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 [[ "${CI_DEBUG:-true}" == [Tt]rue ]] && set -x
  12
  13 IMAGE_URL=https://launchpad.net/cirros/trunk/0.3.0/+download/cirros-0.3.0-x86_64-disk.img
  14 #if an existing image name is provided in the enviroment, use that one
  15 IMAGE_NAME=${IMAGE_NAME:-cirros}
  16 IMAGE_FILE="${IMAGE_NAME}.img"
  17 IMAGE_FORMAT=qcow2
  18 VM_NAME=doctor_vm1
  19 VM_FLAVOR=m1.tiny
  20 ALARM_NAME=doctor_alarm1
  21 INSPECTOR_PORT=12345
  22 CONSUMER_PORT=12346
  23 DOCTOR_USER=doctor
  24 DOCTOR_PW=doctor
  25 DOCTOR_PROJECT=doctor
  26 #TODO: change back to `_member_` when JIRA DOCTOR-55 is done
  27 DOCTOR_ROLE=admin
  28
  29 SUPPORTED_INSPECTOR_TYPES="sample congress"
  30 INSPECTOR_TYPE=${INSPECTOR_TYPE:-sample}
  31 TOP_DIR=$(cd $(dirname "$0") && pwd)
  32
  33 ssh_opts="-o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no"
  34 as_doctor_user="--os-username $DOCTOR_USER --os-password $DOCTOR_PW
  35                 --os-tenant-name $DOCTOR_PROJECT"
  36
  37 if [[ ! "$SUPPORTED_INSPECTOR_TYPES" =~ "$INSPECTOR_TYPE" ]] ; then
  38     echo "ERROR: INSPECTOR_TYPE=$INSPECTOR_TYPE is not supported."
  39     exit 1
  40 fi
  41
  42 get_compute_host_info() {
  43     # get computer host info which VM boot in
  44     COMPUTE_HOST=$(openstack $as_doctor_user server show $VM_NAME |
  45                    grep "OS-EXT-SRV-ATTR:host" | awk '{ print $4 }')
  46     compute_host_in_undercloud=${COMPUTE_HOST%%.*}
  47     die_if_not_set $LINENO COMPUTE_HOST "Failed to get compute hostname"
  48
  49     if is_installer apex; then
  50         COMPUTE_USER=${COMPUTE_USER:-heat-admin}
  51         COMPUTE_IP=$(sudo ssh $ssh_opts $INSTALLER_IP \
  52              "source stackrc; \
  53              nova show $compute_host_in_undercloud \
  54              | awk '/ ctlplane network /{print \$5}'")
  55     elif is_installer fuel; then
  56         COMPUTE_USER=${COMPUTE_USER:-root}
  57         node_id=$(echo $compute_host_in_undercloud | cut -d "-" -f 2)
  58         COMPUTE_IP=$(sshpass -p r00tme ssh 2>/dev/null $ssh_opts root@${INSTALLER_IP} \
  59              "fuel node|awk -F '|' -v id=$node_id '{if (\$1 == id) print \$5}' |xargs")
  60     elif is_installer local; then
  61         COMPUTE_USER=${COMPUTE_USER:-$(whoami)}
  62         COMPUTE_IP=$(getent hosts "$COMPUTE_HOST" | awk '{ print $1 }')
  63     fi
  64
  65     die_if_not_set $LINENO COMPUTE_IP "Could not resolve $COMPUTE_HOST. Either manually set COMPUTE_IP or enable DNS resolution."
  66     echo "COMPUTE_HOST=$COMPUTE_HOST"
  67     echo "COMPUTE_IP=$COMPUTE_IP"
  68
  69     # verify connectivity to target compute host
  70     ping -c 1 "$COMPUTE_IP"
  71     if [[ $? -ne 0 ]] ; then
  72         die $LINENO "Can not ping to computer host"
  73     fi
  74
  75     # verify ssh to target compute host
  76     ssh $ssh_opts_cpu "$COMPUTE_USER@$COMPUTE_IP" 'exit'
  77     if [[ $? -ne 0 ]] ; then
  78         die $LINENO "Can not ssh to computer host"
  79     fi
  80 }
  81
  82 get_consumer_ip() {
  83     local get_consumer_command="ip route get $COMPUTE_IP | awk '/ src /{print \$NF}'"
  84     if is_installer apex; then
  85         CONSUMER_IP=$(sudo ssh $ssh_opts root@$INSTALLER_IP \
  86                       "$get_consumer_command")
  87     elif is_installer fuel; then
  88         CONSUMER_IP=$(sudo sshpass -p r00tme ssh $ssh_opts root@${INSTALLER_IP} \
  89                       "$get_consumer_command")
  90     elif is_installer local; then
  91         CONSUMER_IP=`$get_consumer_command`
  92     fi
  93     echo "CONSUMER_IP=$CONSUMER_IP"
  94
  95     die_if_not_set $LINENO CONSUMER_IP "Could not get CONSUMER_IP."
