[
https://issues.apache.org/jira/browse/OFBIZ-5042?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13470955#comment-13470955
]
Jacques Le Roux edited comment on OFBIZ-5042 at 10/6/12 9:05 AM:
-----------------------------------------------------------------
==== BETTER COMMENT: WE == COMMUNITY ====
Since the community decided to move most specialpurpose components to Apache
Extra ([see slim down
effort|https://issues.apache.org/jira/browse/OFBIZ/fixforversion/12320551]),
the best place for this component is in [Apache OFBiz
Extra|http://code.google.com/a/apache-extras.org/hosting/search?q=label%3aOFBiz].
Please read [Apache Extra
guidelines|http://community.apache.org/apache-extras/guidelines.html].
Note the point 6:
{quote}
Each Apache Extras project is encouraged to publish their releases and complete
the project tagging information available on Apache Extras.
{quote}
As explained in the guidelines, The Apache OFBiz Extra projects will not be
official part of the Apache OFBiz project. The goals are pragmatic (at least as
I see it):
* Components which are not mandatory (not supposed to be used in each project
powered by OFBiz) can be moved to Apache Extra.
* With limited Apache OFBiz project human resources, it's a way to delegate the
code and releases responsabilities to specific communities
But this does not mean that Apache OFBiz Extra projects will be forgotten or
neglected by the Apache OFBiz project. Now I feel something is missing in the
scheme so far: how to relate an Apache OFBiz Extra project release to an Apache
OFBiz release? I think it should part of the Apache OFBiz Extra project
community to indicate which OFBiz release/s (trunk included) are compatible
with the component. This is kind of part of the guidelines point 6 in my
opininon. This is correctly maintained could be used by the Apache OFBiz
project when releasing, we could at least have a link to the [Apache OFBiz
Extra projects
page|http://code.google.com/a/apache-extras.org/hosting/search?q=label%3aOFBiz].
Same from the [Apache OFBiz project site main page|http://ofbiz.apache.org/],
a reference should be added. I will do so soon, but will wait that at least a
complete usable Apache OFBiz Extra project exists there. The Neogia team made a
first effor but nothing was committed yet, the Solr component could be the
first to engage the movement. With the ApacheConEu coming it could be also a
good way to share about this point.
Also I'd like to have a way to know the value of an Apache OFBiz Extra project,
as experienced by its users. For the moment I see only one way to do this:
recommend it on Google + (see in each component Project information). Of course
this is a very limited way, but at least a mean.
So it would be great if you Paul, Pascal and Jonatan (at least for now, the
community will hopefully grow from comments above) could move the Solr
component into the Apache OFBiz Extra projects. This would begin the dynamic,
completing the pending work initiated by the Neogia team. Hopefully more will
come later. For instance I read a lof about Big Fish, why not sharing in Apache
OFBiz Extra projects also?
was (Author: jacques.le.roux):
== BETTER COMMENT: WE == COMMUNITY ==
Since the community decided to move most specialpurpose components to Apache
Extra ([see slim down
effort|https://issues.apache.org/jira/browse/OFBIZ/fixforversion/12320551]),
the best place for this component is in [Apache OFBiz
Extra|http://code.google.com/a/apache-extras.org/hosting/search?q=label%3aOFBiz].
Please read [Apache Extra
guidelines|http://community.apache.org/apache-extras/guidelines.html].
Note the point 6:
{quote}
Each Apache Extras project is encouraged to publish their releases and complete
the project tagging information available on Apache Extras.
{quote}
As explained in the guidelines, The Apache OFBiz Extra projects will not be
official part of the Apache OFBiz project. The goals are pragmatic (at least as
I see it):
* Components which are not mandatory (not supposed to be used in each project
powered by OFBiz) can be moved to Apache Extra.
