Version: 23.10

2-node manual failover installation

Manual HA is now deprecated. Please do not perform new installations using this procedure.

Prerequisites

Understanding

Before applying this procedure, you should have a good knowledge of Linux OS, of Centreon, and of the Pacemaker-Corosync clustering tools in order to have a proper understanding of what is being done and to be able to correct any mistakes that might occur.

WARNING: Anyone following this procedure does so at their own risk. Under no circumstances shall Centreon be liable for any breakdown or data loss.

Installation of Centreon

The installation of a Centreon-HA cluster can only be done on the basis of a functional installation of Centreon. Before following this procedure, it is therefore essential to have applied this installation procedure through to the end, reserving about 5 GB of free space on the volume group that contains the MySQL data (mount point /var/lib/mysql by default).

The vgs command should return a display of the form below (especially the value under VFree):

  VG                    #PV #LV #SN Attr   VSize   VFree 
  centos_centreon-c1      1   5   0 wz--n- <31,00g <5,00g

WARNING: If this prerequisite is not fulfilled, it will not be possible to synchronize the databases as described in this document.

Definition of server names and IP addresses

In this procedure we will refer to parameters that vary from one installation to another (names and IP addresses of nodes, for example) via the following macros:

@CENTRAL_MASTER_IPADDR@: IP address of the main central server
@CENTRAL_MASTER_NAME@: name of the main central server
@CENTRAL_SLAVE_IPADDR@: IP address of the second central server
@CENTRAL_SLAVE_NAME@: name of the second central server
@DB_MASTER_IPADDR@: IP address of the main database server
@DB_MASTER_NAME@: name of the main database server
@DB_SLAVE_IPADDR@: IP address of the secondary database server
@DB_SLAVE_NAME@: name of the secondary database server
@MYSQL_REPL_USER@: MySQL replication account name (suggested: centreon-repl)
@MYSQL_REPL_PASSWD@: password for this account
@MYSQL_CENTREON_USER@: MySQL account name of Centreon (suggested: centreon)
@MYSQL_CENTREON_PASSWD@: password for this account
@CENTRAL_VIP_IPADDR@: virtual IP address of the cluster
@CENTRAL_VIP_IFNAME@: name of the interface that will carry the VIP
@CENTRAL_VIP_CIDR_NETMASK@: subnet mask expressed in the number of bits without the '/' (example: 24)
@CENTRAL_VIP_BROADCAST_IPADDR@: broadcast address
@DB_VIP_IPADDR@: virtual IP address of the cluster. In the case of a two-node HA, this is the same as the central one.
@DB_VIP_IFNAME@: name of the interface that will carry the VIP. In the case of a two-node HA, this is the same as the central one.
@DB_VIP_CIDR_NETMASK@: subnet mask expressed in number of bits without the '/' (example: 24). In the case of a two-node HA, this is the same as the central one.
@DB_VIP_BROADCAST_IPADDR@: broadcast address. In the case of a two-node HA, this is the same as the central one.

Centreon-broker configuration

Linking to the cbd service

In a standard Centreon installation, the cbd service drives two instances of centreon-broker-daemon:

central-broker-master: also known as "central broker" or "SQL broker", which redirects all I/O from Pollers to databases, to the RRD broker, etc.
central-rrd-master: the RRD broker, which receives its flow from the SQL broker, and whose only function is to write the RRD files used to display the graphs.

In a Centreon-HA cluster, the two broker processes will each be managed via a separate service that will be driven by the cluster:

central-broker-master as the cbd_central_broker resource, linked to the systemd cbd-sql service
central-rrd-master as the clone resource cbd_rrd, linked to the standard Centreon systemd cbd service.

For everything to work properly from this point on, you must now undo the link between central-broker-master and the cbd service ** by checking "no" for the parameter "Linked to cbd service "** in Configuration > Pollers > Broker configuration > central-broker-master in the General tab.

Dual RRD streams

Rather than setting up a real-time replication of RRD data files, the technical choice that was made to allow graphs to be displayed on any node as soon as it becomes master was to duplicate the output stream from central-broker-master to central-rrd-master. This is configured in the same menu as in the previous paragraph, but this time in the Output tab of Configuration > Collectors > Centreon Broker Configuration.

Modify the output "IPv4" by replacing "localhost" with @CENTRAL_MASTER_IPADDR@.

Output IPv4
Name	centreon-broker-master-rrd
Connection port	5670
Host to connect to	`@CENTRAL_MASTER_IPADDR@`

Add a new IPv4 output, similar to the first one and named for example "centreon-broker-slave-rrd", this time pointing to @CENTRAL_SLAVE_IPADDR@.

