[jboss-cvs] JBossAS SVN: r100907 - projects/docs/enterprise/5.0/mod_cluster/en-US.
jboss-cvs-commits at lists.jboss.org
jboss-cvs-commits at lists.jboss.org
Fri Feb 12 00:50:27 EST 2010
Author: gmeintjes
Date: 2010-02-12 00:50:26 -0500 (Fri, 12 Feb 2010)
New Revision: 100907
Modified:
projects/docs/enterprise/5.0/mod_cluster/en-US/Book_Info.xml
projects/docs/enterprise/5.0/mod_cluster/en-US/ConfigurationForJBossAS.xml
projects/docs/enterprise/5.0/mod_cluster/en-US/MigratingFromModJK.xml
projects/docs/enterprise/5.0/mod_cluster/en-US/Revision_History.xml
Log:
Amend as per review JBPAPP-3273
Modified: projects/docs/enterprise/5.0/mod_cluster/en-US/Book_Info.xml
===================================================================
--- projects/docs/enterprise/5.0/mod_cluster/en-US/Book_Info.xml 2010-02-12 02:55:49 UTC (rev 100906)
+++ projects/docs/enterprise/5.0/mod_cluster/en-US/Book_Info.xml 2010-02-12 05:50:26 UTC (rev 100907)
@@ -7,7 +7,7 @@
<productname>JBoss Enterprise Application Platform</productname>
<productnumber>5.0</productnumber>
<edition>1</edition>
- <pubsnumber>2.0</pubsnumber>
+ <pubsnumber>2.1</pubsnumber>
<abstract>
<para>
mod_cluster is an httpd-based load balancer that uses a
Modified: projects/docs/enterprise/5.0/mod_cluster/en-US/ConfigurationForJBossAS.xml
===================================================================
--- projects/docs/enterprise/5.0/mod_cluster/en-US/ConfigurationForJBossAS.xml 2010-02-12 02:55:49 UTC (rev 100906)
+++ projects/docs/enterprise/5.0/mod_cluster/en-US/ConfigurationForJBossAS.xml 2010-02-12 05:50:26 UTC (rev 100907)
@@ -164,20 +164,20 @@
delegateBeanName="HAModClusterService"/>
</programlisting>
<para>
- This configuration mode has the following advantages over the non-clustered mode:
+ This configuration mode has an advantage over the non-clustered mode in that the state of each proxy will be kept in sync on each node in the cluster.
</para>
- <orderedlist>
+<!-- <orderedlist>
<listitem>
<para>
The state of each proxy will be kept in sync on each node in the cluster.
</para>
</listitem>
- <!-- <listitem>
+ <listitem>
<para>
Proxies will be proactively notified of view changes - specifically crashed members, allowing the proxy to gracefully reconfigure itself potentially avoiding failover processing.
</para>
- </listitem> -->
- </orderedlist>
+ </listitem>
+ </orderedlist> -->
<para>
If your application server already uses other clustering services (e.g. session replication), then you should consider running mod_cluster in clustered-mode. If your application server is stateless (i.e. no sessions), or you are not using other clustering services, then non-clustered mode may be more appropriate.