  96 }
  97
  98 download_image() {
  99     #if a different name was provided for the image in the enviroment there's no need to download the image
 100     use_existing_image=false
 101     openstack image list | grep -q " $IMAGE_NAME " && use_existing_image=true
 102
 103     if [[ "$use_existing_image" == false ]] ; then
 104         [ -e "$IMAGE_FILE" ] && return 0
 105         wget "$IMAGE_URL" -o "$IMAGE_FILE"
 106     fi
 107 }
 108
 109 register_image() {
 110     openstack image list | grep -q " $IMAGE_NAME " && return 0
 111     openstack image create "$IMAGE_NAME" \
 112                            --public \
 113                            --disk-format "$IMAGE_FORMAT" \
 114                            --container-format bare \
 115                            --file "$IMAGE_FILE"
 116 }
 117
 118 create_test_user() {
 119     openstack project list | grep -q " $DOCTOR_PROJECT " || {
 120         openstack project create "$DOCTOR_PROJECT"
 121     }
 122     openstack user list | grep -q " $DOCTOR_USER " || {
 123         openstack user create "$DOCTOR_USER" --password "$DOCTOR_PW" \
 124                               --project "$DOCTOR_PROJECT"
 125     }
 126     openstack user role list "$DOCTOR_USER" --project "$DOCTOR_PROJECT" \
 127     | grep -q " $DOCTOR_ROLE " || {
 128         openstack role add "$DOCTOR_ROLE" --user "$DOCTOR_USER" \
 129                            --project "$DOCTOR_PROJECT"
 130     }
 131 }
 132
 133 boot_vm() {
 134     # test VM done with test user, so can test non-admin
 135     openstack $as_doctor_user server list | grep -q " $VM_NAME " && return 0
 136     openstack $as_doctor_user server create --flavor "$VM_FLAVOR" \
 137                             --image "$IMAGE_NAME" \
 138                             "$VM_NAME"
 139     sleep 1
 140 }
 141
 142 create_alarm() {
 143     # get vm_id as test user
 144     ceilometer $as_doctor_user alarm-list | grep -q " $ALARM_NAME " && return 0
 145     vm_id=$(openstack $as_doctor_user server list | grep " $VM_NAME " | awk '{print $2}')
 146     # TODO(r-mibu): change notification endpoint from localhost to the consumer
 147     # IP address (functest container).
 148     ceilometer $as_doctor_user alarm-event-create --name "$ALARM_NAME" \
 149         --alarm-action "http://localhost:$CONSUMER_PORT/failure" \
 150         --description "VM failure" \
 151         --enabled True \
 152         --repeat-actions False \
 153         --severity "moderate" \
 154         --event-type compute.instance.update \
 155         -q "traits.state=string::error; traits.instance_id=string::$vm_id"
 156 }
 157
 158 start_monitor() {
 159     pgrep -f "python monitor.py" && return 0
 160     sudo -E python monitor.py "$COMPUTE_HOST" "$COMPUTE_IP" "$INSPECTOR_TYPE" \
 161         "http://127.0.0.1:$INSPECTOR_PORT/events" > monitor.log 2>&1 &
 162 }
 163
 164 stop_monitor() {
 165     pgrep -f "python monitor.py" || return 0
 166     sudo kill $(pgrep -f "python monitor.py")
 167 }
 168
 169 congress_add_rule() {
 170     name=$1
 171     policy=$2
 172     rule=$3
 173
 174     if ! openstack congress policy rule list $policy | grep -q -e "// Name: $name$" ; then
 175         openstack congress policy rule create --name $name $policy "$rule"
 176     fi
 177 }
 178
 179 congress_del_rule() {
 180     name=$1
 181     policy=$2
 182
 183     if openstack congress policy rule list $policy | grep -q -e "^// Name: $name$" ; then
 184         openstack congress policy rule delete $policy $name
 185     fi
 186 }
 187
 188 congress_setup_rules() {
 189     congress_add_rule host_down classification \
 190         'host_down(host) :-
 191             doctor:events(hostname=host, type="compute.host.down", status="down")'
 192
 193     congress_add_rule active_instance_in_host classification \
 194         'active_instance_in_host(vmid, host) :-
 195             nova:servers(id=vmid, host_name=host, status="ACTIVE")'
 196
 197     congress_add_rule host_force_down classification \
 198         'execute[nova:services.force_down(host, "nova-compute", "True")] :-
 199             host_down(host)'
 200
 201     congress_add_rule error_vm_states classification \
 202         'execute[nova:servers.reset_state(vmid, "error")] :-
 203             host_down(host),
 204             active_instance_in_host(vmid, host)'
 205 }
 206
 207 start_inspector() {
