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="${IMAGE_NAME}.img"
22 ALARM_NAME=doctor_alarm1
28 #TODO: change back to `_member_` when JIRA DOCTOR-55 is done
30 PROFILER_TYPE=${PROFILER_TYPE:-none}
32 TOP_DIR=$(cd $(dirname "$0") && pwd)
34 ssh_opts="-o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no"
35 as_doctor_user="--os-username $DOCTOR_USER --os-password $DOCTOR_PW
36 --os-tenant-name $DOCTOR_PROJECT"
41 get_compute_host_info() {
42 # get computer host info which VM boot in
43 COMPUTE_HOST=$(openstack $as_doctor_user server show $VM_NAME |
44 grep "OS-EXT-SRV-ATTR:host" | awk '{ print $4 }')
45 compute_host_in_undercloud=${COMPUTE_HOST%%.*}
46 die_if_not_set $LINENO COMPUTE_HOST "Failed to get compute hostname"
48 if is_installer apex; then
49 COMPUTE_USER=${COMPUTE_USER:-heat-admin}
50 COMPUTE_IP=$(sudo ssh $ssh_opts $INSTALLER_IP \
52 nova show $compute_host_in_undercloud \
53 | awk '/ ctlplane network /{print \$5}'")
54 elif is_installer fuel; then
55 COMPUTE_USER=${COMPUTE_USER:-root}
56 node_id=$(echo $compute_host_in_undercloud | cut -d "-" -f 2)
57 COMPUTE_IP=$(sshpass -p r00tme ssh 2>/dev/null $ssh_opts root@${INSTALLER_IP} \
58 "fuel node|awk -F '|' -v id=$node_id '{if (\$1 == id) print \$5}' |xargs")
59 elif is_installer local; then
60 COMPUTE_USER=${COMPUTE_USER:-$(whoami)}
61 COMPUTE_IP=$(getent hosts "$COMPUTE_HOST" | awk '{ print $1 }')
64 die_if_not_set $LINENO COMPUTE_IP "Could not resolve $COMPUTE_HOST. Either manually set COMPUTE_IP or enable DNS resolution."
65 echo "COMPUTE_HOST=$COMPUTE_HOST"
66 echo "COMPUTE_IP=$COMPUTE_IP"
68 # verify connectivity to target compute host
69 ping -c 1 "$COMPUTE_IP"
70 if [[ $? -ne 0 ]] ; then
71 die $LINENO "Can not ping to computer host"
74 # verify ssh to target compute host
75 ssh $ssh_opts_cpu "$COMPUTE_USER@$COMPUTE_IP" 'exit'
76 if [[ $? -ne 0 ]] ; then
77 die $LINENO "Can not ssh to computer host"
82 local get_consumer_command="ip route get $COMPUTE_IP | awk '/ src /{print \$NF}'"
83 if is_installer apex; then
84 CONSUMER_IP=$(sudo ssh $ssh_opts root@$INSTALLER_IP \
85 "$get_consumer_command")
86 elif is_installer fuel; then
87 CONSUMER_IP=$(sudo sshpass -p r00tme ssh $ssh_opts root@${INSTALLER_IP} \
88 "$get_consumer_command")
89 elif is_installer local; then
90 CONSUMER_IP=`$get_consumer_command`
92 echo "CONSUMER_IP=$CONSUMER_IP"
94 die_if_not_set $LINENO CONSUMER_IP "Could not get CONSUMER_IP."
98 #if a different name was provided for the image in the enviroment there's no need to download the image
99 use_existing_image=false
100 openstack image list | grep -q " $IMAGE_NAME " && use_existing_image=true
102 if [[ "$use_existing_image" == false ]] ; then
103 [ -e "$IMAGE_FILE" ] && return 0
104 wget "$IMAGE_URL" -o "$IMAGE_FILE"
109 openstack image list | grep -q " $IMAGE_NAME " && return 0
110 openstack image create "$IMAGE_NAME" \
112 --disk-format "$IMAGE_FORMAT" \
113 --container-format bare \
118 openstack project list | grep -q " $DOCTOR_PROJECT " || {
119 openstack project create "$DOCTOR_PROJECT"
121 openstack user list | grep -q " $DOCTOR_USER " || {
122 openstack user create "$DOCTOR_USER" --password "$DOCTOR_PW" \
123 --project "$DOCTOR_PROJECT"
125 openstack user role list "$DOCTOR_USER" --project "$DOCTOR_PROJECT" \
126 | grep -q " $DOCTOR_ROLE " || {
127 openstack role add "$DOCTOR_ROLE" --user "$DOCTOR_USER" \
128 --project "$DOCTOR_PROJECT"
133 # test VM done with test user, so can test non-admin
134 openstack $as_doctor_user server list | grep -q " $VM_NAME " && return 0
135 openstack $as_doctor_user server create --flavor "$VM_FLAVOR" \
136 --image "$IMAGE_NAME" \
142 # get vm_id as test user
143 ceilometer $as_doctor_user alarm-list | grep -q " $ALARM_NAME " && return 0
144 vm_id=$(openstack $as_doctor_user server list | grep " $VM_NAME " | awk '{print $2}')
145 # TODO(r-mibu): change notification endpoint from localhost to the consumer
146 # IP address (functest container).