* With limited Apache OFBiz project human resources, it's a way to delegate the
code and releases responsabilities to specific communities
But this does not mean that Apache OFBiz Extra projects will be forgotten or
neglected by the Apache OFBiz project. Now I feel something is missing in the
scheme so far: how to relate an Apache OFBiz Extra project release to an Apache
OFBiz release? I think it should part of the Apache OFBiz Extra project
community to indicate which OFBiz release/s (trunk included) are compatible
with the component. This is kind of part of the guidelines point 6 in my
opininon. This is correctly maintained could be used by the Apache OFBiz
project when releasing, we could at least have a link to the [Apache OFBiz
Extra projects
page|http://code.google.com/a/apache-extras.org/hosting/search?q=label%3aOFBiz].
Same from the [Apache OFBiz project site main page|http://ofbiz.apache.org/],
a reference should be added. I will do so soon, but will wait that at least a
complete usable Apache OFBiz Extra project exists there. The Neogia team made a
first effor but nothing was committed yet, the Solr component could be the
first to engage the movement. With the ApacheConEu coming it could be also a
good way to share about this point.
Also I'd like to have a way to know the value of an Apache OFBiz Extra project,
as experienced by its users. For the moment I see only one way to do this:
recommend it on Google + (see in each component Project information). Of course
this is a very limited way, but at least a mean.
So it would be great if you Paul, Pascal and Jonatan (at least for now, the
community will hopefully grow from comments above) could move the Solr
component into the Apache OFBiz Extra projects. This would begin the dynamic,
completing the pending work initiated by the Neogia team. Hopefully more will
come later. For instance I read a lof about Big Fish, why not sharing in Apache
OFBiz Extra projects also?
> Apache Solr Implementation
> --------------------------
>
> Key: OFBIZ-5042
> URL: https://issues.apache.org/jira/browse/OFBIZ-5042
> Project: OFBiz
> Issue Type: Improvement
> Components: ALL COMPONENTS
> Reporter: Paul Piper
> Labels: component, feature, solr
> Attachments: solr-v1.00-r4284.zip
>
>
> Apache Solr, a modern search engine, seems to be one of the most commonly
> requested feature within the OFBiz community. With my agency, ilscipio, we
> have done several Apache Solr implementations in the past, and we felt that
> we could contribute this feature to the community. Currently, Solr is greatly
> used within syracus.net (our standalone store) and serves to generate the
> categories & search functionalities. From our team mainly Pascal Proulx,
> Jonatan Soto and I were involved in this feature (albeit more having
> contributed to it. We created a standalone component which can be integrated
> into any running ofbiz implementation. It was tested against OFBiz 09.04 -
> 12.04..
> =====================================================
> Solr Component for Apache Ofbiz
> =====================================================
> This document describes the Ofbiz solr component, an Ofbiz
> (http://ofbiz.apache.org/)
> implementation of the Apache Solr search platform
> (http://lucene.apache.org/solr/).
> The solr component includes an Ofbiz service-based wrapper layer to the
> Apache Solr
> webapp queries as well as the native Apache Solr web interface itself.
> Currently, the solr component focuses on Product data.
> Note: This document is a work in progress; information is subject to change.
> Contents:
> 1. Installation
> 2. Configuration
> 3. Data Indexing
> 4. Data Querying
> 5. Implementation Concerns
> 6. Known Bugs, Limitations and Issues
> -----------------------------------------------------
> 1. Installation
> -----------------------------------------------------
> To install solr in an Ofbiz setting, simply extract the solr directory and
> and sub-folders to the hot-deploy folder.
> Afterward, the solr home system property (solr.solr.home) must be set to the
> value
> hot-deploy/solr manually using one of the following methods:
> * solr.solr.home in batch/script file:
> Add the parameter "-Dsolr.solr.home=hot-deploy/solr" to the Java command
> invocation for ofbiz.jar.
> e.g.:
> "%JAVA_HOME%\bin\java" -Xms128M -Xmx512M -XX:MaxPermSize=512m
> -Dsolr.solr.home=hot-deploy/solr -jar ofbiz.jar
> * solr.solr.home in Ant build configuration:
> In your root Ofbiz build.xml file, add the element "<jvmarg
> value="-Dsolr.solr.home=hot-deploy/solr"/>"
> to the "<java jar="ofbiz.jar"...>" invocation of the appropriate Ant
> target(s) (run, start, run-install, etc.).
