charm-percona-cluster/hooks/percona_utils.py

1017 lines
33 KiB
Python

''' General utilities for percona '''
import subprocess
from subprocess import Popen, PIPE
import socket
import tempfile
import os
import shutil
import uuid
from functools import partial
from charmhelpers.core.decorators import retry_on_exception
from charmhelpers.core.host import (
lsb_release,
mkdir,
service,
pwgen,
CompareHostReleases,
)
from charmhelpers.core.hookenv import (
charm_dir,
unit_get,
relation_ids,
related_units,
relation_get,
relation_set,
local_unit,
service_name,
config,
log,
DEBUG,
INFO,
WARNING,
ERROR,
cached,
status_set,
network_get_primary_address,
application_version_set,
is_leader,
leader_get,
leader_set,
)
from charmhelpers.core.unitdata import kv
from charmhelpers.fetch import (
apt_install,
filter_installed_packages,
get_upstream_version,
)
from charmhelpers.contrib.network.ip import (
get_address_in_network,
get_ipv6_addr,
is_ip,
is_ipv6,
)
from charmhelpers.contrib.database.mysql import (
MySQLHelper,
)
from charmhelpers.contrib.hahelpers.cluster import (
is_clustered,
distributed_wait,
)
from charmhelpers.contrib.openstack.utils import (
make_assess_status_func,
pause_unit,
resume_unit,
is_unit_paused_set,
)
# NOTE: python-mysqldb is installed by charmhelpers.contrib.database.mysql so
# hence why we import here
from MySQLdb import (
OperationalError
)
KEY = "keys/repo.percona.com"
REPO = """deb http://repo.percona.com/apt {release} main
deb-src http://repo.percona.com/apt {release} main"""
SEEDED_MARKER = "{data_dir}/seeded"
HOSTS_FILE = '/etc/hosts'
DEFAULT_MYSQL_PORT = 3306
INITIAL_CLUSTERED_KEY = 'initial-cluster-complete'
# NOTE(ajkavanagh) - this is 'required' for the pause/resume code for
# maintenance mode, but is currently not populated as the
# charm_check_function() checks whether the unit is working properly.
REQUIRED_INTERFACES = {}
class LeaderNoBootstrapUUIDError(Exception):
"""Raised when the leader doesn't have set the bootstrap-uuid attribute"""
def __init__(self):
super(LeaderNoBootstrapUUIDError, self).__init__(
"the leader doesn't have set the bootstrap-uuid attribute")
class InconsistentUUIDError(Exception):
"""Raised when the leader and the unit have different UUIDs set"""
def __init__(self, leader_uuid, unit_uuid):
super(InconsistentUUIDError, self).__init__(
"Leader UUID ('%s') != Unit UUID ('%s')" % (leader_uuid,
unit_uuid))
class DesyncedException(Exception):
'''Raised if PXC unit is not in sync with its peers'''
pass
class FakeOSConfigRenderer(object):
"""This class is to provide to register_configs() as a 'fake'
OSConfigRenderer object that has a complete_contexts method that returns
an empty list. This is so that the pause/resume framework can be used
from charmhelpers that requires configs to be able to run.
This is a bit of a hack, but via Python's duck-typing enables the function
to work.
"""
def complete_contexts(self):
return []
def determine_packages():
if CompareHostReleases(lsb_release()['DISTRIB_CODENAME']) >= 'wily':
