The search_api_solr module manages its dependencies and class loader via composer. So if you simply downloaded this module from drupal.org you have to delete it and install it again via composer!
Simply change into the Drupal directory and use composer to install search_api_solr:
cd $DRUPAL
composer require drupal/search_api_solr
This module provides an implementation of the Search API which uses an Apache Solr search server for indexing and searching. Before enabling or using this module, you'll have to carefully read all the instructions given here.
The minimum support version Solr version is 6.4. Any version below might work
if you use your own Solr config or if you enable the optional
search_api_solr_legacy sub-module that is included in this module.
In general it is highly recommended to run Solr in cloud mode (Solr Cloud)!
To setup Solr in Cloud mode locally you can either follow the instructions of the Apache Solr Reference Guide or use docker-compose with https://github.com/docker-solr/docker-solr-examples/blob/master/docker-compose/docker-compose.yml
The preferred way for local development is to use DDev where you can easily add the pre-defined solr-cloud service.
Once Solr Cloud is running with DDev you don't need to deal with any configset
files like described in the sections below. Just enable the
search_api_solr_admin sub-module and configure the Search API Server to use
the Solr Cloud Connector with Basic Auth. The username "solr" and the password
"SolrRocks" are pre-configured in .ddev/solr-cloud/security.json. Now you
create or update your collection any time by clicking the "Upload Configset"
button on the Serch API server details page. Or automate things using
ddev drush search_api_solr:upload-configset SERVER_ID
Check the Apache Solr (Cloud) Integration for DDEV-Local README for more detailed instructions.
In order for this module to work, you need to set up a Solr server. For this, you can either purchase a server from a web Solr hosts or set up your own Solr server on your web server (if you have the necessary rights to do so). If you want to use a hosted solution, a number of companies are listed on the module's project page. Otherwise, please follow the instructions in this section.
Note: A more detailed set of instructions is available at:
- https://lucene.apache.org/solr/guide/8_4/installing-solr.html
- https://lucene.apache.org/solr/guide/8_4/taking-solr-to-production.html
- https://lucene.apache.org/solr/guide/ - list of other version specific guides
As a pre-requisite for running your own Solr server, you'll need a Java JRE.
Download the latest version of Solr 8.x from
https://lucene.apache.org/solr/downloads.html and unpack the archive
somewhere outside of your web server's document tree. The unpacked Solr
directory is named $SOLR in these instructions.
Note: Solr 6.x is still supported by search_api_solr but strongly discouraged. That version has been declared end-of-life by the Apache Solr project and is thus no longer supported by them.
Before creating the Solr core ($CORE) you will have to make sure it uses
the proper configuration files. They aren't always static but vary on your
Drupal setup.
But the Search API Solr Search module will create the correct configs for you!
- Make sure you have Apache Solr started and accessible (i.e. via port 8983). You can start it without having a core configured at this stage.
- Visit Drupal configuration (/admin/config/search/search-api) and create a new Search API Server according to the search_api documentation using "Solr" as Backend and the connector that matches your setup. Input the correct core name (which you will create at step 4, below).
- Download the config.zip from the server's details page or by using
drush solr-gscwith proper options, for example for a server named "my_solr_server":drush solr-gsc my_solr_server config.zip 8.4. - Copy the config.zip to the Solr server and extract. The unpacked
configuration directory is named
$CONFin these instructions.
Now you can create a Solr core using this config-set on a running Solr server. There're different ways to do so. For most Linux distributions you can run
sudo -u solr $SOLR/bin/solr create_core -c $CORE -d $CONF -n $CORE
You will see something like
$ sudo -u solr /opt/solr/bin/solr create_core -c test-core -d /tmp/solr-conf -n test-core
Copying configuration to new core instance directory:
/var/solr/data/test-core
If you're forced to create the core before you can run Drupal to generate the
config-set you could also use the appropriate jump-start config-set you'll
find in the jump-start directory of this module.
You must not create a core without a proper drupal config-set! If you do so - even by accident - you won't recognize it immediately. But you'll run into trouble like this soon: SolrException: Can not use FieldCache on multivalued field: boost_document
Note: Every time you add a new language to your Drupal instance or add a custom
Solr Field Type you have to update your core configuration files. Using the
example above they will be located in /var/solr/data/test-core/conf. The Drupal
admin UI should inform you about the requirement to update the configuration.
Reload the core after updating the config using
curl -k http://localhost:8983/solr/admin/cores?action=RELOAD&core=$CORE on
the command line or enable the search_api_admin sub-module to do it from the
Drupal admin UI.
