mciecier 7da0d18f31 Fix update run to not run update on excluded nodes
Change[1] modified how --limit option, used together with skiplist,
works in ussuri and onwards. For overcloud update run the change poses
some problems, because when update run is triggered --limit option is
used to run control sequence of nodes being updated. As a result
skiplisted nodes can be added to update run unintentionally.

With this fix `excluded_overcloud` group is added to --limit flag to
exclude all nodes in skiplist when update run is triggered.
`excluded_overcloud` group is added only when skiplist contains any host

Resolves: rhbz#2166224

[1]https: //

Change-Id: Icd782739003f4ee77895b0c028500cc6de808ead
2023-03-23 09:13:26 +01:00

3626 lines
131 KiB

# Copyright 2015 Red Hat, Inc.
# Licensed under the Apache License, Version 2.0 (the "License"); you may
# not use this file except in compliance with the License. You may obtain
# a copy of the License at
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
# WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
# License for the specific language governing permissions and limitations
# under the License.
import collections
from collections import abc as collections_abc
import configparser
import copy
import csv
import datetime
import errno
import getpass
import glob
import hashlib
import json
import logging
import multiprocessing
import netaddr
import openstack
import os
import os.path
import prettytable
import pwd
import re
import shutil
import socket
import subprocess
import sys
import tarfile
import tempfile
import textwrap
import time
import yaml
from heatclient.common import event_utils
from heatclient.common import template_utils
from heatclient.common import utils as heat_utils
from heatclient.exc import HTTPNotFound
from osc_lib import exceptions as oscexc
from osc_lib.i18n import _
from oslo_concurrency import processutils
from heatclient import exc as hc_exc
from urllib import error as url_error
from urllib import parse as url_parse
from urllib import request
from tenacity import retry
from tenacity.stop import stop_after_attempt, stop_after_delay
from tenacity.wait import wait_fixed
from tripleo_common.image import image_uploader
from tripleo_common.image import kolla_builder
from tripleo_common.utils import plan as plan_utils
from tripleo_common.utils import heat as tc_heat_utils
from tripleo_common.utils import stack as stack_utils
from tripleo_common import update
from tripleoclient import constants
from tripleoclient import exceptions
from tripleoclient import heat_launcher
import warnings
warnings.simplefilter("ignore", UserWarning)
import ansible_runner # noqa
from ansible.parsing.dataloader import DataLoader # noqa
from ansible.inventory.manager import InventoryManager # noqa
LOG = logging.getLogger(__name__ + ".utils")
_local_orchestration_client = None
_heat_pid = None
class Pushd(object):
"""Simple context manager to change directories and then return."""
def __init__(self, directory):
"""This context manager will enter and exit directories.
>>> with Pushd(directory='/tmp'):
... with open('file', 'w') as f:
... f.write('test')
:param directory: path to change directory to
:type directory: `string`
self.dir = directory
self.pwd = self.cwd = os.getcwd()
def __enter__(self):
self.cwd = os.getcwd()
return self
def __exit__(self, *args):
if self.pwd != self.cwd:
class TempDirs(object):
"""Simple context manager to manage temp directories."""
def __init__(self, dir_path=None, dir_prefix='tripleo', cleanup=True,
"""This context manager will create, push, and cleanup temp directories.
>>> with TempDirs() as t:
... with open('file', 'w') as f:
... f.write('test')
... print(t)
... os.mkdir('testing')
... with open(os.path.join(t, 'file')) as w:
... print(
... with open('testing/file', 'w') as f:
... f.write('things')
... with open(os.path.join(t, 'testing/file')) as w:
... print(
:param dir_path: path to create the temp directory
:type dir_path: `string`
:param dir_prefix: prefix to add to a temp directory
:type dir_prefix: `string`
:param cleanup: when enabled the temp directory will be
removed on exit.
:type cleanup: `boolean`
:param chdir: Change to/from the created temporary dir on enter/exit.
:type chdir: `boolean`
# NOTE(cloudnull): kwargs for tempfile.mkdtemp are created
# because args are not processed correctly
# in py2. When we drop py2 support (cent7)
# these args can be removed and used directly
# in the `tempfile.mkdtemp` function.
tempdir_kwargs = dict()
if dir_path:
tempdir_kwargs['dir'] = dir_path
if dir_prefix:
tempdir_kwargs['prefix'] = dir_prefix
self.dir = tempfile.mkdtemp(**tempdir_kwargs)
self.pushd = Pushd(directory=self.dir)
self.cleanup = cleanup
self.chdir = chdir
def __enter__(self):
if self.chdir:
return self.dir
def __exit__(self, *args):
if self.chdir:
if self.cleanup:
LOG.warning("Not cleaning temporary directory [ %s ]" % self.dir)
def clean(self):
shutil.rmtree(self.dir, ignore_errors=True)"Temporary directory [ %s ] cleaned up" % self.dir)
def _encode_envvars(env):
"""Encode a hash of values.
:param env: A hash of key=value items.
:type env: `dict`.
for key, value in env.items():
env[key] = str(value)
return env
def makedirs(dir_path):
"""Recursively make directories and log the interaction.
:param dir_path: full path of the directories to make.
:type dir_path: `string`
:returns: `boolean`
except FileExistsError:
'Directory "{}" was not created because it'
' already exists.'.format(
return False
LOG.debug('Directory "{}" was created.'.format(dir_path))
return True
def compare_limit_nodes_with_excluded_hosts(limit_nodes, excluded_hosts):
"""Compare limit nodes list with excluded hosts list.
This will compare the limit nodes list with the excluded hosts list
and raise an Exception if there is a match to inform user that the
host is in both lists. If user marks a host in limit option as
excluded with exclamatory mark (!) then it will be ignored during
the comparison.
:param limit_hosts: list of hosts to limit the playbook to.
:type limit_hosts: `list`
:param excluded_hosts: list of hosts to exclude from the playbook.
:type excluded_hosts: `list`
RuntimeError if there is a match between the limit nodes and the
excluded hosts.
if limit_nodes and excluded_hosts:
matching_hosts_found = []
for limit_node in re.split(':', limit_nodes):
if not limit_node.startswith('!') and limit_node in \
if matching_hosts_found:
raise RuntimeError(
'Hosts {} are in both the limit nodes and the '
'excluded hosts list. Please remove it from one of '
'the lists.'.format(','.join(matching_hosts_found))
def playbook_limit_parse(limit_nodes):
"""Return a parsed string for limits.
This will sanitize user inputs so that we guarantee what is provided is
expected to be functional. If limit_nodes is None, this function will
return None.
:returns: String
if not limit_nodes:
return limit_nodes
return ':'.join([i.strip() for i in re.split(',| |:', limit_nodes) if i])
def playbook_verbosity(self):
"""Return an integer for playbook verbosity levels.
:param self: Class object used to interpret the runtime state.
:type self: Object
:returns: Integer
return 3
if self.app_args.verbose_level <= 1:
return 0
return self.app_args.verbose_level
def run_ansible_playbook(playbook, inventory, workdir, playbook_dir=None,
connection='smart', output_callback='tripleo_dense',
ssh_user='root', key=None, module_path=None,
limit_hosts=None, tags=None, skip_tags=None,
verbosity=0, quiet=False, extra_vars=None,
extra_vars_file=None, plan='overcloud',
gathering_policy='smart', extra_env_variables=None,
ansible_cfg=None, ansible_timeout=30,
timeout=None, forks=None,
ignore_unreachable=False, rotate_log=False):
"""Simple wrapper for ansible-playbook.
:param playbook: Playbook filename.
:type playbook: String
:param inventory: Either proper inventory file, or a coma-separated list.
:type inventory: String
:param workdir: Location of the working directory.
:type workdir: String
:param playbook_dir: Location of the playbook directory.
(defaults to workdir).
:type playbook_dir: String
:param connection: Connection type (local, smart, etc).
:type connection: String
:param output_callback: Callback for output format. Defaults to
:type output_callback: String
:param callback_whitelist: Comma separated list of callback plugins.
Defaults to
Custom output_callback is also whitelisted.
:type callback_whitelist: String
:param ssh_user: User for the ssh connection.
:type ssh_user: String
:param key: Private key to use for the ssh connection.
:type key: String
:param module_path: Location of the ansible module and library.
:type module_path: String
:param limit_hosts: Limit the execution to the hosts.
:type limit_hosts: String
:param tags: Run specific tags.
:type tags: String
:param skip_tags: Skip specific tags.
:type skip_tags: String
:param verbosity: Verbosity level for Ansible execution.
:type verbosity: Integer
:param quiet: Disable all output (Defaults to False)
:type quiet: Boolean
:param extra_vars: Set additional variables as a Dict or the absolute
path of a JSON or YAML file type.
