# 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
#TODO: change back to `_member_` when JIRA DOCTOR-55 is done
DOCTOR_ROLE=admin
-SUPPORTED_INSTALLER_TYPES="apex local"
-INSTALLER_TYPE=${INSTALLER_TYPE:-apex}
+SUPPORTED_INSTALLER_TYPES="apex fuel local"
+INSTALLER_TYPE=${INSTALLER_TYPE:-local}
INSTALLER_IP=${INSTALLER_IP:-none}
ssh_opts="-o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no"
get_compute_host_info() {
# get computer host info which VM boot in
COMPUTE_HOST=$(openstack $as_doctor_user server show $VM_NAME |
- grep "OS-EXT-SRV-ATTR:host" | awk '{ print $4 }' |
- awk -F '.' '{print $1}')
+ grep "OS-EXT-SRV-ATTR:host" | awk '{ print $4 }')
+ compute_host_in_undercloud=${COMPUTE_HOST%%.*}
if [[ -z "$COMPUTE_HOST" ]] ; then
echo "ERROR: failed to get compute hostname"
exit 1
fi
+
if [[ "$INSTALLER_TYPE" == "apex" ]] ; then
COMPUTE_USER=${COMPUTE_USER:-heat-admin}
if [[ "$INSTALLER_IP" == "none" ]] ; then
fi
COMPUTE_IP=$(sudo ssh $ssh_opts $INSTALLER_IP \
"source stackrc; \
- nova show $COMPUTE_HOST \
+ nova show $compute_host_in_undercloud \
| awk '/ ctlplane network /{print \$5}'")
+ elif [[ "$INSTALLER_TYPE" == "fuel" ]] ; then
+ COMPUTE_USER=${COMPUTE_USER:-root}
+ if [[ "$INSTALLER_IP" == "none" ]] ; then
+ instack_mac=$(sudo virsh domiflist fuel-opnfv | awk '/pxebr/{print $5}')
+ INSTALLER_IP=$(/usr/sbin/arp -e | grep ${instack_mac} | awk '{print $1}')
+ fi
+ node_id=$(echo $compute_host_in_undercloud | cut -d "-" -f 2)
+ COMPUTE_IP=$(sshpass -p r00tme ssh 2>/dev/null $ssh_opts root@${INSTALLER_IP} \
+ "fuel node|awk -F '|' -v id=$node_id '{if (\$1 == id) print \$5}' |xargs")
elif [[ "$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 $COMPUTE_HOST. Either manually set COMPUTE_IP or enable DNS resolution."
- exit 1
- fi
fi
+ if [[ -z "$COMPUTE_IP" ]]; then
+ echo "ERROR: Could not resolve $COMPUTE_HOST. Either manually set COMPUTE_IP or enable DNS resolution."
+ exit 1
+ fi
+ echo "COMPUTE_HOST=$COMPUTE_HOST"
+ echo "COMPUTE_IP=$COMPUTE_IP"
+
# verify connectivity to target compute host
ping -c 1 "$COMPUTE_IP"
if [[ $? -ne 0 ]] ; then
sudo chown $(whoami):$(whoami) instack_key
chmod 400 instack_key
ssh_opts_cpu+=" -i instack_key"
+ elif [[ "$INSTALLER_TYPE" == "fuel" ]] ; then
+ sshpass -p r00tme scp $ssh_opts root@${INSTALLER_IP}:.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
echo "INSTALLER_TYPE set to 'local'. Assuming SSH keys already exchanged with $COMPUTE_HOST"
fi
}
get_consumer_ip() {
- #get network of the compute_ip address
- network_pre=${COMPUTE_IP%.*}
- network=${network_pre}.0/24
-
- #if there is a dedicated route, use it
- dev=$(ip route |grep $network | awk '{print $3}' |head -n 1)
+ CONSUMER_IP=$(sudo ssh $ssh_opts root@$INSTALLER_IP \
+ "ip route get $COMPUTE_IP | awk '/ src /{print \$NF}'")
+ echo "CONSUMER_IP=$CONSUMER_IP"
- #if there is no route, use default route
- [[ -z $dev ]] && dev=$(ip route |grep ^default | awk '{print $3}' |head -n 1)
-
- CONSUMER_IP=$(ip addr show $dev |grep inet | grep $network_pre | awk '{print $2}' | cut -d'/' -f1)
+ if [[ -z "$CONSUMER_IP" ]]; then
+ echo "ERROR: Could not get CONSUMER_IP."
+ exit 1
+ fi
}
download_image() {
-q "traits.state=string::error; traits.instance_id=string::$vm_id"
}
+print_log() {
+ log_file=$1
+ echo "$log_file:"
+ sed -e 's/^/ /' "$log_file"
+}
start_monitor() {
pgrep -f "python monitor.py" && return 0
stop_monitor() {
pgrep -f "python monitor.py" || return 0
sudo kill $(pgrep -f "python monitor.py")
- cat monitor.log
+ print_log monitor.log
}
start_inspector() {
stop_inspector() {
pgrep -f "python inspector.py" || return 0
kill $(pgrep -f "python inspector.py")
- cat inspector.log
+ print_log inspector.log
}
start_consumer() {
pgrep -f "python consumer.py" && return 0
python consumer.py "$CONSUMER_PORT" > consumer.log 2>&1 &
+ # NOTE(r-mibu): create tunnel to the installer node, so that we can
+ # avoid some network problems dpends on infra and installers.
+ # This tunnel will be terminated by stop_consumer() or after 10 mins passed.
+ TUNNEL_COMMAND="sudo ssh $ssh_opts $INSTALLER_IP -R $CONSUMER_PORT:localhost:$CONSUMER_PORT sleep 600"
+ $TUNNEL_COMMAND > ssh_tunnel.log 2>&1 < /dev/null &
}
stop_consumer() {
pgrep -f "python consumer.py" || return 0
kill $(pgrep -f "python consumer.py")
- cat consumer.log
+ print_log consumer.log
+ kill $(pgrep -f "$TUNNEL_COMMAND")
+ print_log ssh_tunnel.log
}
wait_for_vm_launch() {
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 - $2; if (d < 1 && d > 0) print d " OK"; else print d " NG"}'
}
if [ -z "$host_status" ] ; then
echo "ERROR: host_status not reported by: nova show $VM_NAME"
exit 1
- elif [[ "$host_status" != "$expected_state" ]] ; then
+ elif [[ "$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: $expected_state"
exit 1
- else
- echo "$VM_NAME showing host_status: $host_status"
fi
}
echo "waiting disabled compute host back to be enabled..."
python ./nova_force_down.py "$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"
+ scp $ssh_opts_cpu "$COMPUTE_USER@$COMPUTE_IP:disable_network.log" .
+ print_log disable_network.log
openstack $as_doctor_user server list | grep -q " $VM_NAME " && openstack $as_doctor_user server delete "$VM_NAME"
sleep 1
echo "creating VM..."
boot_vm
wait_for_vm_launch
+openstack $as_doctor_user server show $VM_NAME
echo "get computer host info and prepare to ssh..."
get_compute_host_info
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