-
url: https://blueprints.launchpad.net/congress/+spec/push-type-datasource-driver
system: Launchpad
-#-
-# url: https://review.openstack.org/#/c/314915/
-# system: Gerrit
-
url: https://blueprints.launchpad.net/cinder/+spec/mark-services-down
system: Launchpad
-
url: https://blueprints.launchpad.net/python-cinderclient/+spec/mark-service-down-cli
system: Launchpad
-#-
-# url: https://bugs.launchpad.net/neutron/+bug/1513144
-# system: Launchpad-bug
+-
+ url: https://bugs.launchpad.net/neutron/+bug/1513144
+ system: Launchpad-bug
+-
+ url: https://review.openstack.org/314915
+ system: Gerrit
+ # shortest path
+-
+ url: https://review.openstack.org/424868/
+ system: Gerrit
+ # endswith '/'
+-
+ url: https://review.openstack.org/#/c/430428
+ system: Gerrit
+ # comment path
+-
+ url: https://review.openstack.org/424340
+ system: Gerrit
+ # WIP
+++ /dev/null
-.. This work is licensed under a Creative Commons Attribution 4.0 International License.
-.. http://creativecommons.org/licenses/by/4.0
-
-Summary and conclusion
-======================
-
-The Doctor project aimed at detailing NFVI fault management and NFVI maintenance
-requirements. These are indispensable operations for an Operator, and extremely
-necessary to realize telco-grade high availability. High availability is a large
-topic; the objective of Doctor is not to realize a complete high availability
-architecture and implementation. Instead, Doctor limited itself to addressing
-the fault events in NFVI, and proposes enhancements necessary in VIM, e.g.
-OpenStack, to ensure VNFs availability in such fault events, taking a Telco VNFs
-application level management system into account.
-
-The Doctor project performed a robust analysis of the requirements from NFVI
-fault management and NFVI maintenance operation, concretely found out gaps in
-between such requirements and the current implementation of OpenStack, and
-proposed potential development plans to fill out such gaps in OpenStack.
-Blueprints are already under investigation and the next step is to fill out
-those gaps in OpenStack by code development in the coming releases.
-
-..
- vim: set tabstop=4 expandtab textwidth=80:
realization for a NFVI fault management and maintenance solution in
open source software.
-:History:
-
- ========== =====================================================
- Date Description
- ========== =====================================================
- 02.12.2014 Project creation
- 14.04.2015 Initial version of the deliverable uploaded to Gerrit
- 18.05.2015 Stable version of the Doctor deliverable
- 25.02.2016 Updated version for the Brahmaputra release
- 26.09.2016 Updated version for the Colorado release
- xx.xx.2017 Updated version for the Danube release
- ========== =====================================================
-
.. raw:: latex
\newpage
--- /dev/null
+##############################################################################
+# Copyright (c) 2017 NEC Corporation and others.
+#
+# All rights reserved. This program and the accompanying materials
+# are made available under the terms of the Apache License, Version 2.0
+# which accompanies this distribution, and is available at
+# http://www.apache.org/licenses/LICENSE-2.0
+##############################################################################
+
+import os
+
+from keystoneauth1.identity import v2
+from keystoneauth1.identity import v3
+
+
+def get_identity_auth():
+ auth_url = os.environ['OS_AUTH_URL']
+ username = os.environ['OS_USERNAME']
+ password = os.environ['OS_PASSWORD']
+ user_domain_name = os.environ.get('OS_USER_DOMAIN_NAME')
+ project_name = os.environ.get('OS_PROJECT_NAME') or os.environ.get('OS_TENANT_NAME')
+ project_domain_name = os.environ.get('OS_PROJECT_DOMAIN_NAME')
+ if auth_url.endswith('v3'):
+ return v3.Password(auth_url=auth_url,
+ username=username,
+ password=password,
+ user_domain_name=user_domain_name,
+ project_name=project_name,
+ project_domain_name=project_domain_name)
+ else:
+ return v2.Password(auth_url=auth_url,
+ username=username,
+ password=password,
