2 ##############################################################################
3 # Copyright (c) 2016 NEC Corporation and others.
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 ##############################################################################
13 [[ "${CI_DEBUG:-true}" == [Tt]rue ]] && set -x
15 IMAGE_URL=https://launchpad.net/cirros/trunk/0.3.0/+download/cirros-0.3.0-x86_64-disk.img
16 #if an existing image name is provided in the enviroment, use that one
17 IMAGE_NAME=${IMAGE_NAME:-cirros}
18 IMAGE_FILE="${IMAGE_NAME}.img"
22 ALARM_NAME=doctor_alarm1
28 #TODO: change back to `_member_` when JIRA DOCTOR-55 is done
31 TOP_DIR=$(cd $(dirname "$0") && pwd)
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 $DOCTOR_PROJECT"
40 get_compute_host_info() {
41 # get computer host info which VM boot in
42 COMPUTE_HOST=$(openstack $as_doctor_user server show $VM_NAME |
43 grep "OS-EXT-SRV-ATTR:host" | awk '{ print $4 }')
44 compute_host_in_undercloud=${COMPUTE_HOST%%.*}
45 die_if_not_set $LINENO COMPUTE_HOST "Failed to get compute hostname"
47 if is_installer apex; then
48 COMPUTE_USER=${COMPUTE_USER:-heat-admin}
49 COMPUTE_IP=$(sudo ssh $ssh_opts $INSTALLER_IP \
51 nova show $compute_host_in_undercloud \
52 | awk '/ ctlplane network /{print \$5}'")
53 elif is_installer fuel; then
54 COMPUTE_USER=${COMPUTE_USER:-root}
55 node_id=$(echo $compute_host_in_undercloud | cut -d "-" -f 2)
56 COMPUTE_IP=$(sshpass -p r00tme ssh 2>/dev/null $ssh_opts root@${INSTALLER_IP} \
57 "fuel node|awk -F '|' -v id=$node_id '{if (\$1 == id) print \$5}' |xargs")
58 elif is_installer local; then
59 COMPUTE_USER=${COMPUTE_USER:-$(whoami)}
60 COMPUTE_IP=$(getent hosts "$COMPUTE_HOST" | awk '{ print $1 }')
63 die_if_not_set $LINENO COMPUTE_IP "Could not resolve $COMPUTE_HOST. Either manually set COMPUTE_IP or enable DNS resolution."
64 echo "COMPUTE_HOST=$COMPUTE_HOST"
65 echo "COMPUTE_IP=$COMPUTE_IP"
67 # verify connectivity to target compute host
68 ping -c 1 "$COMPUTE_IP"
69 if [[ $? -ne 0 ]] ; then
70 die $LINENO "Can not ping to computer host"
73 # verify ssh to target compute host
74 ssh $ssh_opts_cpu "$COMPUTE_USER@$COMPUTE_IP" 'exit'
75 if [[ $? -ne 0 ]] ; then
76 die $LINENO "Can not ssh to computer host"
81 local get_consumer_command="ip route get $COMPUTE_IP | awk '/ src /{print \$NF}'"
82 if is_installer apex; then
83 CONSUMER_IP=$(sudo ssh $ssh_opts root@$INSTALLER_IP \
84 "$get_consumer_command")
85 elif is_installer fuel; then
86 CONSUMER_IP=$(sudo sshpass -p r00tme ssh $ssh_opts root@${INSTALLER_IP} \
87 "$get_consumer_command")
88 elif is_installer local; then
89 CONSUMER_IP=`$get_consumer_command`
91 echo "CONSUMER_IP=$CONSUMER_IP"
93 die_if_not_set $LINENO CONSUMER_IP "Could not get CONSUMER_IP."
