This plugin allows to import CouchBase data to Solr. It uses the Cross-datacenter Replication (XDCR) feature of Couchbase Server 2.0 to transfer data continuously.
Java
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
docs
ivy
solr-conf
src
zk-conf
.classpath
.gitignore
.project
LICENSE
README.md
build.properties
build.xml
ivy.xml

README.md

solr-couchbase-plugin

This plugin allows to import CouchBase data to Solr. It uses the Cross-datacenter Replication (XDCR) feature of Couchbase Server 2.0 to transfer data continuously.

Plugin configuration

Dependencies

  • Copy this plugin from solr-shared-libs/ to the directory:
<solr_home>/solr/lib-couchbase/
  • Add dependencies from solr-war-libs/ directory to solr.war and remove the older commons-io dependency version (2.3). Dependencies should be located in the war file under:
<solr-war>/WEB-INF/lib/

solr.xml

An additional line to solr.xml should be added, to inform Solr about new dependencies which should be included in the Solr's classpath. Add the following line to solr.xml file:

<str name="sharedLib">${sharedLib:lib-couchbase}</str>

The whole solr.xml file should look as follows:

<solr>
  <str name="sharedLib">${sharedLib:lib-couchbase}</str>

  <solrcloud>
    <str name="host">${host:}</str>
    <int name="hostPort">${jetty.port:8983}</int>
    <str name="hostContext">${hostContext:solr}</str>
    <int name="zkClientTimeout">${zkClientTimeout:30000}</int>
    <bool name="genericCoreNodeNames">${genericCoreNodeNames:true}</bool>
  </solrcloud>

  <shardHandlerFactory name="shardHandlerFactory"
    class="HttpShardHandlerFactory">
    <int name="socketTimeout">${socketTimeout:0}</int>
    <int name="connTimeout">${connTimeout:0}</int>
  </shardHandlerFactory>

</solr>

solrconfig.xml

It is required to configure Couchbase buckets to index data from in the solrconfig.xml file under /couchbase RequestHandler. Whole RequestHandler configuration should look as following:

<requestHandler name="/couchbase" class="org.apache.solr.couchbase.CouchbaseRequestHandler">
  <lst name="params">
    <str name="username">admin</str>
    <str name="password">admin123</str>
    <int name="port">9876</int>
    <int name="numVBuckets">1024</int>
    <bool name="commitAfterBatch">false</bool>
    <bool name="optimize">false</bool>
    
    <!-- Optional, setting for Couchbase XDCR remote cluster and replication -->
    <lst name="couchbaseServer">
    	<str name="ipAddress">127.0.0.1</str>
        <str name="couchbaseUsername">Administrator</str>
        <str name="couchbasePassword">password</str>
        <str name="clusterName">Solr</str>

        <lst name="bucketInfo">
        	<str name="fromBucketName">test</str>
            <str name="toBucketName">default</str>
        </lst>

        <lst name="bucketInfo">
        	<str name="fromBucketName">program</str>
       		<str name="toBucketName">default</str>
        </lst>
    </lst>
  </lst>
  
  <lst name="bucket">
    <str name="name">default</str>
    <str name="splitpath">/</str>
    <lst name="fieldmappings">
      <str name="name">name:/name</str>
      <str name="city">city_s:/city</str>
      <str name="code">code_s:/code</str>
      <str name="country">country_s:/country</str>
      <str name="phone">phone_s:/phone</str>
      <str name="website">url:/website</str>
      <str name="type">type_s:/type</str>
      <str name="updated">last_modified:/updated</str>
      <str name="description">description:/description</str>
      <str name="address">address_s:/address</str>
      <str name="geo">geo_s:/geo</str>
    </lst>
  </lst>
</requestHandler>
  • params - a list of params required to configure this plugin.

    • username - A valid Couchbase server username

    • password - A valid Couchbase server password

    • port - A port number on which this plugin will register itself as a Couchbase replica.

    • numVBuckets - A number of VBuckets used by this Couchbase replica. Couchbase Server on Mac OS X uses 64 vBuckets as opposed to the 1024 vBuckets used by other platforms. Couchbase clusters with mixed platforms are not supported. It is required that numVBuckets is identical on the Couchbase server and Solr plugin.

    • commitAfterBatch - A flag specifying whether this plugin should commit documents to Solr after every batch of documents or when all the documents are retrieved from Couchbase.

    • couchbaseServer - Optional, a list with attributes that are required for creating Couchbase XDCR remote cluster and/or XDCR replication.

    • ipAddress - The IP address at where this plugin is running. Required if you have couchbaseServer field.

    • couchbaseUsername - The username of target Couchbase instance. With this field the plugin can get authorized by Couchbase. Required if you have couchbaseServer field.

    • couchbasePassword - The password of target Couchbase instance. With this field the plugin can get authorized by Couchbase. Required if you have couchbaseServer field.

    • clusterName - The cluster name that you want to create Couchbase XDCR remote cluster with. Required if you have couchbaseServer field.