> e.g.:
> <target name="start" description="Start OFBiz">
> <java jar="ofbiz.jar" fork="true">
> <jvmarg value="${memory.initial.param}"/>
> <jvmarg value="${memory.max.param}"/>
> <jvmarg value="${memory.maxpermsize.param}"/>
> <jvmarg value="-Dsolr.solr.home=hot-deploy/solr"/>
> </java>
> </target>
> ***
> It may be possible to specify solr home using other methods (JNDI, web.xml),
> but at the time
> of this writing, this was the most reliable method known.
> -----------------------------------------------------
> 2. Configuration
> -----------------------------------------------------
> The solr component can run out-of-the-box without configuration, but many
> files, settings and interfaces allow custom settings. Some of these include:
> * Ofbiz configurations:
> ** System properties:
> *** ofbiz.solr.eca.enabled - Global solr ECA toggling boolean (true/false,
> see Data Indexing)
> ** Config files:
> *** solr/config/solrconfig.properties - Ofbiz solr service behavior control
> *** ofbiz-component.xml - Standard Ofbiz component config
> * Apache Solr configurations:
> ** System properties:
> *** solr.solr.home - Solr home (see Installation)
> ** Config files:
> *** solr.xml - Base solr config
> *** conf/schema.xml - Solr index schema
> *** webapp/WEB-INF/web.xml - Dual Ofbiz/Solr webapp config
> ** Interfaces:
> *** /solr/admin/ - Webapp admin interface (see below)
> ***
> It is possible to set extensive native Solr configuration using
> the admin webapp interface noted above. It should be accessible at the
> address:
> http://localhost:8080/solr/admin/
> (where 8080 is your server's http port)
> Please refer to the Apache Solr documentation for usage of this interface
> and other native Solr configuration details.
> -----------------------------------------------------
> 3. Data Indexing
> -----------------------------------------------------
> The solr component indexes data such as Products into the Apache Solr database
> using services defined in the file:
> servicesdef/solrservices.xml
> The initial indexing may need to be performed or scheduled manually, but
> subsequent indexing
> may be partially or fully automated, though automated methods are disabled by
> default and must be enabled.
> Note that in general, solr services can only successfully run in contexts
> where the solr webapp is
> loaded and accessible.
> There are two methods for indexing data:
> * Index rebuilding service (rebuildSolrIndex):
> The rebuildSolrIndex is the most important data import service. It reindexes
> all Ofbiz Products existing in the system into the solr index.
> rebuildSolrIndex MUST be run
> AT LEAST once after installation and also following any data load operation
> that loads new products using the Ofbiz
> "install" starting mode (run-install, load-demo, etc.).
> To do this, one can simply use the Webtools backend to invoke the service
> manually
> (e.g.,
> /webtools/control/setSyncServiceParameters?SERVICE_NAME=rebuildSolrIndex&POOL_NAME=pool&_RUN_SYNC_=Y)
> or summon the Webtools Job Scheduler.
> Once the initial indexing has been performed, one can then use the Job
> Scheduler to
> invoke rebuildSolrIndex on a regular basis (every hour, every midnight, etc.)
> to update the Solr index.
> * ECAs/SECAs (addToSolr, for Product data):
> Although the rebuildSolrIndex is always necessary for the initial data
> import, one may also
> use ECAs and SECAs to import subsequent data changes automatically at every
> individual data (e.g. Product)
> update instead of running rebuildSolrIndex periodically. This is done by
> defining ECAs or SECAs that
> trigger the addToSolr service.
> The addToSolr service simply accepts a single "instance" parameter, a
> GenericValue. At the time of this writing,
> any entity value having a valid "productId" field designating a Product value
> may be passed; this will trigger reindexing
> for the specific product.