# NOTE(beisner): Use recommended mysql-client package
# https://launchpad.net/bugs/1476845
# https://launchpad.net/bugs/1571789
# NOTE(coreycb): This will install percona-xtradb-cluster-server-5.6
# for >= wily and percona-xtradb-cluster-server-5.7 for >= bionic.
return [
'percona-xtradb-cluster-server',
]
else:
return [
'percona-xtradb-cluster-server-5.5',
'percona-xtradb-cluster-client-5.5',
]
def seeded():
''' Check whether service unit is already seeded '''
return os.path.exists(SEEDED_MARKER.format(data_dir=resolve_data_dir()))
def mark_seeded():
''' Mark service unit as seeded '''
with open(SEEDED_MARKER.format(data_dir=resolve_data_dir()),
'w') as seeded:
seeded.write('done')
def setup_percona_repo():
''' Configure service unit to use percona repositories '''
with open('/etc/apt/sources.list.d/percona.list', 'w') as sources:
sources.write(REPO.format(release=lsb_release()['DISTRIB_CODENAME']))
subprocess.check_call(['apt-key', 'add', KEY])
def resolve_hostname_to_ip(hostname):
"""Resolve hostname to IP
@param hostname: hostname to be resolved
@returns IP address or None if resolution was not possible via DNS
"""
try:
import dns.resolver
except ImportError:
apt_install(filter_installed_packages(['python-dnspython']),
fatal=True)
import dns.resolver
if config('prefer-ipv6'):
if is_ipv6(hostname):
return hostname
query_type = 'AAAA'
elif is_ip(hostname):
return hostname
else:
query_type = 'A'
# This may throw an NXDOMAIN exception; in which case
# things are badly broken so just let it kill the hook
answers = dns.resolver.query(hostname, query_type)
if answers:
return answers[0].address
def is_sufficient_peers():
"""Sufficient number of expected peers to build a complete cluster
If min-cluster-size has been provided, check that we have sufficient
number of peers as expected for a complete cluster.
If not defined assume a single unit.
@returns boolean
"""
min_size = config('min-cluster-size')
if min_size:
log("Checking for minimum of {} peer units".format(min_size),
level=DEBUG)
# Include this unit
units = 1
for rid in relation_ids('cluster'):
units += len(related_units(rid))
if units < min_size:
log("Insufficient number of peer units to form cluster "
"(expected=%s, got=%s)" % (min_size, units), level=INFO)
return False
else:
log("Sufficient number of peer units to form cluster {}"
"".format(min_size, level=DEBUG))
return True
else:
log("min-cluster-size is not defined, race conditions may occur if "
"this is not a single unit deployment.", level=WARNING)
return True
def get_cluster_hosts():
"""Get the bootstrapped cluster peers
Determine the cluster peers that have bootstrapped and return the list
hosts. Secondarily, update the hosts file with IPv6 address name
resolution.
The returned host list is intended to be used in the
wsrep_cluster_address=gcomm:// setting. Therefore, the hosts must have
already been bootstrapped. If an un-bootstrapped host happens to be first
in the list, mysql will fail to start.
@side_effect update_hosts_file called for IPv6 hostname resolution
@returns list of hosts
"""
hosts_map = {}
local_cluster_address = get_cluster_host_ip()
# We need to add this localhost dns name to /etc/hosts along with peer
# hosts to ensure percona gets consistently resolved addresses.
if config('prefer-ipv6'):
addr = get_ipv6_addr(exc_list=[config('vip')], fatal=True)[0]
hosts_map = {addr: socket.gethostname()}
hosts = []
for relid in relation_ids('cluster'):
for unit in related_units(relid):
rdata = relation_get(unit=unit, rid=relid)
# NOTE(dosaboy): see LP: #1599447
cluster_address = rdata.get('cluster-address',
rdata.get('private-address'))
if config('prefer-ipv6'):
hostname = rdata.get('hostname')
if not hostname or hostname in hosts:
log("(unit=%s) Ignoring hostname '%s' provided by cluster "
"relation for addr %s" %
(unit, hostname, cluster_address), level=DEBUG)
continue
else:
log("(unit=%s) hostname '%s' provided by cluster relation "
"for addr %s" % (unit, hostname, cluster_address),
level=DEBUG)
hosts_map[cluster_address] = hostname
host = hostname
else:
host = resolve_hostname_to_ip(cluster_address)