:type extra_vars: Either a Dict or the absolute path of JSON or YAML
:param extra_vars_file: Set additional ansible variables using an
extravar file.
:type extra_vars_file: Dictionary
:param plan: Plan name (Defaults to "overcloud").
:type plan: String
:param gathering_policy: This setting controls the default policy of
fact gathering ('smart', 'implicit', 'explicit').
:type gathering_facts: String
:param extra_env_variables: Dict option to extend or override any of the
default environment variables.
:type extra_env_variables: Dict
:param parallel_run: Isolate playbook execution when playbooks are to be
executed with multi-processing.
:type parallel_run: Boolean
:param ansible_cfg: Path to an ansible configuration file. One will be
generated in the artifact path if this option is None.
:type ansible_cfg: String
:param ansible_timeout: Timeout for ansible connections.
:type ansible_timeout: int
:param reproduce_command: Enable or disable option to reproduce ansible
commands upon failure. This option will produce
a bash script that can reproduce a failing
playbook command which is helpful for debugging
and retry purposes.
:type reproduce_command: Boolean
:param timeout: Timeout for ansible to finish playbook execution (minutes).
:type timeout: int
:param rotate_log: Enable or disable option to rotate ansible.log
:type rotate_log: Boolean
def _playbook_check(play):
if not os.path.exists(play):
play = os.path.join(playbook_dir, play)
if not os.path.exists(play):
raise RuntimeError('No such playbook: {}'.format(play))
LOG.debug('Ansible playbook {} found'.format(play))
return play
def _inventory(inventory):
if inventory:
if isinstance(inventory, str):
# check is file path
if os.path.exists(inventory):
return inventory
elif isinstance(inventory, dict):
inventory = yaml.safe_dump(
inv_file = ansible_runner.utils.dump_artifact(
os.chmod(inv_file, 0o600)
return inv_file
def _running_ansible_msg(playbook, timeout=None):
if timeout and timeout > 0:
return ('Running Ansible playbook with timeout %sm: %s,' %
(timeout, playbook))
return ('Running Ansible playbook: %s,' % playbook)
if not playbook_dir:
playbook_dir = workdir
# Ensure that the ansible-runner env exists
runner_env = os.path.join(workdir, 'env')
if extra_vars_file:
runner_extra_vars = os.path.join(runner_env, 'extravars')
with open(runner_extra_vars, 'w') as f:
f.write(yaml.safe_dump(extra_vars_file, default_flow_style=False))
if timeout and timeout > 0:
settings_file = os.path.join(runner_env, 'settings')
timeout_value = timeout * 60
if os.path.exists(settings_file):
with open(settings_file, 'r') as f:
settings_object = yaml.safe_load(
settings_object['job_timeout'] = timeout_value
settings_object = {'job_timeout': timeout_value}
with open(settings_file, 'w') as f:
f.write(yaml.safe_dump(settings_object, default_flow_style=False))
if isinstance(playbook, (list, set)):
verified_playbooks = [_playbook_check(play=i) for i in playbook]
playbook = os.path.join(workdir, 'tripleo-multi-playbook.yaml')
with open(playbook, 'w') as f:
[{'import_playbook': i} for i in verified_playbooks],
_running_ansible_msg(playbook, timeout) +
' multi-playbook execution: {}'
' Working directory: {},'
' Playbook directory: {}'.format(
playbook = _playbook_check(play=playbook)
_running_ansible_msg(playbook, timeout) +
' Working directory: {},'
' Playbook directory: {}'.format(
if limit_hosts:
'Running ansible with the following limit: {}'.format(
ansible_fact_path = os.path.join(
if output_callback not in callback_whitelist.split(','):
callback_whitelist = ','.join([callback_whitelist, output_callback])
if not forks:
forks = min(multiprocessing.cpu_count() * 4, 100)
env = dict()
'-o UserKnownHostsFile={} '
'-o StrictHostKeyChecking=no '
'-o ControlMaster=auto '
'-o ControlPersist=30m '
'-o ServerAliveInterval=64 '
'-o ServerAliveCountMax=1024 '
'-o Compression=no '
'-o TCPKeepAlive=yes '
'-o VerifyHostKeyDNS=no '
'-o ForwardX11=no '
'-o ForwardAgent=yes '
'-o PreferredAuthentications=publickey '
env['ANSIBLE_FORKS'] = forks
env['ANSIBLE_TIMEOUT'] = ansible_timeout
env['ANSIBLE_SCP_IF_SSH'] = True
env['ANSIBLE_REMOTE_USER'] = ssh_user
env['ANSIBLE_STDOUT_CALLBACK'] = output_callback
env['ANSIBLE_COLLECTIONS_PATHS'] = '/usr/share/ansible/collections'
env['ANSIBLE_CALLBACKS_ENABLED'] = callback_whitelist
env['ANSIBLE_TRANSPORT'] = connection
# Set var handling for better performance
env['ANSIBLE_GATHER_SUBSET'] = '!all,min'
if connection == 'local':
env['ANSIBLE_PYTHON_INTERPRETER'] = sys.executable
if gathering_policy in ('smart', 'explicit', 'implicit'):
env['ANSIBLE_GATHERING'] = gathering_policy
if module_path:
env['ANSIBLE_LIBRARY'] = ':'.join(
[env['ANSIBLE_LIBRARY'], module_path]
env['TRIPLEO_PLAN_NAME'] = plan
get_uid = int(os.getenv('SUDO_UID', os.getuid()))
user_pwd = pwd.getpwuid(get_uid)
except (KeyError, TypeError):
home = constants.CLOUD_HOME_DIR
home = user_pwd.pw_dir
env['ANSIBLE_LOG_PATH'] = os.path.join(home, 'ansible.log')
if key:
# NOTE(cloudnull): Re-apply the original environment ensuring that
# anything defined on the CLI is set accordingly.
if extra_env_variables:
if not isinstance(extra_env_variables, dict):
msg = "extra_env_variables must be a dict"
raise SystemError(msg)
if 'ANSIBLE_CONFIG' not in env and not ansible_cfg:
ansible_cfg = os.path.join(workdir, 'ansible.cfg')
config = configparser.ConfigParser()
if os.path.isfile(ansible_cfg):
if 'defaults' not in config.sections():
config.set('defaults', 'internal_poll_interval', '0.01')
with open(ansible_cfg, 'w') as f:
env['ANSIBLE_CONFIG'] = ansible_cfg
elif 'ANSIBLE_CONFIG' not in env and ansible_cfg:
env['ANSIBLE_CONFIG'] = ansible_cfg
command_path = None
with TempDirs(chdir=False) as ansible_artifact_path:
r_opts = {
'private_data_dir': workdir,
'project_dir': playbook_dir,
'inventory': _inventory(inventory),
'envvars': _encode_envvars(env=env),
'playbook': playbook,
'verbosity': verbosity,
'quiet': quiet,
'extravars': extra_vars,
'fact_cache': ansible_fact_path,
'fact_cache_type': 'jsonfile',
'artifact_dir': ansible_artifact_path,
'rotate_artifacts': 256
if skip_tags:
r_opts['skip_tags'] = skip_tags
if tags:
r_opts['tags'] = tags
if limit_hosts:
r_opts['limit'] = limit_hosts
if parallel_run:
r_opts['directory_isolation_base_path'] = ansible_artifact_path
if rotate_log and os.path.exists(env['ANSIBLE_LOG_PATH']):
runner_config = ansible_runner.runner_config.RunnerConfig(**r_opts)
# NOTE(cloudnull): overload the output callback after prepare
# to define the specific format we want.
# This is only required until PR
# is merged and released. After this PR has been
# made available to us, this line should be removed.
runner_config.env['ANSIBLE_STDOUT_CALLBACK'] = \
runner = ansible_runner.Runner(config=runner_config)
if reproduce_command:
command_path = os.path.join(
with open(command_path, 'w') as f:
f.write('#!/usr/bin/env bash\n')
f.write('echo -e "Exporting environment variables"\n')
for key, value in r_opts['envvars'].items():
f.write('export {}="{}"\n'.format(key, value))
f.write('echo -e "Running Ansible command"\n')
args = '{} "$@"\n'.format(' '.join(runner_config.command))
# Single quote the dict passed to -e
args = re.sub('({.*})', '\'\\1\'', args)
os.chmod(command_path, 0o750)
status, rc =
# NOTE(cloudnull): After a playbook executes, ensure the log
# file, if it exists, was created with
# appropriate ownership.
_log_path = r_opts['envvars']['ANSIBLE_LOG_PATH']
if os.path.isfile(_log_path):
os.chown(_log_path, get_uid, -1)
# Save files we care about
with open(os.path.join(workdir, 'stdout'), 'w') as f:
for output in 'status', 'rc':
val = getattr(runner, output)
if val:
with open(os.path.join(workdir, output), 'w') as f:
if rc != 0:
if rc == 4 and ignore_unreachable:'Ignoring unreachable nodes')
err_msg = (
'Ansible execution failed. playbook: {},'
' Run Status: {},'
' Return Code: {}'.format(
if command_path:
err_msg += (
', To rerun the failed command manually execute the'
' following script: {}'.format(
if not quiet:
raise RuntimeError(err_msg)
'Ansible execution success. playbook: {}'.format(
def convert(data):
"""Recursively converts dictionary keys,values to strings."""