    • bucketInfo - Optional, a list with attributes that are required for crteating Couchbase XDCR replication.

    • fromBucketName - The Couchbase bucket name where you want to pull data from. Required if you have bucketInfo field.

    • toBucketName - The destination bucket name where you want to push data to. Basically you have to use the names that are specified in "bucket" list. Required if you have bucketInfo field.

  • bucket - a list with bucket parameters required to perform a synchronisation with Couchbase. Multiple lists of this type are allowed.

    • name - Bucket name - must be unique
    • splitpath - a list with paths to the fields in JSON Object on which the original Couchbase JSON document will be split up to extract embedded documents. This is a single String where paths are saparated with "|".
    • fieldmappings - a list with field names mapping for Couchbase documents, before indexing them into Solr. List element's name must be unique. At least one field mapping must be provided. Value should be solr_field_name:couchbase_field_path. The ‘json-path’ is a required part. 'target-field-name' is the name of the field in the input Solr document. It is optional and it is automatically derived from the input json.
    • More informations about splitpath and fieldmappings usage can be found here: https://lucidworks.com/blog/indexing-custom-json-data/
    • Wildcards - Instead of specifying all the field names in fieldmappings explicitly , it is possible to specify a wildcard '*' or a wildwildcard '**' to map fields automatically. The constraint is that wild cards can be only used in the end of the json-path. The split path cannot use wildcards. The following are example wildcard path mappings:

Example Wildcards:

f=/docs/* : maps all the fields under docs and in the name as given in json
f=/docs/** : maps all the fields under docs and its children in the name as given in json
f=searchField:/docs/* : maps all fields under /docs to a single field called ‘searchField’
f=searchField:/docs/** : maps all fields under /docs and its children to searchField 

Example Bucket Configuration:

<lst name="bucket">
  <str name="name">default</str>
  <str name="splitpath">/</str>
  <lst name="fieldmappings">
    <str name="name">name:/name</str>
    <str name="city">city_s:/city</str>
    <str name="code">code_s:/code</str>
    <str name="country">country_s:/country</str>
    <str name="phone">phone_s:/phone</str>
    <str name="website">url:/website</str>
    <str name="type">type_s:/type</str>
    <str name="updated">last_modified:/updated</str>
    <str name="description">description:/description</str>
    <str name="address">address_s:/address</str>
    <str name="geo">geo_s:/geo</str>
  </lst>
</lst>

Example JSON:

{
"person" : { 
            "name" : "John",
            "age" : 22
           },
"value" : 100
}

Example field mappings for above JSON:

<lst name="test-fieldmappings">
  <str name="person">person:/person</str>
  <str name="name">name:/person/name</str>
  <str name="age">age_i:/person/age</str>
  <str name="value">value_i:/value</str>
</lst>

Couchbase XDCR

In Couchbase admin panel, under XDCR tab following settings should be configured:

Remote Cluster

This Solr plugin should be configured as Couchbase's Remote Cluster. Click on 'Create Cluster Reference' button and fill in cluster data.

  • 'Cluster Name' can be any name.
  • 'IP/hostname' should be <ip_address>:port and the port number is the port on which this request handler will register itself as a Couchbase Replica. It is specified in CouchbaseRequestHandler's params list as 'port'.
  • Username and password should be the same as those provided in solrconfig.xml file, as the replica starts with the same credentials as the Couchbase server.

Replication

Click on 'Create Replication' button.

  • Select a Couchbase Cluster and a Bucket to replicate from.
  • Select configured Solr's Plugin Remote Cluster and add a Bucket name. This Bucket name must match the name specified in the bucket name parameter in the solrconfig.xml file.
  • Under Advaced settings XDCR Prootcol should be set to 'Version 1' value.

Multiple collections

To sunchronize documents from Couchbase to multiple Solr collections, each collection needs to have this plugin configured in it's solrconfig.xml file and also a replication in Couchbase created. Each collection plugin must be started manually. If this is done, there will be one Couchbase replica running for each Solr collection.

Running Solr Couchbase Plugin

To run this plugin, simply perform all actions described in Configuration section of this instruction, and run Solr. To run single instance of Solr, execute ant run-solr command in the main directory. To run Solr in Cloud mode, execute ant solr-cloud command in the main directory. When solr is started execute GET request to URL

http://<solr_address>/solr/collection1/couchbase?action=start

This will start the plugin, which will register in Couchbase as its replica and the data synchronisation as well as indexing into Solr should start. It assumes that Couchbase server is running and Cross-Datacenter Replication is set as described in Couchbase XDCR section of this document.

To stop the plugin, exegute GET request to the URL

http://<solr_address>/solr/collection1/couchbase?action=stop

In a case, when all configured Solr instances with this plugin are restarted, Couchbase XDCR must be configured again. This is because every time a Couchbase replica is started, it acquires new pool UUID which is used in communications with Couchbase server.