a0983a4ed7
This reverts commit 3f7e74ab24.
After identifying how to workaround the performance issues on the
undercloud, let's put this back in. Enabling innodb_file_per_table is
important for operators to be able to better manage their databases.
Change-Id: I435de381a0f0e3ef221e498f442335cdce3fb818
Depends-On: I77507c638237072e38d9888aff3da884aeff0b59
Closes-Bug: #1660722
248 lines
9.8 KiB
Puppet
248 lines
9.8 KiB
Puppet
# Copyright 2016 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
|
|
#
|
|
# http://www.apache.org/licenses/LICENSE-2.0
|
|
#
|
|
# 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.
|
|
#
|
|
# == Class: tripleo::profile::pacemaker::database::mysql
|
|
#
|
|
# MySQL with Pacemaker profile for tripleo
|
|
#
|
|
# === Parameters
|
|
#
|
|
# [*bootstrap_node*]
|
|
# (Optional) The hostname of the node responsible for bootstrapping tasks
|
|
# Defaults to hiera('mysql_short_bootstrap_node_name')
|
|
#
|
|
# [*bind_address*]
|
|
# (Optional) The address that the local mysql instance should bind to.
|
|
# Defaults to $::hostname
|
|
#
|
|
# [*gmcast_listen_addr*]
|
|
# (Optional) This variable defines the address on which the node listens to
|
|
# connections from other nodes in the cluster.
|
|
# Defaults to hiera('mysql_bind_host')
|
|
#
|
|
# [*step*]
|
|
# (Optional) The current step in deployment. See tripleo-heat-templates
|
|
# for more details.
|
|
# Defaults to hiera('step')
|
|
#
|
|
# [*pcs_tries*]
|
|
# (Optional) The number of times pcs commands should be retried.
|
|
# Defaults to hiera('pcs_tries', 20)
|
|
#
|
|
class tripleo::profile::pacemaker::database::mysql (
|
|
$bootstrap_node = hiera('mysql_short_bootstrap_node_name'),
|
|
$bind_address = $::hostname,
|
|
$gmcast_listen_addr = hiera('mysql_bind_host'),
|
|
$step = hiera('step'),
|
|
$pcs_tries = hiera('pcs_tries', 20),
|
|
) {
|
|
if $::hostname == downcase($bootstrap_node) {
|
|
$pacemaker_master = true
|
|
} else {
|
|
$pacemaker_master = false
|
|
}
|
|
|
|
# use only mysql_node_names when we land a patch in t-h-t that
|
|
# switches to autogenerating these values from composable services
|
|
# The galera node names need to match the pacemaker node names... so if we
|
|
# want to use FQDNs for this, the cluster will not finish bootstrapping,
|
|
# since all the nodes will be marked as slaves. For now, we'll stick to the
|
|
# short name which is already registered in pacemaker until we get around
|
|
# this issue.
|
|
$galera_node_names_lookup = hiera('mysql_short_node_names', hiera('mysql_node_names', $::hostname))
|
|
if is_array($galera_node_names_lookup) {
|
|
$galera_nodes = downcase(join($galera_node_names_lookup, ','))
|
|
} else {
|
|
$galera_nodes = downcase($galera_node_names_lookup)
|
|
}
|
|
$galera_nodes_count = count(split($galera_nodes, ','))
|
|
|
|
$mysqld_options = {
|
|
'mysqld' => {
|
|
'skip-name-resolve' => '1',
|
|
'binlog_format' => 'ROW',
|
|
'default-storage-engine' => 'innodb',
|
|
'innodb_autoinc_lock_mode' => '2',
|
|
'innodb_locks_unsafe_for_binlog'=> '1',
|
|
'innodb_file_per_table' => 'ON',
|
|
'query_cache_size' => '0',
|
|
'query_cache_type' => '0',
|
|