Output IPv4
Name	centreon-broker-slave-rrd
Connection port	5670
Host to connect to	`@CENTRAL_SLAVE_IPADDR@`

Export the configuration

Once the actions in the two previous paragraphs have been performed, the configuration must be exported (the first three boxes for the export of the "Central" Poller) for it to take effect.

These actions must be performed either on both nodes or only on @CENTRAL_MASTER_NAME@ and then the broker configuration files must be copied to @CENTRAL_SLAVE_NAME@.

rsync -a /etc/centreon-broker/*json @CENTRAL_SLAVE_IPADDR@:/etc/centreon-broker/

Modification of the `cbd` reload command

This may not be known to all Centreon users, but every time a reload of the Central Poller configuration is done via the interface, the broker service (cbd) is reloaded (not just centengine), hence the "Centreon Broker reload command" parameter in Configuration > Pollers > Central.

As explained above, the broker processes are split between two services: cbd for the RRD broker, cbd-sql for the central broker. In the context of a centreon-ha cluster, the service that must be reloaded during the configuration export is cbd-sql, not cbd. You must therefore apply the value service cbd-sql reload to the "Centreon Broker reload command" parameter.

Synchronize Centreon configuration files (PHP, Perl, Gorgone)

rsync -av /etc/centreon/* root@@CENTRAL_SLAVE_IPADDR@:/etc/centreon/

System preparation

Before getting to the actual cluster setup, a few preparatory steps are necessary at OS level.

Note: unless otherwise stated, each of the following steps must be performed on both central nodes.

Tuning the network configuration

To improve the reliability of the cluster, and since Centreon HA only runs on IP v4, it is recommended to apply the following tuning to all servers of the Centreon platform:

Alma / RHEL / Oracle Linux 8

cat >> /etc/sysctl.conf <<EOF
net.ipv6.conf.all.disable_ipv6 = 1
net.ipv6.conf.default.disable_ipv6 = 1
net.ipv4.tcp_retries2 = 3
net.ipv4.tcp_keepalive_time = 200
net.ipv4.tcp_keepalive_probes = 2
net.ipv4.tcp_keepalive_intvl = 2
EOF
systemctl restart NetworkManager

To avoid problems with the management of fixed and virtual IP addresses, it is necessary to declare the fixed IP address of each server in the network configuration.

To do this, execute the following commands on each node:

For the server @CENTRAL_MASTER_NAME@:

nmcli con mod @CENTRAL_VIP_IFNAME@ +ipv4.addresses "@CENTRAL_MASTER_IPADDR@/@CENTRAL_VIP_CIDR_NETMASK@"
nmcli con up @CENTRAL_VIP_IFNAME@

For the server @CENTRAL_SLAVE_NAME@:

nmcli con mod @CENTRAL_VIP_IFNAME@ +ipv4.addresses "@CENTRAL_SLAVE_IPADDR@/@CENTRAL_VIP_CIDR_NETMASK@"
nmcli con up @CENTRAL_VIP_IFNAME@

You can verify on each node if the IP address is declared with this command on the Central node, for example:

cat /etc/sysconfig/network-scripts/ifcfg-@CENTRAL_VIP_IFNAME@

The result is like this for @CENTRAL_MASTER_NAME@:

...
DEVICE=@CENTRAL_VIP_IFNAME@
ONBOOT=yes
IPADDR=@CENTRAL_MASTER_IPADDR@
PREFIX=23
...

Name resolution

In order to ensure that the cluster functions properly even if the name resolution service fails, the nodes must know each other through the /etc/hosts file. In the case of a two-node cluster, it is not necessary to declare the hosts @DB_MASTER_NAME@ and @DB_MASTER_SLAVE@, since they are the same as the central hosts.

cat >/etc/hosts <<"EOF"
127.0.0.1 localhost localhost.localdomain localhost4 localhost4.localdomain4
@CENTRAL_MASTER_IPADDR@ @CENTRAL_MASTER_NAME@
@CENTRAL_SLAVE_IPADDR@ @CENTRAL_SLAVE_NAME@
EOF

In the rest of this document, we will refer to the primary node as the primary node and the secondary node as the secondary node. This distinction is purely arbitrary; the roles can of course be exchanged once the installation is complete.

Installing the packages

Centreon offers the centreon-ha package, which provides all the scripts and dependencies needed to run a Centreon cluster.

