ALARM_NAME=doctor_alarm1
INSPECTOR_PORT=12345
CONSUMER_PORT=12346
-TEST_USER=demo
-TEST_PW=demo
-TEST_PROJECT=demo
-TEST_ROLE=_member_
+DOCTOR_USER=doctor
+DOCTOR_PW=doctor
+DOCTOR_PROJECT=doctor
+#TODO: change back to `_member_` when JIRA DOCTOR-55 is done
+DOCTOR_ROLE=admin
SUPPORTED_INSTALLER_TYPES="apex local"
INSTALLER_TYPE=${INSTALLER_TYPE:-apex}
INSTALLER_IP=${INSTALLER_IP:-none}
-COMPUTE_HOST=${COMPUTE_HOST:-overcloud-novacompute-0}
-COMPUTE_IP=${COMPUTE_IP:-none}
-COMPUTE_USER=${COMPUTE_USER:-heat-admin}
+COMPUTE_USER=${COMPUTE_USER:-none}
ssh_opts="-o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no"
if [[ ! "$SUPPORTED_INSTALLER_TYPES" =~ "$INSTALLER_TYPE" ]] ; then
exit 1
fi
-prepare_compute_ssh() {
- ssh_opts_cpu="$ssh_opts"
+get_compute_host_info() {
+ (
+ change_to_doctor_user
+ # get computer host info which VM boot in
+ export COMPUTE_HOST=$(openstack server show $VM_NAME | \
+ grep "OS-EXT-SRV-ATTR:host" | awk '{ print $4 }' |
+ awk -F '.' '{print $1}')
+ )
if [[ "$INSTALLER_TYPE" == "apex" ]] ; then
+ COMPUTE_USER=${COMPUTE_USER:-heat-admin}
if [[ "$INSTALLER_IP" == "none" ]] ; then
instack_mac=$(sudo virsh domiflist instack | awk '/default/{print $5}')
INSTALLER_IP=$(/usr/sbin/arp -e | grep ${instack_mac} | awk '{print $1}')
fi
-
- if [[ "$COMPUTE_IP" == "none" ]] ; then
- COMPUTE_IP=$(sudo ssh $ssh_opts $INSTALLER_IP \
- "source stackrc; \
- nova show $COMPUTE_HOST \
- | awk '/ ctlplane network /{print \$5}'")
+ COMPUTE_IP=$(sudo ssh $ssh_opts $INSTALLER_IP \
+ "source stackrc; \
+ openstack server show $COMPUTE_HOST \
+ | awk '/ ctlplane network /{print \$5}'")
+ elif [[ "$INSTALLER_TYPE" == "local" ]] ; then
+ COMPUTE_USER=${COMPUTE_USER:-$(whoami)}
+ COMPUTE_IP=$(getent hosts "$COMPUTE_HOST" | awk '{ print $1 }')
+ if [[ -z "$COMPUTE_IP" ]]; then
+ echo "ERROR: Could not resolve $COMPUTE_HOST. Either manually set COMPUTE_IP or enable DNS resolution."
+ exit 1
fi
+ fi
- # get ssh key from installer node
+ # verify connectivity to target compute host
+ ping -c 1 "$COMPUTE_IP"
+ if [[ $? -ne 0 ]] ; then
+ echo "ERROR: can not ping to computer host"
+ exit 1
+ fi
+}
+
+prepare_compute_ssh() {
+ ssh_opts_cpu="$ssh_opts"
+
+ # get ssh key from installer node
+ if [[ "$INSTALLER_TYPE" == "apex" ]] ; then
sudo scp $ssh_opts root@"$INSTALLER_IP":/home/stack/.ssh/id_rsa instack_key
- sudo chown $(whoami):$(whoami) instack_key
- chmod 400 instack_key
- ssh_opts_cpu+=" -i instack_key"
elif [[ "$INSTALLER_TYPE" == "local" ]] ; then
- if [[ "$COMPUTE_IP" == "none" ]] ; then
- COMPUTE_IP=$(getent hosts "$COMPUTE_HOST" | awk '{ print $1 }')
- if [[ -z "$COMPUTE_IP" ]]; then
- echo "ERROR: Could not resolve $COMPUTE_HOST. Either manually set COMPUTE_IP or enable DNS resolution."
