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 [[ "${CI_DEBUG:-true}" == [Tt]rue ]] && set -x
  12
  13 IMAGE_URL=https://launchpad.net/cirros/trunk/0.3.0/+download/cirros-0.3.0-x86_64-disk.img
  14 IMAGE_NAME=cirros
  15 IMAGE_FILE="${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 fuel local"
  29 INSTALLER_TYPE=${INSTALLER_TYPE:-local}
  30 INSTALLER_IP=${INSTALLER_IP:-none}
  31
  32 SUPPORTED_INSPECTOR_TYPES="sample congress"
  33 INSPECTOR_TYPE=${INSPECTOR_TYPE:-sample}
  34
  35 ssh_opts="-o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no"
  36 as_doctor_user="--os-username $DOCTOR_USER --os-password $DOCTOR_PW
  37                 --os-tenant-name $DOCTOR_PROJECT"
  38
  39 if [[ ! "$SUPPORTED_INSTALLER_TYPES" =~ "$INSTALLER_TYPE" ]] ; then
  40     echo "ERROR: INSTALLER_TYPE=$INSTALLER_TYPE is not supported."
  41     exit 1
  42 fi
  43
  44 if [[ ! "$SUPPORTED_INSPECTOR_TYPES" =~ "$INSPECTOR_TYPE" ]] ; then
  45     echo "ERROR: INSPECTOR_TYPE=$INSPECTOR_TYPE is not supported."
  46     exit 1
  47 fi
  48
  49 get_installer_ip() {
  50     if [[ "$INSTALLER_TYPE" == "apex" ]] ; then
  51         if [[ "$INSTALLER_IP" == "none" ]] ; then
  52             instack_mac=$(sudo virsh domiflist instack | awk '/default/{print $5}')
  53             INSTALLER_IP=$(/usr/sbin/arp -e | grep ${instack_mac} | awk '{print $1}')
  54         fi
  55     elif [[ "$INSTALLER_TYPE" == "fuel" ]] ; then
  56         if [[ "$INSTALLER_IP" == "none" ]] ; then
  57             instack_mac=$(sudo virsh domiflist fuel-opnfv | awk '/pxebr/{print $5}')
  58             INSTALLER_IP=$(/usr/sbin/arp -e | grep ${instack_mac} | awk '{print $1}')
  59         fi
  60     fi
  61
  62     if [[ "$INSTALLER_TYPE" -ne "local" ]] ; then
  63         if [[ -z "$INSTALLER_IP" ]] ; then
  64             echo "ERROR: no installer ip"
  65             exit 1
  66         fi
  67     fi
  68 }
  69
  70 prepare_ssh_to_cloud() {
  71     ssh_opts_cpu="$ssh_opts"
  72
  73     # get ssh key from installer node
  74     if [[ "$INSTALLER_TYPE" == "apex" ]] ; then
  75         sudo scp $ssh_opts root@"$INSTALLER_IP":/home/stack/.ssh/id_rsa instack_key
  76         sudo chown $(whoami):$(whoami) instack_key
  77         chmod 400 instack_key
  78         ssh_opts_cpu+=" -i instack_key"
  79     elif [[ "$INSTALLER_TYPE" == "fuel" ]] ; then
  80         sshpass -p r00tme scp $ssh_opts root@${INSTALLER_IP}:.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 $COMPUTE_HOST"
  86     fi
  87 }
  88
  89 get_compute_host_info() {
  90     # get computer host info which VM boot in
  91     COMPUTE_HOST=$(openstack $as_doctor_user server show $VM_NAME |
  92                    grep "OS-EXT-SRV-ATTR:host" | awk '{ print $4 }')
  93     compute_host_in_undercloud=${COMPUTE_HOST%%.*}
  94     if [[ -z "$COMPUTE_HOST" ]] ; then
  95         echo "ERROR: failed to get compute hostname"
  96         exit 1
  97     fi
  98
  99     if [[ "$INSTALLER_TYPE" == "apex" ]] ; then
 100         COMPUTE_USER=${COMPUTE_USER:-heat-admin}
 101         COMPUTE_IP=$(sudo ssh $ssh_opts $INSTALLER_IP \
 102              "source stackrc; \
 103              nova show $compute_host_in_undercloud \
 104              | awk '/ ctlplane network /{print \$5}'")
 105     elif [[ "$INSTALLER_TYPE" == "fuel" ]] ; then
 106         COMPUTE_USER=${COMPUTE_USER:-root}