> By default, the addToSolr service implementation (and any service intended
> for use with ECAs) is disabled globally and
> succeeds silently. It can be enabled by setting the "solr.eca.enabled=true"
> in "config/solrconfig.properties".
> This property is provided for easy toggling. It can also be specified on the
> command line or using ant through
> the true/false "ofbiz.solr.eca.enabled" system property (which has priority
> over file-based "solr.eca.enabled") in the same
> way described for the solr home system property (see Installation); this
> allows toggling them per Ant target.
> Some stock/default/example ECAs are provided and are functional once enabled.
> They can be found in:
> entitydef/eecas.xml
> Custom ECAs may also be added to this file, and custom SECAs may be added to
> the file:
> servicedef/secas.xml
> Special care may need to be taken to ensure the addToSolr is not triggered
> during "install" sequences
> (run-install/load-demo) because it requires the solr webapp to be accessible.
> addToSolr will automatically
> attempt to prevent reindexing in cases where the solr webapp is unavailable.
> However, it is possible that
> this behavior results in needlessly prevented reindexing (and thus lost
> updates) in some unknown complex server
> configurations. If this is the case, this behavior can be disabled by setting
> the option
> "solr.eca.useSolrWebappLoadedCheck=false" in the properties file
> "config/solrconfig.properties".
>
> However, if the above useSolrWebappLoadedCheck option is specified, you will
> have to manually disable or comment the
> ECA definitions/services to prevent them from executing during data-load
> operations; the "solr.eca.enabled" property and "ofbiz.solr.eca.enabled"
> system property can be used for this purpose. One easy way is to set
> "solr.eca.enabled=true" in
> the property files and then specify "-Dofbiz.solr.eca.enabled=false" for any
> relevant
> Ant build install/run-install/load-demo targets.
> Note that simply commenting the ECAs/SECAs or preventing their inclusion using
> ofbiz-component.xml may eliminate some service call overhead compared to
> using the property toggling. Ideally,
> a method exploiting ofbiz-component.xml or eecas.xml to prevent ECAs from
> running in inappropriate contexts
> would be ideal; however, at the time of this writing, such a method could not
> be established.
> ***
> To reset and clear the solr index completely, it is sufficient to shut down
> the server and delete the folders:
> solr/data/index
> solr/data/spellchecker
> and any other such folders (must delete the whole folders, not only contents).
> -----------------------------------------------------
> 4. Data Querying
> -----------------------------------------------------
> Solr queries can be done using two methods:
> * Solr Ofbiz services:
> Simply invoke (manually or in code) the query services found in the file:
> servicesdef/solrservices.xml
> These include solrProductsSearch, solrKeywordSearch and others. Note that in
> general,
> solr services can only successfully run in contexts where the solr webapp is
> loaded and accessible.
> * Solr native admin webapp interface:
> One can also perform native Solr queries and diagnostics using the standard
> admin interface, accessible as described under Configuration. Please refer
> to the Apache Solr documentation for usage of this interface.
> -----------------------------------------------------
> 5. Implementation Concerns
> -----------------------------------------------------
> * The structure of the solr/webapp directory closely mirrors the contents of
> the solr.war
> distribution; however, please note that it contains some Ofbiz-specific
> modifications such
> as those found in web.xml. The same is true for other folders such as
> solr/conf. Therefore,
> any library updates must be done with care to preserve these modifications.
> -----------------------------------------------------
> 6. Known Bugs, Limitations and Issues
> -----------------------------------------------------
> * In general, solr services can only successfully run in contexts where the
> solr webapp
> is loaded and accessible.
> * ECA indexing may be problematic due to data loading and unknown complex
> server configurations;
> various toggling options are provided to work around this issue (see Data
> Indexing for details).
> In the future - and ideally - this could be addressed using
> ofbiz-component.xml (or eecas.xml) directives.
> * The indexing services often produce the log warning "Problem reading
> product features" due to their
> use of the stock Ofbiz getProductFeatureSet service.
> ***
> Please report any other issues encountered. Thank you.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