 208     if [[ "$INSPECTOR_TYPE" == "sample" ]] ; then
 209         pgrep -f "python inspector.py" && return 0
 210         python inspector.py "$INSPECTOR_PORT" > inspector.log 2>&1 &
 211     elif [[ "$INSPECTOR_TYPE" == "congress" ]] ; then
 212         nova_api_min_version="2.11"
 213         nova_api_version=$(openstack congress datasource list | \
 214                            grep nova | grep -Po "(?<='api_version': ')[^']*")
 215         if ! is_set nova_api_version; then
 216             nova_api_version="2.0"
 217         fi
 218         if [[ "$nova_api_version" < "$nova_api_min_version" ]]; then
 219             die $LINENO "Congress Nova datasource API version < $nova_api_min_version ($nova_api_version)"
 220         fi
 221         openstack congress driver list | grep -q " doctor "
 222         openstack congress datasource list | grep -q " doctor " || {
 223             openstack congress datasource create doctor doctor
 224         }
 225         congress_setup_rules
 226     fi
 227 }
 228
 229 stop_inspector() {
 230     if [[ "$INSPECTOR_TYPE" == "sample" ]] ; then
 231         pgrep -f "python inspector.py" || return 0
 232         kill $(pgrep -f "python inspector.py")
 233     elif [[ "$INSPECTOR_TYPE" == "congress" ]] ; then
 234         congress_del_rule host_force_down classification
 235         congress_del_rule error_vm_states classification
 236         congress_del_rule active_instance_in_host classification
 237         congress_del_rule host_down classification
 238     fi
 239 }
 240
 241 start_consumer() {
 242     pgrep -f "python consumer.py" && return 0
 243     python consumer.py "$CONSUMER_PORT" > consumer.log 2>&1 &
 244
 245     # NOTE(r-mibu): create tunnel to the controller nodes, so that we can
 246     # avoid some network problems dpends on infra and installers.
 247     # This tunnel will be terminated by stop_consumer() or after 10 mins passed.
 248     if ! is_installer local; then
 249         if is_installer apex; then
 250             CONTROLLER_IPS=$(sudo ssh $ssh_opts $INSTALLER_IP \
 251                              "source stackrc; \
 252                              nova list | grep ' overcloud-controller-[0-9] ' \
 253                              | sed -e 's/^.*ctlplane=//' -e 's/ *|\$//'")
 254         elif is_installer fuel; then
 255             CONTROLLER_IPS=$(sshpass -p r00tme ssh 2>/dev/null $ssh_opts root@${INSTALLER_IP} \
 256                             "fuel node | grep controller | cut -d '|' -f 5|xargs")
 257         fi
 258
 259         die_if_not_set $LINENO CONTROLLER_IPS "Could not get CONTROLLER_IPS."
 260         for ip in $CONTROLLER_IPS
 261         do
 262             forward_rule="-R $CONSUMER_PORT:localhost:$CONSUMER_PORT"
 263             tunnel_command="sudo ssh $ssh_opts_cpu $COMPUTE_USER@$ip $forward_rule sleep 600"
 264             $tunnel_command > "ssh_tunnel.${ip}.log" 2>&1 < /dev/null &
 265         done
 266     fi
 267 }
 268
 269 stop_consumer() {
 270     pgrep -f "python consumer.py" || return 0
 271     kill $(pgrep -f "python consumer.py")
 272
 273     # NOTE(r-mibu): terminate tunnels to the controller nodes
 274     if ! is_installer local; then
 275         for ip in $CONTROLLER_IPS
 276         do
 277             forward_rule="-R $CONSUMER_PORT:localhost:$CONSUMER_PORT"
 278             tunnel_command="sudo ssh $ssh_opts_cpu $COMPUTE_USER@$ip $forward_rule sleep 600"
 279             kill $(pgrep -f "$tunnel_command")
 280         done
 281     fi
 282 }
 283
 284 wait_for_vm_launch() {
 285     echo "waiting for vm launch..."
 286
 287     count=0
 288     while [[ ${count} -lt 60 ]]
 289     do
 290         state=$(openstack $as_doctor_user server list | grep " $VM_NAME " | awk '{print $6}')
 291         if [[ "$state" == "ACTIVE" ]]; then
 292             # NOTE(cgoncalves): sleeping for a bit to stabilize
 293             # See python-openstackclient/functional/tests/compute/v2/test_server.py:wait_for_status
 294             sleep 5
 295             return 0
 296         fi
 297         if [[ "$state" == "ERROR" ]]; then
 298             die $LINENO "vm state is ERROR"
 299         fi
 300         count=$(($count+1))
 301         sleep 1
 302     done
 303     die $LINENO "Time out while waiting for VM launch"
 304 }
 305
 306 inject_failure() {
 307     echo "disabling network of compute host [$COMPUTE_HOST] for 3 mins..."