if isinstance(data, str):
return str(data)
if isinstance(data, collections_abc.Mapping):
return dict(map(convert, data.items()))
if isinstance(data, collections_abc.Iterable):
return type(data)(map(convert, data))
return data
def bracket_ipv6(address):
"""Put a bracket around address if it is valid IPv6
Return it unchanged if it is a hostname or IPv4 address.
socket.inet_pton(socket.AF_INET6, address)
return "[%s]" % address
except socket.error:
return address
def is_valid_ip(ip):
"""Return True if the IP is either v4 or v6
Return False if invalid.
return netaddr.valid_ipv4(ip) or netaddr.valid_ipv6(ip)
def is_loopback(host):
"""Return True of the IP or the host is a loopback
Return False if not.
loopbacks = ['127', '::1']
for item in loopbacks:
if host.startswith(item):
return True
return False
def get_host_ips(host, socket_type=None):
"""Lookup an host to return a list of IPs.
:param host: Host to lookup
:type host: string
:param socket_type: Type of a socket (e.g. socket.AF_INET, socket.AF_INET6)
:type socket_type: string
ips = set()
if socket_type:
socket_types = (socket_type,)
socket_types = (socket.AF_INET, socket.AF_INET6)
for t in socket_types:
res = socket.getaddrinfo(host, None, t, socket.SOCK_STREAM)
except socket.error:
nips = set([x[4][0] for x in res])
return list(ips)
def get_single_ip(host, allow_loopback=False, ip_version=4):
"""Translate an hostname into a single IP address if it is a valid IP.
:param host: IP or hostname or FQDN to lookup
:type host: string
:param allow_loopback: Whether or not a loopback IP can be returned.
Defaults is False.
:type allow_loopback: boolean
Return the host unchanged if it is already an IPv4 or IPv6 address.
ip = host
if not is_valid_ip(host):
socket_type = socket.AF_INET6 if ip_version == 6 else socket.AF_INET
ips = get_host_ips(host, socket_type=socket_type)
if not ips:
raise exceptions.LookupError('No IP was found for the host: '
'%s' % host)
ip = ips[0]
if len(ips) > 1:
raise exceptions.LookupError('More than one IP was found for the '
'host %s: %s' % (host, ips))
if not allow_loopback and is_loopback(ip):
raise exceptions.LookupError('IP address for host %s is a loopback'
' IP: %s' % (host, ip))
if not is_valid_ip(ip):
raise exceptions.LookupError('IP address for host %s is not a '
'valid IP: %s' % (host, ip))
return ip
def write_env_file(env_data, env_file, registry_overwrites):
"""Write the tht env file as yaml"""
data = {'parameter_defaults': env_data}
if registry_overwrites:
data['resource_registry'] = registry_overwrites
with open(env_file, "w") as f:
dumper = yaml.dumper.SafeDumper
dumper.ignore_aliases = lambda self, data: True
yaml.dump(data, f, default_flow_style=False, Dumper=dumper)
def store_cli_param(command_name, parsed_args):
"""write the cli parameters into an history file"""
# The command name is the part after "openstack" with spaces. Switching
# to "-" makes it easier to read. "openstack undercloud install" will be
# stored as "undercloud-install" for example.
command_name = command_name.replace(" ", "-")
history_path = os.path.join(constants.CLOUD_HOME_DIR, '.tripleo')
os.mkdir(history_path, 0o700)
int(os.environ.get('SUDO_UID', -1)),
int(os.environ.get('SUDO_GID', -1)))
except OSError as e:
if e.errno != errno.EEXIST:
messages = _("Unable to create the .tripleo directory: "
"{0}, {1}").format(history_path, e)
raise IOError(messages)
if os.path.isdir(history_path):
history_file_path = os.path.join(history_path, 'history')
with open(history_file_path, 'a') as history:
args = parsed_args.__dict__.copy()
used_args = ', '.join('%s=%s' % (key, value)
for key, value in args.items())
history.write(' '.join([str(,
str(command_name), used_args, "\n"]))
int(os.environ.get('SUDO_UID', -1)),
int(os.environ.get('SUDO_GID', -1)))
except IOError as e:
messages = _("Unable to write into TripleO history file: "
"{0}, {1}").format(history_path, e)
raise IOError(messages)
raise exceptions.InvalidConfiguration(_("Target path %s is not a "
"directory") % history_path)
def create_tempest_deployer_input(config_name='tempest-deployer-input.conf',
config = configparser.ConfigParser()
# Create required sections
for section in ('auth', 'compute', 'compute-feature-enabled', 'identity',
'image', 'network', 'object-storage',
'volume', 'volume-feature-enabled'):
# Dynamic credentials means tempest will create the required credentials if
# a test requires a new account to work, tempest will create one just for
# that test
config.set('auth', 'use_dynamic_credentials', 'true')
# Does the test environment support obtaining instance serial console
# output? (default: true)
# set in [nova.serial_console]->enabled
config.set('compute-feature-enabled', 'console_output', 'false')
# Name of the backend1 (must be declared in cinder.conf)
# (default: 'BACKEND_1')
# set in [cinder]->enabled_backends
config.set('volume', 'backend1_name', 'tripleo_iscsi')
# Update bootable status of a volume Not implemented on icehouse
# (default: false)
# python-cinderclient supports set-bootable
config.set('volume-feature-enabled', 'bootable', 'true')
# Fix region value because TripleO is using non-standard value
for section in ('compute', 'identity', 'image', 'network',
'object-storage', 'volume'):
config.set(section, 'region', 'regionOne')
if output_dir:
config_path = os.path.join(output_dir, config_name)
config_path = config_name
with open(config_path, 'w+') as config_file:
def wait_for_stack_ready(orchestration_client, stack_name, marker=None,
action='CREATE', nested_depth=2,
"""Check the status of an orchestration stack
Get the status of an orchestration stack and check whether it is complete
or failed.
:param orchestration_client: Instance of Orchestration client
:type orchestration_client: heatclient.v1.client.Client
:param stack_name: Name or UUID of stack to retrieve
:type stack_name: string
:param marker: UUID of the last stack event before the current action
:type marker: string
:param action: Current action to check the stack for COMPLETE
:type action: string
:param verbose: Whether to print events
:type verbose: boolean
:param nested_depth: Max depth to look for events
:type nested_depth: int
:param max_retries: Number of retries in the case of server problems
:type max_retries: int
log = logging.getLogger(__name__ + ".wait_for_stack_ready")
stack = get_stack(orchestration_client, stack_name)
if not stack:
return False
stack_name = "%s/%s" % (stack.stack_name,
retries = 0
while retries <= max_retries:
stack_status, msg = event_utils.poll_for_events(
orchestration_client, stack_name, action=action,
poll_period=5, marker=marker, out=sys.stdout,
return stack_status == '%s_COMPLETE' % action
except hc_exc.HTTPException as e:
if e.code in [500, 503, 504]:
retries += 1
log.warning("Server issue while waiting for stack to be ready."
" Attempting retry {} of {}".format(retries,
time.sleep(retries * 5)
log.error("Error occured while waiting for stack to be ready.")
raise e
raise RuntimeError(
"wait_for_stack_ready: Max retries {} reached".format(max_retries))
def get_stack_output_item(stack, item):
if not stack:
return None
for output in stack.to_dict().get('outputs', {}):
if output['output_key'] == item:
return output['output_value']
# item not found in outputs
return None
def get_stack_saved_output_item(output, working_dir):
outputs_dir = os.path.join(working_dir, 'outputs')
output_path = os.path.join(outputs_dir, output)
if not os.path.isfile(output_path):
return None
with open(output_path) as f:
return yaml.safe_load(
def get_overcloud_endpoint(working_dir):
return get_stack_saved_output_item('KeystoneURL', working_dir)
def get_service_ips(stack):
service_ips = {}
for output in stack.to_dict().get('outputs', {}):
service_ips[output['output_key']] = output['output_value']
return service_ips
def get_endpoint_map(working_dir):
endpoint_map = get_stack_saved_output_item('EndpointMap', working_dir)
if not endpoint_map:
endpoint_map = {}
return endpoint_map
def get_excluded_ip_addresses(working_dir):
return get_stack_saved_output_item(
'BlacklistedIpAddresses', working_dir)
def get_excluded_hostnames(working_dir):
return get_stack_saved_output_item(
'BlacklistedHostnames', working_dir)
def get_role_net_ip_map(working_dir):
return get_stack_saved_output_item(
'RoleNetIpMap', working_dir)
def get_stack(orchestration_client, stack_name):
"""Get the ID for the current deployed overcloud stack if it exists.
