- Introduction
- Available Interceptors
- Static Membership
- Attributes
- Common Attributes
- org.apache.catalina.tribes.group.interceptors.DomainFilterInterceptor Attributes
- org.apache.catalina.tribes.group.interceptors.MessageDispatchInterceptor Attributes
- org.apache.catalina.tribes.group.interceptors.TcpFailureDetector Attributes
- org.apache.catalina.tribes.group.interceptors.TcpPingInterceptor Attributes
- org.apache.catalina.tribes.group.interceptors.ThroughputInterceptor Attributes
- Nested Components
The Channel Interceptor object
Table of Contents
Introduction
Apache Tribes supports an interceptor architecture to intercept both messages and membership notifications. This architecture allows decoupling of logic and opens the way for some very kewl feature add ons.
Available Interceptors
org.apache.catalina.tribes.group.interceptors.TcpFailureDetector
org.apache.catalina.tribes.group.interceptors.ThroughputInterceptor
org.apache.catalina.tribes.group.interceptors.MessageDispatchInterceptor
org.apache.catalina.tribes.group.interceptors.NonBlockingCoordinator
org.apache.catalina.tribes.group.interceptors.OrderInterceptor
org.apache.catalina.tribes.group.interceptors.SimpleCoordinator
org.apache.catalina.tribes.group.interceptors.StaticMembershipInterceptor
org.apache.catalina.tribes.group.interceptors.TwoPhaseCommitInterceptor
org.apache.catalina.tribes.group.interceptors.DomainFilterInterceptor
org.apache.catalina.tribes.group.interceptors.FragmentationInterceptor
org.apache.catalina.tribes.group.interceptors.GzipInterceptor
org.apache.catalina.tribes.group.interceptors.TcpPingInterceptor
Static Membership
In addition to dynamic discovery, Apache Tribes also supports static membership, with membership verification.
To achieve this add the org.apache.catalina.tribes.group.interceptors.StaticMembershipInterceptor
underneath the org.apache.catalina.tribes.group.interceptors.TcpFailureDetector
interceptor.
Inside the StaticMembershipInterceptor
you can add the static members you wish to have.
The TcpFailureDetector
will do a health check on the static members,and also monitor them for crashes
so they will have the same level of notification mechanism as the members that are automatically discovered.
<Interceptor className="org.apache.catalina.tribes.group.interceptors.StaticMembershipInterceptor">
<LocalMember className="org.apache.catalina.tribes.membership.StaticMember"
port="4000"
securePort="-1"
host="tomcat01.mydomain.com"
domain="staging-cluster"
uniqueId="{1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,1}"/>
<Member className="org.apache.catalina.tribes.membership.StaticMember"
port="5678"
securePort="-1"
host="tomcat01.mydomain.com"
domain="staging-cluster"
uniqueId="{0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15}"/>
</Interceptor>
Attributes
Common Attributes
Attribute | Description |
---|---|
className | Required, as there is no default |
optionFlag |
If you want the interceptor to trigger on certain message depending on the message's option flag,
you can setup the interceptors flag here.
The default value is 0 , meaning this interceptor will trigger on all messages.
|
org.apache.catalina.tribes.group.interceptors.DomainFilterInterceptor Attributes
Attribute | Description |
---|---|
domain |
The logical cluster domain that this Interceptor accepts.
Two different type of values are possible: 1. Regular string values like "staging-domain" or "tomcat-cluster" will be converted into bytes using ISO-8859-1 encoding. 2. byte array in string form, for example {216,123,12,3} |
org.apache.catalina.tribes.group.interceptors.MessageDispatchInterceptor Attributes
Attribute | Description |
---|---|
optionFlag |
The default and hard coded value is 8 (org.apache.catalina.tribes.Channel.SEND_OPTIONS_ASYNCHRONOUS) .
The dispatcher will trigger on this value only, as it is predefined by Tribes.
|
alwaysSend |
What behavior should be executed when the dispatch queue is full. If true (default), then the message is
is sent synchronously, if false an error is thrown.
|
maxQueueSize |
Size in bytes of the dispatch queue, the default value is 1024*1024*64 (64MB) sets the maximum queue size for the dispatch queue
if the queue fills up, one can trigger the behavior, if alwaysSend is set to true, the message will be sent synchronously
if the flag is false, an error is thrown
|
maxThreads | The maximum number of threads in this pool, default is 10. |
maxSpareThreads | The number of threads to keep in the pool, default is 2. |
keepAliveTime | Maximum number of milliseconds of until Idle thread terminates. Default value is 5000(5 seconds). |
org.apache.catalina.tribes.group.interceptors.TcpFailureDetector Attributes
Attribute | Description |
---|---|
connectTimeout | Specifies the timeout, in milliseconds, to use when attempting a TCP connection to the suspect node. Default is 1000. |
performSendTest | If true is set, send a test message to the suspect node. Default is true. |
performReadTest |
If true is set, read the response of the test message that sent. Default is false.
Note: if performSendTest is false, this attribute will have no effect.
|
readTestTimeout | Specifies the timeout, in milliseconds, to use when performing a read test to the suspicious node. Default is 5000. |
removeSuspectsTimeout | The maximum time(in seconds) for remove from removeSuspects. Member of removeSuspects will be automatically removed after removeSuspectsTimeout. If a negative value specified, the removeSuspects members never be removed until disappeared really. If the attribute is not provided, a default of 300 seconds (5 minutes) is used. |
org.apache.catalina.tribes.group.interceptors.TcpPingInterceptor Attributes
Attribute | Description |
---|---|
interval | If useThread == true, defines the interval of sending a ping message. default is 1000 ms. |
useThread | Flag of whether to start a thread for sending a ping message. If set to true, this interceptor will start a local thread for sending a ping message. if set to false, channel heartbeat will send a ping message. default is false. |
org.apache.catalina.tribes.group.interceptors.ThroughputInterceptor Attributes
Attribute | Description |
---|---|
interval |
Defines the interval in number of messages when we are to report the throughput statistics.
The report is logged to the org.apache.juli.logging.LogFactory.getLog(ThroughputInterceptor.class)
logger under the INFO level.
Default value is to report every 10000 messages.
|
Nested Components
LocalMember:
Static member that is the local member of the static cluster group.
Member:
Static member that add to the static cluster group.
Nested element StaticMember Attributes
Attribute | Description |
---|---|
className |
Only one implementation available:org.apache.catalina.tribes.membership.StaticMember
|
port | The port that this static member listens to for cluster messages |
securePort |
The secure port this static member listens to for encrypted cluster messages
default value is -1 , this value means the member is not listening on a secure port
|
host |
The host (or network interface) that this static member listens for cluster messages.
Three different type of values are possible: 1. IP address in the form of "216.123.1.23" 2. Hostnames like "tomcat01.mydomain.com" or "tomcat01" as long as they resolve correctly 3. byte array in string form, for example {216,123,12,3} |
domain |
The logical cluster domain for that this static member listens for cluster messages.
Two different type of values are possible: 1. Regular string values like "staging-domain" or "tomcat-cluster" will be converted into bytes using ISO-8859-1 encoding. 2. byte array in string form, for example {216,123,12,3} |
uniqueId |
A universally uniqueId for this static member.
The values must be 16 bytes in the following form: 1. byte array in string form, for example {0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15} |