Red Hat Training

A Red Hat training course is available for Red Hat JBoss Web Server

HTTP Connectors Load Balancing Guide

Red Hat JBoss Web Server 2.1

HTTP load balancing for JBoss Enterprise Application Platform and Red Hat JBoss Web Server 2.1.0

Misha Husnain Ali

Red Hat Engineering Content Services

Mandar Joshi

Red Hat Engineering Content Services

Gemma Sheldon

Red Hat Engineering Content Services

Abstract

Read this guide to install and configure JBoss Enterprise Application Platform and Red Hat JBoss Web Server HTTP connectors: mod_jk, mod_cluster, ISAPI, and NSAPI. This guide also discusses clustering and load-balancing using these connectors.

Part I. Apache Tomcat Connector

Chapter 1. Overview

1.1. Overview

Apache is a well-known web server which can be extended using plug-ins. The Apache Tomcat Connector mod_jk is a plug-in designed to allow request forwarding from Apache httpd Server to a Servlet container. The module also supports load-balancing HTTP calls to a set of Servlet containers while maintaining sticky sessions.

1.2. Download and Install

Apache httpd is included in the JBoss Enterprise Web Server binary you download from https://access.redhat.com.
mod_jk is included in the native installation binaries for JBoss Enterprise Application Platform and JBoss Enterprise Web Server.
Follow the procedures in the JBoss Enterprise Application Platform or JBoss Enterprise Web Server Installation Guide to download and install the correct platform and native binaries.
For supported configurations, see the JBoss Enterprise Web Server Supported Configurations page.

Chapter 2. Configure Load Balancing Using Apache and mod_jk

2.1. Configure Load Balancing Using Apache and mod_jk

Follow the tasks in this chapter to correctly configure load balancing using Apache and the mod_jk connector.

Procedure 2.1. Configure Apache to Load mod_jk

Ensure that Apache and mod_jk are installed (see Section 1.2, “Download and Install”).
To configure Apache to Load mod_jk:
  1. Open HTTPD_DIST/conf/httpd.conf and add a single line at the end of the file.
    # Include mod_jk's specific configuration file  
    Include conf/mod-jk.conf
    
  2. Create a new file named HTTPD_DIST/conf/mod-jk.conf
  3. Add the following configuration to the mod-jk.conf file.

    Important

    The LoadModule directive must reference the mod_jk library directory location applicable to the native binary you installed.

    Note

    The JkMount directive specifies which URLs Apache should forward to the mod_jk module. Based on the directive's configuration, mod_jk forwards the received URL onto the correct Servlet containers.
    To enable Apache to serve static content (or PHP content) directly, and only use the load balancer for Java applications, the suggested configuration specifies all requests with URL path /application/* are sent to the mod_jk load-balancer.
    Only mod_jk is used as a load balancer, forward all URLs to mod_jk by specifying /* in the directive.
    # Load mod_jk module
    # Specify the filename of the mod_jk lib
    LoadModule jk_module modules/mod_jk.so
     
    # Where to find workers.properties
    JkWorkersFile conf/workers.properties
    
    # Where to put jk logs
    JkLogFile logs/mod_jk.log
     
    # Set the jk log level [debug/error/info]
    JkLogLevel info 
     
    # Select the log format
    JkLogStampFormat  "[%a %b %d %H:%M:%S %Y]"
     
    # JkOptions indicates to send SSK KEY SIZE
    JkOptions +ForwardKeySize +ForwardURICompat -ForwardDirectories
     
    # JkRequestLogFormat
    JkRequestLogFormat "%w %V %T"
                   
    # Mount your applications
    JkMount /application/* loadbalancer
     
    # Add shared memory.
    # This directive is present with 1.2.10 and
    # later versions of mod_jk, and is needed for
    # for load balancing to work properly
    JkShmFile logs/jk.shm 
                  
    # Add jkstatus for managing runtime data
    <Location /jkstatus/>
        JkMount status
        Order deny,allow
        Deny from all
        Allow from 127.0.0.1
    </Location>
    
  4. Optional: JKMountFile Directive

    In addition to the JkMount directive, use the JkMountFile directive to specify a mount points configuration file. The configuration file contains multiple Tomcat forwarding URL mappings.
    1. Navigate to HTTPD_DIST/conf.
    2. Create a file named uriworkermap.properties.
    3. Specify the URL to forward and the worker name using the following syntax example as a guide.
      The example block will configure mod_jk to forward requests to /jmx-console and /web-console to Apache.
      The syntax required takes the form /url=worker_name.
      # Simple worker configuration file
      
      # Mount the Servlet context to the ajp13 worker
      /jmx-console=loadbalancer
      /jmx-console/*=loadbalancer
      /web-console=loadbalancer
      /web-console/*=loadbalancer
      
    4. In HTTPD_DIST/conf/mod-jk.conf, append the following directive.
      # Use external file for mount points.
      # It will be checked for updates each 60 seconds.
      # The format of the file is: /url=worker
      # /examples/*=loadbalancer
      JkMountFile conf/uriworkermap.properties

2.2. Configuring Worker Nodes in mod_jk

2.2.1. Configuring Worker Nodes in mod_jk

Configure two mod_jk Worker node definitions in a weighted round robin configuration with sticky sessions active between two servlet containers.

Procedure 2.2. Configure mod_jk Worker Nodes

As a prerequisite, understand the format of the workers.properties directives, as specified in Section A.1, “workers.properties” and configure mod_jk (see Section 2.1, “Configure Load Balancing Using Apache and mod_jk”).
To configure mod_jk worker nodes:
  1. Navigate to HTTPD_DIST/conf/.
  2. Create a file named workers.properties.
  3. Append the following information into the workers.properties file.
    # Define list of workers that will be used
    # for mapping requests
    worker.list=loadbalancer,status
    
    # Define Node1
    # modify the host as your host IP or DNS name.
    worker.node1.port=8009
    worker.node1.host=node1.mydomain.com
    worker.node1.type=ajp13
    worker.node1.ping_mode=A
    worker.node1.lbfactor=1 
    
    # Define Node2
    # modify the host as your host IP or DNS name.
    worker.node2.port=8009
    worker.node2.host=node2.mydomain.com
    worker.node2.type=ajp13
    worker.node2.ping_mode=A
    worker.node2.lbfactor=1
    
    # Load-balancing behavior
    worker.loadbalancer.type=lb
    worker.loadbalancer.balance_workers=node1,node2
    worker.loadbalancer.sticky_session=1
    
    # Status worker for managing load balancer
    worker.status.type=status

2.3. Configuring Apache Tomcat to Work with mod_jk

Tomcat is configured to use mod_jk by default. Specifically, see the conf/server.xml file, which contains the following code for this purpose:
<connector port="8009" protocol="AJP/1.3" redirectPort="8443" />
Set the jvmRoute attribute at your engine as follows:
<Engine name="Catalina" jvmRoute="node1" >
The jvmRoute attribute value must match the worker name set in workers.properties.
This default Tomcat configuration is ready for immediate use with mod_jk.