On all :

Alma Linux 8
RHEL 8
Oracle Linux 8

dnf config-manager --enable ha
dnf install centreon-ha-web

subscription-manager repos --enable rhel-8-for-x86_64-highavailability-rpms
dnf install centreon-ha-web

dnf config-manager --enable ol8_addons
dnf install centreon-ha-web

SSH key exchange

To allow servers to exchange files and commands, it is necessary to set up the possibility to connect through SSH from one server to the other for the users:

mysql
centreon

There are two ways to exchange SSH keys:

Using sh-copy-id: using this command requires the ability to validate authentication with a password. However, it is not desirable to define a password for the service accounts we are talking about here. If this method is used anyway, we recommend that you remove the password after the exchange, using the commands passwd -d centreon and passwd -d mysql.
By manually copying the public key into ~/.ssh/authorized_keys. This method is to be preferred, but to work properly, only the owner of the file must be able to read it.

This is the second method, which will be proposed below.

Centreon account

This procedure is to be applied on the two central nodes. To begin, switch to the bash environment of centreon:

su - centreon

Then run these commands on the two central nodes:

ssh-keygen -t ed25519 -a 100
cat ~/.ssh/id_ed25519.pub | tee ~/.ssh/authorized_keys

After running these commands on both nodes, copy the contents of the file that was displayed under the cat command and paste it into the ~/.ssh/authorized_keys file on the other node and then apply the correct permissions to the file (still as centreon):

chmod 600 ~/.ssh/authorized_keys

The key exchange must then be validated by a first connection that will accept the signature of the SSH server (still as centreon):

ssh @CENTRAL_MASTER_NAME@
ssh @CENTRAL_SLAVE_NAME@

Then exit the centreon session using exit or Ctrl-D.

Mysql account

For the mysql account, the procedure differs somewhat because this user does not normally have a home directory or the ability to open a Shell session. This procedure must also be applied to the two central nodes.

systemctl stop mysql
mkdir /home/mysql
chown mysql: /home/mysql
usermod -d /home/mysql mysql
usermod -s /bin/bash mysql
touch /var/log/mysqld.log
chown mysql. /var/log/mysqld.log
systemctl start mysql

Then switch to the bash environment of mysql and generate the SSH keys:

su - mysql
ssh-keygen -t ed25519 -a 100
cat ~/.ssh/id_ed25519.pub | tee ~/.ssh/authorized_keys

After running these commands on both nodes, copy the contents of the file and paste it into ~/.ssh/authorized_keys on the other node and then apply the correct permissions to the file (still as mysql):

chmod 600 ~/.ssh/authorized_keys

The key exchange must then be validated by a first connection that will accept the signature of the SSH server by typing "yes" (still as mysql):

ssh @DB_MASTER_NAME@
ssh @DB_SLAVE_NAME@

Then exit the mysql session using exit or Ctrl-D.

Setting up MySQL replication

For the two nodes to be interchangeable at any time, the two databases must be continuously replicated. To do this, we will set up a Master-Slave replication.

Since Centreon 22.04, MariaDB Replication is now based on GTID.

Note: unless otherwise stated, each of the following steps must be performed on both database nodes.

MySQL configuration

To begin with, we need to tune the MySQL configuration, which will be concentrated in the single file /etc/my.cnf.d/server.cnf. By default, the [server] section of this file is empty, which is where the following lines should be pasted:

[server]
sql_mode = 'NO_AUTO_CREATE_USER,NO_ENGINE_SUBSTITUTION'
server-id=1 # SET TO 1 FOR MASTER AND 2 FOR SLAVE
#read_only
log-bin=mysql-bin
binlog-do-db=centreon
binlog-do-db=centreon_storage
innodb_flush_log_at_trx_commit=1
sync_binlog=1
binlog_format=MIXED
slave_compressed_protocol=1
datadir=/var/lib/mysql
pid-file=/var/lib/mysql/mysql.pid
skip-slave-start
log-slave-updates
gtid_strict_mode=ON
expire_logs_days=7
ignore-db-dir=lost+found

# Tuning standard Centreon
innodb_file_per_table=1
open_files_limit=32000
key_buffer_size=256M
sort_buffer_size=32M
join_buffer_size=4M
thread_cache_size=64
read_buffer_size=512K
read_rnd_buffer_size=256K
max_allowed_packet=64M
# Uncomment for 4 Go Ram
#innodb_buffer_pool_size=512M
# Uncomment for 8 Go Ram
#innodb_buffer_pool_size=1G

Important: the value of server-id must be different from one server to another, so that they can identify each other correctly. The values 1 => Master and 2 => Slave are not mandatory, but are recommended.

N.B: Remember to uncomment (remove the '#' at the beginning of the line) the innodb_buffer_pool_size parameter that corresponds to your platform.

For these modifications to be taken into account, you must restart the database server:

systemctl restart mariadb

Use the following command to make sure the reboot was successful:

systemctl status mariadb

Warning: The centreon.cnf file will no longer be taken into account, so if any parameters were customized there, they must be transferred to server.cnf.