'bind-address' => $bind_address,
|
|
'max_connections' => hiera('mysql_max_connections'),
|
|
'open_files_limit' => '-1',
|
|
'wsrep_on' => 'ON',
|
|
'wsrep_provider' => '/usr/lib64/galera/libgalera_smm.so',
|
|
'wsrep_cluster_name' => 'galera_cluster',
|
|
'wsrep_cluster_address' => "gcomm://${galera_nodes}",
|
|
'wsrep_slave_threads' => '1',
|
|
'wsrep_certify_nonPK' => '1',
|
|
'wsrep_max_ws_rows' => '131072',
|
|
'wsrep_max_ws_size' => '1073741824',
|
|
'wsrep_debug' => '0',
|
|
'wsrep_convert_LOCK_to_trx' => '0',
|
|
'wsrep_retry_autocommit' => '1',
|
|
'wsrep_auto_increment_control' => '1',
|
|
'wsrep_drupal_282555_workaround'=> '0',
|
|
'wsrep_causal_reads' => '0',
|
|
'wsrep_sst_method' => 'rsync',
|
|
'wsrep_provider_options' => "gmcast.listen_addr=tcp://${gmcast_listen_addr}:4567;",
|
|
}
|
|
}
|
|
|
|
# remove_default_accounts parameter will execute some mysql commands
|
|
# to remove the default accounts created by MySQL package.
|
|
# We need MySQL running to run the commands successfully, so better to
|
|
# wait step 2 before trying to run the commands.
|
|
if $step >= 2 and $pacemaker_master {
|
|
$remove_default_accounts = true
|
|
} else {
|
|
$remove_default_accounts = false
|
|
}
|
|
|
|
class { '::tripleo::profile::base::database::mysql':
|
|
bootstrap_node => $bootstrap_node,
|
|
manage_resources => false,
|
|
remove_default_accounts => $remove_default_accounts,
|
|
mysql_server_options => $mysqld_options,
|
|
}
|
|
|
|
if $step >= 1 and $pacemaker_master and hiera('stack_action') == 'UPDATE' {
|
|
tripleo::pacemaker::resource_restart_flag { 'galera-master':
|
|
subscribe => File['mysql-config-file'],
|
|
}
|
|
}
|
|
|
|
if $step >= 2 {
|
|
pacemaker::property { 'galera-role-node-property':
|
|
property => 'galera-role',
|
|
value => true,
|
|
tries => $pcs_tries,
|
|
node => $::hostname,
|
|
}
|
|
if $pacemaker_master {
|
|
pacemaker::resource::ocf { 'galera' :
|
|
ocf_agent_name => 'heartbeat:galera',
|
|
op_params => 'promote timeout=300s on-fail=block',
|
|
master_params => '',
|
|
meta_params => "master-max=${galera_nodes_count} ordered=true",
|
|
resource_params => "additional_parameters='--open-files-limit=16384' enable_creation=true wsrep_cluster_address='gcomm://${galera_nodes}'",
|
|
tries => $pcs_tries,
|
|
location_rule => {
|
|
resource_discovery => 'exclusive',
|
|
score => 0,
|
|
expression => ['galera-role eq true'],
|
|
},
|
|
require => [Class['::mysql::server'],
|
|
Pacemaker::Property['galera-role-node-property']],
|
|
before => Exec['galera-ready'],
|
|
}
|
|
exec { 'galera-ready' :
|
|
command => '/usr/bin/clustercheck >/dev/null',
|
|
timeout => 30,
|
|
tries => 180,
|
|
try_sleep => 10,
|
|
environment => ['AVAILABLE_WHEN_READONLY=0'],
|
|
require => Exec['create-root-sysconfig-clustercheck'],
|
|
}
|
|
# We add a clustercheck db user and we will switch /etc/sysconfig/clustercheck
|
|
# to it in a later step. We do this only on one node as it will replicate on
|
|
# the other members. We also make sure that the permissions are the minimum necessary
|
|
mysql_user { 'clustercheck@localhost':
|
|
ensure => 'present',
|
|
password_hash => mysql_password(hiera('mysql_clustercheck_password')),