Caller is responsible for checking if return is None
stack = orchestration_client.stacks.get(stack_name)
return stack
except HTTPNotFound:
def get_rc_params(working_dir):
rc_params = {}
rc_params['password'] = get_stack_saved_output_item(
'AdminPassword', working_dir)
rc_params['region'] = get_stack_saved_output_item(
'KeystoneRegion', working_dir)
return rc_params
def check_ceph_ansible(resource_registry, stage):
"""Fail if ceph-ansible is still passed
If any of the ceph-ansible related resources are part of the
Ceph services path, then the overcloud deploy (or the stack
update) should fail, unless they are included in the context
of Update/Upgrade/Converge, where these environments are still
relevant and required.
if not resource_registry or stage not in "DeployOvercloud":
# for each Ceph related service, fail if ceph-ansible is part
# of the provided path
for name, path in resource_registry.items():
if 'Ceph' in name and 'ceph-ansible' in path:
raise exceptions.InvalidConfiguration('The Ceph deployment is not '
'available anymore using '
'ceph-ansible. If you want '
'to deploy Ceph, please add '
'the cephadm environment '
def check_deployed_ceph_stage(environment):
"""Raises an exception if Ceph is being deployed without DeployedCeph:True.
If Ceph is not being deployed or DeployedCeph is true, then return
nothing, so the program that calls this function can continue without
error. This function also looks for the external Ceph Heat resource to
make sure in this scenario an error is not raised regardless of the
DeployedCeph boolean value.
resource_registry = environment.get('resource_registry', {})
if not resource_registry:
ceph_external = environment.get('resource_registry', {}).get(
'OS::TripleO::Services::CephExternal', 'OS::Heat::None')
if ceph_external != "OS::Heat::None":
# it's not an external Ceph cluster, let's evaluate the DeployedCeph param
# and the Ceph resources provided
deployed_ceph = environment.get('parameter_defaults',
{}).get('DeployedCeph', False)
# for each ceph resource, if the path contains cephadm and the DeployedCeph
# boolean is not True, raise an exception and guide the operator through
# the right path of deploying ceph
for name, path in resource_registry.items():
if 'Ceph' in name and 'cephadm' in path and not deployed_ceph:
raise exceptions.InvalidConfiguration('Ceph deployment is not '
'available anymore during '
'overcloud deploy. If you '
'want to deploy Ceph, '
'please see "openstack '
' overcloud ceph deploy '
'--help" to deploy ceph '
' before deploying the '
'overcloud and then include '
'the cephadm environment '
def check_ceph_fsid_matches_env_files(old_env, environment):
"""Check CephClusterFSID against proposed env files
There have been cases where operators inadvertenly changed the
CephClusterFSID on a stack update, which is unsupported by both
Ceph and openstack.
For this reason we need to check that the existing deployed Ceph
cluster ID present in the stack is consistent with the value of
the environment, raising an exception if they are different.
env_ceph_fsid = environment.get('parameter_defaults',
{}).get('CephClusterFSID', False)
stack_ceph_fsid = old_env.get('parameter_defaults',
{}).get('CephClusterFSID', False)
if bool(env_ceph_fsid) and env_ceph_fsid != stack_ceph_fsid:
raise exceptions.InvalidConfiguration('The CephFSID environment value '
' ({}) does not match the stack '
' configuration value ({}).'
' Ensure the CephClusterFSID '
' param is properly configured '
' in the storage environment '
' files.'
def check_swift_and_rgw(old_env, env, stage):
"""Check that Swift and RGW aren't both enabled in the overcloud
When Ceph is deployed by TripleO using the default cephadm environment
file, the RGW component is included by default and deployed on both
greenfield and brownfield deployments.
However, if an overcloud upgrade is run and Swift was already deployed,
the RGW resource shouldn't replace Swift and -e cephadm-rbd-only.yaml
should be passed.
For this reason we need to check if Swift was previously enabled, and
fail if the RGW resource is passed.
rgw_env = env.get('resource_registry',
allowed_stage = re.compile("(Upgrade|Update)Prepare", re.I)
# if the RGW resource isn't passed or we're not in the upgrade context
# there's no need to run this check
if not re.match(allowed_stage, stage) or rgw_env == 'OS::Heat::None':
sw = old_env.get('resource_registry',
# RGW is present in the env list and swift was previously deployed
if sw != "OS::Heat::None":
raise exceptions.InvalidConfiguration('Both Swift and RGW resources '
'detected. '
'Ensure you have only one of '
'them enabled (or provide the '
'cephadm-rbd-only.yaml '
'environment file to exclude '
def check_network_plugin(output_dir, env):
"""Disallow upgrade if change in network plugin detected
If the undercloud is upgraded with a change in network plugin
i.e ovs to ovn or ovn to ovs it will break the undercloud as
just switching is not enough it needs network resources to be
migrated, so we detect if there is change in network and block
the upgrade
neutron_env = env.get('resource_registry',
# Neutron is not deployed so just return
if neutron_env == "OS::Heat::None":
parameters = env.get('parameter_defaults', {})
file_name = constants.TRIPLEO_STATIC_INVENTORY
inventory_path = os.path.join(output_dir, file_name)
if not os.path.isfile(inventory_path):
message = (_("The %s inventory file is missing. Without it "
"network plugin change can't be detected, and upgrade "
"will have issues if there is a change" % inventory_path))
raise exceptions.InvalidConfiguration(message)
with open(inventory_path, 'r') as f:
inventory_data = yaml.safe_load(f)
if ('neutron_ovs_agent' in inventory_data and
'ovn' in parameters.get('NeutronMechanismDrivers')):
message = _("Network Plugin mismatch detected, "
"Upgrade from ml2 ovs to ml2 ovn is not allowed")
raise exceptions.InvalidConfiguration(message)
elif ("ovn_controller" in inventory_data and
"openvswitch" in parameters.get('NeutronMechanismDrivers')):
message = _("Network Plugin mismatch detected, "
"Upgrade from ml2 ovn to ml2 ovs is not allowed")
raise exceptions.InvalidConfiguration(message)
def check_service_vips_migrated_to_service(environment):
registry = environment.get('resource_registry', {})
removed_resources = {'OS::TripleO::Network::Ports::RedisVipPort',
msg = ("Resources 'OS::TripleO::Network::Ports::RedisVipPort' and "
"'OS::TripleO::Network::Ports::OVNDBsVipPort' can no longer be "
"used. Service VIPs has been moved to the service definition "
"template. To configure a specific IP address use the parameters "
"'RedisVirtualFixedIPs' and/or 'OVNDBsVirtualFixedIPs'. To control"
"the network or subnet for VIP allocation set up the "
"'ServiceNetMap' and/or 'VipSubnetMap' parameters with the desired "
"network and/or subnet for the service.")
for resource in removed_resources:
if (resource in registry and
registry.get(resource) != 'OS::Heat::None'):
raise exceptions.InvalidConfiguration(msg)
def check_neutron_resources(environment):
registry = environment.get('resource_registry', {})
msg = ("Resource {} maps to type {} and the Neutron "
"service is not available when using ephemeral Heat. "
"The generated environments from "
"'openstack overcloud baremetal provision' and "
"'openstack overcloud network provision' must be included "
"with the deployment command.")
for rsrc, rsrc_type in registry.items():
if (type(rsrc_type) == str and
raise exceptions.InvalidConfiguration(msg.format(rsrc, rsrc_type))
def remove_known_hosts(overcloud_ip):
"""For a given IP address remove SSH keys from the known_hosts file"""
known_hosts = os.path.join(constants.CLOUD_HOME_DIR, '.ssh/known_hosts')
if os.path.exists(known_hosts):
command = ['ssh-keygen', '-R', overcloud_ip, '-f', known_hosts]
def file_checksum(filepath, hash_algo='sha512'):
"""Calculate sha512 checksum on file
:param filepath: Full path to file (e.g. /home/stack/image.qcow2)
:type filepath: string
:param hash_algo: name of the hash algorithm, 'sha512' by default
:type hash_algo: string
:returns: hexadecimal hash of the file
RuntimeError if the 'hash_algo' value isn't supported.
ValueError if the path isn't pointing to a regular file.
if not os.path.isfile(filepath):
raise ValueError(_("The given file {0} is not a regular "
if hash_algo not in constants.FIPS_COMPLIANT_HASHES:
raise RuntimeError(
"The requested hash algorithm (%s) is not supported." % hash_algo)
checksum =
with open(filepath, 'rb') as f:
while True:
fragment =
if not fragment:
return checksum.hexdigest()
def ensure_run_as_normal_user():
"""Check if the command runs under normal user (EUID!=0)"""
if os.geteuid() == 0:
raise exceptions.RootUserExecution(_(
'This command cannot run under root user.'