Securing the database

Access to databases should be restricted as strictly as possible. The command mysql_secure_installation is used to remove any accesses not protected by passwords and the test database. Run this command and be guided by the default choices. Be careful to choose a password that does not belong to any dictionary.

mysql_secure_installation

Creating the centreon account

To be able to administer MySQL users, you must first connect as root (with the password entered in the previous paragraph):

mysql -p

Then, in the MySQL prompt, paste the commands below (changing the IP addresses and passwords appropriately).

CREATE USER '@MYSQL_CENTREON_USER@'@'@CENTRAL_SLAVE_IPADDR@' IDENTIFIED BY '@MYSQL_CENTREON_PASSWD@';
GRANT ALL PRIVILEGES ON centreon.* TO '@MYSQL_CENTREON_USER@'@'@CENTRAL_SLAVE_IPADDR@';
GRANT ALL PRIVILEGES ON centreon_storage.* TO '@MYSQL_CENTREON_USER@'@'@CENTRAL_SLAVE_IPADDR@';

CREATE USER '@MYSQL_CENTREON_USER@'@'@CENTRAL_MASTER_IPADDR@' IDENTIFIED BY '@MYSQL_CENTREON_PASSWD@';
GRANT ALL PRIVILEGES ON centreon.* TO '@MYSQL_CENTREON_USER@'@'@CENTRAL_MASTER_IPADDR@';
GRANT ALL PRIVILEGES ON centreon_storage.* TO '@MYSQL_CENTREON_USER@'@'@CENTRAL_MASTER_IPADDR@';

Creating the replication account

Still in the MySQL prompt (see previous paragraph) create the user @MYSQL_REPL_USER@, dedicated to replication, using the following commands:

CREATE USER '@MYSQL_REPL_USER@'@'localhost' IDENTIFIED BY '@MYSQL_REPL_PASSWD@';
GRANT PROCESS, RELOAD, SHUTDOWN, SUPER, REPLICATION CLIENT, REPLICATION SLAVE ON *.* TO '@MYSQL_REPL_USER@'@'localhost';

CREATE USER '@MYSQL_REPL_USER@'@'@CENTRAL_SLAVE_IPADDR@' IDENTIFIED BY '@MYSQL_REPL_PASSWD@';
GRANT PROCESS, RELOAD, SHUTDOWN, SUPER, REPLICATION CLIENT, REPLICATION SLAVE ON *.* TO '@MYSQL_REPL_USER@'@'@DB_SLAVE_IPADDR@';

CREATE USER '@MYSQL_REPL_USER@'@'@CENTRAL_MASTER_IPADDR@' IDENTIFIED BY '@MYSQL_REPL_PASSWD@';
GRANT PROCESS, RELOAD, SHUTDOWN, SUPER, REPLICATION CLIENT, REPLICATION SLAVE ON *.* TO '@MYSQL_REPL_USER@'@'@DB_MASTER_IPADDR@';

Configuring environment variables for MySQL scripts

The file /etc/centreon-ha/mysql-resources.sh contains environment variables that must be adapted to the current installation by replacing the macros with the appropriate value.

#!/bin/bash

###############################
# Database access credentials #
###############################

DBHOSTNAMEMASTER='@DB_MASTER_NAME@'
DBHOSTNAMESLAVE='@DB_SLAVE_NAME@'
DBREPLUSER='@MYSQL_REPL_USER@'
DBREPLPASSWORD='@MYSQL_REPL_PASSWD@'
DBROOTUSER='@MYSQL_REPL_USER@'
DBROOTPASSWORD='@MYSQL_REPL_PASSWD@'
CENTREON_DB='centreon'
CENTREON_STORAGE_DB='centreon_storage'

##########################################
# Manual failover additional informations #
##########################################

CENTRAL_VIP_IPADDR='@CENTRAL_VIP_IPADDR@'
CENTRAL_VIP_IFNAME='@CENTRAL_VIP_IFNAME@'
CENTRAL_VIP_CIDR_NETMASK='@CENTRAL_VIP_CIDR_NETMASK@'
CENTRAL_VIP_BROADCAST_IPADDR='@CENTRAL_VIP_BROADCAST_IPADDR@'
MYSQL_VIP_IPADDR='@DB_VIP_IPADDR@'
MYSQL_VIP_IFNAME='@DB_VIP_IFNAME@'
MYSQL_VIP_CIDR_NETMASK='@DB_VIP_CIDR_NETMASK@'
MYSQL_VIP_BROADCAST_IPADDR='@DB_VIP_BROADCAST_IPADDR@'

To make sure that the last steps have been done correctly, and that the correct names, logins and passwords have been entered in the configuration file, run the command :

/usr/share/centreon-ha/bin/mysql-check-status.sh

The expected outcome is:

Connection Status '@DB_MASTER_NAME@' [OK]
Connection Status '@DB_SLAVE_NAME@' [OK]
Slave Thread Status [KO]
Error reports:
    No slave (maybe because we cannot check a server).
Position Status [SKIP]
!Error reports:
    Skip because we can't identify a unique slave.