147 ceilometer $as_doctor_user alarm-event-create --name "$ALARM_NAME" \
148 --alarm-action "http://localhost:$CONSUMER_PORT/failure" \
149 --description "VM failure" \
151 --repeat-actions False \
152 --severity "moderate" \
153 --event-type compute.instance.update \
154 -q "traits.state=string::error; traits.instance_id=string::$vm_id"
158 pgrep -f "python monitor.py" && return 0
159 sudo -E python monitor.py "$COMPUTE_HOST" "$COMPUTE_IP" "$INSPECTOR_TYPE" \
160 "http://127.0.0.1:$INSPECTOR_PORT/events" > monitor.log 2>&1 &
164 pgrep -f "python monitor.py" || return 0
165 sudo kill $(pgrep -f "python monitor.py")
169 pgrep -f "python consumer.py" && return 0
170 python consumer.py "$CONSUMER_PORT" > consumer.log 2>&1 &
172 # NOTE(r-mibu): create tunnel to the controller nodes, so that we can
173 # avoid some network problems dpends on infra and installers.
174 # This tunnel will be terminated by stop_consumer() or after 10 mins passed.
175 if ! is_installer local; then
176 if is_installer apex; then
177 CONTROLLER_IPS=$(sudo ssh $ssh_opts $INSTALLER_IP \
179 nova list | grep ' overcloud-controller-[0-9] ' \
180 | sed -e 's/^.*ctlplane=//' -e 's/ *|\$//'")
181 elif is_installer fuel; then
182 CONTROLLER_IPS=$(sshpass -p r00tme ssh 2>/dev/null $ssh_opts root@${INSTALLER_IP} \
183 "fuel node | grep controller | cut -d '|' -f 5|xargs")
186 die_if_not_set $LINENO CONTROLLER_IPS "Could not get CONTROLLER_IPS."
187 for ip in $CONTROLLER_IPS
189 forward_rule="-R $CONSUMER_PORT:localhost:$CONSUMER_PORT"
190 tunnel_command="sudo ssh $ssh_opts_cpu $COMPUTE_USER@$ip $forward_rule sleep 600"
191 $tunnel_command > "ssh_tunnel.${ip}.log" 2>&1 < /dev/null &
197 pgrep -f "python consumer.py" || return 0
198 kill $(pgrep -f "python consumer.py")
200 # NOTE(r-mibu): terminate tunnels to the controller nodes
201 if ! is_installer local; then
202 for ip in $CONTROLLER_IPS
204 forward_rule="-R $CONSUMER_PORT:localhost:$CONSUMER_PORT"
205 tunnel_command="sudo ssh $ssh_opts_cpu $COMPUTE_USER@$ip $forward_rule sleep 600"
206 kill $(pgrep -f "$tunnel_command")
211 wait_for_vm_launch() {
212 echo "waiting for vm launch..."
215 while [[ ${count} -lt 60 ]]
217 state=$(openstack $as_doctor_user server list | grep " $VM_NAME " | awk '{print $6}')
218 if [[ "$state" == "ACTIVE" ]]; then
219 # NOTE(cgoncalves): sleeping for a bit to stabilize
220 # See python-openstackclient/functional/tests/compute/v2/test_server.py:wait_for_status
224 if [[ "$state" == "ERROR" ]]; then
225 die $LINENO "vm state is ERROR"
230 die $LINENO "Time out while waiting for VM launch"
234 echo "disabling network of compute host [$COMPUTE_HOST] for 3 mins..."
