# http://www.apache.org/licenses/LICENSE-2.0
##############################################################################
-[[ "${CI_DEBUG:-true}" == "true" ]] && set -x
+[[ "${CI_DEBUG:-true}" == [Tt]rue ]] && set -x
IMAGE_URL=https://launchpad.net/cirros/trunk/0.3.0/+download/cirros-0.3.0-x86_64-disk.img
IMAGE_NAME=cirros
SUPPORTED_INSTALLER_TYPES="apex local"
INSTALLER_TYPE=${INSTALLER_TYPE:-apex}
INSTALLER_IP=${INSTALLER_IP:-none}
-COMPUTE_USER=${COMPUTE_USER:-none}
+
ssh_opts="-o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no"
+as_doctor_user="--os-username $DOCTOR_USER --os-password $DOCTOR_PW
+ --os-tenant-name $DOCTOR_PROJECT"
if [[ ! "$SUPPORTED_INSTALLER_TYPES" =~ "$INSTALLER_TYPE" ]] ; then
echo "ERROR: INSTALLER_TYPE=$INSTALLER_TYPE is not supported."
fi
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 $1}')
- )
+ # get computer host info which VM boot in
+ COMPUTE_HOST=$(openstack $as_doctor_user server show $VM_NAME |
+ grep "OS-EXT-SRV-ATTR:host" | awk '{ print $4 }')
+ if [[ -z "$COMPUTE_HOST" ]] ; then
+ echo "ERROR: failed to get compute hostname"
+ exit 1
+ fi
if [[ "$INSTALLER_TYPE" == "apex" ]] ; then
+ compute_host_in_undercloud=${COMPUTE_HOST%%.*}
COMPUTE_USER=${COMPUTE_USER:-heat-admin}
if [[ "$INSTALLER_IP" == "none" ]] ; then
instack_mac=$(sudo virsh domiflist instack | awk '/default/{print $5}')
fi
COMPUTE_IP=$(sudo ssh $ssh_opts $INSTALLER_IP \
"source stackrc; \
- openstack server show $COMPUTE_HOST \
+ nova show $compute_host_in_undercloud \
| awk '/ ctlplane network /{print \$5}'")
elif [[ "$INSTALLER_TYPE" == "local" ]] ; then
COMPUTE_USER=${COMPUTE_USER:-$(whoami)}
exit 1
fi
fi
+ echo "COMPUTE_HOST=$COMPUTE_HOST"
+ echo "COMPUTE_IP=$COMPUTE_IP"
# verify connectivity to target compute host
ping -c 1 "$COMPUTE_IP"
fi
}
+get_consumer_ip() {
+ CONSUMER_IP=$(ip route get $COMPUTE_IP | awk '/ src /{print $NF}')
+ echo "CONSUMER_IP=$CONSUMER_IP"
+
+ if [[ -z "$CONSUMER_IP" ]]; then
+ echo "ERROR: Could not get CONSUMER_IP."
+ exit 1
+ fi
+}
+
download_image() {
[ -e "$IMAGE_FILE" ] && return 0
wget "$IMAGE_URL" -o "$IMAGE_FILE"
}
}
-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() {
- (
- # test VM done with test user, so can test non-admin
- 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
- )
-
+ # test VM done with test user, so can test non-admin
+ openstack $as_doctor_user server list | grep -q " $VM_NAME " && return 0
+ openstack $as_doctor_user server create --flavor "$VM_FLAVOR" \
+ --image "$IMAGE_NAME" \
+ "$VM_NAME"
+ sleep 1
}
create_alarm() {
- (
- # 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"
- )
+ # get vm_id as test user
+ ceilometer $as_doctor_user alarm-list | grep -q " $ALARM_NAME " && return 0
+ vm_id=$(openstack $as_doctor_user server list | grep " $VM_NAME " | awk '{print $2}')
+ ceilometer $as_doctor_user alarm-event-create --name "$ALARM_NAME" \
+ --alarm-action "http://$CONSUMER_IP:$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"
}
wait_for_vm_launch() {
echo "waiting for vm launch..."