It is important to check that the first two connection tests are OK.

Switching to read-only

Now that everything is set up properly, we need to enable read_only mode on both servers by uncommenting (i.e. removing the # at the beginning of the line) this statement in the /etc/my.cnf.d/server.cnf file:

Main node

[server]
server-id=1
read_only
log-bin=mysql-bin

Secondary node

[server]
server-id=2
read_only
log-bin=mysql-bin

Then apply this change by restarting MariaDB on both nodes:

systemctl restart mariadb

Authorization of centreon and mysql accounts

For the scripts to run correctly through the ssh connections via the centreon and mysql accounts, they must be given additional rights.

For the centreon account, edit the file /etc/sudoers.d/centreon-cluster and add the following lines:

CENTREON   ALL = NOPASSWD: /usr/bin/systemctl reload centreon
CENTREON   ALL = NOPASSWD: /usr/bin/systemctl restart centreon
CENTREON   ALL = NOPASSWD: /usr/bin/systemctl stop centreon
CENTREON   ALL = NOPASSWD: /usr/bin/systemctl start centreon

CENTREON   ALL = NOPASSWD: /usr/bin/nmcli

For the mysql account, edit the file /etc/sudoers.d/centreon-cluster-db and add the following line:

MYSQL   ALL = NOPASSWD: /usr/bin/nmcli

Synchronize databases and start MySQL replication

To synchronize the databases, stop the mysql service on the secondary node to overwrite its data with that of the main server.

Run the following command on the secondary node:

systemctl stop mariadb

In some cases, systemd may fail to stop the mysql service. To be sure, check that the following command returns no lines:

ps -ef | grep mysql[d]

If a mysqld process is still running, then the following command should be run to stop it (and provide the mysql root password when prompted):

mysqladmin -p shutdown

Once the service is stopped on the secondary node, run the synchronization script from the main node:

/usr/share/centreon-ha/bin/mysql-sync-bigdb.sh

This script does the following:

checks that mysql is stopped on the secondary node
stops mysql on the primary node
mounts an LVM snapshot on the volume group that supports the /var/lib/mysql partition
starts mysql on the primary node
stores the current position in the binary logs
disables the MySQL global variable read_only on the master node (i.e. the master node is now allowed to write to its database)
synchronizes all data files (excluding the mysql system database) by overwriting the files on the secondary node
unmounts the LVM snapshot
creates the replication thread that will keep the data updated on the secondary node

This script is very verbose, and not everything that is displayed is understandable, but to make sure that it has run to completion, just make sure that the end looks like this:

Umount and Delete LVM snapshot
  Logical volume "dbbackupdatadir" successfully removed
Start MySQL Slave
OK
Start Replication
Id	User	Host	db	Command	Time	State	Info	Progress
3	@MYSQL_REPL_USER@	@CENTRAL_MASTER_NAME@:33084	NULL	Query	0	init	show processlist	0.000

If all went well, then the mysql-check-status.sh command should return an error-free result:

/usr/share/centreon-ha/bin/mysql-check-status.sh

Expected outcome:

Connection Status '@CENTRAL_MASTER_NAME@' [OK]
Connection Status '@CENTRAL_SLAVE_NAME@' [OK]
Slave Thread Status [OK]
Position Status [OK]

Creating the MySQL VIP

The VIP will be added to the @DB_VIP_IFNAME@ interface configuration on the master node:

nmcli con mod "@DB_VIP_IFNAME@" +ipv4.addresses "@DB_VIP_IPADDR@/@DB_VIP_CIDR_NETMASK@"
nmcli connection up "@DB_VIP_IFNAME@"

Then run the following script on one of the two MySQL nodes to check if the VIP is correctly mounted on the MySQL master server (this script can be run on either of the two nodes and it will detect by itself which is the master):

/usr/share/centreon-ha/bin/move-mysql-vip-to-mysql-master.sh

The result should be like this; if not, use the script to move the VIP to the correct place:

The VIP address is already in the right place. Nothing to do.

Setting up the Centreon cluster

Note: unless otherwise noted, each of the following steps must be performed on both central nodes.