Chapter 3. WebSocket on Tomcat

3.1. About WebSocket

WebSocket is a web technology that provides bi-directional, full duplex, messages to be instantly distributed between the client and server over a single TCP socket connection. A full duplex communication allows two-way communication simultaneously.
The container provides an implementations of the WebSockets 1.0 JSR 356 API. To use the API, you must run Java 7+ and configure the APR or NIO2 HTTP/1.1 connectors of the web container.
JSR 356 is a standard for WebSocket API for Java. Developers can use JSR 356 API for creating WebSocket applications independent of the implementation. The WebSocket API is purely event driven.
Developers can use JSR 356 Java API for WebSocket to integrate WebSockets in applications on the server side as well as on the java client side. Tomcat 7 and 8 implement WebSocket protocol that adheres to JSR-356 standard.
A Java client uses JSR 356 compliant client implementation to connect to a WebSocket server. For web clients, WebSocket JavaScript API can be used to communicate with WebSocket server. The only difference between a WebSocket client and a WebSocket server is the method in which they are connected. A WebSocket client is a WebSocket point from which the connection to a peer originates. While a WebSocket server is WebSocket endpoint which is already published and awaits connections from peers.
Some of the examples where WebSocket can be used are banking applications, chat applications, multiplayer applications, and social networking applications.

3.2. Implementing WebSocket on Tomcat

The WebSocket on Tomcat configuration requires individual configuration of the following:
  • Configuring write timeout
  • Configuring incoming binary messages
  • Configuring incoming text messages
  • Configuring additional programmatic deployment
  • Configuring callbacks for asynchronous writes
  • Configuring timeout for IO operations while establishing the connections
Configuring write timeout

You can change the write timeout in blocking mode by using the org.apache.tomcat.websocket.BLOCKING_SEND_TIMEOUT property. The property accepts values in milliseconds. The default value is 20000 milliseconds (20 seconds).

Configuring incoming binary messages

To configure incoming binary messages, MessageHandler.Partial must be defined. If MessageHandler.Partial is not defined then incoming binary messages must be buffered so that the entire message is delivered in a single call to MessageHandler.Whole.

The default buffer size for binary messages is 8192 bytes. You can change the buffer size for a web application by changing the value of the servlet context initializing parameter org.apache.tomcat.websocket.binaryBufferSize.
Configuring incoming text messages

To configure incoming text messages, MessageHandler.Partial must be defined. If MessageHandler.Partial is not defined then incoming text messages must be buffered so that the entire message is delivered in a single call to MessageHandler.Whole.

The default buffer size for text messages is 8192 bytes. You can change the buffer size for a web application by changing the value of the servlet context initializing parameter org.apache.tomcat.websocket.textBufferSize.
Configuring additional programmatic deployment

Java WebSocket specification 1.0 does not allow programmatic deployment after the first endpoint has started a WebSocket handshake. However, Tomcat by default allows additional programmatic deployment. Additional programmatic deployment can be done by using the servlet context initialization parameter org.apache.tomcat.websocket.noAddAfterHandshake.

Set the system property org.apache.tomcat.websocket.STRICT_SPEC_COMPLIANCE to true to change the default setting.
Configuring callbacks for asynchronous writes

Callbacks for asynchronous writes need to be performed on a different thread to the thread that initiated the write. The container thread pool is not exposed via the Servlet API. Hence the WebSocket implementation has to provide its own thread pool.

The following servlet context initialization parameters control the thread pool:
  • org.apache.tomcat.websocket.executorCoreSize
    The core size of the executor thread pool. If not set, the default of 0 (zero) is used.
  • org.apache.tomcat.websocket.executorMaxSize
    The maximum permitted size of the executor thread pool. If not set, the default of 10 is used.
  • org.apache.tomcat.websocket.executorKeepAliveTimeSeconds
    The maximum time an idle thread will remain in the executor thread pool until it is terminated. If not specified, the default of 60 seconds is used.
Configuring timeout for IO operations while establishing the connections

The timeout for IO operations while establishing the connections is controlled by userProperties of the provided javax.websocket.ClientEndpointConfig. You can change timeout by changing the org.apache.tomcat.websocket.IO_TIMEOUT_MS property. The property accepts the values in milliseconds. The default value is 5000 (5 seconds).

To connect WebSocket client to secure server endpoints, the client SSL configuration is controlled by userProperties of the provided javax.websocket.ClientEndpointConfig.
The following user properties are supported:
  • org.apache.tomcat.websocket.SSL_CONTEXT
  • org.apache.tomcat.websocket.SSL_PROTOCOLS
  • org.apache.tomcat.websocket.SSL_TRUSTSTORE
  • org.apache.tomcat.websocket.SSL_TRUSTSTORE_PWD
The default truststore password is changeit. The org.apache.tomcat.websocket.SSL_TRUSTSTORE and org.apache.tomcat.websocket.SSL_TRUSTSTORE_PWD properties are ignored if the org.apache.tomcat.websocket.SSL_CONTEXT property is set.

Part II. JBoss HTTP Connector

Chapter 4. Overview

4.1. Overview

The JBoss HTTP Connector mod_cluster is a reduced configuration, intelligent load-balancing solution for JBoss Enterprise Application Platform, based on technology originally developed by the JBoss mod_cluster community project.
The JBoss HTTP connector load-balances HTTP requests to JBoss Enterprise Application Platform and JBoss Enterprise Web Server worker nodes, utilizing Apache as the proxy server. It serves as a load balancing solution for Tomcat in JBoss Enterprise Web Server as well as for JBoss Enterprise Application Platform.

4.2. Key Features

Table 4.1.  Features

Feature Description
Apache HTTP Server-based The JBoss HTTP Connector mod-cluster uses Apache as the proxy server.
Real-time load-balancing calculation
The JBoss HTTP Connector mod_cluster creates a feedback network between the worker nodes and the proxy server. The mod_cluster service is deployed on each of the worker nodes. This service feeds real time load information to the proxy server. The proxy server then allocates work, based on the current load on each worker node. This real time adaptive load distribution results in increased optimization of resources.
The information reported by the worker nodes and the load-balancing policy used by the proxy are customizable.
Routing based on real-time application life cycle The JBoss HTTP Connector mod_cluster service deployed on the worker nodes relays application lifecycle events to the proxy server. This allows the server to dynamically update its routing table. When an application is undeployed on a node, the proxy server does notroute traffic for that application to that node
Automatic Proxy Discovery
The proxy server is configured to announce its presence via UDP multicast. New worker nodes discover the proxy server and automatically add themselves to the load-balancing cluster. This reduces the configuration and maintenance needed. When UDP multicast is not available or is undesirable, worker nodes are configured with a static list of proxies.
Multiple Protocol Support The JBoss HTTP Connector mod_cluster uses either HTTP, HTTPS, or Apache JServ Protocol (AJP) for communication between the proxy and the worker nodes.

4.3. Components

Proxy Server
On the proxy server, the JBoss HTTP Connector mod-cluster consists of four Apache modules.

Table 4.2.  Components

Component Description
mod_slotmem.so The Shared Memory Manager module shares the real time worker node information with multiple Apache server processes
mod_manager.so
The Cluster Manager module receives and acknowledges messages from nodes, including worker node registrations, worker node load data, and worker node application life cycle events
mod_proxy_cluster.so The Proxy Balancer Module handles request routing to cluster nodes. The Proxy Balancer selects the appropriate destination node based on application location in the cluster, current state of each of the cluster nodes, and the Session ID (if a request is part of an established session).
mod_advertise.so
The Proxy Advertisement Module broadcasts the existence of the proxy server via UDP multicast messages. The server advertisement messages contain the IP address and port number where the proxy is listening for responses from nodes that want to join the load-balancing cluster.
see Section 5.1.1, “Apache Modules” for detailed information about the available modules including user-configurable parameters.
Worker Node Components
Worker Node Service
The mod_cluster offered as part of JBoss Enterprise Web Server is used instead of a JBoss HTTP Connector client service for the worker node service. The mod_cluster in JBoss Enterprise Web Server consists of the following two parts:
  • Native - Apache HTTP Server module (balancer logic)
  • Java - Tomcat6/7 Listener library (worker logic)

