This repo will holds a basic solr config, schema and xslt to use as a starting point for future projects.
It is now dependent on the discoverygarden GSearch extensions--which includes the Joda time library.
If one wishes to index Drupal content and users, one might process the conf/data-import-config.xml.erb
into conf/data-import-config.xml
. It takes three parameters:
drupal_dbname
drupal_db_username
drupal_db_password
The 4.10.x branch is setup to utilize solr 4.10ish configurations. These have been successfully tested with solr 4.10.3.
One major difference between solr 3.x and 4.x is when deploying configs they used to go to /usr/local/fedora/solr now they get deployed to /usr/local/fedora/solr/collection1 instead.
You also have to copy the jars from solr/example/lib/ext/* to $CATALINA_HOME/webapps/solr/WEB-INF/lib/ for more information see https://wiki.apache.org/solr/SolrLogging#Using_the_example_logging_setup_in_containers_other_than_Jetty
Requires Gsearch 2.8+
See the wiki page for installation details.
In our gsearch fork as of version 2.9.0, we allow for an addition custom_parameters.properties
file to be placed beside the foxmlToSolr.xslt
file (or whatever the "top-level" XSLT is named, when deployed).
Parameter | Default | Description |
---|---|---|
index_ancestors |
false |
Boolean flag: true to produce an ancestors_ms field (as used by the islandora_collection_search module); otherwise, false avoid generating. Also, note: When migrating objects between collections, it would be necessary to update all descendents to ensure their list of ancestors reflect the current state... We do this in the islandora_collection_search module when migrating, instead of reindexing all the descendents whenever indexing an object (updating a collection label would be fairly expensive if we blindly reindexed). |
index_ancestors_models |
false |
Boolean flag: true to produce an ancestors_models_ms field (as used by the islandora_child_filter module; otherwise, false avoid generating. NOTE: A triplestore with Sparql 1.1 is required for the query to work. |
maintain_dataset_latest_version_flag |
false |
Boolean flag: true to produce a mmv_is_latest_b field, as used by the islandora_research_data module; however, we are not given the opportunity to promote another version should the latest be purged (similarly described in the README.md for islandora_research_data ). NOTE: Due to how this is calculated, it requires an RI which supports Sparql 1.1 (such as Blazegraph). |
index_compound_sequence |
true |
Boolean flag: false to remove RELS_EXT_isSequenceNumberOf<pid>_literal* and RELS_EXT_http://islandora.ca/ontology/relsext#isSequenceNumberOf<pid>_literal_* fields, as used by the islandora_solr_table_of_contents module. NOTE: It is a known fact this field on large repositories will cause significant performance issues with he Luke Request Handler due to the fact it creates multiple Solr fields for each compound relationship. |
index_checksums |
false |
Boolean flag: true to produce fields used by the islandora_checksum_duplicate_files module; otherwise false to avoid generating. NOTE: Checksums will need to be enabled either on a repository level or via the islandora_checksum module to have any effect. |