# Add only cluster peers who have set bootstrap-uuid
# An indiction they themselves are bootstrapped.
# Un-bootstrapped hosts in gcom lead mysql to fail to start
# if it happens to be the first address in the list
# Also fix strange bug when executed from actions where the local
# unit is returned in related_units. We do not want the local IP
# in the gcom hosts list.
if (rdata.get('bootstrap-uuid') and
host not in hosts and
host != local_cluster_address):
hosts.append(host)
if hosts_map:
update_hosts_file(hosts_map)
# Return a sorted list to avoid uneccessary restarts
hosts.sort()
return hosts
SQL_SST_USER_SETUP = ("GRANT {permissions} ON *.* "
"TO 'sstuser'@'localhost' IDENTIFIED BY '{password}'")
SQL_SST_USER_SETUP_IPV6 = ("GRANT {permissions} "
"ON *.* TO 'sstuser'@'ip6-localhost' IDENTIFIED "
"BY '{password}'")
def get_db_helper():
return MySQLHelper(rpasswdf_template='/var/lib/charm/%s/mysql.passwd' %
(service_name()),
upasswdf_template='/var/lib/charm/%s/mysql-{}.passwd' %
(service_name()))
def configure_sstuser(sst_password):
# xtrabackup 2.4 (introduced in Bionic) needs PROCESS privilege for backups
permissions = [
"RELOAD",
"LOCK TABLES",
"REPLICATION CLIENT"
]
if CompareHostReleases(lsb_release()['DISTRIB_CODENAME']) >= 'bionic':
permissions.append('PROCESS')
m_helper = get_db_helper()
m_helper.connect(password=m_helper.get_mysql_root_password())
m_helper.execute(SQL_SST_USER_SETUP.format(
permissions=','.join(permissions),
password=sst_password)
)
m_helper.execute(SQL_SST_USER_SETUP_IPV6.format(
permissions=','.join(permissions),
password=sst_password)
)
# TODO: mysql charmhelper
def configure_mysql_root_password(password):
''' Configure debconf with root password '''
dconf = Popen(['debconf-set-selections'], stdin=PIPE)
# Set both percona and mysql password options to cover
# both upstream and distro packages.
packages = ["percona-server-server", "mysql-server",
"percona-xtradb-cluster-server"]
m_helper = get_db_helper()
root_pass = m_helper.get_mysql_root_password(password)
for package in packages:
dconf.stdin.write("%s %s/root_password password %s\n" %
(package, package, root_pass))
dconf.stdin.write("%s %s/root_password_again password %s\n" %
(package, package, root_pass))
dconf.communicate()
dconf.wait()
# TODO: Submit for charmhelper
def relation_clear(r_id=None):
''' Clears any relation data already set on relation r_id '''
settings = relation_get(rid=r_id,
unit=local_unit())
for setting in settings:
if setting not in ['public-address', 'private-address']:
settings[setting] = None
relation_set(relation_id=r_id,
**settings)
def update_hosts_file(map):
"""Percona does not currently like ipv6 addresses so we need to use dns
names instead. In order to make them resolvable we ensure they are in
/etc/hosts.
See https://bugs.launchpad.net/galera/+bug/1130595 for some more info.
"""
with open(HOSTS_FILE, 'r') as hosts:
lines = hosts.readlines()
log("Updating %s with: %s (current: %s)" % (HOSTS_FILE, map, lines),
level=DEBUG)
newlines = []
for ip, hostname in map.items():
if not ip or not hostname:
continue
keepers = []
for line in lines:
_line = line.split()
if len(line) < 2 or not (_line[0] == ip or hostname in _line[1:]):
keepers.append(line)
else:
log("Marking line '%s' for update or removal" % (line.strip()),
level=DEBUG)
lines = keepers
newlines.append("%s %s\n" % (ip, hostname))
lines += newlines
with tempfile.NamedTemporaryFile(delete=False) as tmpfile:
with open(tmpfile.name, 'w') as hosts:
for line in lines:
hosts.write(line)
os.rename(tmpfile.name, HOSTS_FILE)
os.chmod(HOSTS_FILE, 0o644)
def assert_charm_supports_ipv6():
"""Check whether we are able to support charms ipv6."""
_release = lsb_release()['DISTRIB_CODENAME'].lower()
if CompareHostReleases(_release) < "trusty":
raise Exception("IPv6 is not supported in the charms for Ubuntu "
"versions less than Trusty 14.04")
def unit_sorted(units):
"""Return a sorted list of unit names."""
return sorted(
units, lambda a, b: cmp(int(a.split('/')[-1]), int(b.split('/')[-1])))
def install_mysql_ocf():
dest_dir = '/usr/lib/ocf/resource.d/percona/'
for fname in ['ocf/percona/mysql_monitor']:
src_file = os.path.join(charm_dir(), fname)
if not os.path.isdir(dest_dir):
os.makedirs(dest_dir)
dest_file = os.path.join(dest_dir, os.path.basename(src_file))
if not os.path.exists(dest_file):
log('Installing %s' % dest_file, level='INFO')
shutil.copy(src_file, dest_file)
else:
log("'%s' already exists, skipping" % dest_file, level='INFO')
def get_wsrep_value(key):
m_helper = get_db_helper()
try:
m_helper.connect(password=m_helper.get_mysql_root_password())
except OperationalError:
log("Could not connect to db", DEBUG)
return None
cursor = m_helper.connection.cursor()
ret = None
try:
cursor.execute("show status like '%s'" % (key))
ret = cursor.fetchall()
except:
log("Failed to get '%s'", ERROR)
return None
finally:
cursor.close()
if ret:
return ret[0][1]
return None
def is_leader_bootstrapped():
""" Check that the leader is bootstrapped and has set required settings
:side_effect: calls leader_get
:returns: boolean
"""
check_settings = ['bootstrap-uuid', 'mysql.passwd', 'root-password',
'sst-password', 'leader-ip']
leader_settings = leader_get()