4.4. Limitations

The JBoss HTTP Connector mod_cluster uses shared memory to keep the nodes description, the shared memory is created at the startup of httpd and the structure of each item is fixed. Therefore, when defining proxy server and worker node properties, make sure to follow these character limits:
  • Maximum Alias length: 100 characters (Alias corresponds to the network name of the respective virtual host; the name is defined in the Host element)
  • Maximum context length: 40 characters (for example, if myapp.war is deployed in /myapp, then /myapp is the context)
  • Maximum balancer name length: 40 characters (the balancer property in mbean)
  • Maximum JVMRoute string length: 80 character (JVMRoute in the <Engine> element)
  • Maximum domain name length: 20 characters (the domain property in mbean)
  • Maximum hostname length for a node: 64 characters (hostname address in the <Connector> element)
  • Maximum port length for a node: 7 characters (8009 is 4 characters, the port property in the <Connector> element)
  • Maximum scheme length for a node: 6 characters (possible values are http, https, ajp, the protocol of the connector)
  • Maximum cookie name length: 30 characters (the header cookie name for session ID default value: JSESSIONID from org.apache.catalina.Globals.SESSION_COOKIE_NAME)
  • Maximum path name length: 30 characters (the parameter name for the session ID default value: JSESSIONID from org.apache.catalina.Globals.SESSION_PARAMETER_NAME)
  • Maximum length of a session ID: 120 characters (session ID resembles the following: BE81FAA969BF64C8EC2B6600457EAAAA.node01)

Chapter 5. Proxy Server Components

5.1. Apache Modules

5.1.1. Apache Modules

Read this section for expanded definitions of the Apache proxy server modules discussed in Section 4.3, “Components” .

5.1.2. mod_manager.so

The Cluster Manager module, mod_manager, receives and acknowledges messages from nodes, including worker node registrations, worker node load data, and worker node application life cycle events.
LoadModule manager_module modules/mod_manager.so
Configurable directives in the <VirtualHost> element are as follows:
EnableMCPMReceive
Allows the VirtualHost to receive the mod_cluster Protocol Message (MCPM) from nodes. Add one EnableMCPMRecieve attribute to the httpd configuration to allow mod_cluster to operate correctly. EnableMCPMRecieve must be added in the VirtualHost configuration, at the location where advertise is configured.
MaxMCMPMaxMessSize
Defines the maximum size of mod_cluster Management Protocol (MCMP) messages. The default value for this is calculate from other Max directives. The minumum value for this is 1024.
AllowDisplay
Toggles the additional display on the mod_cluster-manager main page. The default value is off which causes only the versions to display on the mod_cluster-manager main page.
AllowCmd
Toggles permissions for commands using mod_cluster-manager URL. The default value is on, which allows commands.
ReduceDisplay
Toggles the reduction of information displayed on the mod_cluster-manager page. Reducing the information allows more nodes to display on the page. The default value is off which allows all the available information to display.
MemManagerFile
Defines the location for the files in which mod_manager stores configuration details. mod_manager also uses this location for generated keys for shared memory and lock files. This must be an absolute path name. It is recommended that this path be on a local drive, and not a NFS share. The default value is /logs/.
Maxcontext
The maximum number of contexts JBoss mod_cluster will use. The default value is 100.
Maxnode
The maximum number of worker nodes JBoss mod_cluster will use. The default value is 20.
Maxhost
The maximum number of hosts (aliases) JBoss mod_cluster will use. This is also the maximum number of load balancers. The default value is 10.
Maxsessionid
The maximum number of active session identifiers stored. A session is considered inactive when no information is received from that session within five minutes. The default value is 0, which disables this logic.
ManagerBalancerName
The name of the load balancer to use when the worker node does not provide a load balancer name. The default value is mycluster.
PersistSlots
When set to on, nodes, aliases and contexts are persisted in files. The default value is off.
CheckNonce
When set to on, session identifiers are checked to ensure that they are unique, and have not occurred before. The default is on.

Warning

Setting this directive to off can leave your server vulnerable to replay attacks.
SetHandler mod_cluster-manager
Defines a handler to display information about worker nodes in the cluster. This is defined in the Location element:
<Location $LOCATION>
  SetHandler mod_cluster-manager
  Order deny,allow
  Deny from all
  Allow from 127.0.0.1
</Location>
When accessing the $LOCATION defined in the Location element in your browser, you will see something like the following. (In this case, $LOCATION was also defined as mod_cluster-handler.)
Transferred corresponds to the POST data sent to the worker node. Connected corresponds to the number of requests that had been processed when this status page was requested. Sessions corresponds to the number of active sessions. This field is not present when Maxsessionid is 0.

5.1.3. mod_proxy_cluster.so

The Proxy Balancer Module, mod_proxy_cluster, handles the routing of requests to cluster nodes. The Proxy Balancer selects the appropriate node to forward the request to, based on application location in the cluster, current state of each of the cluster nodes, and the Session ID (if a request is part of an established session).
LoadModule proxy_cluster_module modules/mod_proxy_cluster.so
You can also define the following related directives in the <VirtualHost> element to change load balancing behavior.

mod_proxy_cluster directives

CreateBalancers
Defines how load balancers are created in the Apache HTTP Server virtual hosts. The following values are valid in CreateBalancers:
0
Create load balancers in all virtual hosts defined in Apache HTTP Server. Remember to configure the load balancers in the ProxyPass directive.
1
Do not create balancers. When using this value, you must also define the load balancer name in the ProxyPass or ProxyPassMatch.
2
Create only the main server. This is the default value for CreateBalancers.
UseAlias
Defines whether to check that the defined Alias corresponds to the ServerName. The following values are valid for UseAlias:
0
Ignore Alias information from worker nodes. This is the default value for UseAlias.
1
Verify that the defined alias corresponds to a worker node's server name.
LBstatusRecalTime
Defines the interval in seconds between the proxy calculating the status of a worker node. The default interval is 5 seconds.
ProxyPassMatch; ProxyPass
ProxyPass maps remote servers into the local server namespace. If the local server has an address http://local.com/, then the following ProxyPass directive would convert a local request for http://local.com/requested/file1 into a proxy request for http://worker.local.com/file1.
ProxyPass /requested/ http://worker.local.com/
ProxyPassMatch uses Regular Expressions to match local paths to which the proxied URL should apply.
For either directive, ! indicates that a specified path is local, and a request for that path should not be routed to a remote server. For example, the following directive specifies that .gif files should be served locally.
ProxyPassMatch ^(/.*\.gif)$ !

5.1.4. mod_advertise.so

The Proxy Advertisement Module, mod_advertise.so, broadcasts the existence of the proxy server via UDP multicast messages. The server advertisement messages contain the IP address and port number where the proxy is listening for responses from nodes that wish to join the load-balancing cluster.
This module must be defined alongside mod_manager in the VirtualHost element. Its identifier in the following code snippet is advertise_module.
LoadModule advertise_module modules/mod_advertise.so
mod_advertise also takes the following directives:
ServerAdvertise
Defines how the advertising mechanism is used.
When set to On, the advertising mechanism is used to tell worker nodes to send status information to this proxy. You can also specify a hostname and port with the following syntax: ServerAdvertise On http://hostname:port/. This is only required when using a name-based virtual host, or when a virtual host is not defined.
The default value is Off. When set to off, the proxy does not advertise its location.
AdvertiseGroup
Defines the multicast address to advertise on. The syntax is AdvertiseGroup address:port, where address should correspond to AdvertiseGroupAddress, and port should correspond to AdvertisePort in your worker nodes.
If your worker node is JBoss Enterprise Application Platform-based, and the -u switch is used at startup, the default AdvertiseGroupAddress is the value passed via the -u switch.
The default value is 224.0.1.105:23364. If port is not specified, the default port specified is 23364.
AdvertiseFrequency
The interval (in seconds) between multicast messages advertising the IP address and port. The default value is 10.
AdvertiseSecurityKey
Defines a string used to identify the JBoss HTTP Connector mod_cluster in JBoss Web. By default this directive is not set and no information is sent.
AdvertiseManagerUrl
Defines the URL that the worker node should use to send information to the proxy server. By default this directive is not set and no information is sent.
AdvertiseBindAddress
Defines the address and port over which to send multicast messages. The syntax is AdvertiseBindAddress address:port. This allows an address to be specified on machines with multiple IP addresses. The default value is 0.0.0.0:23364.

