-#!/bin/bash -ex
+#!/bin/bash -e
##############################################################################
# Copyright (c) 2016 NEC Corporation and others.
#
# http://www.apache.org/licenses/LICENSE-2.0
##############################################################################
+[[ "${CI_DEBUG:-true}" == "true" ]] && set -x
+
IMAGE_URL=https://launchpad.net/cirros/trunk/0.3.0/+download/cirros-0.3.0-x86_64-disk.img
IMAGE_NAME=cirros
IMAGE_FILE="${IMAGE_NAME}.img"
ALARM_NAME=doctor_alarm1
INSPECTOR_PORT=12345
CONSUMER_PORT=12346
+DOCTOR_USER=doctor
+DOCTOR_PW=doctor
+DOCTOR_PROJECT=doctor
+DOCTOR_ROLE=_member_
SUPPORTED_INSTALLER_TYPES="apex local"
INSTALLER_TYPE=${INSTALLER_TYPE:-apex}
--file "$IMAGE_FILE"
}
+create_test_user() {
+ openstack user list | grep -q " $DOCTOR_USER " || {
+ openstack user create "$DOCTOR_USER" --password "$DOCTOR_PW"
+ }
+ openstack project list | grep -q " $DOCTOR_PROJECT " || {
+ openstack project create "$DOCTOR_PROJECT"
+ }
+ openstack user role list "$DOCTOR_USER" --project "$DOCTOR_PROJECT" \
+ | grep -q " $DOCTOR_ROLE " || {
+ openstack role add "$DOCTOR_ROLE" --user "$DOCTOR_USER" \
+ --project "$DOCTOR_PROJECT"
+ }
+}
+
+change_to_doctor_user() {
+ export OS_USERNAME="$DOCTOR_USER"
+ export OS_PASSWORD="$DOCTOR_PW"
+ export OS_PROJECT_NAME="$DOCTOR_PROJECT"
+ export OS_TENANT_NAME="$DOCTOR_PROJECT"
+}
+
boot_vm() {
- nova list | grep -q " $VM_NAME " && return 0
- nova boot --flavor "$VM_FLAVOR" \
- --image "$IMAGE_NAME" \
- "$VM_NAME"
- sleep 1
+ (
+ # test VM done with test user, so can test non-admin
+ change_to_doctor_user
+ nova list | grep -q " $VM_NAME " && return 0
+ nova boot --flavor "$VM_FLAVOR" \
+ --image "$IMAGE_NAME" \
+ "$VM_NAME"
+ sleep 1
+ )
+
}
create_alarm() {
- ceilometer alarm-list | grep -q " $ALARM_NAME " && return 0
- vm_id=$(nova list | grep " $VM_NAME " | awk '{print $2}')
- ceilometer alarm-event-create --name "$ALARM_NAME" \
- --alarm-action "http://localhost:$CONSUMER_PORT/failure" \
- --description "VM failure" \
- --enabled True \
- --repeat-actions False \
- --severity "moderate" \
- --event-type compute.instance.update \
- -q "traits.state=string::error; traits.instance_id=string::$vm_id"
+ (
+ # get vm_id as test user
+ change_to_doctor_user
+ ceilometer alarm-list | grep -q " $ALARM_NAME " && return 0
+ vm_id=$(nova list | grep " $VM_NAME " | awk '{print $2}')
+ ceilometer alarm-event-create --name "$ALARM_NAME" \
+ --alarm-action "http://localhost:$CONSUMER_PORT/failure" \
+ --description "VM failure" \
+ --enabled True \
+ --repeat-actions False \
+ --severity "moderate" \
+ --event-type compute.instance.update \
+ -q "traits.state=string::error; traits.instance_id=string::$vm_id"
+ )
}
+
start_monitor() {
pgrep -f "python monitor.py" && return 0
sudo python monitor.py "$COMPUTE_HOST" "$COMPUTE_IP" \
wait_for_vm_launch() {
echo "waiting for vm launch..."