' Switch to a normal user.'))
def get_deployment_user():
"""Return the user name which is used to deploy the cloud"""
return getpass.getuser()
def capabilities_to_dict(caps):
"""Convert the Node's capabilities into a dictionary."""
if not caps:
return {}
return dict([key.split(':', 1) for key in caps.split(',')])
def dict_to_capabilities(caps_dict):
"""Convert a dictionary into a string with the capabilities syntax."""
return ','.join(["%s:%s" % (key, value)
for key, value in caps_dict.items()
if value is not None])
def node_get_capabilities(node):
"""Get node capabilities."""
return capabilities_to_dict('capabilities'))
def node_add_capabilities(bm_client, node, **updated):
"""Add or replace capabilities for a node."""
caps = node_get_capabilities(node)
converted_caps = dict_to_capabilities(caps)['capabilities'] = converted_caps
bm_client.node.update(node.uuid, [{'op': 'add',
'path': '/properties/capabilities',
'value': converted_caps}])
return caps
def assign_and_verify_profiles(bm_client, flavors,
assign_profiles=False, dry_run=False):
"""Assign and verify profiles for given flavors.
:param bm_client: ironic client instance
:param flavors: map flavor name -> (flavor object, required count)
:param assign_profiles: whether to allow assigning profiles to nodes
:param dry_run: whether to skip applying actual changes (only makes sense
if assign_profiles is True)
:returns: tuple (errors count, warnings count)
log = logging.getLogger(__name__ + ".assign_and_verify_profiles")
predeploy_errors = 0
predeploy_warnings = 0
# nodes available for deployment and scaling (including active)
bm_nodes = {node.uuid: node
for node in bm_client.node.list(maintenance=False,
if node.provision_state in ('available', 'active')}
# create a pool of unprocessed nodes and record their capabilities
free_node_caps = {uu: node_get_capabilities(node)
for uu, node in bm_nodes.items()}
# TODO(dtantsur): use command-line arguments to specify the order in
# which profiles are processed (might matter for assigning profiles)
profile_flavor_used = False
for flavor_name, (flavor, scale) in flavors.items():
if not scale:
log.debug("Skipping verification of flavor %s because "
"none will be deployed", flavor_name)
profile = flavor.get_keys().get('capabilities:profile')
# If there's only a single flavor, then it's expected for it to have
# no profile assigned.
if not profile and len(flavors) > 1:
predeploy_errors += 1
'Error: The %s flavor has no profile associated', flavor_name)
'Recommendation: assign a profile with openstack flavor '
'set --property "capabilities:profile"="PROFILE_NAME" %s',
profile_flavor_used = True
# first collect nodes with known profiles
assigned_nodes = [uu for uu, caps in free_node_caps.items()
if caps.get('profile') == profile]
required_count = scale - len(assigned_nodes)
if required_count < 0:
log.warning('%d nodes with profile %s won\'t be used '
'for deployment now', -required_count, profile)
predeploy_warnings += 1
required_count = 0
elif required_count > 0 and assign_profiles:
# find more nodes by checking XXX_profile capabilities that are
# set by ironic-inspector or manually
capability = '%s_profile' % profile
more_nodes = [
uu for uu, caps in free_node_caps.items()
# use only nodes without a know profile
if not caps.get('profile') and
caps.get(capability, '').lower() in ('1', 'true') and
# do not assign profiles for active nodes
bm_nodes[uu].provision_state == 'available'
required_count -= len(more_nodes)
for uu in assigned_nodes:
# make sure these nodes are not reused for other profiles
node_caps = free_node_caps.pop(uu)
# save profile for newly assigned nodes, but only if we
# succeeded in finding enough of them
if not required_count and not node_caps.get('profile'):
node = bm_nodes[uu]
if not dry_run:
node_add_capabilities(bm_client, node, profile=profile)'Node %s was assigned profile %s', uu, profile)
log.debug('Node %s has profile %s', uu, profile)
if required_count > 0:
"Error: only %s of %s requested ironic nodes are tagged "
"to profile %s (for flavor %s)",
scale - required_count, scale, profile, flavor_name
"Recommendation: tag more nodes using $ openstack baremetal "
"node set --properties capabilities=profile:%s, <NODE ID>",
predeploy_errors += 1
nodes_without_profile = [uu for uu, caps in free_node_caps.items()
if not caps.get('profile')]
if nodes_without_profile and profile_flavor_used:
predeploy_warnings += 1
"There are %d ironic nodes with no profile that will "
"not be used: %s", len(nodes_without_profile),
', '.join(nodes_without_profile)
return predeploy_errors, predeploy_warnings
def add_deployment_plan_arguments(parser):
"""Add deployment plan arguments (flavors and scales) to a parser"""
# TODO(d0ugal): Deprecated in Newton. Remove these in U.
parser.add_argument('--control-scale', type=int,
help=_('New number of control nodes.'))
parser.add_argument('--compute-scale', type=int,
help=_('New number of compute nodes.'))
parser.add_argument('--ceph-storage-scale', type=int,
help=_('New number of ceph storage nodes.'))
parser.add_argument('--block-storage-scale', type=int,
help=_('New number of cinder storage nodes.'))
parser.add_argument('--swift-storage-scale', type=int,
help=_('New number of swift storage nodes.'))
help=_('Nova flavor to use for control nodes.'))
help=_('Nova flavor to use for compute nodes.'))
help=_('Nova flavor to use for ceph storage nodes.'))
help=_('Nova flavor to use for cinder storage nodes'))
help=_('Nova flavor to use for swift storage nodes'))
def get_roles_info(parsed_args):
"""Get flavor name and scale for all deployment roles.
:returns: dict role name -> (flavor name, scale)
return {
'control': (parsed_args.control_flavor, parsed_args.control_scale),
'compute': (parsed_args.compute_flavor, parsed_args.compute_scale),
'ceph-storage': (parsed_args.ceph_storage_flavor,
'block-storage': (parsed_args.block_storage_flavor,
'swift-storage': (parsed_args.swift_storage_flavor,
def _csv_to_nodes_dict(nodes_csv):
"""Convert CSV to a list of dicts formatted for os_cloud_config
Given a CSV file in the format below, convert it into the
structure expected by os_cloud_config JSON files.
pm_type, pm_addr, pm_user, pm_password, mac
data = []
for row in csv.reader(nodes_csv):
node = {
"pm_user": row[2],
"pm_addr": row[1],
"pm_password": row[3],
"pm_type": row[0],
"mac": [
node['pm_port'] = row[5]
except IndexError:
return data
def parse_env_file(env_file, file_type=None):
if file_type == 'json' or'.json'):
nodes_config = json.load(env_file)
elif file_type == 'csv' or'.csv'):
nodes_config = _csv_to_nodes_dict(env_file)
nodes_config = yaml.safe_load(env_file)
raise exceptions.InvalidConfiguration(
_("Invalid file extension for %s, must be json, yaml or csv") %
if 'nodes' in nodes_config:
nodes_config = nodes_config['nodes']
return nodes_config
def prompt_user_for_confirmation(message, logger, positive_response='y'):
"""Prompt user for a y/N confirmation
Use this function to prompt the user for a y/N confirmation
with the provided message. The [y/N] should be included in
the provided message to this function to indicate the expected
input for confirmation. You can customize the positive response if
y/N is not a desired input.
:param message: Confirmation string prompt
:param logger: logger object used to write info logs
:param positive_response: Beginning character for a positive user input
:return: boolean true for valid confirmation, false for all others
if not sys.stdin.isatty():
logger.error(_('User interaction required, cannot confirm.'))
return False
prompt_response = sys.stdin.readline().lower()
if not prompt_response.startswith(positive_response):
'User did not confirm action so taking no action.'))
return False'User confirmed action.'))
return True
except KeyboardInterrupt: # ctrl-c
'User did not confirm action (ctrl-c) so taking no action.'))
except EOFError: # ctrl-d
'User did not confirm action (ctrl-d) so taking no action.'))
return False
def replace_links_in_template_contents(contents, link_replacement):
"""Replace get_file and type file links in Heat template contents
If the string contents passed in is a Heat template, scan the
template for 'get_file' and 'type' occurrences, and replace the
file paths according to link_replacement dict. (Key/value in
link_replacement are from/to, respectively.)