5.1.5. mod_proxy.so

A standard Apache HTTP Server module. This module lets the server act as proxy for data transferred over AJP (Apache JServe Protocol), FTP (File Transfer Protocol), CONNECT (for SSL, the Secure Sockets Layer), and HTTP (Hyper Text Transfer Protocol). This module does not require additional configuration. Its identifier is proxy_module.
Mod_proxy directives such as ProxyIOBufferSize are used to configure mod_cluster.

5.1.6. mod_proxy_ajp.so

A standard Apache HTTP Server module that provides support for AJP (Apache JServe Protocol) proxying. Mod_proxy.so is required to use this module.

5.1.7. mod_slotmem.so

Mod_slotmen does not require any configuration directives.

5.2. Proxy Server Components Installation and Default Configuration

5.2.1. Proxy Server Components Installation and Default Configuration

In JBoss Enterprise Web Server 2.1, mod_cluster is configured correctly for httpd by default. See the Chapter 6, Configure Basic Proxy Server to set a custom configuration.
For more information on configuring Tomcat worker node with mod_cluster, see Section 7.2.1, “Configure a Tomcat Worker Node”.

Chapter 6. Configure Basic Proxy Server

6.1. Basic Proxy Configuration Overview

Proxy server configuration consists of one mandatory and one optional portion:
  1. Configure a Proxy Server listener to receive worker node connection requests and worker node feedback.
  2. Optional: Disable server advertisement.
Server Advertisement

The proxy server advertises itself using UDP multicast. When UDP multicast is available on the network between the proxy server and the worker nodes, Server Advertisement adds worker nodes without further configuration required on the proxy server, and minimal configuration on the worker nodes.

If UDP multicast is not available or undesirable, configure the worker nodes with a static list of proxy servers, as detailed in Section 8.1, “Static Proxy Configuration”. In either case, the proxy server does not need to be configured with a list of worker nodes.

6.2. Configure a Load-balancing Proxy Using the HTTP Connector

Prerequisites

The following are a series of prerequisites for this procedure:

Procedure 6.1. Configure a Load-balancing Proxy Using the HTTP Connector

To configure JBoss Enterprise Web Server service to act as a load-balancing proxy using the JBoss HTTP Connector:
  1. Create a Listen Directive for the Proxy Server

    Edit the configuration file JBOSS_EWS_DIST/httpd/conf.d/JBoss_HTTP.conf and add the following:
    Listen IP_ADDRESS:PORT_NUMBER
    Where IP_ADDRESS is the IP address of a server network interface to communicate with the worker nodes, and PORT_NUMBER is the port on that interface to listen on.

    Note

    The port PORT_NUMBER must be open on the server firewall for incoming TCP connections.

    Example 6.1. Example Listen Directive

    Listen 10.33.144.3:6666
  2. Create Virtual Host

    Add the following to the file JBOSS_EWS_DIST/httpd/conf.d/JBoss_HTTP.conf:
    <VirtualHost IP_ADDRESS:PORT_NUMBER> 
    
       <Directory />
          Order deny,allow
          Deny from all
          Allow from 10.33.144.
       </Directory>
    
       KeepAliveTimeout 60
       MaxKeepAliveRequests 0
    
       ManagerBalancerName mycluster
       AdvertiseFrequency 5
       EnableMCPMReceive On
    
    </VirtualHost>
    Where IP_ADDRESS and PORT_NUMBER are the values from the Listen directive.
  3. Optional: Disable Server Advertisement

    The AdvertiseFrequency directive, set to five seconds, makes the server to periodically send server advertisement messages via UDP multicast.
    These server advertisement messages contain the IP_ADDRESS and PORT_NUMBER specified in the VirtualHost definition. Worker nodes configured to respond to server advertisements use this information to register themselves with the proxy server.
    To disable server advertisement, add the following directive to the VirtualHost definition:
    ServerAdvertise Off
    If server advertisements are disabled, or UDP multicast is not available on the network between the proxy server and the worker nodes, configure worker nodes with a static list of proxy servers. See Section 8.1, “Static Proxy Configuration” for directions.
  4. Restart the JBoss Enterprise Web Server Apache service

    See the JBoss Enterprise Web Server documentation for detailed directions.

Chapter 7. Install Node with Basic Configuration

7.1. Worker Node Requirements

Supported Worker Node types

  • JBoss Enterprise Web Server Tomcat service

Note

JBoss Enterprise Web Server Tomcat worker nodes support a subset of JBoss HTTP Connector functionality.

JBoss HTTP Connector Enterprise Web Server Node Limitations

  • Non-clustered mode only.
  • Only one load metric can be used at a time when calculating the load balance factor.

7.2. Install and Configure a Worker Node

7.2.1. Configure a Tomcat Worker Node

Follow this procedure to install the JBoss HTTP Connector on a JBoss Enterprise Web Server node and configure it for non-clustered operation.

Note

The supplied instructions are valid for Tomcat version 6 and 7 and do not require any changes for either of these versions.
Prerequisites

The following is a list of prerequisites for this task:

  • Install a supported JBoss Enterprise Web Server.
  • Understand the Proxy Configuration parameters discussed in Appendix B: Java Properties Reference.

Procedure 7.1. Configure a Worker Node for Tomcat

  1. Add a Listener to Tomcat

    Add the following Listener element beneath the other Listener elements in JBOSS_EWS_DIST/tomcat6/conf/server.xml.
    <Listener className="org.jboss.modcluster.container.catalina.standalone.ModClusterListener" advertise="true" stickySession="true" stickySessionForce="false" stickySessionRemove="true" />
  2. Give the Worker a Unique Identity

    Edit JBOSS_EWS_DIST/tomcat6/conf/server.xml and add a jvmRoute attribute and value to the Engine element, as shown:
    <Engine name="Catalina" defaultHost="localhost" jvmRoute="worker01">
  3. Optional Step: Configure Firewall for Proxy Server Advertisements

    A proxy server using the JBoss HTTP Connector can advertise itself via UDP multicast. To receive these multicast messages, open port 23364 for UDP connections on the worker node's firewall.
    For Linux Users
    /sbin/iptables -A INPUT -m state --state NEW -m udp -p udp --dport
     23364 -j ACCEPT 
    -m comment -comment "receive mod_cluster proxy server advertisements"
    If Automatic Proxy Discovery is not used, configure worker nodes with a static list of proxies. In this case you can safely ignore the following warning message:
    [warning] mod_advertise: ServerAdvertise Address or Port not defined, Advertise disabled!!!
  4. Set STATUS MCMP Messages frequency