Note: There's file called solrcore.properties within the set of generated
config files. If you need to fine tune some setting you should do it within this
file if possible instead of modifying solrconf.xml.
Afterwards, go to http://localhost:8983/solr/#/$CORE in your web browser to
ensure Solr is running correctly.
CAUTION! For production sites, it is vital that you somehow prevent outside access to the Solr server. Otherwise, attackers could read, corrupt or delete all your indexed data. Using the server as described below WON'T prevent this by default! If it is available, the probably easiest way of preventing this is to disable outside access to the ports used by Solr through your server's network configuration or through the use of a firewall. Other options include adding basic HTTP authentication or renaming the solr/ directory to a random string of characters and using that as the path.
For configuring indexes and searches you have to follow the documentation of search_api.
Instead of a single core you have to create a collection in your Solr Cloud instance. To do so you have to read the Solr handbook.
- Create a Search API Server according to the search_api documentation using "Solr" or "Multilingual Solr" as Backend and the "Solr Cloud" or "Solr Cloud with Basic Auth" Connector.
- Download the config.zip from the server's details page or by using
drush solr-gsc - Deploy the config.zip via zookeeper.
Note: The paths where the config.zip needs to be extracted to might differ from
the instructions above as well. For some distributions a directory like
/var/solr or /usr/local/solr exists.
This module contains a jump-start directory where you'll find a
docker-compose.yml files for various Solr versions. These use default
config-sets that will work for most drupal use-cases.
This variant is suitable for evaluation and development purposes.
These config-sets are also suitable for standard production use-cases without the need for advanced features or customizations.
Whenever you update your Solr installation it is recommended that you generate a new config-set and deploy it. The deployment depends on the installation variation you choose before. It is also recommended to re-index yur content after an update. But if it is a minor update it should be save to just queue all contentfor re-indexing.
When performing a major version update like from Solr 6 to Solr 8 it is recommended to delete the core or collection and recreate it like described in the installation instructions above.
All Search API datatypes are supported by using appropriate Solr datatypes for indexing them.
The "direct" parse mode for queries will result in the keys being directly used as the query to Solr using the Standard Parse Mode.
Adding Devel module (and optionally, addons like Kint) provides the site with a Solr Query Debugger and shows how content gets indexed.
Regarding third-party features, the following are supported:
- autocomplete
- Introduced by module: search_api_solr_autocomplete
- Lets you add autocompletion capabilities to search forms on the site.
- facets
- Introduced by module: facet
- Allows you to create facetted searches for dynamically filtering search results.
- more like this
- Introduced by module: search_api
- Lets you display items that are similar to a given one. Use, e.g., to create a "More like this" block for node pages build with Views.
- multisite
- Introduced by module: search_api_solr
- spellcheck
- Introduced by module: search_api_solr
- Views integration provided by search_api_spellcheck
- attachments
- Introduced by module: search_api_attachments
- location
- Introduced by module: search_api_location
- NLP
- Introduced by module: search_api_solr_nlp
- Adds more fulltext field types based on natural language processing, for example field types that filter all word which aren't nouns. This is great for auto completion.
If you feel some service option is missing, or have other ideas for improving this implementation, please file a feature request in the project's issue queue, at https://drupal.org/project/issues/search_api_solr.
Please consider that, since Solr handles tokenizing, stemming and other preprocessing tasks, activating any preprocessors in a search index' settings is usually not needed or even cumbersome. If you are adding an index to a Solr server you should therefore then disable all processors which handle such classic preprocessing tasks.
If you create a new index, such processors won't be offered anymore since 8.x-2.0.
But the remaining processors are useful and should be activated. For example the HTML filter or the Highlighting processor.
By default the Highlighting processor provided by Search API uses PHP to create
highlighted snippets or an excerpt based on the entities loaded from the
database. Solr itself can do that much better, especially for different
languages. If you check Retrieve result data from Solr and Highlight retrieved data on the index edit page, the Highlighting processor will use
this data directly and bypass it's own logic. To do the highlighting, Solr will
use the configuration of the Highlighting processor.
The communication details between Drupal and Solr is implemented by connectors. This module includes:
- Standard Connector
- BasicAuth Connector
- Solr Cloud Connector
- Solr Cloud BasicAuth Connector
There are service provider specific connectors available, for example from Acquia, Pantheon, hosted solr, platform.sh, and others. Please contact your provider for details if you don't run your own Solr server.