</para>
Modified: projects/docs/enterprise/5.0/mod_cluster/en-US/MigratingFromModJK.xml
===================================================================
--- projects/docs/enterprise/5.0/mod_cluster/en-US/MigratingFromModJK.xml 2010-02-12 02:55:49 UTC (rev 100906)
+++ projects/docs/enterprise/5.0/mod_cluster/en-US/MigratingFromModJK.xml 2010-02-12 05:50:26 UTC (rev 100907)
@@ -97,7 +97,7 @@
-
</entry>
<entry>
- The nodes with <literal>loadfactor = 0</literal> are standby nodes that will be used when there are no other nodes available
+ The nodes with <literal>loadfactor</literal> = 0 are standby nodes that will be used when there are no other nodes available
</entry>
</row>
<row>
@@ -105,7 +105,7 @@
socket_timeout
</entry>
<entry>
- nodeTimeout
+ <varname>nodeTimeout</varname>
</entry>
<entry>
Default: 10 seconds
@@ -138,7 +138,7 @@
connection_pool_minsize
</entry>
<entry>
- smax
+ <literal>smax</literal>
</entry>
<entry>
The default is max
@@ -149,10 +149,10 @@
connection_pool_timeout
</entry>
<entry>
- ttl
+ <varname>ttl</varname>
</entry>
<entry>
- Time-to-live when over smax connections. Default: 60 seconds
+ <literal>Time-to-live</literal> when over smax connections. Default: 60 seconds
</entry>
</row>
<row>
@@ -160,7 +160,7 @@
-
</entry>
<entry>
- workerTimeout
+ <varname>workerTimeout</varname>
</entry>
<entry>
Max time to wait for a free worker. Default: 1 second
@@ -171,7 +171,7 @@
retries
</entry>
<entry>
- maxAttempts
+ <varname>maxAttempts</varname>
</entry>
<entry>
Max retries before returning an error. Default: 3
@@ -204,7 +204,7 @@
max_packet_size
</entry>
<entry>
- iobuffersize/receivebuffersize
+ <varname>iobuffersize/receivebuffersize</varname>
</entry>
<entry>
Not supported in this version. Use <varname>ProxyIOBufferSize</varname>
@@ -292,7 +292,7 @@
prepost_timeout
</entry>
<entry>
- ping
+ <varname>ping</varname>
</entry>
<entry>
Default: 10 seconds
@@ -320,8 +320,8 @@
To migrate from mod_proxy to mod_cluster, refer to http://www.jboss.org/mod_cluster/native/mod_proxy.html.
</para>
-<!-- <para>
- As mod_cluster is a sophisticated balancer migration from mod_proxy to mod_cluster is strait forward. mod_cluster replaces a reverse proxy with loadbalancing. A reversed proxy is configured like:
+ <para>
+ As mod_cluster is a sophisticated balancer, migration from mod_proxy to mod_cluster is strait forward. mod_cluster replaces a reverse proxy with loadbalancing. A reversed proxy is configured as follows:
</para>
<programlisting><![CDATA[
ProxyRequests Off
@@ -335,313 +335,311 @@
ProxyPassReverse /foo http://foo.example.com/bar
]]></programlisting>
<para>
- All the general proxy parameters could be used in mod_cluster they work like in mod_proxy, only the balancers and the workers definitions are slightly different.
+ All the general proxy parameters could be used in mod_cluster as they work in mod_proxy. The balancers and the workers definitions differ slightly as detailed in the following tables.
</para>
- <para>
- Workers:
- </para>
+
<table>
- <title>Worker Property Transfer</title>
+ <title>Worker Properties</title>
<tgroup align="left" cols="3" colsep="1" rowsep="1">
<colspec colname="c1"></colspec>
<colspec colname="c2"></colspec>
- <colspec colname="c3"></colspec>
+ <colspec colname="c3" colwidth="3*"></colspec>
<thead>
<row>
<entry>
- mod_jk worker property
+ <classname>Mod_proxy</classname> Parameter
</entry>
<entry>
- <classname>mod_cluster configuration</classname> Parameter
+ <classname>ClusterListener</classname> parameter
</entry>
<entry>
- Remarks
+ Remarks
</entry>
</row>
</thead>
<tbody>
<row>
<entry>
- host
+ min
</entry>
<entry>
-
</entry>
<entry>
- It is read from the <Connector/> Address information
+ Not supported in this version
</entry>
</row>
<row>
<entry>