# Is the leader bootstrapped?
for setting in check_settings:
if leader_settings.get(setting) is None:
log("Leader is NOT bootstrapped {}: {}".format(setting,
leader_settings.get('bootstrap-uuid')), DEBUG)
return False
log("Leader is bootstrapped uuid: {}".format(
leader_settings.get('bootstrap-uuid')), DEBUG)
return True
def clustered_once():
"""Determine if the cluster has ever bootstrapped completely
Check unittest.kv if the cluster has bootstrapped at least once.
@returns boolean
"""
# Run is_bootstrapped once to guarantee kvstore is up to date
is_bootstrapped()
kvstore = kv()
return kvstore.get(INITIAL_CLUSTERED_KEY, False)
def is_bootstrapped():
"""Determine if each node in the cluster has been bootstrapped and the
cluster is complete with the expected number of peers.
Check that each node in the cluster, including this one, has set
bootstrap-uuid on the cluster relation.
Having min-cluster-size set will guarantee is_bootstrapped will not
return True until the expected number of peers are bootstrapped. If
min-cluster-size is not set, it will check peer relations to estimate the
expected cluster size. If min-cluster-size is not set and there are no
peers it must assume the cluster is bootstrapped in order to allow for
single unit deployments.
@returns boolean
"""
min_size = config('min-cluster-size')
if not min_size:
units = 1
for relation_id in relation_ids('cluster'):
units += len(related_units(relation_id))
min_size = units
if not is_sufficient_peers():
return False
elif min_size > 1:
uuids = []
for relation_id in relation_ids('cluster'):
units = related_units(relation_id) or []
units.append(local_unit())
for unit in units:
if not relation_get(attribute='bootstrap-uuid',
rid=relation_id,
unit=unit):
log("{} is not yet clustered".format(unit),
DEBUG)
return False
else:
bootstrap_uuid = relation_get(attribute='bootstrap-uuid',
rid=relation_id,
unit=unit)
if bootstrap_uuid:
uuids.append(bootstrap_uuid)
if len(uuids) < min_size:
log("Fewer than minimum cluster size: "
"{} percona units reporting clustered".format(min_size),
DEBUG)
return False
elif len(set(uuids)) > 1:
raise Exception("Found inconsistent bootstrap uuids: "
"{}".format((uuids)))
else:
log("All {} percona units reporting clustered".format(min_size),
DEBUG)
# Set INITIAL_CLUSTERED_KEY as the cluster has fully bootstrapped
kvstore = kv()
if not kvstore.get(INITIAL_CLUSTERED_KEY, False):
kvstore.set(key=INITIAL_CLUSTERED_KEY, value=True)
kvstore.flush()
return True
def bootstrap_pxc():
"""Bootstrap PXC
On systemd systems systemctl bootstrap-pxc mysql does not work.
Run service mysql bootstrap-pxc to bootstrap."""
service('stop', 'mysql')
bootstrapped = service('bootstrap-pxc', 'mysql')
if not bootstrapped:
try:
cmp_os = CompareHostReleases(
lsb_release()['DISTRIB_CODENAME']
)
if cmp_os < 'bionic':