+ tenant_name=project_name)
import threading
import time
+from keystoneauth1 import session
import novaclient.client as novaclient
-import nova_force_down
+import identity_auth
LOG = doctor_log.Logger('doctor_inspector').getLogger()
class DoctorInspectorSample(object):
- NOVA_API_VERSION = '2.11'
+ NOVA_API_VERSION = '2.34'
NUMBER_OF_CLIENTS = 50
# TODO(tojuvone): This could be enhanced in future with dynamic
# reuse of self.novaclients when all threads in use and
def __init__(self):
self.servers = collections.defaultdict(list)
self.novaclients = list()
+ auth=identity_auth.get_identity_auth()
+ sess=session.Session(auth=auth)
# Pool of novaclients for redundant usage
for i in range(self.NUMBER_OF_CLIENTS):
- self.novaclients.append(novaclient.Client(self.NOVA_API_VERSION,
- os.environ['OS_USERNAME'],
- os.environ['OS_PASSWORD'],
- os.environ['OS_TENANT_NAME'],
- os.environ['OS_AUTH_URL'],
- connection_pool=True))
+ self.novaclients.append(
+ novaclient.Client(self.NOVA_API_VERSION, session=sess,
+ connection_pool=True))
# Normally we use this client for non redundant API calls
self.nova=self.novaclients[0]
self.nova.servers.list(detailed=False)
threads.append(t)
for t in threads:
t.join()
- # NOTE: We use our own client here instead of this novaclient for a
- # workaround. Once keystone provides v2.1 nova api endpoint
- # in the service catalog which is configured by OpenStack
- # installer, we can use this:
- #
- # self.nova.services.force_down(hostname, 'nova-compute', True)
- #
- nova_force_down.force_down(hostname)
+ self.nova.services.force_down(hostname, 'nova-compute', True)
LOG.info('doctor mark host(%s) down at %s' % (hostname, time.time()))
#!/bin/bash
+INSPECTOR_PORT=${INSPECTOR_PORT:-12345}
+
function start_inspector_sample {
pgrep -f "python inspector.py" && return 0
- # TODO (r-mibu): make sample inspector use keystone v3 api
- OS_AUTH_URL=${OS_AUTH_URL/v3/v2.0} \
python inspector.py "$INSPECTOR_PORT" > inspector.log 2>&1 &
}
get_installer_ip
installer_get_ssh_keys
get_controller_ips
+
+ # NOTE: while executing command as doctor user,
+ # 'OS_PROJECT_ID' env parameter make openstack clients confused.
+ unset OS_PROJECT_ID
}
function get_compute_ip_from_hostname {
import time
from congressclient.v1 import client
-from keystoneclient import session as ksc_session
-from keystoneclient.auth.identity import v2
+
+import identity_auth
# NOTE: icmp message with all zero data (checksum = 0xf7ff)
# see https://tools.ietf.org/html/rfc792
LOG = doctor_log.Logger('doctor_monitor').getLogger()
+
class DoctorMonitorSample(object):
interval = 0.1 # second
raise Exception("Inspector type '%s' not supported", args.inspector_type)
self.hostname = args.hostname
- self.inspector_url = args.inspector_url
self.inspector_type = args.inspector_type
self.ip_addr = args.ip or socket.gethostbyname(self.hostname)
- if self.inspector_type == 'congress':
- auth = v2.Password(auth_url=os.environ['OS_AUTH_URL'],
- username=os.environ['OS_USERNAME'],
- password=os.environ['OS_PASSWORD'],
- tenant_name=os.environ['OS_TENANT_NAME'])
- self.session = ksc_session.Session(auth=auth)
-
- congress = client.Client(session=self.session, service_type='policy')
+ if self.inspector_type == 'sample':
+ self.inspector_url = 'http://127.0.0.1:12345/events'
+ elif self.inspector_type == 'congress':
+ auth=identity_auth.get_identity_auth()
+ sess=session.Session(auth=auth)
+ congress = client.Client(session=sess, service_type='policy')
ds = congress.list_datasources()['results']
doctor_ds = next((item for item in ds if item['driver'] == 'doctor'),
None)
parser.add_argument('inspector_type', metavar='INSPECTOR_TYPE', type=str, nargs='?',
help='inspector to report',
default='sample')
- parser.add_argument('inspector_url', metavar='INSPECTOR_URL', type=str, nargs='?',
- help='inspector url to report error',
- default='http://127.0.0.1:12345/events')
return parser.parse_args()
+++ /dev/null
-##############################################################################
-# Copyright (c) 2016 NEC Corporation and others.