 107         node_id=$(echo $compute_host_in_undercloud | cut -d "-" -f 2)
 108         COMPUTE_IP=$(sshpass -p r00tme ssh 2>/dev/null $ssh_opts root@${INSTALLER_IP} \
 109              "fuel node|awk -F '|' -v id=$node_id '{if (\$1 == id) print \$5}' |xargs")
 110     elif [[ "$INSTALLER_TYPE" == "local" ]] ; then
 111         COMPUTE_USER=${COMPUTE_USER:-$(whoami)}
 112         COMPUTE_IP=$(getent hosts "$COMPUTE_HOST" | awk '{ print $1 }')
 113     fi
 114
 115     if [[ -z "$COMPUTE_IP" ]]; then
 116         echo "ERROR: Could not resolve $COMPUTE_HOST. Either manually set COMPUTE_IP or enable DNS resolution."
 117         exit 1
 118     fi
 119     echo "COMPUTE_HOST=$COMPUTE_HOST"
 120     echo "COMPUTE_IP=$COMPUTE_IP"
 121
 122     # verify connectivity to target compute host
 123     ping -c 1 "$COMPUTE_IP"
 124     if [[ $? -ne 0 ]] ; then
 125         echo "ERROR: can not ping to computer host"
 126         exit 1
 127     fi
 128
 129     # verify ssh to target compute host
 130     ssh $ssh_opts_cpu "$COMPUTE_USER@$COMPUTE_IP" 'exit'
 131     if [[ $? -ne 0 ]] ; then
 132         echo "ERROR: can not ssh to computer host"
 133         exit 1
 134     fi
 135 }
 136
 137 get_consumer_ip() {
 138     CONSUMER_IP=$(sudo ssh $ssh_opts root@$INSTALLER_IP \
 139                   "ip route get $COMPUTE_IP | awk '/ src /{print \$NF}'")
 140     echo "CONSUMER_IP=$CONSUMER_IP"
 141
 142     if [[ -z "$CONSUMER_IP" ]]; then
 143         echo "ERROR: Could not get CONSUMER_IP."
 144         exit 1
 145     fi
 146 }
 147
 148 download_image() {
 149     [ -e "$IMAGE_FILE" ] && return 0
 150     wget "$IMAGE_URL" -o "$IMAGE_FILE"
 151 }
 152
 153 register_image() {
 154     openstack image list | grep -q " $IMAGE_NAME " && return 0
 155     openstack image create "$IMAGE_NAME" \
 156                            --public \
 157                            --disk-format "$IMAGE_FORMAT" \
 158                            --container-format bare \
 159                            --file "$IMAGE_FILE"
 160 }
 161
 162 create_test_user() {
 163     openstack project list | grep -q " $DOCTOR_PROJECT " || {
 164         openstack project create "$DOCTOR_PROJECT"
 165     }
 166     openstack user list | grep -q " $DOCTOR_USER " || {
 167         openstack user create "$DOCTOR_USER" --password "$DOCTOR_PW" \
 168                               --project "$DOCTOR_PROJECT"
 169     }
 170     openstack user role list "$DOCTOR_USER" --project "$DOCTOR_PROJECT" \
 171     | grep -q " $DOCTOR_ROLE " || {
 172         openstack role add "$DOCTOR_ROLE" --user "$DOCTOR_USER" \
 173                            --project "$DOCTOR_PROJECT"
 174     }
 175 }
 176
 177 boot_vm() {
 178     # test VM done with test user, so can test non-admin
 179     openstack $as_doctor_user server list | grep -q " $VM_NAME " && return 0
 180     openstack $as_doctor_user server create --flavor "$VM_FLAVOR" \
 181                             --image "$IMAGE_NAME" \
 182                             "$VM_NAME"
 183     sleep 1
 184 }
 185
 186 create_alarm() {
 187     # get vm_id as test user
 188     ceilometer $as_doctor_user alarm-list | grep -q " $ALARM_NAME " && return 0
 189     vm_id=$(openstack $as_doctor_user server list | grep " $VM_NAME " | awk '{print $2}')
 190     # TODO(r-mibu): change notification endpoint from localhost to the consumer
 191     # IP address (functest container).