It's highly recommended that you don't modify the schema.xml and solrconfig.xml files manually because this module dynamically generates them for you.
Most features that can be configured within these config files are reflected by drupal configs that could be handled via drupal's own config management.
You can also create your own Solr field types by providing additional field config YAML files. Have a look at this module's config folder to see examples.
Such field types can target a specific Solr version and a "domain". For example "Apple" means two different things in a "fruits" domain or a "computer" domain.
When displaying search results from Solr in Views using the Search API Views integration, you have the choice to fetch the displayed values from Solr by enabling "Retrieve result data from Solr" on the server edit page. Otherwise Solr will only return the IDs and Search API loads the values from the database.
If you decide to retrieve the values from Solr you have to enable "Skip item access checks" in the query options in the views advanced settings. Otherwise the database objects will be loaded again for this check. It's obvious that you have to apply required access checks during indexing in this setup. For example using the corresponding processor or by having different indexes for different user roles.
In general it's recommended to disable the Views cache. By default the Solr search index is updated asynchronously from Drupal, and this interferes with the Views cache. Having the cache enabled will cause stale results to be shown, and new content might not show up at all.
In case you really need caching (for example because you are showing some search results on your front page) then you use the 'Search API (time based)' cache plugin. This will make sure the cache is cleared at certain time intervals, so your results will remain relevant. This can work well for views that have no exposed filters and are set up by site administrators.
Since 8.x-2.0 in combination with Solr 6.6 or higher you can also use the 'Search API (tag based)' cache. But in this case you need to ensure that you enable "Finalize index before first search" and "Wait for commit after last finalization" in the "Solr specific index options".
But be aware that this will slow down the first search after any modification to an index. So you have to choose if no caching or tag based caching in combination with finalization is the better solution for your setup. The decision depends on how frequent index modification happen or how expensive your queries are.
If you index some drupal fields multiple times in the same index and modify the single values differently via our API before the values get indexed, you'll notice that Views will randomly output the same value for all of these fields if you enabled "Retrieve result data from Solr". In this case you have to enable the "Solr dummy fields" processor and add as many dummy fields to the index as you require. Afterwards you should manipulate these fields via API.
Facetting on fulltext fields is not yet supported. We recommend the use of string fields for that purpose.
Trie based field types were deprecated in Solr 6 and with Solr 7 we switched to the point based equivalents. But lucene doesn't support a mincount of "0" for these field types. We recommend the use of string fields instead of numeric ones for that purpose.
If updating from Search API Solr 8.x-1.x or from Solr versions before 7 to Solr 7 or 8, check your Search API index' field configurations to avoid these errors that will lead to exceptions and zero results.
Support is currently provided via our issue queue or on https://drupalchat.me/channel/search.
Whenever you need to enhance the functionality you should do it using the API instead of extending the SearchApiSolrBackend class!
To customize connection-specific things you should provide your own implementation of the \Drupal\search_api_solr\SolrConnectorInterface.
A lot of customization can be achieved using YAML files and drupal's configuration management.
We leverage the solarium library. You can also interact with solarium's API using our hooks and callbacks or via event listeners. This way you can for example add any solr specific parameter to a query you need.
But if you create Search API Queries by yourself in code there's an easier way. You can simply set the required parameter as option prefixed by 'solr_param_'.
So these two lines are "similar":
$search_api_query->setOption('solr_param_mm', '75%');
$solarium_query->setParam('mm', '75%');
Our test suite includes integration tests that require a real Solr server. This requirement can't be provided by the drupal.org test infrastructure. Therefore we leverage github workflows for our tests and had to establish a more complex workflow:
- open an issue on drupal.org as usual
- upload the patch for being reviewed to that issue on drupal.org as usual
- fork https://github.com/mkalkbrenner/search_api_solr
- apply your patch and file a PR on github
- add a link to the github PR to the drupal.org issue
The PR on github will automatically be tested on github and the test results will be reflected in the PR conversation.
This module comes with a suite of automated tests. To execute those, you just need to have a (correctly configured) Solr instance running at the following address:
http://localhost:8983/solr/drupal
This represents a core named "drupal" in a default installation of Solr.
As long as you're changes don't modify the config-set generation you could
leverage docker, too. You'll find ready to use docker-compose files in the
jump-start directory.
The tests themselves could be started by running something like this in your drupal folder:
phpunit -c core --group search_api_solr
(The exact command varies on your setup and paths.)