Good evening... and rats...(kind of).
I'd just started putting jk2_get_most_suitable_worker() into pseudocode
myself, but glad that it has finally been done in any case. For those that
haven't looked at the code, it will come as a surprise to learn the
magnitude of the decision making process.
Thanks to you and Angus.
Norm
----- Original Message -----
From: <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Thursday, April 15, 2004 4:32 PM
Subject: cvs commit: jakarta-tomcat-connectors/jk/xdocs/jk2 configwebcom.xml
> mturk 2004/04/14 23:32:33
>
> Modified: jk/xdocs/jk2 configwebcom.xml
> Log:
> Update few missing docs. Thanks to Agnus Mezick.
>
> Revision Changes Path
> 1.17 +109 -37
jakarta-tomcat-connectors/jk/xdocs/jk2/configwebcom.xml
>
> Index: configwebcom.xml
> ===================================================================
> RCS file:
/home/cvs/jakarta-tomcat-connectors/jk/xdocs/jk2/configwebcom.xml,v
> retrieving revision 1.16
> retrieving revision 1.17
> diff -u -r1.16 -r1.17
> --- configwebcom.xml 13 Mar 2004 07:33:24 -0000 1.16
> +++ configwebcom.xml 15 Apr 2004 06:32:33 -0000 1.17
> @@ -156,26 +156,26 @@
> <th>Default</th>
> <th>Description</th>
> </tr>
> - <tr>
> - <td>file</td>
> - <td>No default value</td>
> - <td>Name of the file that will be mmapped to
use as shared memory, If set to 'anonymous' use the anonymous shered
memory</td>
> - </tr>
> - <tr>
> - <td>size</td>
> - <td>No default value</td>
> - <td>Deprecated. Size of the file.</td>
> - </tr>
> - <tr>
> - <td>slots</td>
> - <td>256</td>
> - <td>Number of shared memory slots. Set to the
number of child processes</td>
> - </tr>
> - <tr>
> - <td>useMemory</td>
> - <td>0</td>
> - <td>Use process memory instead of shared
memory. Useful for single child mpm's</td>
> - </tr>
> + <tr>
> + <td>file</td>
> + <td>No default value</td>
> + <td>Name of the file that will be mmapped to
use as shared memory, If set to 'anonymous' use the anonymous shered
memory</td>
> + </tr>
> + <tr>
> + <td>size</td>
> + <td>No default value</td>
> + <td>Deprecated. Size of the file.</td>
> + </tr>
> + <tr>
> + <td>slots</td>
> + <td>256</td>
> + <td>Number of shared memory slots. Set to the
number of child processes</td>
> + </tr>
> + <tr>
> + <td>useMemory</td>
> + <td>0</td>
> + <td>Use process memory instead of shared
memory. Useful for single child mpm's</td>
> + </tr>
> </table>
> </p>
> </section>
> @@ -346,10 +346,18 @@
> <tr>
> <td>lb_factor</td>
> <td>1</td>
> - <td>
> - Load balancing factor to use. At this moment, it'll be set on the
worker,
> - but in future it should be possible to use lb on a channel level.
> - </td>
> + <td>Load balancing factor to use. The lower
the lb_factor the more often that tomcat will get requests but see
> + "level" below.</td>
> + </tr>
> + <tr>
> + <td>level</td>
> + <td>1</td>
> + <td>Worker Priority. Valid values are 0-3.
The functioning workers with the lowest level
> + will be checked for the lowest
lb_value, and if found will be run. The upper level workers are
> + only checked upon failure of all
workers on ALL of the levels below them. This is very
> + useful for implementing failover
withing a cluster. You could set the tomcat server local
> + on the same machine as the apache
instance to level 0 and all of the other workers to level 1.
> + This would cause apache to only use the
external tomcats when the local tomcat is down.</td>
> </tr>
> <tr>
> <td>group</td>
> @@ -359,8 +367,13 @@
> <tr>
> <td>tomcatId</td>
> <td>Automatically set to the localname (
host:port )</td>
> - <td>Must match the JVM route on tomcat
Engine, for load balancing</td>
> + <td>Must match the JVM route on tomcat the
server.xml Engine element, for load balancing</td>
> </tr>
> + <tr>
> + <td>route</td>
> + <td>Automatically set to the localname (
host:port )</td>
> + <td>Same as tomcatId</td>
> + </tr>
> </table>
> </p>
> </subsection>
> @@ -409,7 +422,8 @@
> </p>
> </subsection>
> <subsection name="ajp13">
> - <p>Default worker</p>
> + <p>Default worker. If a property is in both the worker
and the channel, you only need to define it in one place.