 192     ceilometer $as_doctor_user alarm-event-create --name "$ALARM_NAME" \
 193         --alarm-action "http://localhost:$CONSUMER_PORT/failure" \
 194         --description "VM failure" \
 195         --enabled True \
 196         --repeat-actions False \
 197         --severity "moderate" \
 198         --event-type compute.instance.update \
 199         -q "traits.state=string::error; traits.instance_id=string::$vm_id"
 200 }
 201
 202 print_log() {
 203     log_file=$1
 204     echo "$log_file:"
 205     sed -e 's/^/    /' "$log_file"
 206 }
 207
 208 start_monitor() {
 209     pgrep -f "python monitor.py" && return 0
 210     sudo -E python monitor.py "$COMPUTE_HOST" "$COMPUTE_IP" "$INSPECTOR_TYPE" \
 211         "http://127.0.0.1:$INSPECTOR_PORT/events" > monitor.log 2>&1 &
 212 }
 213
 214 stop_monitor() {
 215     pgrep -f "python monitor.py" || return 0
 216     sudo kill $(pgrep -f "python monitor.py")
 217     print_log monitor.log
 218 }
 219
 220 congress_add_rule() {
 221     name=$1
 222     policy=$2
 223     rule=$3
 224
 225     if ! openstack congress policy rule list $policy | grep -q -e "// Name: $name$" ; then
 226         openstack congress policy rule create --name $name $policy "$rule"
 227     fi
 228 }
 229
 230 congress_del_rule() {
 231     name=$1
 232     policy=$2
 233
 234     if openstack congress policy rule list $policy | grep -q -e "^// Name: $name$" ; then
 235         openstack congress policy rule delete $policy $name
 236     fi
 237 }
 238
 239 congress_setup_rules() {
 240     congress_add_rule host_down classification \
 241         'host_down(host) :-
 242             doctor:events(hostname=host, type="compute.host.down", status="down")'
 243
 244     congress_add_rule active_instance_in_host classification \
 245         'active_instance_in_host(vmid, host) :-
 246             nova:servers(id=vmid, host_name=host, status="ACTIVE")'
 247
 248     congress_add_rule host_force_down classification \
 249         'execute[nova:services.force_down(host, "nova-compute", "True")] :-
 250             host_down(host)'
 251
 252     congress_add_rule error_vm_states classification \
 253         'execute[nova:servers.reset_state(vmid, "error")] :-
 254             host_down(host),
 255             active_instance_in_host(vmid, host)'
 256 }
 257
 258 start_inspector() {
 259     if [[ "$INSPECTOR_TYPE" == "sample" ]] ; then
 260         pgrep -f "python inspector.py" && return 0
 261         python inspector.py "$INSPECTOR_PORT" > inspector.log 2>&1 &
 262     elif [[ "$INSPECTOR_TYPE" == "congress" ]] ; then
 263         nova_api_min_version="2.11"
 264         nova_api_version=$(openstack congress datasource list | \
 265                            grep nova | grep -Po "(?<='api_version': ')[^']*")
 266         [[ -z $nova_api_version ]] && nova_api_version="2.0"
 267         if [[ "$nova_api_version" < "$nova_api_min_version" ]]; then
 268             echo "ERROR: Congress Nova datasource API version < $nova_api_min_version ($nova_api_version)"
 269             exit 1
 270         fi
 271         openstack congress driver list | grep -q " doctor "
 272         openstack congress datasource list | grep -q " doctor " || {
 273             openstack congress datasource create doctor doctor
 274         }
 275         congress_setup_rules
 276     fi
 277 }
 278
 279 stop_inspector() {
 280     if [[ "$INSPECTOR_TYPE" == "sample" ]] ; then
 281         pgrep -f "python inspector.py" || return 0
 282         kill $(pgrep -f "python inspector.py")
 283         print_log inspector.log
 284     elif [[ "$INSPECTOR_TYPE" == "congress" ]] ; then
 285         congress_del_rule host_force_down classification
 286         congress_del_rule error_vm_states classification
 287         congress_del_rule active_instance_in_host classification
 288         congress_del_rule host_down classification
 289     fi
 290 }
 291
 292 start_consumer() {
 293     pgrep -f "python consumer.py" && return 0
 294     python consumer.py "$CONSUMER_PORT" > consumer.log 2>&1 &
 295
 296     # NOTE(r-mibu): create tunnel to the controller nodes, so that we can
 297     # avoid some network problems dpends on infra and installers.