235 cat > disable_network.sh << 'END_TXT'
237 dev=$(sudo ip a | awk '/ @COMPUTE_IP@\//{print $7}')
239 sudo ip link set $dev down
240 echo "doctor set host down at" $(date "+%s.%N")
242 sudo ip link set $dev up
245 sed -i -e "s/@COMPUTE_IP@/$COMPUTE_IP/" disable_network.sh
246 chmod +x disable_network.sh
247 scp $ssh_opts_cpu disable_network.sh "$COMPUTE_USER@$COMPUTE_IP:"
248 ssh $ssh_opts_cpu "$COMPUTE_USER@$COMPUTE_IP" 'nohup ./disable_network.sh > disable_network.log 2>&1 &'
251 profile_performance_poc() {
252 triggered=$(grep "^doctor set host down at" disable_network.log |\
253 sed -e "s/^.* at //")
254 vmdown=$(grep "doctor mark vm.* error at" inspector.log |tail -n 1 |\
255 sed -e "s/^.* at //")
256 hostdown=$(grep "doctor mark host.* down at" inspector.log |\
257 sed -e "s/^.* at //")
259 #calculate the relative interval to triggered(T00)
260 export DOCTOR_PROFILER_T00=0
261 export DOCTOR_PROFILER_T01=$(echo "($detected-$triggered)*1000/1" |bc)
262 export DOCTOR_PROFILER_T03=$(echo "($vmdown-$triggered)*1000/1" |bc)
263 export DOCTOR_PROFILER_T04=$(echo "($hostdown-$triggered)*1000/1" |bc)
264 export DOCTOR_PROFILER_T09=$(echo "($notified-$triggered)*1000/1" |bc)
266 python profiler-poc.py
269 calculate_notification_time() {
270 if ! grep -q "doctor consumer notified at" consumer.log ; then
271 die $LINENO "Consumer hasn't received fault notification."
274 #keep 'at' as the last keyword just before the value, and
275 #use regex to get value instead of the fixed column
276 detected=$(grep "doctor monitor detected at" monitor.log |\
277 sed -e "s/^.* at //")
278 notified=$(grep "doctor consumer notified at" consumer.log |\
279 sed -e "s/^.* at //")
281 if [[ "$PROFILER_TYPE" == "poc" ]]; then
282 profile_performance_poc
285 echo "$notified $detected" | \
288 if (d < 1 && d > 0) { print d " OK"; exit 0 }
289 else { print d " NG"; exit 1 }
293 check_host_status() {
296 host_status_line=$(openstack $as_doctor_user --os-compute-api-version 2.16 \
297 server show $VM_NAME | grep "host_status")
298 host_status=$(echo $host_status_line | awk '{print $4}')
299 die_if_not_set $LINENO host_status "host_status not reported by: nova show $VM_NAME"
300 if [[ "$expected_state" =~ "$host_status" ]] ; then
301 echo "$VM_NAME showing host_status: $host_status"
303 die $LINENO "host_status:$host_status not equal to expected_state: $expected_state"
314 echo "waiting disabled compute host back to be enabled..."
315 python ./nova_force_down.py "$COMPUTE_HOST" --unset
317 check_host_status "UP"
318 scp $ssh_opts_cpu "$COMPUTE_USER@$COMPUTE_IP:disable_network.log" .
320 openstack $as_doctor_user server list | grep -q " $VM_NAME " && openstack $as_doctor_user server delete "$VM_NAME"
322 alarm_id=$(ceilometer $as_doctor_user alarm-list | grep " $ALARM_NAME " | awk '{print $2}')
324 [ -n "$alarm_id" ] && ceilometer $as_doctor_user alarm-delete "$alarm_id"
327 image_id=$(openstack image list | grep " $IMAGE_NAME " | awk '{print $2}')
329 #if an existing image was used, there's no need to remove it here
330 if [[ "$use_existing_image" == false ]] ; then
331 [ -n "$image_id" ] && openstack image delete "$image_id"
333 openstack role remove "$DOCTOR_ROLE" --user "$DOCTOR_USER" \
334 --project "$DOCTOR_PROJECT"
335 openstack project delete "$DOCTOR_PROJECT"
336 openstack user delete "$DOCTOR_USER"
344 echo "Note: doctor/tests/run.sh has been executed."
345 git log --oneline -1 || true # ignore even you don't have git installed
349 source $TOP_DIR/functions-common
350 source $TOP_DIR/lib/installer
351 source $TOP_DIR/lib/inspector
355 echo "preparing VM image..."
359 echo "creating test user..."
362 echo "creating VM..."
366 echo "get computer host info..."
367 get_compute_host_info
369 echo "creating alarm..."
370 #TODO: change back to use, network problems depends on infra and installers
374 echo "starting doctor sample components..."
380 echo "injecting host failure..."
384 check_host_status "(DOWN|UNKNOWN)"
385 calculate_notification_time