# NOTE(jamespage): execute under systemd-run to ensure
# that the bootstrap-pxc mysqld does
# not end up in the juju unit daemons
# cgroup scope.
cmd = ['systemd-run', '--service-type=forking',
'service', 'mysql', 'bootstrap-pxc']
subprocess.check_call(cmd)
else:
service('start', 'mysql@bootstrap')
except subprocess.CalledProcessError as e:
msg = 'Bootstrap PXC failed'
error_msg = '{}: {}'.format(msg, e)
status_set('blocked', msg)
log(error_msg, ERROR)
raise Exception(error_msg)
if CompareHostReleases(lsb_release()['DISTRIB_CODENAME']) < 'bionic':
# To make systemd aware mysql is running after a bootstrap
service('start', 'mysql')
log("Bootstrap PXC Succeeded", DEBUG)
def notify_bootstrapped(cluster_rid=None, cluster_uuid=None):
if cluster_rid:
rids = [cluster_rid]
else:
rids = relation_ids('cluster')
if not rids:
log("No relation ids found for 'cluster'", level=INFO)
return
if not cluster_uuid:
cluster_uuid = get_wsrep_value('wsrep_cluster_state_uuid')
if not cluster_uuid:
cluster_uuid = str(uuid.uuid4())
log("Could not determine cluster uuid so using '%s' instead" %
(cluster_uuid), INFO)
log("Notifying peers that percona is bootstrapped (uuid=%s)" %
(cluster_uuid), DEBUG)
for rid in rids:
relation_set(relation_id=rid, **{'bootstrap-uuid': cluster_uuid})
if is_leader():
leader_set(**{'bootstrap-uuid': cluster_uuid})
def update_bootstrap_uuid():
"""This function verifies if the leader has set the bootstrap-uuid
attribute to then check it against the running cluster uuid, if the check
succeeds the bootstrap-uuid field is set in the cluster relation.
:returns: True if the cluster UUID was updated, False if the local UUID is
empty.
"""
lead_cluster_state_uuid = leader_get('bootstrap-uuid')
if not lead_cluster_state_uuid:
log('Leader has not set bootstrap-uuid', level=DEBUG)
raise LeaderNoBootstrapUUIDError()
wsrep_ready = get_wsrep_value('wsrep_ready') or ""
log("wsrep_ready: '%s'" % wsrep_ready, DEBUG)
if wsrep_ready.lower() in ['on', 'ready']:
cluster_state_uuid = get_wsrep_value('wsrep_cluster_state_uuid')
else:
cluster_state_uuid = None
if not cluster_state_uuid:
log("UUID is empty: '%s'" % cluster_state_uuid, level=DEBUG)
return False
elif lead_cluster_state_uuid != cluster_state_uuid:
# this may mean 2 things:
# 1) the units have diverged, which it's bad and we do stop.
# 2) cluster_state_uuid could not be retrieved because it
# hasn't been bootstrapped, mysqld is stopped, etc.
log('bootstrap uuid differs: %s != %s' % (lead_cluster_state_uuid,
cluster_state_uuid),
level=ERROR)
raise InconsistentUUIDError(lead_cluster_state_uuid,
cluster_state_uuid)
for rid in relation_ids('cluster'):
notify_bootstrapped(cluster_rid=rid,
cluster_uuid=cluster_state_uuid)
return True
def cluster_in_sync():
'''
Determines whether the current unit is in sync
with the rest of the cluster
'''
ready = get_wsrep_value('wsrep_ready') or False
sync_status = get_wsrep_value('wsrep_local_state') or 0
if ready and int(sync_status) in [2, 4]:
return True
return False
def charm_check_func():
"""Custom function to assess the status of the current unit
@returns (status, message) - tuple of strings if an issue
"""
@retry_on_exception(num_retries=10,
base_delay=2,
exc_type=DesyncedException)
def _cluster_in_sync():
'''Helper func to wait for a while for resync to occur
@raise DesynedException: raised if local unit is not in sync
with its peers
'''
if not cluster_in_sync():
raise DesyncedException()
min_size = config('min-cluster-size')
# Ensure that number of peers > cluster size configuration
if not is_sufficient_peers():
return ('blocked', 'Insufficient peers to bootstrap cluster')
if min_size and int(min_size) > 1:
# Once running, ensure that cluster is in sync
# and has the required peers
if not is_bootstrapped():
return ('waiting', 'Unit waiting for cluster bootstrap')
elif cluster_ready():
try:
_cluster_in_sync()
return ('active', 'Unit is ready and clustered')
except DesyncedException:
return ('blocked', 'Unit is not in sync')
else:
return ('waiting', 'Unit waiting on hacluster relation')
else:
return ('active', 'Unit is ready')
@cached
def resolve_data_dir():
_release = lsb_release()['DISTRIB_CODENAME'].lower()
if CompareHostReleases(_release) < 'vivid':
return '/var/lib/mysql'
else:
return '/var/lib/percona-xtradb-cluster'
@cached
def resolve_cnf_file():
_release = lsb_release()['DISTRIB_CODENAME'].lower()
if CompareHostReleases(_release) < 'vivid':
return '/etc/mysql/my.cnf'
else:
return '/etc/mysql/percona-xtradb-cluster.conf.d/mysqld.cnf'
def register_configs():
"""Return a OSConfigRenderer object.