 298     # This tunnel will be terminated by stop_consumer() or after 10 mins passed.
 299     if [[ "$INSTALLER_TYPE" == "apex" ]] ; then
 300         CONTROLLER_IPS=$(sudo ssh $ssh_opts $INSTALLER_IP \
 301                          "source stackrc; \
 302                          nova list | grep ' overcloud-controller-[0-9] ' \
 303                          | sed -e 's/^.*ctlplane=//' -e 's/ *|\$//'")
 304     fi
 305     if [[ -z "$CONTROLLER_IPS" ]]; then
 306         echo "ERROR: Could not get CONTROLLER_IPS."
 307         exit 1
 308     fi
 309     for ip in $CONTROLLER_IPS
 310     do
 311         forward_rule="-R $CONSUMER_PORT:localhost:$CONSUMER_PORT"
 312         tunnel_command="sudo ssh $ssh_opts_cpu $COMPUTE_USER@$ip $forward_rule sleep 600"
 313         $tunnel_command > "ssh_tunnel.${ip}.log" 2>&1 < /dev/null &
 314     done
 315 }
 316
 317 stop_consumer() {
 318     pgrep -f "python consumer.py" || return 0
 319     kill $(pgrep -f "python consumer.py")
 320     print_log consumer.log
 321
 322     # NOTE(r-mibu): terminate tunnels to the controller nodes
 323     for ip in $CONTROLLER_IPS
 324     do
 325         forward_rule="-R $CONSUMER_PORT:localhost:$CONSUMER_PORT"
 326         tunnel_command="sudo ssh $ssh_opts_cpu $COMPUTE_USER@$ip $forward_rule sleep 600"
 327         kill $(pgrep -f "$tunnel_command")
 328         print_log "ssh_tunnel.${ip}.log"
 329     done
 330 }
 331
 332 wait_for_vm_launch() {
 333     echo "waiting for vm launch..."
 334
 335     count=0
 336     while [[ ${count} -lt 60 ]]
 337     do
 338         state=$(openstack $as_doctor_user server list | grep " $VM_NAME " | awk '{print $6}')
 339         [[ "$state" == "ACTIVE" ]] && return 0
 340         [[ "$state" == "ERROR" ]] && echo "vm state is ERROR" && exit 1
 341         count=$(($count+1))
 342         sleep 1
 343     done
 344     echo "ERROR: time out while waiting for vm launch"
 345     exit 1
 346 }
 347
 348 inject_failure() {
 349     echo "disabling network of compute host [$COMPUTE_HOST] for 3 mins..."
 350     cat > disable_network.sh << 'END_TXT'
 351 #!/bin/bash -x
 352 dev=$(sudo ip a | awk '/ @COMPUTE_IP@\//{print $7}')
 353 sleep 1
 354 sudo ip link set $dev down
 355 sleep 180
 356 sudo ip link set $dev up
 357 sleep 1
 358 END_TXT
 359     sed -i -e "s/@COMPUTE_IP@/$COMPUTE_IP/" disable_network.sh
 360     chmod +x disable_network.sh
 361     scp $ssh_opts_cpu disable_network.sh "$COMPUTE_USER@$COMPUTE_IP:"
 362     ssh $ssh_opts_cpu "$COMPUTE_USER@$COMPUTE_IP" 'nohup ./disable_network.sh > disable_network.log 2>&1 &'
 363 }
 364
 365 calculate_notification_time() {
 366     detected=$(grep "doctor monitor detected at" monitor.log | awk '{print $5}')
 367     notified=$(grep "doctor consumer notified at" consumer.log | awk '{print $5}')
 368     if ! grep -q "doctor consumer notified at" consumer.log ; then
 369         echo "ERROR: consumer hasn't received fault notification."