-#
-# All rights reserved. This program and the accompanying materials
-# are made available under the terms of the Apache License, Version 2.0
-# which accompanies this distribution, and is available at
-# http://www.apache.org/licenses/LICENSE-2.0
-##############################################################################
-
-import argparse
-import json
-import os
-
-from keystoneclient.v2_0 import client
-import requests
-
-
-def force_down(hostname, force_down=True):
- keystone = client.Client(username=os.environ['OS_USERNAME'],
- password=os.environ['OS_PASSWORD'],
- tenant_name=os.environ['OS_TENANT_NAME'],
- auth_url=os.environ['OS_AUTH_URL'])
-
- for service in keystone.auth_ref['serviceCatalog']:
- if service['type'] == 'compute':
- base_url = service['endpoints'][0]['internalURL']
- break
-
- url = '%s/os-services/force-down' % base_url.replace('/v2/', '/v2.1/')
- data = {
- 'forced_down': force_down,
- 'binary': 'nova-compute',
- 'host': hostname,
- }
- headers = {
- 'Content-Type': 'application/json',
- 'Accept': 'application/json',
- 'X-Auth-Token': keystone.auth_ref['token']['id'],
- 'X-OpenStack-Nova-API-Version': '2.11',
- }
-
- print requests.put(url, data=json.dumps(data), headers=headers)
-
-
-def get_args():
- parser = argparse.ArgumentParser(description='Doctor Test Cleaner')
- parser.add_argument('hostname', metavar='HOSTNAME', type=str, nargs='?',
- help='a nova-compute hostname to force down')
- parser.add_argument('--unset', action='store_true', default=False,
- help='unset force_down flag')
- return parser.parse_args()
-
-
-def main():
- args = get_args()
- force_down(args.hostname, not(args.unset))
-
-
-if __name__ == '__main__':
- main()
VM_FLAVOR=m1.tiny
#if VM_COUNT set, use that instead
VM_COUNT=${VM_COUNT:-1}
+NET_NAME=doctor_net
+NET_CIDR=192.168.168.0/24
ALARM_BASENAME=doctor_alarm
-INSPECTOR_PORT=12345
CONSUMER_PORT=12346
DOCTOR_USER=doctor
DOCTOR_PW=doctor
TOP_DIR=$(cd $(dirname "$0") && pwd)
as_doctor_user="--os-username $DOCTOR_USER --os-password $DOCTOR_PW
- --os-tenant-name $DOCTOR_PROJECT"
+ --os-project-name $DOCTOR_PROJECT --os-tenant-name $DOCTOR_PROJECT"
+# NOTE: ceilometer command still requires '--os-tenant-name'.
+#ceilometer="ceilometer ${as_doctor_user/--os-project-name/--os-tenant-name}"
+ceilometer="ceilometer $as_doctor_user"
# Functions
openstack user create "$DOCTOR_USER" --password "$DOCTOR_PW" \
--project "$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"
+ openstack role show "$DOCTOR_ROLE" || {
+ openstack role create "$DOCTOR_ROLE"
}
+ openstack role add "$DOCTOR_ROLE" --user "$DOCTOR_USER" \
+ --project "$DOCTOR_PROJECT"
# tojuvone: openstack quota show is broken and have to use nova
# https://bugs.launchpad.net/manila/+bug/1652118
# Note! while it is encouraged to use openstack client it has proven
# quite buggy.
# QUOTA=$(openstack quota show $DOCTOR_PROJECT)
- DOCTOR_QUOTA=$(nova quota-show --tenant DOCTOR_PROJECT)
+ DOCTOR_QUOTA=$(nova quota-show --tenant $DOCTOR_PROJECT)
# We make sure that quota allows number of instances and cores
OLD_INSTANCE_QUOTA=$(echo "${DOCTOR_QUOTA}" | grep " instances " | \
awk '{print $4}')
boot_vm() {
# test VM done with test user, so can test non-admin
+
+ if ! openstack $as_doctor_user network show $NET_NAME; then
+ openstack $as_doctor_user network create $NET_NAME
+ fi
+ if ! openstack $as_doctor_user subnet show $NET_NAME; then
+ openstack $as_doctor_user subnet create $NET_NAME \
+ --network $NET_NAME --subnet-range $NET_CIDR --no-dhcp
+ fi
+ net_id=$(openstack $as_doctor_user network show $NET_NAME -f value -c id)
+
servers=$(openstack $as_doctor_user server list)
for i in `seq $VM_COUNT`; do
echo "${servers}" | grep -q " $VM_BASENAME$i " && continue
openstack $as_doctor_user server create --flavor "$VM_FLAVOR" \
- --image "$IMAGE_NAME" \
- "$VM_BASENAME$i"
+ --image "$IMAGE_NAME" --nic net-id=$net_id "$VM_BASENAME$i"
done
sleep 1
}
create_alarm() {
# get vm_id as test user
- alarm_list=$(ceilometer $as_doctor_user alarm-list)
+ alarm_list=$($ceilometer alarm-list)
vms=$(openstack $as_doctor_user server list)
for i in `seq $VM_COUNT`; do
echo "${alarm_list}" | grep -q " $ALARM_BASENAME$i " || {
vm_id=$(echo "${vms}" | grep " $VM_BASENAME$i " | awk '{print $2}')