- port
+ max
</entry>
<entry>
-
</entry>
<entry>
- It is read from the <Connector/> Port information
+ mod_cluster uses mod_proxy default value
</entry>
</row>
<row>
<entry>
- type
+ smax
</entry>
<entry>
- -
+ <varname>smax</varname>
</entry>
<entry>
- It is read from the <Connector/> Protocol information
+ Same as mod_proxy
</entry>
</row>
<row>
<entry>
- route
+ ttl
</entry>
<entry>
- -
+ <varname>ttl</varname>
</entry>
<entry>
- It is read from the <Engine/> JVMRoute information
+ Same as mod_proxy
</entry>
</row>
<row>
<entry>
- domain
+ acquire
</entry>
<entry>
- domain
+ <varname>workerTimeout</varname>
</entry>
<entry>
- Not supported in this version
+ Same as mod_proxy acquire but in seconds
</entry>
</row>
<row>
<entry>
- redirect
+ disablereuse
</entry>
<entry>
-
</entry>
<entry>
- The nodes with <literal>loadfactor = 0</literal> are standby nodes that will be used when there are no other nodes available
+ mod_cluster will disable the node in case of error and the <classname>ClusterListener</classname> will for the reuse via the STATUS message
</entry>
</row>
<row>
<entry>
- socket_timeout
+ flushPackets
</entry>
<entry>
- nodeTimeout
+ <varname>flushPackets</varname>
</entry>
<entry>
- Default: 10 seconds
+ Same as mod_proxy
</entry>
</row>
<row>
<entry>
- socket_keepalive
+ flushwait
</entry>
<entry>
- -
+ <varname>flushwait</varname>
</entry>
<entry>
- <classname>KEEP_ALIVE</classname> os is always on in mod_cluster
+ Same as mod_proxy
</entry>
</row>
<row>
<entry>
- connection_pool_size
+ keepalive
</entry>
<entry>
-
</entry>
<entry>
- The max size is calculated to be <literal>AP_MPMQ_MAX_THREADS+1</literal> (max)
+ Always on: <literal>OS KEEP_ALIVE</literal> is always used. Use connectionTimeout in the <![CDATA[<Connector>]]> if needed
</entry>
</row>
<row>
<entry>
- connection_pool_minsize
+ lbset
</entry>
<entry>
- smax
+ -
</entry>
<entry>
- The default is max
+ Not supported
</entry>
</row>
<row>
<entry>
- connection_pool_timeout
+ ping
</entry>
<entry>
- ttl
+ ping
</entry>
<entry>
- Time-to-live when over smax connections. Default: 60 seconds
+ Same as mod_proxy Default value 10 seconds
</entry>
</row>
<row>
<entry>
- -
+ lbfactor
</entry>
<entry>
- workerTimeout
+ -
</entry>
<entry>
- Max time to wait for a free worker. Default: 1 second
+ The load factor is received by mod_cluster from a calculated value in the <classname>ClusterListener</classname>
</entry>
</row>
<row>
<entry>
- retries
+ redirect
</entry>
<entry>
- maxAttempts
- </entry>
- <entry>
- Max retries before returning an error. Default: 3
- </entry>
- </row>
- <row>
- <entry>
- recovery_options
- </entry>
- <entry>
-
</entry>
<entry>
- mod_cluster behaves like mod_jk with value <varname>7</varname>
+ Not supported <varname>lbfactor</varname> set to 0 makes a standby node
</entry>
</row>
<row>
<entry>
- fail_on_status
+ retry
</entry>
<entry>
-
</entry>
<entry>
- Not supported
+ <classname>ClusterListener</classname> will test when the node is back online
</entry>
</row>
<row>
<entry>
- max_packet_size
+ route
</entry>
<entry>
- iobuffersize/receivebuffersize
+ <classname>JVMRoute</classname>
</entry>
<entry>
- Not supported in this version. Use <varname>ProxyIOBufferSize</varname>
+ In fact <application>JBossWEB</application> via the <classname>JVMRoute</classname> in the Engine will add it
</entry>
</row>
<row>
<entry>
- max_reply_timeouts
+ status
</entry>
<entry>
-
</entry>
<entry>
- Not supported
+ mod_cluster has a finer status handling: by context via the ENABLE/STOP/DISABLE/REMOVE application messages. <literal>hot-standby</literal> is done by <varname>lbfactor</varname> = 0 and Error by <varname>lbfactor</varname> = 1 both values are sent in STATUS message by the <classname>ClusterListener</classname>.