- (
- # 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
- )
+ count=0
+ while [[ ${count} -lt 60 ]]
+ do
+ state=$(openstack $as_doctor_user server list | grep " $VM_NAME " | awk '{print $6}')
+ [[ "$state" == "ACTIVE" ]] && return 0
+ [[ "$state" == "ERROR" ]] && echo "vm state is ERROR" && exit 1
+ count=$(($count+1))
+ sleep 1
+ done
+ echo "ERROR: time out while waiting for vm launch"
+ exit 1
}
inject_failure() {
echo "disabling network of compute host [$COMPUTE_HOST] for 3 mins..."
cat > disable_network.sh << 'END_TXT'
#!/bin/bash -x
-dev=$(sudo ip route | awk '/^default/{print $5}')
+dev=$(sudo ip a | awk '/ @COMPUTE_IP@\//{print $7}')
sleep 1
sudo ip link set $dev down
sleep 180
sudo ip link set $dev up
sleep 1
END_TXT
+ sed -i -e "s/@COMPUTE_IP@/$COMPUTE_IP/" disable_network.sh
chmod +x disable_network.sh
scp $ssh_opts_cpu disable_network.sh "$COMPUTE_USER@$COMPUTE_IP:"
ssh $ssh_opts_cpu "$COMPUTE_USER@$COMPUTE_IP" 'nohup ./disable_network.sh > disable_network.log 2>&1 &'
calculate_notification_time() {
detected=$(grep "doctor monitor detected at" monitor.log | awk '{print $5}')
notified=$(grep "doctor consumer notified at" consumer.log | awk '{print $5}')
+ if ! grep -q "doctor consumer notified at" consumer.log ; then
+ echo "ERROR: consumer hasn't received fault notification."
+ exit 1
+ fi
echo "$notified $detected" | \
awk '{d = $1 - $2; if (d < 1 && d > 0) print d " OK"; else print d " NG"}'
}
check_host_status() {
- expect_state=$1
- (
- change_to_doctor_user
+ expected_state=$1
- host_status_line=$(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 $4}')
- if [ -z "$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: $host_status"
- fi
- )
+ host_status_line=$(openstack $as_doctor_user --os-compute-api-version 2.16 \
+ server show $VM_NAME | grep "host_status")
+ host_status=$(echo $host_status_line | awk '{print $4}')
+ if [ -z "$host_status" ] ; then
+ echo "ERROR: host_status not reported by: nova show $VM_NAME"
+ exit 1
+ elif [[ "$expected_state" =~ "$host_status" ]] ; then
+ echo "$VM_NAME showing host_status: $host_status"
+ else
+ echo "ERROR: host_status:$host_status not equal to expected_state: $expected_state"
+ exit 1
+ fi
}
cleanup() {
echo "waiting disabled compute host back to be enabled..."
python ./nova_force_down.py "$COMPUTE_HOST" --unset
- sleep 180
+ sleep 240
check_host_status "UP"
ssh $ssh_opts_cpu "$COMPUTE_USER@$COMPUTE_IP" \
"[ -e disable_network.log ] && cat disable_network.log"
+
+ openstack $as_doctor_user server list | grep -q " $VM_NAME " && openstack $as_doctor_user server delete "$VM_NAME"
+ sleep 1
+ alarm_id=$(ceilometer $as_doctor_user alarm-list | grep " $ALARM_NAME " | awk '{print $2}')
+ sleep 1
+ [ -n "$alarm_id" ] && ceilometer $as_doctor_user alarm-delete "$alarm_id"
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"
echo "creating test user..."
create_test_user
-echo "creating VM and alarm..."
+echo "creating VM..."
boot_vm
wait_for_vm_launch
-create_alarm
+openstack $as_doctor_user server show $VM_NAME
echo "get computer host info and prepare to ssh..."
get_compute_host_info
prepare_compute_ssh
+echo "creating alarm..."
+get_consumer_ip
+create_alarm
+
echo "starting doctor sample components..."
start_monitor
start_inspector
sleep 60
echo "injecting host failure..."
inject_failure
-sleep 10
+sleep 60
-check_host_status "DOWN"
+check_host_status "(DOWN|UNKNOWN)"
calculate_notification_time
echo "done"
Code Review / doctor.git / blobdiff