    The tomcat worker nodes send status messages periodically to Apache HTTP Server balancer containing their current load status. The default frequency of these messages is 10 seconds. With many active worker nodes which are in hundreds, the STATUS MCMP Messages might increase the traffic congestion on Apache HTTP Server network.
    To set the MCMP message frequency, modify the org.jboss.modcluster.container.catalina.status-frequency property. By default, the property accepts values in seconds*10. Example, value = 1 means 10 seconds.
    -Dorg.jboss.modcluster.container.catalina.status-frequency=6
Apache HTTP server advertising is available via UDP multicast. This is not available by default on most systems. To enable server advertising, configure the firewall settings as follows to allow multicast UDP on the required ports:

Procedure 7.2. Configuring Firewall on Microsoft Windows using PowerShell

  1. Switch off firewall for debug purpose to determine whether the current network behavior is related to the firewall configuration.
    Start-Process "$psHome\powershell.exe" -Verb Runas -ArgumentList '-command "NetSh Advfirewall set allprofiles state off"'
  2. Allow UDP connections on port 23364. For example:
    Start-Process "$psHome\powershell.exe" -Verb Runas -ArgumentList '-command "NetSh Advfirewall firewall add rule name="UDP Port 23364" dir=in  action=allow protocol=UDP localport=23364"'
    Start-Process "$psHome\powershell.exe" -Verb Runas -ArgumentList '-command "NetSh Advfirewall firewall add rule name="UDP Port 23364" dir=out action=allow protocol=UDP localport=23364"'

Procedure 7.3. Configure the Firewall on Red Hat Enterprise Linux 7 to Allow mod_cluster Advertising

  • To allow mod_cluster advertising on Red Hat Enterprise Linux 7, you must enable the UDP port in the firewal as follows:
    firewall-cmd --permanent --zone=public --add-port=23364/udp

    Note

    224.0.1.105:23364 is the default address and port for mod_cluster balancer advertising UDP multicast.

Chapter 8. Advanced Configuration

8.1. Static Proxy Configuration

Server advertisement allows worker nodes to dynamically discover and register themselves with proxy servers. If UDP broadcast is not available or server advertisement is disabled then worker nodes must be configured with a static list of proxy server addresses and ports.
Use the following procedure to configure a JBoss Enterprise Web Server worker node to operate with a static list of proxy servers.
Prerequisites

Ensure that the following prerequisites are satisfied before initiating the procedure.

Procedure 8.1. Configure Web Server Worker Node with Static Proxy List

  1. Disable Dynamic Proxy Discovery

    Edit the file JBOSS_EWS_DIST/tomcat6/conf/server.xml. and set the advertise property of the ModClusterListener to false:
  2. Define a mod_cluster listener

    Add a <Listener> element to the server.xml file.
    <Listener className="org.jboss.modcluster.container.catalina.standalone.ModClusterListener" advertise="false" stickySession="true" stickySessionForce="false" stickySessionRemove="true"/>
  3. Create a static proxy server list

    Add a comma separated list of proxies in the form of IP_ADDRESS:PORT as the proxyList property of the ModClusterListener <Listener> element.

    Example 8.1. Example Static Proxy List

    <Listener className="org.jboss.modcluster.container.catalina.standalone.ModClusterListener" advertise="false" stickySession="true" stickySessionForce="false" stickySessionRemove="true" proxyList="10.33.144.3:6666,10.33.144.1:6666"/>

Chapter 9. Configuring httpd for SSL connections

9.1. Configuring httpd for SSL connections

Procedure 9.1. To configure httpd for ssl connections:

  1. Install mod_ssl using the following command:
    # rpm -qa | grep mod_ssl
  2. Edit the HTTPD_HOME/conf.d/ssl.conf file and add ServerName, SSLCertificateFile, and SSLCertificateKeyFile
     Example SSL configuration
        <VirtualHost _default_:443>
    #ServerName www.example.com:443
    SSLCertificateFile /etc/pki/tls/certs/localhost.crt
    SSLCertificateKeyFile /etc/pki/tls/private/localhost.key
    
    1. ServerName must match the Common Name (CN) of the SSL certificate. If the ServerName does not match the CN, the client browsers display a message "domain mismatch".
    2. The SSLCertificateFile is the private key associate with the certificate (the public key).
    3. Verify that the Listen directive in the ssl.conf file is correct as per your setup. For example, if an IP address is specified, it must match the IP address the httpd service is bound to.
  3. Restart httpd using the following command:
    # service httpd restart

Part III. Online Certificate Status Protocol

Chapter 10. Online Certificate Status Protocol

10.1. About Online Certificate Status Protocol

Online Certificate Status Protocol is a technology which allows web browsers and web servers to communicate over a secured connection. In this the encrypted data is sent from one side and decrypted by the other side before processing. The web browser and the web server both encrypt and decrypt the data.
During communication with a web server, the server presents a set of credentials in the form of certificate. The browser then checks the certificate for its validity and sends a request for certificate status information. The server sends back a status as current, expired, or unknown. The certificate specifies syntax for communication and contains control information such as start time and end time, address information to access an OCSP responder. The web server can use an OCSP responder, it has been configured for, or the one listed in the certificate, to check the status. OCSP allows a grace period for expired certificates which allows access to a server for a limited time before renewing the certificate.
Online Certificate Status Protocol overcomes the limitation of older method, Certificate Revocation List (CRL). For more information on OCSP, see the Red Hat Certificate System Admin Guide and https://access.redhat.com/site/articles/417843.

10.2. Using Online Certificate Status Protocol for httpd

Before you use Online Certificate Status Protocol for https, ensure you have configured httpd for SSL connections (see Section 9.1, “Configuring httpd for SSL connections”) as prerequisite.
To use Online Certificate Status Protocol (OCSP) for httpd, ensure that Certificate Authority (CA) and OCSP Responder is configured properly.
For more information on how to configure CA, see the Managing Certificates and Certificate Authorities section in the Linux Domain Identity, Authentication, and Policy Guide.
For more information on how to configure OCSP Responder, see the Configuring OCSP Responders section in the Linux Domain Identity, Authentication, and Policy Guide.

Note

The Certificate Authority must be able to append the following attributes to the Certificate:
[ usr_cert ]
...
authorityInfoAccess=OCSP;URI:http://HOST:PORT 
...
[ v3_OCSP ]
basicConstraints = CA:FALSE
keyUsage = nonRepudiation, digitalSignature, keyEncipherment
extendedKeyUsage = OCSP Signing
Note that HOST and PORT will need to be changed to match your responder that you set up later.

10.3. Configure httpd to validate OCSP certificates

Before configuring httpd to validate OCSP certificates ensure Certificate Authority (CA) and OCSP Responder is configured properly. Follow this procedure to perform OCSP validation of client certificates:

Procedure 10.1. To httpd to validate OCSP certificates

  • Use the SSLOCSPEnable attribute to enable OCSP validation.
    # Require valid client certificates (mutual auth)
      SSLVerifyClient require
      SSLVerifyDepth  3
      # Enable OCSP
      SSLOCSPEnable on
      SSLOCSPDefaultResponder http://10.10.10.25:3456
      SSLOCSPOverrideResponder on

10.4. Verify the configuration

You can use OpenSSL to verify your configuration.
# openssl ocsp -issuer cacert.crt -cert client.cert -url http://HOST:PORT -CA ocsp_ca.cert -VAfile ocsp.cert

Note

  • -issuer is the Certificate Authority certificate.
  • -cert is the Client certificate which you want to verify.
  • -url is the http server validating Certificate (OCSP).
  • -CA is the CA certificate for verifying the httpd server certificate.
  • -VAfile is the OCSP Responder certificate.