- while true
- do
- state=$(nova list | grep " $VM_NAME " | awk '{print $6}')
- [[ "$state" == "ACTIVE" ]] && return 0
- sleep 1
- done
+
+ (
+ # get VM state as test user
+ change_to_doctor_user
+
+ count=0
+ while [[ ${count} -lt 60 ]]
+ do
+ state=$(nova list | grep " $VM_NAME " | awk '{print $6}')
+ [[ "$state" == "ACTIVE" ]] && return 0
+ [[ "$state" == "ERROR" ]] && echo "vm state is ERROR" && exit 1
+ count=$(($count+1))
+ sleep 1
+ done
+ echo "ERROR: time out while waiting for vm launch"
+ exit 1
+ )
}
inject_failure() {
- echo "disabling network of comupte host [$COMPUTE_HOST] for 3 mins..."
+ echo "disabling network of compute host [$COMPUTE_HOST] for 3 mins..."
cat > disable_network.sh << 'END_TXT'
#!/bin/bash -x
-dev=$(/usr/sbin/ip route | awk '/^default/{print $5}')
+dev=$(sudo ip route | awk '/^default/{print $5}')
sleep 1
-echo sudo ip link set $dev down
+sudo ip link set $dev down
sleep 180
-echo sudo ip link set $dev up
+sudo ip link set $dev up
sleep 1
END_TXT
chmod +x disable_network.sh
}
calculate_notification_time() {
- detect=$(grep "doctor monitor detected at" monitor.log | awk '{print $5}')
+ detected=$(grep "doctor monitor detected at" monitor.log | awk '{print $5}')
notified=$(grep "doctor consumer notified at" consumer.log | awk '{print $5}')
- duration=$(echo "$notified $detect" | awk '{print $1 - $2 }')
- echo "$notified $detect" | \
- awk '{d = $1 - $2; if (d < 1 ) print d " OK"; else print d " NG"}'
+ echo "$notified $detected" | \
+ awk '{d = $1 - $2; if (d < 1 && d > 0) print d " OK"; else print d " NG"}'
+}
+
+check_host_status_down() {
+ (
+ change_to_doctor_user
+
+ host_status_line=$(nova show $VM_NAME | grep "host_status")
+ [[ $? -ne 0 ]] && {
+ echo "ERROR: host_status not configured for owner in Nova policy.json"
+ }
+
+ host_status=$(echo $host_status_line | awk '{print $4}')
+ [[ "$host_status" == "DOWN" ]] && {
+ echo "$VM_NAME showing host_status: $host_status"
+ }
+ echo "ERROR: host_status not reported by: nova show $VM_NAME"
+ )
}
cleanup() {
python ./nova_force_down.py "$COMPUTE_HOST" --unset
sleep 1
- nova delete "$VM_NAME"
- sleep 1
- alarm_id=$(ceilometer alarm-list | grep " $ALARM_NAME " | awk '{print $2}')
- sleep 1
- [ -n "$alarm_id" ] && ceilometer alarm-delete "$alarm_id"
- sleep 1
+ (
+ change_to_doctor_user
+ nova list | grep -q " $VM_NAME " && nova delete "$VM_NAME"
+ sleep 1
+ alarm_id=$(ceilometer alarm-list | grep " $ALARM_NAME " | awk '{print $2}')
+ sleep 1
+ [ -n "$alarm_id" ] && ceilometer alarm-delete "$alarm_id"
+ sleep 1
+ )
image_id=$(glance image-list | grep " $IMAGE_NAME " | awk '{print $2}')
sleep 1
[ -n "$image_id" ] && glance image-delete "$image_id"
+ openstack role remove "$DOCTOR_ROLE" --user "$DOCTOR_USER" \
+ --project "$DOCTOR_PROJECT"
+ openstack project delete "$DOCTOR_PROJECT"
+ openstack user delete "$DOCTOR_USER"
+
#TODO: add host status check via nova admin api
echo "waiting disabled compute host back to be enabled..."
sleep 180
start_inspector
start_consumer
+echo "creating test user..."
+create_test_user
+
echo "creating VM and alarm..."
boot_vm
create_alarm
inject_failure
sleep 10
+check_host_status_down
calculate_notification_time
echo "done"
Code Review / doctor.git / blobdiff