 370         exit 1
 371     fi
 372     echo "$notified $detected" | \
 373         awk '{d = $1 - $2; if (d < 1 && d > 0) print d " OK"; else print d " NG"}'
 374 }
 375
 376 check_host_status() {
 377     expected_state=$1
 378
 379     host_status_line=$(openstack $as_doctor_user --os-compute-api-version 2.16 \
 380                        server show $VM_NAME | grep "host_status")
 381     host_status=$(echo $host_status_line | awk '{print $4}')
 382     if [ -z "$host_status" ] ; then
 383         echo "ERROR: host_status not reported by: nova show $VM_NAME"
 384         exit 1
 385     elif [[ "$expected_state" =~ "$host_status" ]] ; then
 386         echo "$VM_NAME showing host_status: $host_status"
 387     else
 388         echo "ERROR: host_status:$host_status not equal to expected_state: $expected_state"
 389         exit 1
 390     fi
 391 }
 392
 393 cleanup() {
 394     set +e
 395     echo "cleanup..."
 396     stop_monitor
 397     stop_inspector
 398     stop_consumer
 399
 400     echo "waiting disabled compute host back to be enabled..."
 401     python ./nova_force_down.py "$COMPUTE_HOST" --unset
 402     sleep 240
 403     check_host_status "UP"
 404     scp $ssh_opts_cpu "$COMPUTE_USER@$COMPUTE_IP:disable_network.log" .
 405     print_log disable_network.log
 406
 407     openstack $as_doctor_user server list | grep -q " $VM_NAME " && openstack $as_doctor_user server delete "$VM_NAME"
 408     sleep 1
 409     alarm_id=$(ceilometer $as_doctor_user alarm-list | grep " $ALARM_NAME " | awk '{print $2}')
 410     sleep 1
 411     [ -n "$alarm_id" ] && ceilometer $as_doctor_user alarm-delete "$alarm_id"
 412     sleep 1
 413
 414     image_id=$(openstack image list | grep " $IMAGE_NAME " | awk '{print $2}')
 415     sleep 1
 416     [ -n "$image_id" ] && openstack image delete "$image_id"
 417     openstack role remove "$DOCTOR_ROLE" --user "$DOCTOR_USER" \
 418                               --project "$DOCTOR_PROJECT"
 419     openstack project delete "$DOCTOR_PROJECT"
 420     openstack user delete "$DOCTOR_USER"
 421 }
 422
 423
 424 echo "Note: doctor/tests/run.sh has been executed."
 425
 426 trap cleanup EXIT
 427
 428 echo "preparing test env..."
 429 get_installer_ip
 430 prepare_ssh_to_cloud
 431
 432 echo "preparing VM image..."
 433 download_image
 434 register_image
 435
 436 echo "creating test user..."
 437 create_test_user
 438
 439 echo "creating VM..."
 440 boot_vm
 441 wait_for_vm_launch
 442 openstack $as_doctor_user server show $VM_NAME
 443
 444 echo "get computer host info..."
 445 get_compute_host_info
 446
 447 echo "creating alarm..."
 448 get_consumer_ip
 449 create_alarm
 450
 451 echo "starting doctor sample components..."
 452 start_inspector
 453 start_monitor
 454 start_consumer
 455
 456 sleep 60
 457 echo "injecting host failure..."
 458 inject_failure
 459 sleep 60
 460
 461 check_host_status "(DOWN|UNKNOWN)"
 462 calculate_notification_time
 463
 464 echo "done"