Part IV. Using Jsvc with JBoss Enterprise Web Server

Chapter 11. Jsvc

11.1. About Jsvc

Jsvc is a set of libraries and applications that facilitates running Java applications on UNIX. In the context of JBoss Enterprise Web Server 2.1, Jsvc allows Tomcat to switch identities. Using Jsvc, Tomcat can perform root user level operations and then revert to a non-privileged user. Jsvc is mainly used for running Tomcat as a service in the background.
For JBoss Enterprise Server 2.1, Jsvc files are available at the following locations:

For Red Hat Enterprise Linux:

  • jboss-ews-2.1/extras/jsvc
  • jboss-ews-2.1/tomcat6/bin/jsvc
  • jboss-ews-2.1/tomcat7/bin/jsvc

Note

Jsvc in Tomcat 6 and Tomcat 7 folders are symlinks to extra/jsvc in the following way:
./jboss-ews-2.1/tomcat6/bin/jsvc -> ../../extras/jsvc
./jboss-ews-2.1/tomcat7/bin/jsvc -> ../../extras/jsvc

For Solaris:

  • ./jboss-ews-2.1/sbin/jsvc
  • ./jboss-ews-2.1/share/apache-tomcat-6.0.41/bin/jsvc
  • ./jboss-ews-2.1/share/apache-tomcat-7.0.54/bin/jsvc

Note

Jsvc in apache-tomcat-6.0.41 and apache-tomcat-7.0.54 folders are symlinks to sbin/jsvc in the following way:
./jboss-ews-2.1/share/apache-tomcat-7.0.54/bin/jsvc -> ../../../sbin/jsvc
./jboss-ews-2.1/share/apache-tomcat-6.0.41/bin/jsvc -> ../../../sbin/jsvc

11.2. Use Jsvc with Tomcat 6 and 7

11.2.1. Run Jsvc with Tomcat 6 and 7

Run the following command to run Jsvc with Tomcat 6 and 7:
/opt/jboss-ews-2.1/share/tomcat<VERSION>/bin/daemon.sh start

11.2.2. Configure Jsvc with Tomcat 6 and 7

The following parameters can be set when running the daemon.sh script to run Jsvc:

Table 11.1.  daemon.sh configuration parameters used by Tomcat 6 and 7

Parameter Name Environment Variable Default Value Description
--java-home JAVA_HOME Based on the value of the PATH variable. The Java home directory location.
--catalina-home CATALINA_HOME Determined by the location of the script. The Tomcat installation directory location.
--catalina-base CATALINA_BASE Based on the value of the PATH variable. The directory that contains the specific configuration and set up information if multiple servers are using the same installation.
--catalina-pid - $CATALINA_BASE/logs/catalina-daemon.pid The file where the process ID (PID) for the running instance of Tomcat is stored.
--tomcat-user - tomcat The user Tomcat uses.
--service-start-wait-time - This is a wrapper to the --wait parameter. The --wait parameter accepts values in seconds.

Chapter 12. Working Examples

12.1. Complete Working Example

Following are a set of example configuration files for a complete working example.
Load Balancer

A proxy server listening on localhost:

LoadModule slotmem_module modules/mod_slotmem.so
LoadModule manager_module modules/mod_manager.so
LoadModule proxy_cluster_module modules/mod_proxy_cluster.so
LoadModule advertise_module modules/mod_advertise.so

Listen 127.0.0.1:6666
<VirtualHost 127.0.0.1:6666>

    <Directory />
        Order deny,allow
        Deny from all
        Allow from 127.0.0.1
    </Directory>

    KeepAliveTimeout 60
    MaxKeepAliveRequests 0

    ManagerBalancerName mycluster
    ServerAdvertise On
    AdvertiseFrequency 5
    EnableMCPMReceive On

<Location /mod_cluster-manager>
    SetHandler mod_cluster-manager
    Order deny,allow
    Deny from all
    Allow from 127.0.0.1
</Location>

</VirtualHost>
Worker Configuration for Tomcat

Edit the $CATALINA_HOME/conf/server.xml file and add the following listener element to configure the worker for Tomcat:

<Listener className="org.jboss.modcluster.container.catalina.standalone.ModClusterListener" advertise="true"/>
Example iptables Firewall Rules

Following are a set of example firewall rules using iptables, for a cluster node on the 192.168.1.0/24 subnet.

/sbin/iptables -I INPUT 5 -p udp -d 224.0.1.0/24 -j ACCEPT -m comment --comment "mod_cluster traffic"
/sbin/iptables -I INPUT 6 -p udp -d 224.0.0.0/4 -j ACCEPT -m comment --comment "JBoss Cluster traffic"
/sbin/iptables -I INPUT 9 -p udp -s 192.168.1.0/24 -j ACCEPT -m comment --comment "cluster subnet for inter-node communication"
/sbin/iptables -I INPUT 10 -p tcp -s 192.168.1.0/24 -j ACCEPT -m comment --comment "cluster subnet for inter-node communication"
/etc/init.d/iptables save

12.2. mod_auth_kerb Example

12.2.1. About the mod_auth_kerb Example

Use the prerequisites and subsequent procedure for a basic example about configuring and running Kerberos authentication with JBoss Enterprise Web Server's httpd and mod_auth_kerb on Red Hat Enterprise Linux.

12.2.2. mod_auth_kerb Example Prerequisites

The following is a list of prerequisites for the working example. Ensure that all prerequisites are met before attempting to use the example instructions.
  • Install mod_auth_kerb on Red Hat Enterprise Linux.
  • Install curl with GSS-negotiated support.
  • Configure and run a Kerberos or LDAP server (for example ApacheDS) on the same host as your JBoss Enterprise Web Server.
  • Create the following LDAP users:
    • Create the user krbtgt:
      dn: uid=krbtgt,ou=Users,dc=example,dc=com
      objectClass: top
      objectClass: person
      objectClass: inetOrgPerson
      objectClass: krb5principal
      objectClass: krb5kdcentry
      cn: KDC Service
      sn: Service
      uid: krbtgt
      userPassword: secret
      krb5PrincipalName: krbtgt/EXAMPLE.COM@EXAMPLE.COM
      krb5KeyVersionNumber: 0
    • Create the user ldap:
      dn: uid=ldap,ou=Users,dc=example,dc=com
      objectClass: top
      objectClass: person
      objectClass: inetOrgPerson
      objectClass: krb5principal
      objectClass: krb5kdcentry
      cn: LDAP
      sn: Service
      uid: ldap
      userPassword: randall
      krb5PrincipalName: ldap/localhost@EXAMPLE.COM
      krb5KeyVersionNumber: 0
    • Create the user HTTP:
      dn: uid=HTTP,ou=Users,dc=example,dc=com
      objectClass: top
      objectClass: person
      objectClass: inetOrgPerson
      objectClass: krb5principal
      objectClass: krb5kdcentry
      cn: HTTP
      sn: Service
      uid: HTTP
      userPassword: secretpwd
      krb5PrincipalName: HTTP/localhost@EXAMPLE.COM
      krb5KeyVersionNumber: 0
    • Create user hnelson (test user):
      dn: uid=hnelson,ou=Users,dc=example,dc=com
      objectClass: top
      objectClass: person
      objectClass: inetOrgPerson
      objectClass: krb5principal
      objectClass: krb5kdcentry
      cn: Horatio Nelson
      sn: Nelson
      uid: hnelson
      userPassword: secret
      krb5PrincipalName: hnelson@EXAMPLE.COM
      krb5KeyVersionNumber: 0

12.2.3. Configure the Kerberos Client

Use the following procedure to configure a Kerberos client for testing purposes:

Procedure 12.1. Configure the Kerberos Client

  1. Create the Kerberos Configuration File

    Create the krb5.conf configuration file in the /etc directory and add the following to the file:
    [logging]
      default = FILE:/var/log/krb5libs.log
      kdc = FILE:/var/log/krb5kdc.log
      admin_server = FILE:/var/log/kadmind.log
    
    [libdefaults]
      default_realm = EXAMPLE.COM
      default_tgs_enctypes = des-cbc-md5,des3-cbc-sha1-kd
      default_tkt_enctypes = des-cbc-md5,des3-cbc-sha1-kd
      dns_lookup_realm = false
      dns_lookup_kdc = false
      allow_weak_crypto = yes
      ticket_lifetime = 24h
      renew_lifetime = 7d
      forwardable = yes  
    
    [realms]
      EXAMPLE.COM = {
        kdc = localhost:60088
        admin_server = localhost:60088
      }
    
    [domain_realm]
      .example.com = EXAMPLE.COM
      example.com = EXAMPLE.COM
  2. Create a Key Tab

    Create a key tab in the /etc/httpd folder with the following contents:
    ktutil
    ktutil:  addent -password -p HTTP/localhost@EXAMPLE.COM -k 0 -e des-cbc-md5
    Password for HTTP/localhost@EXAMPLE.COM: secretpwd
    ktutil:  list
    slot KVNO Principal
    ---- ---- ---------------------------------------------------------------------
       1    0               HTTP/localhost@EXAMPLE.COM
    ktutil:  wkt krb5.keytab
    ktutil:  quit
    
    Under root user:
    chgrp apache /etc/httpd/krb5.keytab
    chmod 640 /etc/httpd/krb5.keytab
  3. Check the Hosts File

    Ensure that the following host configuration is included in the /etc/hosts file:
    127.0.0.1 localhost

12.2.4. Configure mod_auth_kerb

Use the following procedure to configure mod_auth_kerb. As a prerequisite, ensure that the Kerberos Client is configured (see Section 12.2.3, “Configure the Kerberos Client”).

Procedure 12.2. Configure mod_auth_kerb

  • Create the auth_kerb.conf configuration file in the /etc/httpd/conf.d/ folder and add the following information to the file:
    #
    # The mod_auth_kerb module implements Kerberos authentication over
    # HTTP, following the "Negotiate" protocol.
    # 
    
    LoadModule auth_kerb_module modules/mod_auth_kerb.so
    
    <Location /kerberostest>
    #  SSLRequireSSL
      AuthType Kerberos
      AuthName "Kerberos Login"
      KrbMethodNegotiate On
      KrbMethodK5Passwd Off
      KrbAuthRealms EXAMPLE.COM
      KrbServiceName HTTP
      Krb5KeyTab /etc/httpd/krb5.keytab
      require valid-user
    </Location>

12.2.5. Test the Kerberos Authentication

Use the following instructions to test the Kerberos authentication. As a prerequisite for this procedure, ensure that the Kerberos Client is configured (see Section 12.2.3, “Configure the Kerberos Client”).

Procedure 12.3. Test the Kerberos Authentication

  1. Create a Test Page

    Create a test page named auth_kerb_page.html in the $EWS_HOME/httpd/www/html/kerberostest/.
  2. Add the Contents of the Test Page

    Add the following contents to the test page (auth_kerb_page.html:
    <html>
      <body>
        <h1>mod_auth_kerb successfully authenticated!</h1>
    </body>
    </html>
  3. Optional: Set Log Level

    Optionally, set the log level for debugging in the $EWS_HOME/httpd/conf/httpd.conf file.
  4. Start httpd

    As the root user, start the JBoss Enterprise Web Server httpd as follows:
    # $EWS_HOME/httpd/sbin/apachectl start
  5. Test Authentication

    Test the authentication as follows:
    1. Initiate Kerberos authentication for the test user hnelson:
      $ kinit hnelson
    2. View the details for the test user hnelson:
      $ klist
      A result similar to the following appears:
      Ticket cache: FILE:/tmp/krb5cc_18602
      Default principal: hnelson@EXAMPLE.COM
      
      Valid starting     Expires            Service principal
      06/03/13 14:21:13  06/04/13 14:21:13  krbtgt/EXAMPLE.COM@EXAMPLE.COM
              renew until 06/10/13 14:21:13
    3. Testing httpd Kerberos Authentication

      Test httpd Kerberos authentication as follows:
      $ curl --negotiate -u : http://localhost/kerberostest/auth_kerb_page.html
      If working correctly, the following result appears:
      <html>
        <body>
          <h1>mod_auth_kerb successfully authenticated!</h1>
      </body>
      </html>
See http://modauthkerb.sourceforge.net/ for more information about mod_auth_kerb.

Appendix A. Reference

A.1. workers.properties

Apache httpd Server worker nodes are Servlet containers that are mapped to the mod_jk load balancer. The worker nodes are defined in HTTPD_DIST/conf/workers.properties. This file specifies where the different Servlet containers are located, and how calls should be load-balanced across them.
The workers.properties file contains two sections:
Global Properties
This section contains directives that apply to all workers.
Worker Properties
This section contains directives that apply to each individual worker.
Each node is defined using the Worker Properties naming convention. The worker name can only contain alphanumeric characters, limited to [a-z][A-Z][0-9][_-/].
The structure of a Worker Property is worker.worker_name.directive
worker
The constant prefix for all worker properties.
worker_name
The arbitrary name given to the worker. For example: node1, node_01, Node_1.
directive
The specific directive required.
The main directives required to configure worker nodes are described below.

Note

For the full list of worker.properties configuration directives, refer directly to the Apache Tomcat Connector - Reference Guide

worker.properties Global Directives

worker.list
Specifies the list of worker names used by mod_jk. The workers in this list are available to map requests to.

Note

A single node configuration, which is not managed by a load balancer, must be set to worker.list=[worker name].

workers.properties Mandatory Directives

type
Specifies the type of worker, which determines the directives applicable to the worker. The default value is ajp13, which is the preferred worker type to select for communication between the web server and Apache httpd Server.
Other values include ajp14, lb, status.
For detailed information about ajp13, see The Apache Tomcat Connector - AJP Protocol Reference

workers.properties Connection Directives

host
The hostname or IP address of the worker. The worker node must support the ajp13 protocol stack. The default value is localhost.
You can specify the port directive as part of the host directive by appending the port number after the hostname or IP address. For example: worker.node1.host=192.168.2.1:8009 or worker.node1.host=node1.example.com:8009
port
The port number of the remote server instance listening for defined protocol requests. The default value is 8009, which is the default listen port for AJP13 workers. If you are using AJP14 workers, this value must be set to 8011.
ping_mode
Specifies the conditions under which connections are probed for their current network health.
The probe uses an empty AJP13 packet for the CPing, and expects a CPong in return, within a specified timeout.
You specify the conditions by using a combination of the directive flags. The flags are not comma-separated. For example, a correct directive flag set is worker.node1.ping_mode=CI
C (connect)
Specifies the connection is probed once after connecting to the server. You specify the timeout using the connect_timeout directive, otherwise the value for ping_timeout is used.
P (prepost)
Specifies the connection is probed before sending each request to the server. You specify the timeout using the prepost_timeout directive, otherwise the value for ping_timeout is used.
I (interval)
Specifies the connection is probed during regular internal maintenance cycles. You specify the idle time between each interval using the connection_ping_interval directive, otherwise the value for ping_timeout is used.
A (all)
The most common setting, which specifies all directive flags are applied. For information about the *_timeout advanced directives, refer directly to Apache Tomcat Connector - Reference Guide.
ping_timeout
Specifies the time to wait for CPong answers to a CPing connection probe (see ping_mode). The default value is 10000 (milliseconds).

worker.properties Load Balancing Directives

lbfactor
Specifies the load-balancing factor for an individual worker, and is only specified for a member worker of a load balancer.
This directive defines the relative amount of HTTP request load distributed to the worker compared to other workers in the cluster.
A common example where this directive applies is where you want to differentiate servers with greater processing power than others in the cluster. For example, if you require a worker to take three times the load than other workers, specify worker.worker name.lbfactor=3
balance_workers
Specifies the worker nodes that the load balancer must manage. The directive can be used multiple times for the same load balancer, and consists of a comma-separated list of worker names as specified in the workers.properties file.
sticky_session
Specifies whether requests for workers with SESSION IDs are routed back to the same worker. The default is 0 (false). When set to 1 (true), load balancer persistence is enabled.
For example, if you specify worker.loadbalancer.sticky_session=0, each request is load balanced between each node in the cluster. In other words, different requests for the same session will go to different servers based on server load.
If worker.loadbalancer.sticky_session=1, each session is persisted (locked) to one server until the session is terminated, providing that server is available.