If the string contents don't look like a Heat template, return the
contents unmodified.
template = {}
template = yaml.safe_load(contents)
except yaml.YAMLError:
return contents
if not (isinstance(template, dict) and
return contents
template = replace_links_in_template(template, link_replacement)
return yaml.safe_dump(template)
def replace_links_in_template(template_part, link_replacement):
"""Replace get_file and type file links in a Heat template
Scan the template for 'get_file' and 'type' occurrences, and
replace the file paths according to link_replacement
dict. (Key/value in link_replacement are from/to, respectively.)
def replaced_dict_value(key, value):
if ((key == 'get_file' or key == 'type') and
isinstance(value, str)):
return link_replacement.get(value, value)
return replace_links_in_template(value, link_replacement)
def replaced_list_value(value):
return replace_links_in_template(value, link_replacement)
if isinstance(template_part, dict):
return {k: replaced_dict_value(k, v)
for k, v in template_part.items()}
if isinstance(template_part, list):
return list(map(replaced_list_value, template_part))
return template_part
def relative_link_replacement(link_replacement, current_dir):
"""Generate a relative version of link_replacement dictionary.
Get a link_replacement dictionary (where key/value are from/to
respectively), and make the values in that dictionary relative
paths with respect to current_dir.
return {k: os.path.relpath(v, current_dir)
for k, v in link_replacement.items()}
def load_environment_directories(directories):
log = logging.getLogger(__name__ + ".load_environment_directories")
environments = []
for d in directories:
if os.path.exists(d) and d != '.':
log.debug("Environment directory: %s" % d)
for f in sorted(glob.glob(os.path.join(d, '*.yaml'))):
log.debug("Environment directory file: %s" % f)
if os.path.isfile(f):
return environments
def get_key(stack, needs_pair=False):
"""Returns the private key from the local file system.
Searches for and returns the stack private key. If the key is inaccessible
for any reason, the process will fall back to using the users key. If no
key is found, this method will return None.
:params stack: name of the stack to use
:type stack: String
:param needs_pair: Enable key pair search
:type needs_pair: Boolean
:returns: String || None
key_files = list()
stack_dir = get_default_working_dir(stack)
key_files.append(os.path.join(stack_dir, 'ssh_private_key'))
user_dir = os.path.join(constants.CLOUD_HOME_DIR, '.ssh')
key_files.append(os.path.join(user_dir, 'id_rsa_tripleo'))
key_files.append(os.path.join(user_dir, 'id_rsa'))
legacy_dir = os.path.join(constants.DEFAULT_WORK_DIR, '.ssh')
key_files.append(os.path.join(legacy_dir, 'tripleo-admin-rsa'))
for key_file in key_files:
if os.path.exists(key_file):
if needs_pair:
if not os.path.exists('{}.pub'.format(key_file)):
with open(key_file):
return key_file
except IOError:
def get_tripleo_ansible_inventory(inventory_file=None,
if not inventory_file:
inventory_file = os.path.join(
command = ['/usr/bin/tripleo-ansible-inventory',
'--os-cloud', 'undercloud']
if stack:
command.extend(['--stack', stack])
command.extend(['--undercloud-key-file', get_key(stack=stack)])
if ssh_user:
command.extend(['--ansible_ssh_user', ssh_user])
if undercloud_connection:
if inventory_file:
command.extend(['--static-yaml-inventory', inventory_file])
rc = run_command_and_log(LOG, command)
if rc != 0:
message = "Failed to generate inventory"
raise exceptions.InvalidConfiguration(message)
if os.path.exists(inventory_file):
if return_inventory_file_path:
return inventory_file
with open(inventory_file, "r") as f:
inventory =
return inventory
raise exceptions.InvalidConfiguration(_(
"Inventory file %s can not be found.") % inventory_file)
def cleanup_tripleo_ansible_inventory_file(path):
"""Remove the static tripleo-ansible-inventory file from disk"""
if os.path.exists(path):
processutils.execute('/usr/bin/rm', '-f', path)
def get_roles_file_path(working_dir, stack_name):
roles_file = os.path.join(
return roles_file
def get_networks_file_path(working_dir, stack_name):
networks_file = os.path.join(
return networks_file
def get_baremetal_file_path(working_dir, stack_name):
baremetal_file_name = os.path.join(
baremetal_file = (baremetal_file_name
if os.path.exists(baremetal_file_name) else None)
return baremetal_file
def get_vip_file_path(working_dir, stack_name):
vip_file = os.path.join(
return vip_file
def rewrite_ansible_playbook_paths(src, dest):
"""Rewrite relative paths to playbooks in the dest roles file, so that
the path is the absolute path relative to the src roles file
with open(dest, 'r') as f:
wd_roles = yaml.safe_load(
for role_idx, role in enumerate(wd_roles):
for pb_idx, pb_def in enumerate(role.get('ansible_playbooks', [])):
path = rel_or_abs_path_role_playbook(os.path.dirname(src),
'playbook'] = path
with open(dest, 'w') as f:
def copy_to_wd(working_dir, file, stack, kind):
src = os.path.abspath(file)
dest = os.path.join(working_dir,
shutil.copy(src, dest)
if kind == 'baremetal':
rewrite_ansible_playbook_paths(src, dest)
def update_working_dir_defaults(working_dir, args):
stack_name = args.stack
tht_root = os.path.abspath(args.templates)
if isinstance(args.baremetal_deployment, str):
copy_to_wd(working_dir, args.baremetal_deployment, stack_name,
if args.roles_file:
copy_to_wd(working_dir, args.roles_file, stack_name, 'roles')
elif not os.path.exists(
file = os.path.join(tht_root, constants.OVERCLOUD_ROLES_FILE)
copy_to_wd(working_dir, file, stack_name, 'roles')
if args.networks_file:
copy_to_wd(working_dir, args.networks_file, args.stack, 'networks')
elif not os.path.exists(
file = os.path.join(tht_root, constants.OVERCLOUD_NETWORKS_FILE)
copy_to_wd(working_dir, file, stack_name, 'networks')
if args.vip_file:
copy_to_wd(working_dir, args.vip_file, args.stack, 'vips')
elif not os.path.exists(
file = os.path.join(tht_root, constants.OVERCLOUD_VIP_FILE)
copy_to_wd(working_dir, file, stack_name, 'vips')
def build_stack_data(clients, stack_name, template,
files, env_files):
orchestration_client = clients.orchestration
fields = {
'template': template,
'files': files,
'environment_files': env_files,
'show_nested': True
stack_data = {}
result = orchestration_client.stacks.validate(**fields)
if result:
stack_data['environment_parameters'] = result.get(
'Environment', {}).get('parameter_defaults')
flattened = {'resources': {}, 'parameters': {}}
stack_utils._flat_it(flattened, 'Root', result)
stack_data['heat_resource_tree'] = flattened
return stack_data
def archive_deploy_artifacts(log, stack_name, working_dir, ansible_dir=None):
"""Create a tarball of the temporary folders used"""
log.debug(_("Preserving deployment artifacts"))
def get_tar_filename():
return os.path.join(
working_dir, '%s-install-%s.tar.bzip2' %
def tar_filter(info):
"""Tar filter to remove output dir from path"""
return None
leading_path = working_dir[1:] + '/' =, '')
return info
tar_filename = get_tar_filename()
tf =, 'w:bz2')
tf.add(working_dir, recursive=True, filter=tar_filter)
if ansible_dir:
tf.add(ansible_dir, recursive=True,
except tarfile.TarError as ex:
msg = _("Unable to create artifact tarball, %s") % str(ex)
return tar_filename
def jinja_render_files(log, templates, working_dir,
roles_file=None, networks_file=None,
base_path=None, output_dir=None):
python_version = sys.version_info[0]
python_cmd = "python{}".format(python_version)
process_templates = os.path.join(
templates, 'tools/')
args = [python_cmd, process_templates]
args.extend(['--roles-data', roles_file])
args.extend(['--network-data', networks_file])
if base_path:
args.extend(['-p', base_path])
if output_dir:
args.extend(['-o', output_dir])
if run_command_and_log(log, args, working_dir) != 0:
msg = _("Problems generating templates.")