Configuring the Synchronization Service

The centreon-central-sync service requires that, for each node in /etc/centreon-ha/centreon_central_sync.pm, we define the IP address of its correspondent:

So for the server @CENTRAL_MASTER_NAME@ we must have:

our %centreon_central_sync_config = (
    peer_addr => "@CENTRAL_SLAVE_IPADDR@"
);
1;

And for the server @CENTRAL_SLAVE_NAME@ we must have:

our %centreon_central_sync_config = (
    peer_addr => "@CENTRAL_MASTER_IPADDR@"
);
1;

Removing crons

Scheduled cron tasks are executed directly by the gorgone process in the highly available architecture. This ensures that they do not compete with each other on the central nodes. It is therefore necessary to delete them manually:

rm -f /etc/cron.d/centreon
rm -f /etc/cron.d/centstorage
rm -f /etc/cron.d/centreon-auto-disco

Changing rights

The directories /var/log/centreon-engine and /tmp/centreon-autodisco are shared between several processes. It is necessary to modify the rights of the directories and files to ensure the proper functioning of file replication and automatic discovery of services:

chmod 775 /var/log/centreon-engine/
mkdir /var/log/centreon-engine/archives
chown centreon-engine: /var/log/centreon-engine/archives
chmod 775 /var/log/centreon-engine/archives/
chmod 664 /var/log/centreon-engine/*
chmod 664 /var/log/centreon-engine/archives/*

mkdir /tmp/centreon-autodisco/
chown apache: /tmp/centreon-autodisco/
chmod 775 /tmp/centreon-autodisco/

Stop and disable services

The Centreon application services will no longer be launched on server startup as in the case of a standard installation; the clustering services will take care of it. It is therefore necessary to stop and disable these services.

Alma / RHEL / Oracle Linux 8

systemctl stop centengine snmptrapd centreontrapd gorgoned cbd httpd php-fpm centreon
systemctl disable centengine snmptrapd centreontrapd gorgoned cbd httpd php-fpm centreon centreon-central-sync cbd-sql

Redefine service startup rules

Some services should only be started on one node, but for others, it's OK or even desirable to start them on both nodes. We'll start with these services.

PHP service

The php-fpm.service service does not need to be modified, but must be enabled so that it is started automatically when the central servers are started.

Execute this command on both central nodes:

systemctl enable php-fpm

Broker RRD service

The "broker-rrd" process is started via the cbd.service. By default, this is driven by centreon.service, but in this configuration, it must be detached by modifying its definition on the two central nodes:

cat > /usr/lib/systemd/system/cbd.service <<"EOF"
[Unit]
Description=Centreon Broker watchdog
After=syslog.target network.target

[Service]
ExecStart=/usr/sbin/cbwd /etc/centreon-broker/watchdog.json
ExecReload=/bin/kill -HUP $MAINPID
Type=simple
User=centreon-broker
UMask=0002

[Install]
WantedBy=multi-user.target
EOF

Then start it and activate it:

systemctl start cbd
systemctl enable cbd

Creating the Centreon VIP

The VIP will be added to the @CENTRAL_VIP_IFNAME@ interface configuration on the master node. In most cases, the central VIP is the same as the database VIP for a two-node cluster and this action has already been done for the replication.

If not, execute these commands on @CENTRAL_MASTER_NAME@:

nmcli con mod "@CENTRAL_VIP_IFNAME@" +ipv4.addresses "@CENTRAL_VIP_IPADDR@/@CENTRAL_VIP_CIDR_NETMASK@"
nmcli connection up "@CENTRAL_VIP_IFNAME@"

Httpd service

By default, the httpd.service service is independent of centreon.service, but in this configuration, it must be attached to it by modifying its definition on the two central nodes:

Alma / RHEL / Oracle Linux 8

cat > /usr/lib/systemd/system/httpd.service <<"EOF"
[Unit]
Description=The Apache HTTP Server
PartOf=centreon.service
ReloadPropagatedFrom=centreon.service
After=centreon.service

[Service]
Type=notify
Environment=LANG=C

ExecStart=/usr/sbin/httpd $OPTIONS -DFOREGROUND
ExecReload=/usr/sbin/httpd $OPTIONS -k graceful
ExecStop=/usr/sbin/httpd $OPTIONS -k graceful-stop
# Send SIGWINCH for graceful stop
KillSignal=SIGCONT
KillMode=mixed
PrivateTmp=true

[Install]
WantedBy=centreon.service
EOF

Gorgone service

By default, the gorgoned.service service is independent of centreon.service, but in this configuration, it must be attached to it by modifying its definition on the two central nodes:

Alma / RHEL / Oracle Linux 8

cat > /etc/systemd/system/gorgoned.service <<"EOF"
[Unit]
Description=Centreon Gorgone
PartOf=centreon.service
After=httpd.service
ReloadPropagatedFrom=centreon.service