</entry>
</row>
<row>
<entry>
- recovery_time
+ timeout
</entry>
<entry>
- -
+ <varname>nodeTimeout</varname>
</entry>
<entry>
- The <literal>ClusterListener</literal> will tell (via a <varname>STATUS</varname> message) mod_cluster that the node is up again
+ Default wait for ever
</entry>
</row>
<row>
<entry>
- activation
+ ttl
</entry>
<entry>
- -
+ <varname>ttl</varname>
</entry>
<entry>
- mod_cluster receives this information via <varname>ENABLE/DISABLE/STOP</varname> messages
+ Default 60 seconds
</entry>
</row>
+ </tbody>
+ </tgroup>
+ </table>
+
+ <table>
+ <title>Balancer Properties</title>
+ <tgroup align="left" cols="3" colsep="1" rowsep="1">
+ <colspec colname="c1"></colspec>
+ <colspec colname="c2"></colspec>
+ <colspec colname="c3" colwidth="2*"></colspec>
+ <thead>
<row>
<entry>
- distance
+ <classname>Mod_proxy</classname> Parameter
</entry>
<entry>
- -
+ <classname>ClusterListener</classname> parameter
</entry>
<entry>
- mod_cluster handles this via the <varname>loadfactor</varname> logic
+ Remarks
</entry>
</row>
+ </thead>
+ <tbody>
<row>
<entry>
- mount
+ lbmethod
</entry>
<entry>
-
</entry>
<entry>
- The context mounted automatically via the <varname>ENABLE-APP</varname> messages. <varname>ProxyPass</varname> could also be used
+ There is only one load balancing method in mod_cluster; <classname>cluster_byrequests</classname>
</entry>
</row>
<row>
<entry>
- secret
+ maxattempts
</entry>
<entry>
- -
+ <varname>maxAttempts</varname>
</entry>
<entry>
- Not supported
+ Default 3
</entry>
</row>
<row>
<entry>
- connect_timeout
+ nofailover
</entry>
<entry>
- -
+ <varname>stickySessionForce</varname>
</entry>
<entry>
- Not supported. Use <varname>ProxyTimeout</varname> or server <varname>TimeOut</varname> (Default: 300 seconds)
+ Same as mod_proxy
</entry>
</row>
<row>
<entry>
- prepost_timeout
+ stickysession
</entry>
<entry>
- ping
+ <varname>StickySessionCookie/StickySessionPath</varname>
</entry>
<entry>
- Default: 10 seconds
+ The two parameters in the <classname>ClusterListener</classname> are combined into one, behaving as in mod_proxy
</entry>
</row>
<row>
<entry>
- reply_timeout
+ timeout
</entry>
<entry>
- -
+ <varname>workerTimeout</varname>
</entry>
<entry>
- Not supported. Use <varname>ProxyTimeout</varname> or server <varname>TimeOut?</varname> directive (Default: 300 seconds)
+ Default 1 seconds
</entry>
</row>
</tbody>
</tgroup>
- </table> -->
+ </table>
+
</section>
-
-
-
</chapter>
Modified: projects/docs/enterprise/5.0/mod_cluster/en-US/Revision_History.xml
===================================================================
--- projects/docs/enterprise/5.0/mod_cluster/en-US/Revision_History.xml 2010-02-12 02:55:49 UTC (rev 100906)
+++ projects/docs/enterprise/5.0/mod_cluster/en-US/Revision_History.xml 2010-02-12 05:50:26 UTC (rev 100907)
@@ -6,8 +6,8 @@
<simpara>
<revhistory>
<revision>
- <revnumber>2.0</revnumber>
- <date>Thurs 11 Feb 2010</date>
+ <revnumber>2.1</revnumber>
+ <date>Fri 12 Feb 2010</date>
<author>
<firstname>Gary</firstname>
<surname>Meintjes</surname>
More information about the jboss-cvs-commits
mailing list