu disable_network.sh "$COMPUTE_USER@$COMPUTE_IP:"
 231     ssh $ssh_opts_cpu "$COMPUTE_USER@$COMPUTE_IP" 'nohup ./disable_network.sh > disable_network.log 2>&1 &'
 232 }
 233
 234 calculate_notification_time() {
 235     detected=$(grep "doctor monitor detected at" monitor.log | awk '{print $5}')
 236     notified=$(grep "doctor consumer notified at" consumer.log | awk '{print $5}')
 237     echo "$notified $detected" | \
 238         awk '{d = $1 - $2; if (d < 1 && d > 0) print d " OK"; else print d " NG"}'
 239 }
 240
 241 check_host_status() {
 242     expect_state=$1
 243     (
 244         change_to_doctor_user
 245
 246         host_status_line=$(openstack --os-compute-api-version 2.16 server show $VM_NAME | grep "host_status")
 247         if [[ $? -ne 0 ]] ; then
 248             echo "ERROR: host_status not configured for owner in Nova policy.json"
 249             exit 1
 250         fi
 251
 252         host_status=$(echo $host_status_line | awk '{print $4}')
 253         if [ -z "$host_status" ] ; then
 254             echo "ERROR: host_status not reported by: nova show $VM_NAME"
 255             exit 1
 256         elif [[ "$host_status" != "$expect_state" ]] ; then
 257             echo "ERROR: host_status:$host_status not equal to expect_state: $expect_state"
 258             exit 1
 259         else
 260             echo "$VM_NAME showing host_status: $host_status"
 261         fi
 262     )
 263 }
 264
 265 cleanup() {
 266     set +e
 267     echo "cleanup..."
 268     stop_monitor
 269     stop_inspector
 270     stop_consumer
 271
 272     python ./nova_force_down.py "$COMPUTE_HOST" --unset
 273     sleep 1
 274     (
 275         change_to_doctor_user
 276         openstack server list | grep -q " $VM_NAME " && openstack server delete "$VM_NAME"
 277         sleep 1
 278         alarm_id=$(ceilometer alarm-list | grep " $ALARM_NAME " | awk '{print $2}')
 279         sleep 1
 280         [ -n "$alarm_id" ] && ceilometer alarm-delete "$alarm_id"
 281         sleep 1
 282     )
 283     image_id=$(openstack image list | grep " $IMAGE_NAME " | awk '{print $2}')
 284     sleep 1
 285     [ -n "$image_id" ] && openstack image delete "$image_id"
 286     openstack role remove "$DOCTOR_ROLE" --user "$DOCTOR_USER" \
 287                               --project "$DOCTOR_PROJECT"
 288     openstack project delete "$DOCTOR_PROJECT"
 289     openstack user delete "$DOCTOR_USER"
 290
 291     echo "waiting disabled compute host back to be enabled..."
 292     sleep 180
 293     check_host_status "UP"
 294     ssh $ssh_opts_cpu "$COMPUTE_USER@$COMPUTE_IP" \
 295         "[ -e disable_network.log ] && cat disable_network.log"
 296 }
 297
 298
 299 echo "Note: doctor/tests/run.sh has been executed."
 300
 301 trap cleanup EXIT
 302
 303 echo "preparing VM image..."
 304 download_image
 305 register_image
 306
 307 echo "creating test user..."
 308 create_test_user
 309
 310 echo "creating VM and alarm..."
 311 boot_vm
 312 wait_for_vm_launch
 313 create_alarm
 314
 315 echo "get computer host info and prepare to ssh..."
 316 get_compute_host_info
 317 prepare_compute_ssh
 318
 319 echo "starting doctor sample components..."
 320 start_monitor
 321 start_inspector
 322 start_consumer
 323
 324 sleep 60
 325 echo "injecting host failure..."
 326 inject_failure
 327 sleep 10
 328
 329 check_host_status "DOWN"
 330 calculate_notification_time
 331
 332 echo "done"