> + The channel passes down the properties to its
worker. (at least in the ajp13-channel.socket linkage)</p>
> <p>
> <table>
> <tr>
> @@ -427,28 +441,39 @@
> </tr>
> <tr>
> <td>tomcatId</td>
> - <td/>
> - <td/>
> + <td>Automatically set to the localname (
host:port )</td>
> + <td>Must match the JVM route on the tomcat
server.xml Engine element, for load balancing</td>
> </tr>
> <tr>
> <td>route</td>
> - <td/>
> - <td/>
> + <td>Automatically set to the localname (
host:port )</td>
> + <td>Same as tomcatId</td>
> </tr>
> <tr>
> <td>group</td>
> - <td/>
> - <td/>
> + <td>lb</td>
> + <td>loadbalanced groups to which this
channel and the associated worker will be added, multivalued. You need to
set it only if you have an advanced setup with multiple clusters.</td>
> + </tr>
> + <tr>
> + <td>lb_factor</td>
> + <td>1</td>
> + <td>Load balancing factor to use. The lower
the lb_factor the more often that tomcat will get requests but see
> + "level" below.</td>
> </tr>
> <tr>
> <td>level</td>
> - <td/>
> - <td/>
> + <td>1</td>
> + <td>Worker Priority. Valid values are 0-3.
The functioning workers with the lowest level
> + will be checked for the lowest
lb_value, and if found will be run. The upper level workers are
> + only checked upon failure of all
workers on ALL of the levels below them. This is very
> + useful for implementing failover
withing a cluster. You could set the tomcat server local
> + on the same machine as the apache
instance to level 0 and all of the other workers to level 1.
> + This would cause apache to only use the
external tomcats when the local tomcat is down.</td>
> </tr>
> <tr>
> <td>channel</td>
> <td/>
> - <td/>
> + <td>Communication channel used by the
worker. Use the full name of the channel (everything between the
[]'s).</td>
> </tr>
> <tr>
> <td>max_connections</td>
> @@ -574,7 +599,7 @@
> <tr>
> <td>level</td>
> <td>INFO</td>
> - <td>Log level. Supported: EMERG, ERROR, INFO,
DEBUG</td>
> + <td>Text of the log level. Strings supported:
EMERG, ERROR, INFO, DEBUG</td>
> </tr>
> </table>
> </p>
> @@ -603,5 +628,52 @@
> <subsection name="logger.apache2">
> <p>Logger used in Apache2 servers, it normally in ends
in error.log </p>
> </subsection>
> + </section>
> + <section name="How Load Balancing Works">
> + <p>The lb_factor and level properties combine to deliver a
flexible static load balancing solution.
> + The level property is used to create up to four pools over
workers in descending priority and lb_factor
> + is used to weight the workers within a pool. The lower the
level the more likely the worker is to be used
> + and the lower the lb_factor the more likely the worker is
to be used. Here is how the algorithm is
> + currently implemented:</p>
> + <p>
> +(Assume that every worker's lb_value is set to their lb_factor)<br/>
> +<br/><source>
> +if (loadbalancer has a route) and (stickysession=1){<br/>
> + worker= loadbanlancer.getWorkerForRoute(route)<br/>
> + if worker.hasRedirect<br/>
> + redirect=worker.redirect<br/>
> + else if !worker.hasError<br/>
> + return worker<br/>
> + }<br/>
> +<br/>
> +selectedWorker=null<br/>
> +foreach lb_level {<br/>
> + foreach worker in the level {<br/>
> + if worker.isNotWorking<br/>
> + continue<br/>
> + if selectedWorker == null<br/>
> + selectedWorker = worker<br/>
> + continue<br/>
> + if worker.lb_value < selectedWorker.lb_value<br/>
> + selectedWorker = worker<br/>
> + continue<br/>
> + }<br/>
> + if selectedWorker !=null<br/>
> + break<br/>
> +}<br/>
> +<br/>
> +Reenable workers in error state if the timeout has passed()<br/>
> +<br/>
> +if selectedWorker !=null {<br/>
> + selectedWorker.lb_value += selectedWorker.lb_factor<br/>
> + if selectedWorker.lb_value > 255 {<br/>
> + foreach worker in
load_balancer.workers[selectedWorker.level]<br/>
> + worker.lb_value=worker.lb_factor<br/>
> + }<br/>
> + }<br/>
> + return selectedWorker<br/>
> +}<br/></source>
> +</p>
> +
> </section>
> </document>
>
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