|
|
require => Exec['galera-ready'],
|
|
}
|
|
mysql_grant { 'clustercheck@localhost/*.*':
|
|
ensure => 'present',
|
|
options => ['GRANT'],
|
|
privileges => ['PROCESS'],
|
|
table => '*.*',
|
|
user => 'clustercheck@localhost',
|
|
}
|
|
|
|
# We create databases and users for services at step 2 as well. This ensures
|
|
# Galara is up before those get created
|
|
Exec['galera-ready'] -> Mysql_database<||>
|
|
Exec['galera-ready'] -> Mysql_user<||>
|
|
|
|
}
|
|
# This step is to create a sysconfig clustercheck file with the root user and empty password
|
|
# on the first install only (because later on the clustercheck db user will be used)
|
|
# We are using exec and not file in order to not have duplicate definition errors in puppet
|
|
# when we later set the the file to contain the clustercheck data
|
|
exec { 'create-root-sysconfig-clustercheck':
|
|
command => "/bin/echo 'MYSQL_USERNAME=root\nMYSQL_PASSWORD=\'\'\nMYSQL_HOST=localhost\n' > /etc/sysconfig/clustercheck",
|
|
unless => '/bin/test -e /etc/sysconfig/clustercheck && grep -q clustercheck /etc/sysconfig/clustercheck',
|
|
}
|
|
xinetd::service { 'galera-monitor' :
|
|
port => '9200',
|
|
server => '/usr/bin/clustercheck',
|
|
per_source => 'UNLIMITED',
|
|
log_on_success => '',
|
|
log_on_failure => 'HOST',
|
|
flags => 'REUSE',
|
|
service_type => 'UNLISTED',
|
|
user => 'root',
|
|
group => 'root',
|
|
require => Exec['create-root-sysconfig-clustercheck'],
|
|
}
|
|
}
|
|
|
|
if $step >= 4 or ( $step >= 3 and $pacemaker_master ) {
|
|
# At this stage we are guaranteed that the clustercheck db user exists
|
|
# so we switch the resource agent to use it.
|
|
$mysql_clustercheck_password = hiera('mysql_clustercheck_password')
|
|
file { '/etc/sysconfig/clustercheck' :
|
|
ensure => file,
|
|
mode => '0600',
|
|
owner => 'root',
|
|
group => 'root',
|
|
content => "MYSQL_USERNAME=clustercheck\n
|
|
MYSQL_PASSWORD='${mysql_clustercheck_password}'\n
|
|
MYSQL_HOST=localhost\n",
|
|
}
|
|
}
|
|
|
|
if $step >= 5 {
|
|
# We now make sure that the root db password is set to a random one
|
|
# At first installation /root/.my.cnf will be empty and we connect without a root
|
|
# password. On second runs or updates /root/.my.cnf will already be populated
|
|
# with proper credentials. This step happens on every node because this sql
|
|
# statement does not automatically replicate across nodes.
|
|
$mysql_root_password = hiera('mysql::server::root_password')
|
|
$galera_set_pwd = "/bin/touch /root/.my.cnf && \
|
|
/bin/echo \"UPDATE mysql.user SET Password = PASSWORD('${mysql_root_password}') WHERE user = 'root'; \
|
|
flush privileges;\" | \
|
|
/bin/mysql --defaults-extra-file=/root/.my.cnf -u root"
|
|
exec { 'galera-set-root-password':
|
|
command => $galera_set_pwd,
|
|
}
|
|
file { '/root/.my.cnf' :
|
|
ensure => file,
|
|
mode => '0600',
|
|
owner => 'root',
|
|
group => 'root',
|
|
content => "[client]
|
|
user=root
|
|
password=\"${mysql_root_password}\"
|
|
|
|
[mysql]
|
|
user=root
|
|
password=\"${mysql_root_password}\"",
|
|
require => Exec['galera-set-root-password'],
|
|
}
|
|
}
|
|
|
|
}
|