However, ceph-mon wasn't written using OSConfigRenderer objects to do the
config files, so this just returns an empty OSConfigRenderer object.
@returns empty FakeOSConfigRenderer object.
"""
return FakeOSConfigRenderer()
def services():
"""Return a list of services that are managed by this charm.
@returns [services] - list of strings that are service names.
"""
# NOTE(jamespage): Native systemd variants of the packagin
# use mysql@bootstrap to seed the cluster
# however this is cleared after a reboot,
# so dynamically check to see if this active
if service('is-active', 'mysql@bootstrap'):
return ['mysql@bootstrap']
return ['mysql']
def assess_status(configs):
"""Assess status of current unit
Decides what the state of the unit should be based on the current
configuration.
SIDE EFFECT: calls set_os_workload_status(...) which sets the workload
status of the unit.
Also calls status_set(...) directly if paused state isn't complete.
@param configs: a templating.OSConfigRenderer() object
@returns None - this function is executed for its side-effect
"""
assess_status_func(configs)()
if pxc_installed():
application_version_set(get_upstream_version(determine_packages()[0]))
def assess_status_func(configs):
"""Helper function to create the function that will assess_status() for
the unit.
Uses charmhelpers.contrib.openstack.utils.make_assess_status_func() to
create the appropriate status function and then returns it.
Used directly by assess_status() and also for pausing and resuming
the unit.
NOTE(ajkavanagh) ports are not checked due to race hazards with services
that don't behave sychronously w.r.t their service scripts. e.g.
apache2.
@param configs: a templating.OSConfigRenderer() object
@return f() -> None : a function that assesses the unit's workload status
"""
return make_assess_status_func(
configs, REQUIRED_INTERFACES,
charm_func=lambda _: charm_check_func(),
services=services(), ports=None)
def pause_unit_helper(configs):
"""Helper function to pause a unit, and then call assess_status(...) in
effect, so that the status is correctly updated.
Uses charmhelpers.contrib.openstack.utils.pause_unit() to do the work.
@param configs: a templating.OSConfigRenderer() object
@returns None - this function is executed for its side-effect
"""
_pause_resume_helper(pause_unit, configs)
def resume_unit_helper(configs):
"""Helper function to resume a unit, and then call assess_status(...) in
effect, so that the status is correctly updated.
Uses charmhelpers.contrib.openstack.utils.resume_unit() to do the work.
@param configs: a templating.OSConfigRenderer() object
@returns None - this function is executed for its side-effect
"""
_pause_resume_helper(resume_unit, configs)
def _pause_resume_helper(f, configs):
"""Helper function that uses the make_assess_status_func(...) from
charmhelpers.contrib.openstack.utils to create an assess_status(...)
function that can be used with the pause/resume of the unit
@param f: the function to be used with the assess_status(...) function
@returns None - this function is executed for its side-effect
"""
# TODO(ajkavanagh) - ports= has been left off because of the race hazard
# that exists due to service_start()
f(assess_status_func(configs),
services=services(),
ports=None)
def create_binlogs_directory():
if not pxc_installed():
log("PXC not yet installed. Not setting up binlogs", DEBUG)
return
binlogs_directory = os.path.dirname(config('binlogs-path'))
data_dir = resolve_data_dir() + '/'
if binlogs_directory.startswith(data_dir):
raise Exception("Configured binlogs directory (%s) must not be inside "
"mysql data dir" % (binlogs_directory))
if not os.path.isdir(binlogs_directory):
mkdir(binlogs_directory, 'mysql', 'mysql', 0o750)
def get_cluster_host_ip():
"""Get the this host's IP address for use with percona cluster peers
@returns IP to pass to cluster peers
"""
cluster_network = config('cluster-network')
if cluster_network:
cluster_addr = get_address_in_network(cluster_network, fatal=True)
else:
try:
cluster_addr = network_get_primary_address('cluster')
except NotImplementedError:
# NOTE(jamespage): fallback to previous behaviour
cluster_addr = resolve_hostname_to_ip(
unit_get('private-address')
)
return cluster_addr
def cluster_ready():
"""Determine if each node in the cluster is ready to respond to client
requests.