97 #if a different name was provided for the image in the enviroment there's no need to download the image
98 use_existing_image=false
99 openstack image list | grep -q " $IMAGE_NAME " && use_existing_image=true
101 if [[ "$use_existing_image" == false ]] ; then
102 [ -e "$IMAGE_FILE" ] && return 0
103 wget "$IMAGE_URL" -o "$IMAGE_FILE"
108 openstack image list | grep -q " $IMAGE_NAME " && return 0
109 openstack image create "$IMAGE_NAME" \
111 --disk-format "$IMAGE_FORMAT" \
112 --container-format bare \
117 openstack project list | grep -q " $DOCTOR_PROJECT " || {
118 openstack project create "$DOCTOR_PROJECT"
120 openstack user list | grep -q " $DOCTOR_USER " || {
121 openstack user create "$DOCTOR_USER" --password "$DOCTOR_PW" \
122 --project "$DOCTOR_PROJECT"
124 openstack user role list "$DOCTOR_USER" --project "$DOCTOR_PROJECT" \
125 | grep -q " $DOCTOR_ROLE " || {
126 openstack role add "$DOCTOR_ROLE" --user "$DOCTOR_USER" \
127 --project "$DOCTOR_PROJECT"
132 # test VM done with test user, so can test non-admin
133 openstack $as_doctor_user server list | grep -q " $VM_NAME " && return 0
134 openstack $as_doctor_user server create --flavor "$VM_FLAVOR" \
135 --image "$IMAGE_NAME" \
141 # get vm_id as test user
142 ceilometer $as_doctor_user alarm-list | grep -q " $ALARM_NAME " && return 0
143 vm_id=$(openstack $as_doctor_user server list | grep " $VM_NAME " | awk '{print $2}')
144 # TODO(r-mibu): change notification endpoint from localhost to the consumer
145 # IP address (functest container).
146 ceilometer $as_doctor_user alarm-event-create --name "$ALARM_NAME" \
147 --alarm-action "http://localhost:$CONSUMER_PORT/failure" \
148 --description "VM failure" \
150 --repeat-actions False \
151 --severity "moderate" \
152 --event-type compute.instance.update \
153 -q "traits.state=string::error; traits.instance_id=string::$vm_id"
157 pgrep -f "python monitor.py" && return 0
158 sudo -E python monitor.py "$COMPUTE_HOST" "$COMPUTE_IP" "$INSPECTOR_TYPE" \
159 "http://127.0.0.1:$INSPECTOR_PORT/events" > monitor.log 2>&1 &
163 pgrep -f "python monitor.py" || return 0
164 sudo kill $(pgrep -f "python monitor.py")
168 pgrep -f "python consumer.py" && return 0
169 python consumer.py "$CONSUMER_PORT" > consumer.log 2>&1 &
171 # NOTE(r-mibu): create tunnel to the controller nodes, so that we can
172 # avoid some network problems dpends on infra and installers.
173 # This tunnel will be terminated by stop_consumer() or after 10 mins passed.
174 if ! is_installer local; then
175 if is_installer apex; then
176 CONTROLLER_IPS=$(sudo ssh $ssh_opts $INSTALLER_IP \
178 nova list | grep ' overcloud-controller-[0-9] ' \
179 | sed -e 's/^.*ctlplane=//' -e 's/ *|\$//'")
180 elif is_installer fuel; then
181 CONTROLLER_IPS=$(sshpass -p r00tme ssh 2>/dev/null $ssh_opts root@${INSTALLER_IP} \
182 "fuel node | grep controller | cut -d '|' -f 5|xargs")
185 die_if_not_set $LINENO CONTROLLER_IPS "Could not get CONTROLLER_IPS."
186 for ip in $CONTROLLER_IPS
188 forward_rule="-R $CONSUMER_PORT:localhost:$CONSUMER_PORT"
189 tunnel_command="sudo ssh $ssh_opts_cpu $COMPUTE_USER@$ip $forward_rule sleep 600"
190 $tunnel_command > "ssh_tunnel.${ip}.log" 2>&1 < /dev/null &
196 pgrep -f "python consumer.py" || return 0
197 kill $(pgrep -f "python consumer.py")
199 # NOTE(r-mibu): terminate tunnels to the controller nodes
200 if ! is_installer local; then
201 for ip in $CONTROLLER_IPS
203 forward_rule="-R $CONSUMER_PORT:localhost:$CONSUMER_PORT"
204 tunnel_command="sudo ssh $ssh_opts_cpu $COMPUTE_USER@$ip $forward_rule sleep 600"
205 kill $(pgrep -f "$tunnel_command")
210 wait_for_vm_launch() {
211 echo "waiting for vm launch..."