[Service]
EnvironmentFile=/etc/sysconfig/gorgoned
ExecStart=/usr/bin/perl /usr/bin/gorgoned $OPTIONS
Type=simple
User=centreon-gorgone

[Install]
WantedBy=centreon.service
EOF

Centreon-central-sync service

This service is specific to Centreon HA. Its function is to replicate configuration changes, add images via the interface, etc.

cat > /usr/lib/systemd/system/centreon-central-sync.service <<"EOF"
[Unit]
Description=Centreon Central Sync (failover only)
PartOf=centreon.service
ReloadPropagatedFrom=centreon.service
After=gorgoned.service

[Service]
EnvironmentFile=/etc/sysconfig/centreon_central_sync
ExecStart=/usr/share/centreon-ha/bin/centreon_central_sync $OPTIONS
ExecReload=/bin/kill -HUP $MAINPID
Type=simple
User=centreon

[Install]
WantedBy=centreon.service
EOF

Broker SQL service

cat > /usr/lib/systemd/system/cbd-sql.service <<"EOF"
[Unit]
Description=Centreon SQL Broker service
PartOf=centreon.service
ReloadPropagatedFrom=centreon.service
After=centreon-central-sync.service

[Service]
EnvironmentFile=/etc/sysconfig/cbd_sql
ExecStart=/usr/sbin/cbd $OPTIONS
ExecReload=/bin/kill -HUP $MAINPID
Type=simple
User=centreon-broker
UMask=0002

[Install]
WantedBy=centreon.service
EOF

Centengine service

cat > /usr/lib/systemd/system/centengine.service <<"EOF"
[Unit]
Description=Centreon Engine
PartOf=centreon.service
ReloadPropagatedFrom=centreon.service
After=cbd-sql.service

[Service]
ExecStart=/usr/sbin/centengine /etc/centreon-engine/centengine.cfg
ExecReload=/bin/kill -HUP $MAINPID
Type=simple
User=centreon-engine

[Install]
WantedBy=centreon.service
EOF

Centreontrapd service

cat > /usr/lib/systemd/system/centreontrapd.service <<"EOF"
[Unit]
Description=Centreon Trapd Daemon is a Centreon program that manages traps
PartOf=centreon.service
After=centreon.service
ReloadPropagatedFrom=centreon.service

[Service]
EnvironmentFile=/etc/sysconfig/centreontrapd
ExecStart=/usr/share/centreon/bin/centreontrapd $OPTIONS
ExecReload=/bin/kill -HUP $MAINPID
Type=simple
User=centreon

[Install]
WantedBy=centreon.service
EOF

Snmptrapd service

cat > /usr/lib/systemd/system/snmptrapd.service <<"EOF"
[Unit]
Description=Simple Network Management Protocol (SNMP) Trap Daemon.
PartOf=centreon.service
After=centreontrapd.service

[Service]
Type=notify
Environment=OPTIONS="-Lsd"
EnvironmentFile=-/etc/sysconfig/snmptrapd
ExecStart=/usr/sbin/snmptrapd $OPTIONS -f
ExecReload=/bin/kill -HUP $MAINPID

[Install]
WantedBy=centreon.service
EOF

Initial launch of all services

Installation of the Centreon virtual IP address

Normally, the VIP has already been mounted previously. Check that this is the case on the primary server defined above (normally @CENTRAL_MASTER_NAME@)

ip a

You should see the virtual IP @CENTRAL_VIP_IPADDR@ attached to the @CENTRAL_VIP_IFNAME@ interface appear in the return of the ip a command.

If not, the virtual IP address must be mounted, for example on @CENTRAL_MASTER_NAME@ :

nmcli con mod "@CENTRAL_VIP_IFNAME@" +ipv4.addresses "@CENTRAL_VIP_IPADDR@/@CENTRAL_VIP_CIDR_NETMASK@"
nmcli con up "@CENTRAL_VIP_IFNAME@"

Taking into account the changes made to the services

To take into account all the previous modifications, and to activate the services (so that starting the centreon.service service starts them all), it is necessary to launch these commands on the two central nodes:

Alma / RHEL / Oracle Linux 8

systemctl daemon-reload
systemctl enable cbd httpd gorgoned centreon-central-sync cbd-sql centengine centreontrapd snmptrapd mariadb

And, finally, start them all via the centreon.service on the node where the VIP was mounted:

systemctl start centreon

Then check the status of all services:

Alma / RHEL / Oracle Linux 8

systemctl status cbd httpd gorgoned centreon-central-sync cbd-sql centengine centreontrapd snmptrapd

After this step, all resources should be active in the same place, and the platform should be functional and redundant. If not, refer to the troubleshooting guide in the next paragraph.