Once cluster_ready returns True it is safe to execute client relation
hooks.
If a VIP is set do not return ready until hacluster relationship is
complete.
@returns boolean
"""
if config("vip") and not is_clustered():
log("Waiting on hacluster to complete clustering, not clustered yet.",
DEBUG)
return False
return is_bootstrapped()
def client_node_is_ready():
"""Determine if the leader node has set client data
@returns boolean
"""
# Bail if this unit is paused
if is_unit_paused_set():
return False
if not cluster_ready():
return False
for rid in relation_ids('shared-db'):
if leader_get(attribute='{}_password'.format(rid)):
return True
for rid in relation_ids('db-admin'):
if leader_get(attribute='{}_password'.format(rid)):
return True
for rid in relation_ids('db'):
if leader_get(attribute='{}_password'.format(rid)):
return True
return False
def leader_node_is_ready():
"""Determine if the leader node is ready to handle client relationship
hooks.
IFF percona is not paused, is installed, this is the leader node and the
cluster is complete.
@returns boolean
"""
# Paused check must run before other checks
# Bail if this unit is paused
if is_unit_paused_set():
return False
return (is_leader() and cluster_ready())
def _get_password(key):
'''Retrieve named password
This function will ensure that a consistent named password
is used across all units in the pxc cluster; the lead unit
will generate or use the root-password configuration option
to seed this value into the deployment.
Once set, it cannot be changed.
@requires: str: named password or None if unable to retrieve
at this point in time
'''
_password = leader_get(key)
if not _password and is_leader():
_password = config(key) or pwgen()
leader_set({key: _password})
return _password
root_password = partial(_get_password, 'root-password')
sst_password = partial(_get_password, 'sst-password')
def pxc_installed():
'''Determine whether percona-xtradb-cluster is installed
@returns: boolean: indicating installation
'''
return os.path.exists('/usr/sbin/mysqld')
def update_root_password():
"""Update root password if needed
:returns: `False` when configured password has not changed
"""
cfg = config()
if not cfg.changed('root-password'):
return False
m_helper = get_db_helper()
# password that needs to be set
new_root_passwd = cfg['root-password'] or root_password()
m_helper.set_mysql_root_password(new_root_passwd)
# check the password was changed
try:
m_helper.connect(user='root', password=new_root_passwd)
m_helper.execute('select 1;')
except OperationalError as ex:
log("Error connecting using new password: %s" % str(ex), level=DEBUG)
log(('Cannot connect using new password, not updating password in '
'the relation'), level=WARNING)
return
def cluster_wait():
''' Wait for operations based on modulo distribution
Use the distributed_wait function to determine how long to wait before
running an operation like restart or cluster join. By setting modulo to
the exact number of nodes in the cluster we get serial operations.
Check for explicit configuration parameters for modulo distribution.
The config setting modulo-nodes has first priority. If modulo-nodes is not
set, check min-cluster-size. Finally, if neither value is set, determine
how many peers there are from the cluster relation.
@side_effect: distributed_wait is called which calls time.sleep()
@return: None
'''
wait = config('known-wait')
if config('modulo-nodes') is not None:
# modulo-nodes has first priority
num_nodes = config('modulo-nodes')
elif config('min-cluster-size'):
# min-cluster-size is consulted next
num_nodes = config('min-cluster-size')
else:
# If nothing explicit is configured, determine cluster size based on
# peer relations
num_nodes = 1
for rid in relation_ids('cluster'):
num_nodes += len(related_units(rid))
distributed_wait(modulo=num_nodes, wait=wait)
def get_wsrep_provider_options():
wsrep_provider_options = []
if config('prefer-ipv6'):
wsrep_provider_options.append('gmcast.listen_addr=tcp://:::4567')
peer_timeout = config('peer-timeout')
if peer_timeout and(not peer_timeout.startswith('PT') or
not peer_timeout.endswith('S')):
raise ValueError("Invalid gcast.peer_timeout value: {}"
.format(peer_timeout))
elif peer_timeout:
wsrep_provider_options.append('gmcast.peer_timeout={}'
.format(config('peer-timeout')))
return ';'.join(wsrep_provider_options)