214 while [[ ${count} -lt 60 ]]
216 state=$(openstack $as_doctor_user server list | grep " $VM_NAME " | awk '{print $6}')
217 if [[ "$state" == "ACTIVE" ]]; then
218 # NOTE(cgoncalves): sleeping for a bit to stabilize
219 # See python-openstackclient/functional/tests/compute/v2/test_server.py:wait_for_status
223 if [[ "$state" == "ERROR" ]]; then
224 die $LINENO "vm state is ERROR"
229 die $LINENO "Time out while waiting for VM launch"
233 echo "disabling network of compute host [$COMPUTE_HOST] for 3 mins..."
234 cat > disable_network.sh << 'END_TXT'
236 dev=$(sudo ip a | awk '/ @COMPUTE_IP@\//{print $7}')
238 sudo ip link set $dev down
240 sudo ip link set $dev up
243 sed -i -e "s/@COMPUTE_IP@/$COMPUTE_IP/" disable_network.sh
244 chmod +x disable_network.sh
245 scp $ssh_opts_cpu disable_network.sh "$COMPUTE_USER@$COMPUTE_IP:"
246 ssh $ssh_opts_cpu "$COMPUTE_USER@$COMPUTE_IP" 'nohup ./disable_network.sh > disable_network.log 2>&1 &'
249 profile_performance_poc() {
250 total=`python -c "print(int(($notified-$detected)*1000))"`
252 export DOCTOR_PROFILER_T00=0
253 export DOCTOR_PROFILER_T09=$((total))
254 python profiler-poc.py
257 calculate_notification_time() {
258 detected=$(grep "doctor monitor detected at" monitor.log | awk '{print $10}')
259 notified=$(grep "doctor consumer notified at" consumer.log | awk '{print $10}')
260 if ! grep -q "doctor consumer notified at" consumer.log ; then
261 die $LINENO "Consumer hasn't received fault notification."
264 if [[ PROFILER == 'poc' ]]; then
265 profile_performance_poc
268 echo "$notified $detected" | \
271 if (d < 1 && d > 0) { print d " OK"; exit 0 }
272 else { print d " NG"; exit 1 }
276 check_host_status() {
279 host_status_line=$(openstack $as_doctor_user --os-compute-api-version 2.16 \
280 server show $VM_NAME | grep "host_status")
281 host_status=$(echo $host_status_line | awk '{print $4}')
282 die_if_not_set $LINENO host_status "host_status not reported by: nova show $VM_NAME"
283 if [[ "$expected_state" =~ "$host_status" ]] ; then
284 echo "$VM_NAME showing host_status: $host_status"
286 die $LINENO "host_status:$host_status not equal to expected_state: $expected_state"
297 echo "waiting disabled compute host back to be enabled..."
298 python ./nova_force_down.py "$COMPUTE_HOST" --unset
300 check_host_status "UP"
301 scp $ssh_opts_cpu "$COMPUTE_USER@$COMPUTE_IP:disable_network.log" .
303 openstack $as_doctor_user server list | grep -q " $VM_NAME " && openstack $as_doctor_user server delete "$VM_NAME"
305 alarm_id=$(ceilometer $as_doctor_user alarm-list | grep " $ALARM_NAME " | awk '{print $2}')
307 [ -n "$alarm_id" ] && ceilometer $as_doctor_user alarm-delete "$alarm_id"
310 image_id=$(openstack image list | grep " $IMAGE_NAME " | awk '{print $2}')
312 #if an existing image was used, there's no need to remove it here
313 if [[ "$use_existing_image" == false ]] ; then
314 [ -n "$image_id" ] && openstack image delete "$image_id"
316 openstack role remove "$DOCTOR_ROLE" --user "$DOCTOR_USER" \
317 --project "$DOCTOR_PROJECT"
318 openstack project delete "$DOCTOR_PROJECT"
319 openstack user delete "$DOCTOR_USER"
327 echo "Note: doctor/tests/run.sh has been executed."
331 source $TOP_DIR/functions-common
332 source $TOP_DIR/lib/installer
333 source $TOP_DIR/lib/inspector
337 echo "preparing VM image..."
341 echo "creating test user..."
344 echo "creating VM..."
348 echo "get computer host info..."
349 get_compute_host_info
351 echo "creating alarm..."
352 #TODO: change back to use, network problems depends on infra and installers
356 echo "starting doctor sample components..."
362 echo "injecting host failure..."
366 check_host_status "(DOWN|UNKNOWN)"
367 calculate_notification_time