Standard manual HA operation

Switch Centreon services from one node to another

This script can be run from any Centreon node, but you must (for now) specify the target node. For example, to change the active node from @CENTRAL_MASTER_NAME@ to @CENTRAL_SLAVE_NAME@:

/usr/share/centreon-ha/bin/set-centreon-master.sh @CENTRAL_SLAVE_NAME@

The expected outcome is:

Stopping centreon.service on @CENTRAL_MASTER_NAME@...
Unmounting VIP on @CENTRAL_MASTER_NAME@...
Adding vip to @CENTRAL_SLAVE_NAME@...
Starting centreon.service on @CENTRAL_SLAVE_NAME@...

If we try to switch to the node that is already active, we will get:

Host @CENTRAL_MASTER_NAME@ is already the current master :-)

Reverse MySQL SLAVE/MASTER roles

This script can be launched from any node. This time, we do not specify the name of the target server:

/usr/share/centreon-ha/bin/mysql-switch-slave-master.sh

The expected outcome is:

Locking master database on @DB_SLAVE_NAME@
Waiting Relay log bin to finish proceed (TIMEOUT = 60sec)
Removing slave thread on @DB_MASTER_NAME@
Recording binlog file and position from @DB_MASTER_NAME@
Unlocking databases on @DB_MASTER_NAME@
Setting and starting slave thread on @DB_SLAVE_NAME@
We have to move the VIP address

Checking database synchronization

The status of MySQL replication can be checked at any time using the command mysql-check-status.sh:

/usr/share/centreon-ha/bin/mysql-check-status.sh

Expected outcome:

Connection Status '@CENTRAL_MASTER_NAME@' [OK]
Connection Status '@CENTRAL_SLAVE_NAME@' [OK]
Slave Thread Status [OK]
Position Status [OK]

Integrating pollers

You can now add your Pollers and start monitoring!

Prerequisites​

Understanding​

Installation of Centreon​

Definition of server names and IP addresses​

Centreon-broker configuration​

Linking to the cbd service​

Dual RRD streams​

Export the configuration​

Modification of the cbd reload command​

Synchronize Centreon configuration files (PHP, Perl, Gorgone)​

System preparation​

Tuning the network configuration​

Name resolution​

Installing the packages​

SSH key exchange​

Centreon account​

Mysql account​

Setting up MySQL replication​

MySQL configuration​

Securing the database​

Creating the centreon account​

Creating the replication account​

Configuring environment variables for MySQL scripts​

Switching to read-only​

Authorization of centreon and mysql accounts​

Synchronize databases and start MySQL replication​

Creating the MySQL VIP​

Setting up the Centreon cluster​

Configuring the Synchronization Service​

Removing crons​

Changing rights​

Stop and disable services​

Redefine service startup rules​

PHP service​

Broker RRD service​

Creating the Centreon VIP​

Httpd service​

Gorgone service​

Centreon-central-sync service​

Broker SQL service​

Centengine service​

Centreontrapd service​

Snmptrapd service​

Initial launch of all services​

Installation of the Centreon virtual IP address​

Taking into account the changes made to the services​

Standard manual HA operation​

Switch Centreon services from one node to another​

Reverse MySQL SLAVE/MASTER roles​

Checking database synchronization​

Integrating pollers​

Prerequisites

Understanding

Installation of Centreon

Definition of server names and IP addresses

Centreon-broker configuration

Linking to the cbd service

Dual RRD streams

Export the configuration

Modification of the `cbd` reload command

Synchronize Centreon configuration files (PHP, Perl, Gorgone)

System preparation

Tuning the network configuration

Name resolution

Installing the packages

SSH key exchange

Centreon account

Mysql account

Setting up MySQL replication

MySQL configuration

Securing the database

Creating the centreon account

Creating the replication account

Configuring environment variables for MySQL scripts

Switching to read-only

Authorization of centreon and mysql accounts

Synchronize databases and start MySQL replication

Creating the MySQL VIP

Setting up the Centreon cluster

Configuring the Synchronization Service

Removing crons

Changing rights

Stop and disable services

Redefine service startup rules

PHP service

Broker RRD service

Creating the Centreon VIP

Httpd service

Gorgone service

Centreon-central-sync service

Broker SQL service

Centengine service

Centreontrapd service

Snmptrapd service

Initial launch of all services

Installation of the Centreon virtual IP address

Taking into account the changes made to the services

Standard manual HA operation

Switch Centreon services from one node to another

Reverse MySQL SLAVE/MASTER roles

Checking database synchronization

Integrating pollers