Appendix B. Java Properties Reference

B.1. Proxy Configuration

The configuration values are sent to proxies under the following conditions:
  • During server startup
  • When a proxy is detected through the advertise mechanism
  • During error recovery, when a proxy's configuration is reset.

Table B.1. Proxy Configuration Values for Tomcat

Value Default Description
stickySession true Specifies whether subsequent requests for a given session should be routed to the same node, if possible.
stickySessionRemove false Specifies whether the httpd proxy should remove session stickiness if the balancer is unable to route a request to the node to which it is stuck. This property is ignored if stickySession is false.
stickySessionForce true Specifies whether the httpd proxy should return an error if the balancer is unable to route a request to the node to which it is stuck. This property is ignored if stickySession is false.
workerTimeout -1 Specifies the number of seconds to wait for a worker to become available to handle a request. When all the workers of a balancer are unusable, mod_cluster will retry after a while (workerTimeout/100) to find an usable worker. A value of -1 indicates that the httpd will not wait for a worker to be available and will return an error if no workers are available.
maxAttempts 1 Specifies the number of times the httpd proxy will attempt to send a given request to a worker before aborting. The minimum value is 1: try once before aborting.
flushPackets false Specifies whether packet flushing is enabled or disabled.
flushWait -1 Specifies the time to wait before flushing packets. A value of -1 means wait forever.
ping 10 Time to wait (in seconds) for a pong answer to a ping.
smax - Specifies the soft maximum idle connection count. The maximum value is determined by the httpd thread configuration (ThreadsPerChild or 1).
ttl 60 Specifies the time (in seconds) idle connections persist, above the smax threshold.
nodeTimeout -1 Specifies the time (in seconds) mod_cluster waits for the back-end server response before returning an error. mod_cluster always uses a cping/cpong before forwarding a request. The connectiontimeout value used by mod_cluster is the ping value.
balancer mycluster Specifies the name of the load-balancer.
loadBalancingGroup - Specifies the load balancing among jvmRoutes within the same load balancing group. A loadBalancingGroup is conceptually equivalent to a mod_jk domain directive.

B.2. mod_cluster Proxy and Proxy Discovery Configuration Attributes

The following tables contain attributes and information about mod_cluster proxy and proxy discovery configuration attributes.

Table B.2. mod_cluster Proxy Discovery Configuration Attributes

Attribute Property Default Value
proxy-list proxyList -
proxy-url proxyURL -
advertise advertise true
advertise-security-key advertiseSecurityKey -
excluded-contexts excludedContexts -
auto-enable-contexts autoEnableContexts true
stop-context-timeout stopContextTimeout 10 seconds (in seconds)
socket-timeout nodeTimeout 20 seconds (in milliseconds)

Note

When nodeTimeout is not defined the ProxyTimeout directive, Proxy is used. If ProxyTimeout is not defined the server timeout (Timeout) is used (default 300 seconds). nodeTimeout, ProxyTimeout or Timeout is set at the socket level.

Table B.3. mod_cluster Proxy Configuration Attributes

Attribute Property Default Value
sticky-session stickySession true
sticky-session-remove stickySessionRemove false
sticky-session-force stickySessionForce true
node-timeout workerTimeout -1
max-attempts maxAttempts 1
flush-packets flushPackets false
flush-wait flushWait -1
ping ping 10
smax smax -1 (uses the default value)
ttl ttl -1 (uses the default value)
domain loadBalancingGroup -
load-balancing-group loadBalancingGroup -

B.3. Load Configuration

The following table contains additional configuration properties that are used when mod_cluster is configured with Tomcat:

Table B.4. Load Configuration for Tomcat

Attribute Default Value Description
loadMetricClass org.jboss.modcluster.load.metric.impl.BusyConnectorsLoadMetric This is the class name of an object that is implementing org.jboss.load.metric.LoadMetric.
loadMetricCapacity 1 This is the capacity of the load metric defined via the loadMetricClass property.
loadHistory 9 This is the number of historic load values that must be considered in the load balance factor computation.
loadDecayFactor 2 This is the factor by which the historic load values decrease in significance.

Appendix C. Revision History

Revision History
Revision 3.0.0-4.1Wed Feb 11 2015Lucas Costi
Updated the Product Name to reflect the new name grouping for the product. No update was made to details in the guide.
Revision 3.0.0-4Tue Oct 07 2014Husnain Husnain Ali
BZ-1122076: Added a note for clarity.
Revision 3.0.0-3Tue Aug 19 2014Rakesh Ghatvisave
BZ-1130972: Commented out EAP subsystem topics.
BZ-1131120: Removed proxy config EAP table, domain attribute and fixed a typo.
Revision 3.0.0-2Mon Jul 14 2014Mandar Joshi
BZ-1115148: Updated version number from 2.0 to 2.1 in the guide.
BZ-1072114: Added topics for using WebSocket on tomcat
BZ-1115956: Ensured that we have consistent style in the docs
BZ-1089687: Corrected the typo in section "Static Proxy Configuration."
BZ-1032725: Updated the property name in table B.1

Legal Notice

Copyright © 2015 RedHat, Inc.
This document is licensed by Red Hat under the Creative Commons Attribution-ShareAlike 3.0 Unported License. If you distribute this document, or a modified version of it, you must provide attribution to Red Hat, Inc. and provide a link to the original. If the document is modified, all Red Hat trademarks must be removed.
Red Hat, as the licensor of this document, waives the right to enforce, and agrees not to assert, Section 4d of CC-BY-SA to the fullest extent permitted by applicable law.
Red Hat, Red Hat Enterprise Linux, the Shadowman logo, JBoss, OpenShift, Fedora, the Infinity logo, and RHCE are trademarks of Red Hat, Inc., registered in the United States and other countries.
Linux® is the registered trademark of Linus Torvalds in the United States and other countries.
Java® is a registered trademark of Oracle and/or its affiliates.
XFS® is a trademark of Silicon Graphics International Corp. or its subsidiaries in the United States and/or other countries.
MySQL® is a registered trademark of MySQL AB in the United States, the European Union and other countries.
Node.js® is an official trademark of Joyent. Red Hat Software Collections is not formally related to or endorsed by the official Joyent Node.js open source or commercial project.
The OpenStack® Word Mark and OpenStack logo are either registered trademarks/service marks or trademarks/service marks of the OpenStack Foundation, in the United States and other countries and are used with the OpenStack Foundation's permission. We are not affiliated with, endorsed or sponsored by the OpenStack Foundation, or the OpenStack community.
All other trademarks are the property of their respective owners.