 308     cat > disable_network.sh << 'END_TXT'
 309 #!/bin/bash -x
 310 dev=$(sudo ip a | awk '/ @COMPUTE_IP@\//{print $7}')
 311 sleep 1
 312 sudo ip link set $dev down
 313 sleep 180
 314 sudo ip link set $dev up
 315 sleep 1
 316 END_TXT
 317     sed -i -e "s/@COMPUTE_IP@/$COMPUTE_IP/" disable_network.sh
 318     chmod +x disable_network.sh
 319     scp $ssh_opts_cpu disable_network.sh "$COMPUTE_USER@$COMPUTE_IP:"
 320     ssh $ssh_opts_cpu "$COMPUTE_USER@$COMPUTE_IP" 'nohup ./disable_network.sh > disable_network.log 2>&1 &'
 321 }
 322
 323 calculate_notification_time() {
 324     detected=$(grep "doctor monitor detected at" monitor.log | awk '{print $10}')
 325     notified=$(grep "doctor consumer notified at" consumer.log | awk '{print $10}')
 326     if ! grep -q "doctor consumer notified at" consumer.log ; then
 327         die $LINENO "Consumer hasn't received fault notification."
 328     fi
 329     echo "$notified $detected" | \
 330         awk '{
 331             d = $1 - $2;
 332             if (d < 1 && d > 0) { print d " OK"; exit 0 }
 333             else { print d " NG"; exit 1 }
 334         }'
 335 }
 336
 337 check_host_status() {
 338     expected_state=$1
 339
 340     host_status_line=$(openstack $as_doctor_user --os-compute-api-version 2.16 \
 341                        server show $VM_NAME | grep "host_status")
 342     host_status=$(echo $host_status_line | awk '{print $4}')
 343     die_if_not_set $LINENO host_status "host_status not reported by: nova show $VM_NAME"
 344     if [[ "$expected_state" =~ "$host_status" ]] ; then
 345         echo "$VM_NAME showing host_status: $host_status"
 346     else
 347         die $LINENO "host_status:$host_status not equal to expected_state: $expected_state"
 348     fi
 349 }
 350
 351 cleanup() {
 352     set +e
 353     echo "cleanup..."
 354     stop_monitor
 355     stop_inspector
 356     stop_consumer
 357
 358     echo "waiting disabled compute host back to be enabled..."
 359     python ./nova_force_down.py "$COMPUTE_HOST" --unset
 360     sleep 240
 361     check_host_status "UP"
 362     scp $ssh_opts_cpu "$COMPUTE_USER@$COMPUTE_IP:disable_network.log" .
 363
 364     openstack $as_doctor_user server list | grep -q " $VM_NAME " && openstack $as_doctor_user server delete "$VM_NAME"
 365     sleep 1
 366     alarm_id=$(ceilometer $as_doctor_user alarm-list | grep " $ALARM_NAME " | awk '{print $2}')
 367     sleep 1
 368     [ -n "$alarm_id" ] && ceilometer $as_doctor_user alarm-delete "$alarm_id"
 369     sleep 1
 370
 371     image_id=$(openstack image list | grep " $IMAGE_NAME " | awk '{print $2}')
 372     sleep 1
 373     #if an existing image was used, there's no need to remove it here
 374     if [[ "$use_existing_image" == false ]] ; then
 375         [ -n "$image_id" ] && openstack image delete "$image_id"
 376     fi
 377     openstack role remove "$DOCTOR_ROLE" --user "$DOCTOR_USER" \
 378                               --project "$DOCTOR_PROJECT"
 379     openstack project delete "$DOCTOR_PROJECT"
 380     openstack user delete "$DOCTOR_USER"
 381
 382     cleanup_installer
 383 }
 384
 385
 386 echo "Note: doctor/tests/run.sh has been executed."
 387
 388 trap cleanup EXIT
 389
 390 source $TOP_DIR/functions-common
 391 source $TOP_DIR/lib/installer
 392
 393 setup_installer
 394
 395 echo "preparing VM image..."
 396 download_image
 397 register_image
 398
 399 echo "creating test user..."
 400 create_test_user
 401
 402 echo "creating VM..."
 403 boot_vm
 404 wait_for_vm_launch
 405
 406 echo "get computer host info..."
 407 get_compute_host_info
 408
 409 echo "creating alarm..."
 410 #TODO: change back to use, network problems depends on infra and installers
 411 #get_consumer_ip
 412 create_alarm
 413
 414 echo "starting doctor sample components..."
 415 start_inspector
 416 start_monitor
 417 start_consumer
 418
 419 sleep 60
 420 echo "injecting host failure..."
 421 inject_failure
 422 sleep 60
 423
 424 check_host_status "(DOWN|UNKNOWN)"
 425 calculate_notification_time
 426
 427 echo "done"