raise exceptions.DeploymentError(msg)
def rewrite_env_path(env_path, tht_root, user_tht_root, log=None):
abs_env_path = os.path.abspath(env_path)
if (abs_env_path.startswith(user_tht_root)
and ((user_tht_root + '/') in env_path
or (user_tht_root + '/') in abs_env_path
or user_tht_root == abs_env_path
or user_tht_root == env_path)):
new_env_path = env_path.replace(user_tht_root + '/', tht_root + '/')
if log:
log.debug("Redirecting env file %s to %s"
% (abs_env_path, new_env_path))
env_path = new_env_path
return env_path, abs_env_path
def process_multiple_environments(created_env_files, tht_root,
log = logging.getLogger(__name__ + ".process_multiple_environments")
env_files = {}
localenv = {}
include_env_in_files = env_files_tracker is not None
# Normalize paths for full match checks
user_tht_root = os.path.normpath(user_tht_root)
tht_root = os.path.normpath(tht_root)
for env_path in created_env_files:
log.debug("Processing environment files %s" % env_path)
env_path, abs_env_path = rewrite_env_path(env_path, tht_root,
user_tht_root, log=log)
files, env = template_utils.process_environment_and_files(
env_path=env_path, include_env_in_files=include_env_in_files)
if env_files_tracker is not None:
except hc_exc.CommandError as ex:
# This provides fallback logic so that we can reference files
# inside the resource_registry values that may be rendered via
# j2.yaml templates, where the above will fail because the
# file doesn't exist in user_tht_root, but it is in tht_root
# See bug
# for details on why this is needed (backwards-compatibility)
log.debug("Error %s processing environment file %s"
% (str(ex), env_path))
# Use the temporary path as it's possible the environment
# itself was rendered via jinja.
with open(env_path, 'r') as f:
env_map = yaml.safe_load(f)
env_registry = env_map.get('resource_registry', {})
env_dirname = os.path.dirname(os.path.abspath(env_path))
for rsrc, rsrc_path in env_registry.items():
# We need to calculate the absolute path relative to
# env_path not cwd (which is what abspath uses).
abs_rsrc_path = os.path.normpath(
os.path.join(env_dirname, rsrc_path))
# If the absolute path matches user_tht_root, rewrite
# a temporary environment pointing at tht_root instead
if (abs_rsrc_path.startswith(user_tht_root) and
((user_tht_root + '/') in abs_rsrc_path or
abs_rsrc_path == user_tht_root)):
new_rsrc_path = abs_rsrc_path.replace(
user_tht_root + '/', tht_root + '/')
log.debug("Rewriting %s %s path to %s"
% (env_path, rsrc, new_rsrc_path))
env_registry[rsrc] = new_rsrc_path
# Skip any resources that are mapping to OS::*
# resource names as these aren't paths
if not rsrc_path.startswith("OS::"):
env_registry[rsrc] = abs_rsrc_path
env_map['resource_registry'] = env_registry
f_name = os.path.basename(os.path.splitext(abs_env_path)[0])
with tempfile.NamedTemporaryFile(dir=tht_root,
prefix="env-%s-" % f_name,
delete=cleanup) as f:
log.debug("Rewriting %s environment to %s"
% (env_path,
f.write(yaml.safe_dump(env_map, default_flow_style=False))
files, env = template_utils.process_environment_and_files(, include_env_in_files=include_env_in_files)
if env_files_tracker is not None:
if files:
log.debug("Adding files %s for %s" % (files, env_path))
# 'env' can be a deeply nested dictionary, so a simple update is
# not enough
localenv = template_utils.deep_update(localenv, env)
return env_files, localenv
def parse_extra_vars(extra_var_strings):
"""Parses extra variables like Ansible would.
Each element in extra_var_strings is like the raw value of -e
parameter of ansible-playbook command. It can either be very
simple 'key=val key2=val2' format or it can be '{ ... }'
representing a YAML/JSON object.
The 'key=val key2=val2' format gets processed as if it was
'{"key": "val", "key2": "val2"}' object, and all YAML/JSON objects
get shallow-merged together in the order as they appear in
extra_var_strings, latter objects taking precedence over earlier
:param extra_var_strings: unparsed value(s) of -e parameter(s)
:type extra_var_strings: list of strings
:returns dict representing a merged object of all extra vars
result = {}
for extra_var_string in extra_var_strings:
invalid_yaml = False
parse_vars = yaml.safe_load(extra_var_string)
except yaml.YAMLError:
invalid_yaml = True
if invalid_yaml or not isinstance(parse_vars, dict):
parse_vars = dict(
item.split('=') for item in extra_var_string.split())
except ValueError:
raise ValueError(
'Invalid format for {extra_var_string}'.format(
return result
def prepend_environment(environment_files, templates_dir, environment):
if not environment_files:
environment_files = []
full_path = os.path.join(templates_dir, environment)
# sanity check it exists before proceeding
if os.path.exists(full_path):
# We need to prepend before the files provided by user.
environment_files.insert(0, full_path)
raise exceptions.InvalidConfiguration(_(
"Expected environment file {0} not found in {1} cannot proceed.")
.format(environment, templates_dir))
return environment_files
def get_hostname(short=False):
"""Returns the local hostname
:param (short): boolean true to run 'hostname -s'
:return string
if short:
cmd = ["hostname", "-s"]
cmd = ["hostname"]
p = subprocess.Popen(cmd, stdout=subprocess.PIPE,
return p.communicate()[0].rstrip().lower()
def get_short_hostname():
"""Returns the local short hostname
:return string
return get_hostname(short=True)
def wait_api_port_ready(api_port, host=''):
"""Wait until an http services becomes available
:param api_port: api service port
:type api_port: integer
:param host: host running the service (default:
:type host: string
:return boolean
log = logging.getLogger(__name__ + ".wait_api_port_ready")
urlopen_timeout = 1
max_retries = 30
count = 0
while count < max_retries:
count += 1
"http://%s:%s/" % (host, api_port), timeout=urlopen_timeout)
return False
except url_error.HTTPError as he:
if he.code == 300:
return True
except url_error.URLError:
except socket.timeout:
"Timeout at attempt {} of {} after {}s waiting for API port..."
.format(count, max_retries, urlopen_timeout))
raise RuntimeError(
"wait_api_port_ready: Max retries {} reached".format(max_retries))
def bulk_symlink(log, src, dst, tmpd='/tmp'):
"""Create bulk symlinks from a directory
:param log: logger instance for logging
:type log: Logger
:param src: dir of directories to symlink
:type src: string
:param dst: dir to create the symlinks
:type dst: string
:param tmpd: temporary working directory to use
:type tmp: string
log.debug("Symlinking %s to %s, via temp dir %s" %
(src, dst, tmpd))
with TempDirs(dir_path=tmpd) as tmp:
for obj in os.listdir(src):
if not os.path.exists(os.path.join(dst, obj)):
tmpf = os.path.join(tmp, obj)
os.symlink(os.path.join(src, obj), tmpf)
os.rename(tmpf, os.path.join(dst, obj))
def run_command_and_log(log, cmd, cwd=None, env=None):
"""Run command and log output
:param log: logger instance for logging
:type log: Logger
:param cmd: command in list form
:type cmd: List
:param cwd: current working directory for execution
:type cmd: String
:param env: modified environment for command run
:type env: List
proc = subprocess.Popen(cmd, stdout=subprocess.PIPE,
stderr=subprocess.STDOUT, shell=False,
cwd=cwd, env=env)
while True:
line = proc.stdout.readline()
except StopIteration:
if line != b'':
if isinstance(line, bytes):
line = line.decode('utf-8')
return proc.wait()
def build_prepare_env(environment_files, environment_directories):
'''Build the environment for container image prepare
:param environment_files: List of environment files to build
environment from
:type environment_files: list
:param environment_directories: List of environment directories to build
environment from
:type environment_directories: list
env_files = []
if environment_directories:
if environment_files:
def get_env_file(method, path):
if not os.path.exists(path):
return '{}'
env_url = heat_utils.normalise_file_path_to_url(path)
return request.urlopen(env_url).read()
return (
env_path_is_object=lambda path: True,
def rel_or_abs_path(file_path, tht_root):
'''Find a file, either absolute path or relative to the t-h-t dir'''
if not file_path:
return None
path = os.path.abspath(file_path)
if not os.path.isfile(path):
path = os.path.abspath(os.path.join(tht_root, file_path))
if not os.path.isfile(path):
raise exceptions.DeploymentError(
"Can't find path %s %s" % (file_path, path))
return path
def fetch_roles_file(roles_file, tht_path=constants.TRIPLEO_HEAT_TEMPLATES):
'''Fetch t-h-t roles data fromm roles_file abs path or rel to tht_path.'''