# TODO(r-mibu): change notification endpoint from localhost to the
# consumer. IP address (functest container).
- ceilometer $as_doctor_user alarm-event-create \
+ $ceilometer alarm-event-create \
--name "$ALARM_BASENAME$i" \
--alarm-action "http://localhost:$CONSUMER_PORT/failure" \
--description "VM failure" \
start_monitor() {
pgrep -f "python monitor.py" && return 0
sudo -E python monitor.py "$COMPUTE_HOST" "$COMPUTE_IP" "$INSPECTOR_TYPE" \
- "http://127.0.0.1:$INSPECTOR_PORT/events" > monitor.log 2>&1 &
+ > monitor.log 2>&1 &
}
stop_monitor() {
}'
}
-wait_ping() {
- local interval=5
- local rounds=$(($1 / $interval))
- for i in `seq $rounds`; do
- ping -c 1 "$COMPUTE_IP"
- if [[ $? -ne 0 ]] ; then
- sleep $interval
- continue
- fi
- return 0
- done
-}
-
check_host_status() {
# Check host related to first Doctor VM is in wanted state
# $1 Expected state
}
unset_forced_down_hosts() {
- for host in $(openstack compute service list --service nova-compute \
- -f value -c Host -c State | sed -n -e '/down$/s/ *down$//p')
+ # for debug
+ openstack compute service list --service nova-compute
+
+ downed_computes=$(openstack compute service list --service nova-compute \
+ -f value -c Host -c State | grep ' down$' \
+ | sed -e 's/ *down$//')
+ echo "downed_computes: $downed_computes"
+ for host in $downed_computes
do
- # TODO (r-mibu): make sample inspector use keystone v3 api
- OS_AUTH_URL=${OS_AUTH_URL/v3/v2.0} \
- python ./nova_force_down.py $host --unset
+ # TODO(r-mibu): use openstack client
+ #openstack compute service set --up $host nova-compute
+ nova service-force-down --unset $host nova-compute
done
echo "waiting disabled compute host back to be enabled..."
wait_until 'openstack compute service list --service nova-compute
-f value -c State | grep -q down' 240 5
+
+ for host in $downed_computes
+ do
+ # TODO(r-mibu): improve 'get_compute_ip_from_hostname'
+ get_compute_ip_from_hostname $host
+ wait_until "! ping -c 1 $COMPUTE_IP" 120 5
+ done
}
collect_logs() {
- unset_forced_down_hosts
- # TODO: We need to make sure the target compute host is back to IP
- # reachable. wait_ping() will be added by tojuvone .
- sleep 110
- scp $ssh_opts_cpu "$COMPUTE_USER@$COMPUTE_IP:disable_network.log" .
+ if [[ -n "$COMPUTE_IP" ]];then
+ scp $ssh_opts_cpu "$COMPUTE_USER@$COMPUTE_IP:disable_network.log" .
+ fi
# TODO(yujunz) collect other logs, e.g. nova, aodh
}
stop_consumer
unset_forced_down_hosts
+ collect_logs
- wait_ping 120
-
- scp $ssh_opts_cpu "$COMPUTE_USER@$COMPUTE_IP:disable_network.log" .
vms=$(openstack $as_doctor_user server list)
vmstodel=""
for i in `seq $VM_COUNT`; do
vmstodel+=" $VM_BASENAME$i"
done
[[ $vmstodel ]] && openstack $as_doctor_user server delete $vmstodel
- alarm_list=$(ceilometer $as_doctor_user alarm-list)
+ alarm_list=$($ceilometer alarm-list)
for i in `seq $VM_COUNT`; do
alarm_id=$(echo "${alarm_list}" | grep " $ALARM_BASENAME$i " |
awk '{print $2}')
- [ -n "$alarm_id" ] && ceilometer $as_doctor_user alarm-delete "$alarm_id"
+ [ -n "$alarm_id" ] && $ceilometer alarm-delete "$alarm_id"
done
+ openstack $as_doctor_user subnet delete $NET_NAME
+ sleep 1
+ openstack $as_doctor_user network delete $NET_NAME
sleep 1
image_id=$(openstack image list | grep " $IMAGE_NAME " | awk '{print $2}')
--project "$DOCTOR_PROJECT"
openstack project delete "$DOCTOR_PROJECT"
openstack user delete "$DOCTOR_USER"
+ # NOTE: remove role only for doctor test.
+ #openstack role delete "$DOCTOR_ROLE"
cleanup_installer
cleanup_inspector
+
+ # NOTE: Temporal log printer.
+ for f in $(find . -name '*.log')
+ do
+ echo
+ echo "[$f]"
+ sed -e 's/^/ | /' $f
+ echo
+ done
}
# Main process
check_host_status "(DOWN|UNKNOWN)" 60
calculate_notification_time
+unset_forced_down_hosts
collect_logs
run_profiler
Code Review / doctor.git / commitdiff