- exit 1
- fi
- fi
-
echo "INSTALLER_TYPE set to 'local'. Assuming SSH keys already exchanged with $COMPUTE_HOST"
fi
- # verify connectivity to target compute host
- ping -c 1 "$COMPUTE_IP"
+ sudo chown $(whoami):$(whoami) instack_key
+ chmod 400 instack_key
+ ssh_opts_cpu+=" -i instack_key"
+
+ # verify ssh to target compute host
+ ssh $ssh_opts_cpu "$COMPUTE_USER@$COMPUTE_IP" 'exit'
+ if [[ $? -ne 0 ]] ; then
+ echo "ERROR: can not ssh to computer host"
+ exit 1
+ fi
}
download_image() {
}
register_image() {
- glance image-list | grep -q " $IMAGE_NAME " && return 0
- glance image-create --name "$IMAGE_NAME" \
- --visibility public \
- --disk-format "$IMAGE_FORMAT" \
- --container-format bare \
- --file "$IMAGE_FILE"
+ openstack image list | grep -q " $IMAGE_NAME " && return 0
+ openstack image create "$IMAGE_NAME" \
+ --public \
+ --disk-format "$IMAGE_FORMAT" \
+ --container-format bare \
+ --file "$IMAGE_FILE"
}
create_test_user() {
- openstack user list | grep -q "$TEST_USER" || {
- openstack user create "$TEST_USER" --password "$TEST_PW"
+ openstack user list | grep -q " $DOCTOR_USER " || {
+ openstack user create "$DOCTOR_USER" --password "$DOCTOR_PW"
}
- openstack project list | grep -q "$TEST_PROJECT" || {
- openstack project create "$TEST_PROJECT"
+ openstack project list | grep -q " $DOCTOR_PROJECT " || {
+ openstack project create "$DOCTOR_PROJECT"
}
- openstack user role list "$TEST_USER" --project "$TEST_PROJECT" \
- | grep -q "$TEST_ROLE" || {
- openstack role add "$TEST_ROLE" --user "$TEST_USER" \
- --project "$TEST_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 " $VM_NAME " && return 0
(
# test VM done with test user, so can test non-admin
- export OS_USERNAME="$TEST_USER"
- export OS_PASSWORD="$TEST_PW"
- export OS_TENANT_NAME="$TEST_PROJECT"
- nova boot --flavor "$VM_FLAVOR" \
- --image "$IMAGE_NAME" \
- "$VM_NAME"
+ change_to_doctor_user
+ openstack server list | grep -q " $VM_NAME " && return 0
+ openstack server create --flavor "$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 " $ALARM_NAME " && return 0
+ vm_id=$(openstack server 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..."
- 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
+
+ (
+ # get VM state as test user
+ change_to_doctor_user
+
+ count=0
+ while [[ ${count} -lt 60 ]]
+ do
+ state=$(openstack server 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() {
awk '{d = $1 - $2; if (d < 1 && d > 0) print d " OK"; else print d " NG"}'
}
-check_host_status_down() {
+check_host_status() {
+ expect_state=$1
(
- # Switching to test user
- export OS_USERNAME="$TEST_USER"
- export OS_PASSWORD="$TEST_PW"
- export OS_TENANT_NAME="$TEST_PROJECT"
+ change_to_doctor_user
- host_status_line=$(nova show $VM_NAME | grep "host_status")
- [[ $? -ne 0 ]] && {
+ host_status_line=$(openstack --os-compute-api-version 2.16 server show $VM_NAME | grep "host_status")
+ if [[ $? -ne 0 ]] ; then
echo "ERROR: host_status not configured for owner in Nova policy.json"
- }
+ exit 1
+ fi
host_status=$(echo $host_status_line | awk '{print $4}')
- [[ "$host_status" == "DOWN" ]] && {
+ if [ -z "$host_status" ] ; then
+ echo "ERROR: host_status not reported by: nova show $VM_NAME"
+ exit 1
+ elif [[ "$host_status" != "$expect_state" ]] ; then
+ echo "ERROR: host_status:$host_status not equal to expect_state: $expect_state"
+ exit 1
+ else
echo "$VM_NAME showing host_status: $host_status"
- }
- echo "ERROR: host_status not reported by: nova show $VM_NAME"
+ fi
)
}
stop_inspector
stop_consumer
- python ./nova_force_down.py "$COMPUTE_HOST" --unset
- sleep 1
- 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 "$image_id"
- openstack role remove "$TEST_ROLE" --user "$TEST_USER" \
- --project "$TEST_PROJECT"
- openstack project delete "$TEST_PROJECT"
- openstack user delete "$TEST_USER"
-
- #TODO: add host status check via nova admin api
echo "waiting disabled compute host back to be enabled..."
+ python ./nova_force_down.py "$COMPUTE_HOST" --unset
sleep 180
+ check_host_status "UP"
ssh $ssh_opts_cpu "$COMPUTE_USER@$COMPUTE_IP" \
"[ -e disable_network.log ] && cat disable_network.log"
+ sleep 1
+
+ (
+ change_to_doctor_user
+ openstack server list | grep -q " $VM_NAME " && openstack server delete "$VM_NAME"
+ sleep 1
+ alarm_id=$(ceilometer alarm-list | grep " $ALARM_NAME " | awk '{print $2}')
+ sleep 1
+ [ -n "$alarm_id" ] && ceilometer alarm-delete "$alarm_id"
+ sleep 1
+ )
+ image_id=$(openstack image list | grep " $IMAGE_NAME " | awk '{print $2}')
+ sleep 1
+ [ -n "$image_id" ] && openstack 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"
}
echo "Note: doctor/tests/run.sh has been executed."
-prepare_compute_ssh
-
trap cleanup EXIT
echo "preparing VM image..."
download_image
register_image
-echo "starting doctor sample components..."
-start_monitor
-start_inspector
-start_consumer
-
echo "creating test user..."
create_test_user
echo "creating VM and alarm..."
boot_vm
-create_alarm
wait_for_vm_launch
+create_alarm
+
+echo "get computer host info and prepare to ssh..."
+get_compute_host_info
+prepare_compute_ssh
+
+echo "starting doctor sample components..."
+start_monitor
+start_inspector
+start_consumer
sleep 60
echo "injecting host failure..."
inject_failure
sleep 10
-check_host_status_down
+check_host_status "DOWN"
calculate_notification_time
echo "done"
Code Review / doctor.git / blobdiff