if not roles_file:
return None
with open(rel_or_abs_path(roles_file, tht_path)) as f:
return yaml.safe_load(f)
def load_config(osloconf, path):
'''Load oslo config from a file path. '''
log = logging.getLogger(__name__ + ".load_config")
conf_params = []
if os.path.isfile(path):
conf_params += ['--config-file', path]
log.warning(_('%s does not exist. Using defaults.') % path)
def configure_logging(log, level, log_file):
'''Mimic oslo_log default levels and formatting for the logger. '''
fhandler = logging.FileHandler(log_file)
formatter = logging.Formatter(
'%(asctime)s.%(msecs)03d %(process)d %(levelname)s '
'%(name)s [ ] %(message)s',
'%Y-%m-%d %H:%M:%S')
if level > 1:
# NOTE(bogdando): we are making an exception to the oslo_log'ish
# default WARN level to have INFO logs as well. Some modules
# produce INFO msgs we want to see and keep by default, like
# pre-flight valiation notes.
def _name_helper(basename, arch=None, platform=None, use_subdir=False):
# NOTE(tonyb): We don't accept a platform with an arch. This caught when
# import the nodes / process args, but lets be a little cautious here
# anyway.
if use_subdir:
delim = '/'
delim = '-'
if arch and platform:
basename = platform + '-' + arch + delim + basename
elif arch:
basename = arch + delim + basename
return basename
def overcloud_kernel(basename, arch=None, platform=None,
return (_name_helper('%s-vmlinuz' % basename, arch=arch,
platform=platform, use_subdir=use_subdir),
def overcloud_ramdisk(basename, arch=None, platform=None,
return (_name_helper('%s-initrd' % basename, arch=arch,
platform=platform, use_subdir=use_subdir),
def overcloud_image(basename, arch=None, platform=None,
return (_name_helper(basename, arch=arch, platform=platform,
def deploy_kernel(basename='agent', arch=None, platform=None,
return _name_helper(basename, arch=arch, platform=platform,
use_subdir=use_subdir) + '.kernel'
def deploy_ramdisk(basename='agent', arch=None, platform=None,
return _name_helper(basename, arch=arch, platform=platform,
use_subdir=use_subdir) + '.ramdisk'
def _candidate_files(node, call):
arch = node.get('arch')
platform = node.get('platform')
if arch:
if platform:
yield call(arch=arch, platform=platform)
yield call(arch=arch)
yield call()
def update_nodes_deploy_data(nodes,
"""Add specific kernel and ramdisk IDs to a node.
Look at all images and update node data with the most specific
deploy_kernel and deploy_ramdisk for the architecture/platform combination.
for node in nodes:
# NOTE(tonyb): Check to see if we have a specific kernel for this node
# and use that. Fall back to the generic image.
if 'kernel_id' not in node:
kernel_locations = list(_candidate_files(node, deploy_kernel))
for kernel in kernel_locations:
path = os.path.join(http_boot, kernel)
if os.path.exists(path):
node['kernel_id'] = 'file://%s/%s' % (
raise RuntimeError('No kernel image provided and none of %s '
'found in %s' % (kernel_locations,
# NOTE(tonyb): As above except for ramdisks
if 'ramdisk_id' not in node:
ramdisk_locations = list(_candidate_files(node, deploy_ramdisk))
for ramdisk in ramdisk_locations:
path = os.path.join(http_boot, ramdisk)
if os.path.exists(path):
node['ramdisk_id'] = 'file://%s/%s' % (
raise RuntimeError('No ramdisk image provided and none of %s '
'found in %s' % (ramdisk_locations,
def get_deployment_python_interpreter(parsed_args):
"""Return correct deployment python interpreter """
if parsed_args.deployment_python_interpreter:
return parsed_args.deployment_python_interpreter
return sys.executable
def run_command(args, env=None, name=None, logger=None):
"""Run the command defined by args and return its output
:param args: List of arguments for the command to be run.
:param env: Dict defining the environment variables. Pass None to use
the current environment.
:param name: User-friendly name for the command being run. A value of
None will cause args[0] to be used.
if logger is None:
logger = LOG
if name is None:
name = args[0]
output = subprocess.check_output(args,
if isinstance(output, bytes):
output = output.decode('utf-8')
return output
except subprocess.CalledProcessError as e:
message = '%s failed: %s' % (name, e.output)
raise RuntimeError(message)
def set_hostname(hostname):
"""Set system hostname to provided hostname
:param hostname: The hostname to set
args = ['sudo', 'hostnamectl', 'set-hostname', hostname]
return run_command(args, name='hostnamectl')
def check_hostname(fix_etc_hosts=True, logger=None):
"""Check system hostname configuration
Rabbit and Puppet require pretty specific hostname configuration. This
function ensures that the system hostname settings are valid before
continuing with the installation.
:param fix_etc_hosts: Boolean to to enable adding hostname to /etc/hosts
if not found.
if logger is None:
logger = LOG'Checking for a FQDN hostname...')
args = ['hostnamectl', '--static']
detected_static_hostname = run_command(args, name='hostnamectl').rstrip()'Static hostname detected as %s', detected_static_hostname)
args = ['hostnamectl', '--transient']
detected_transient_hostname = run_command(args,
name='hostnamectl').rstrip()'Transient hostname detected as %s',
if detected_static_hostname != detected_transient_hostname:
logger.error('Static hostname "%s" does not match transient hostname '
'"%s".', detected_static_hostname,
logger.error('Use hostnamectl to set matching hostnames.')
raise RuntimeError('Static and transient hostnames do not match')
short_hostname = detected_static_hostname.split('.')[0]
if short_hostname == detected_static_hostname:
message = _('Configured hostname is not fully qualified.')
raise RuntimeError(message)
with open('/etc/hosts') as hosts_file:
for line in hosts_file:
# check if hostname is in /etc/hosts
if (not line.lstrip().startswith('#') and
detected_static_hostname in line.split()):
# hostname not found, add it to /etc/hosts
if not fix_etc_hosts:
sed_cmd = (r'sed -i "s/\(\s*\)/\\1%s %s /" '
'/etc/hosts' %
(detected_static_hostname, short_hostname))
args = ['sudo', '/bin/bash', '-c', sed_cmd]
run_command(args, name='hostname-to-etc-hosts')'Added hostname %s to /etc/hosts',
def check_env_for_proxy(no_proxy_hosts=None):
"""Check env proxy settings
:param no_proxy_hosts: array of hosts to check if in no_proxy env var
if no_proxy_hosts is None:
no_proxy_hosts = ['']
http_proxy = os.environ.get('http_proxy')
https_proxy = os.environ.get('https_proxy')
if os.environ.get('no_proxy'):
no_proxy = os.environ.get('no_proxy').split(',')
no_proxy = []
missing_hosts = []
if http_proxy or https_proxy:
missing_hosts = set(no_proxy_hosts) - set(no_proxy)
if missing_hosts:
message = _('http_proxy or https_proxy is set but the following local '
'addresses "{}" may be missing from the no_proxy '
'environment variable').format(','.join(missing_hosts))
raise RuntimeError(message)
def get_read_config(cfg):
"""Return the config read from ini config file(s)"""
config = configparser.ConfigParser()
return config
def getboolean_from_cfg(cfg, param, section="DEFAULT"):
"""Return a parameter from Kolla config"""
return _get_from_cfg(cfg, cfg.getboolean, param, section)
def get_from_cfg(cfg, param, section="DEFAULT"):
"""Return a parameter from Kolla config"""
return _get_from_cfg(cfg, cfg.get, param, section)
def _get_from_cfg(cfg, accessor, param, section):
"""Return a parameter from Kolla config"""
val = accessor(section, param)
except (ValueError, configparser.Error):
raise exceptions.NotFound(_("Unable to find {section}/{option} in "
return val
def get_local_timezone():
info = run_command(['timedatectl'], name='timedatectl')
timezoneline = [tz for tz in info.split('\n') if 'Time zone:' in tz]
if not timezoneline:
LOG.warning('Unable to determine timezone, using UTC')
return 'UTC'
# The line returned is "[whitespace]Time zone: [timezone] ([tz], [offset])"
timezone = timezoneline[0].strip().split(' ')[2]
except IndexError:
LOG.error('Unable to parse timezone from timedatectl, using UTC')
timezone = 'UTC'
return timezone
def check_file_for_enabled_service(env_file):
"""Checks environment file for the said service.
If stack to be to be deployed/updated/upgraded has any deprecated service
enabled, throw a warning about its deprecation and ask the user
whether to proceed with deployment despite deprecation.
For ODL as an example:
:param env_file: The path of the environment file
:type env_file: String
:raises CommandError: If the action is not confirmed
if os.path.exists(env_file):
with open(env_file, "r") as f:
content = yaml.safe_load(f)
deprecated_services_enabled = []
for service in constants.DEPRECATED_SERVICES.keys():
if content["resource_registry"][service] != "OS::Heat::None":
LOG.warning("service " + service + " is enabled in "
+ str(env_file) + ". " +
except (KeyError, TypeError):
# ignore if content["resource_registry"] is empty
if deprecated_services_enabled:
confirm = prompt_user_for_confirmation(
message="Do you still wish to continue with deployment [y/N]",
if not confirm:
raise oscexc.CommandError("Action not confirmed, exiting.")
def check_deprecated_service_is_enabled(environment_files):
for env_file in environment_files:
def reset_cmdline():
"""Run reset to cleanup cmdline"""
# only try to reset if stdout is a terminal, skip if not (e.g. CI)
if not sys.stdout.isatty():
output = ''
output = run_command(['reset', '-I'])
except RuntimeError:
LOG.warning('Unable to reset command line. Try manually running '