New in version 3.0.0.
The Wazuh cluster has been developed in order to get a stronger communication between agents and managers. This means that the cluster is able to synchronize the necessary files between its nodes to allow agents to report events to any manager of the cluster.
The Wazuh cluster provides horizontal scalability to our Wazuh environment, allowing agents to report to any manager belonging to the cluster. This way, we have the ability to process more events than before distributing the load of information between several managers simultaneously.
In addition, a cluster of Wazuh managers is prepared to handle the fall of any manager without affecting its operation, unless it is the master manager. Agents that were reporting to this fallen manager will start to report to another manager of the cluster automatically, without the loss of events.
Finally, the Wazuh cluster is in continuous development and we hope it to include many new features very soon. For example, we refer to the possibility of switching the role of master between all managers to provide high availability.
The Wazuh cluster is managed by a cluster daemon which communicates managers, following a master-client architecture.
The master node is the manager who has the control of the cluster. This means that every configuration done in the master node is pushed to the others managers, and it allows to centralize the following configurations:
Agents registration and deletion.
Configuration of agents grouping.
Centralized configuration for agents, synchronizing its
To sum up, the master node sends to its clients the whole
etc/shared directory contained in the Wazuh installation directory, with
the centralized configuration of each agent ordered by groups, as well as the
client.keys file. These shared files agents to report to any manager of the cluster.
The comunications between nodes of the cluster are made with a self developed protocol which synchronizes those files with a specific interval time, defined in
<cluster> section of Local configuration.
These communications are encrypted with AES providing confidentiality.
Managers which have the client role in the cluster receive the data from the master node, and update their files with the received ones. This way, it’s guaranteed that the shared configuration for the agents is the same in all managers.
On the other hand, client nodes send to the master the
agent-info file of their reporting agents. This file contains the most important information of each agent, and allows to the master node to know in real-time
the connection state of agents, as well as the manager each agent is reporting to.
wazuh-clusterd is the module that manage the synchronization between managers in the cluster. There exits a file located at
logs/cluster.logwhere can be found all the logging messages of this daemon.
wazuh-clusterd-internal is the daemon in charge of monitoring the files to synchronize and managing the cluster database. The logs of this daemon are stored in
Those daemons must be running in all the managers of the cluster in order to ensure everything works properly. The wazuh-clusterd will start the wazuh-clusterd-internal daemon.
Refer to the section Daemons to find out more information about the usage of these daemons.
The cluster can be efficiently controlled from any manager with the cluster_control. This tool allows you to obtain real-time information about any node, as well as the status of the all synchronized files and information about agents connected to the cluster.
The manual about this tool can be found at cluster_control tool.
To run the wazuh-clusterd binary, Python 2.7 is required. If your OS has a lower python version, please check section Run the cluster in CentOS 6
In order to deploy a Wazuh cluster, follow these steps:
For RPM-based distributions:# yum install python-setuptools python-cryptography
For Debian-based distributions:# apt install python-cryptography
Set the properly configuration in all the managers of the cluster.
<cluster>section of the Local configuration it should be set the configuration for the cluster regarding the following considerations.
One manager should be the master and the other ones, the clients. This is specified in the
The key should be the same for all the nodes of the cluster and it must be 32 characters long. To generate a random password you can use the following command:# openssl rand -hex 16
The IP addresses of all nodes of the cluster must be specified in the
<nodes>, including the IP of the local manager. The managers will use the bash command
hostname --all-ip-addressesto find out which IP from the list is theirs. If none of the IPs match with the ones returned by the
hostname --all-ip-addressescommand, an error will be raised.
An example of configuration could be the following.<cluster> <name>cluster01</name> <node_name>manager_centos</node_name> <node_type>master</node_type> <key>nso42FGdswR0805tnVqeww0u3Rubwk2a</key> <interval>2m</interval> <port>1516</port> <bind_addr>0.0.0.0</bind_addr> <nodes> <node>192.168.0.3</node> <node>192.168.0.4</node> <node>192.168.0.5</node> </nodes> <hidden>no</hidden> <disabled>yes</disabled> </cluster>
Agents should be configured for connecting to all the managers of the cluster.
For example, if we have three managers in the cluster with the IP addresses
192.168.0.3-5the configuration in agents should be like this.<client> ... <server> <address>192.168.0.3</address> <port>1514</port> <protocol>udp</protocol> </server> <server> <address>192.168.0.4</address> <port>1514</port> <protocol>tcp</protocol> </server> <server> <address>192.168.0.5</address> <port>1514</port> <protocol>tcp</protocol> </server> ... </client>
To enable the Wazuh cluster, set the field
<cluster>section of the ossec.conf file and restart:
# /var/ossec/bin/ossec-control restart
Since this moment, the cluster should be synchronized and the shared files should be the same in all the managers.
Python 2.6 is the default python version in CentOS6. Since Python 2.7 is required to run the cluster, follow these steps:
Install Python 2.7:
# yum install -y centos-release-scl # yum install -y python27
Enable python2.7 in bash:
# scl enable python27 bash
The default version of
sqlite3library is not compatible but a compiled version of
sqlite3can be found at
/var/ossec/framework/lib. To load this version follow these steps:
chrpath:# yum install -y chrpath
chrpathto remove the reference path to the system’s sqlite3 library:# chrpath --delete /opt/rh/python27/root/usr/lib64/python2.7/lib-dynload/_sqlite3.so
Add the compiled version of sqlite3 to the
LD_LIBRARY_PATHvariable:# export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/var/ossec/framework/lib
# /opt/rh/python27/root/usr/bin/pip2.7 install cryptography
python2.7to start the cluster:
# python2.7 /var/ossec/bin/wazuh-clusterd
Finally, if the cluster is correctly configured, check it’s running:
# ps -aux | grep cluster ossec 6533 0.0 1.4 135424 15128 ? S 07:19 0:00 python2.7 /var/ossec/bin/wazuh-clusterd root 6536 0.0 0.4 158608 4584 ? Ssl 07:19 0:00 /var/ossec/bin/wazuh-clusterd-internal -tmaster ossec 6539 0.0 1.5 136464 15932 ? S 07:19 0:00 python2.7 /var/ossec/bin/wazuh-clusterd root 6556 0.0 0.2 8032 2092 ? S+ 07:21 0:00 grep cluster