This documentation is for a version that has reached its End Of Life. Such versions are no longer supported and don't receive security updates. Consider updating to a newer version.
Indexing eContent data¶
eContent data can be indexed by executing the following command:
1 |
|
eContent Solr configuration uses two cores to keep search services available while indexing. This means that the indexer indexes eContent data in the back core, and then the indexer can be executed again with a swap parameter to swap the cores.
Command options | Results |
---|---|
-c 5000 |
Specifies the chunk size (how many elements are indexed at the same time). |
-r |
Deletes all previous eContent data. |
--siteaccess=import |
Instructs the indexer to take the product data from the temporary tables. |
--live-core |
Indexes directly on the live core (and not on the back core).If you use -r together --live-core , the delete operation takes place before the indexer starts, so the full indexed content is not be available while indexing. |
swap |
Does not index, but swaps the cores instead. |
Additional parameters¶
siso_search.default.index_econtent_languages |
Specifies the languages to index. |
siso_search.default.solr_spellcheck |
Enables or disables the Solr spellchecker (the "Did you mean?" functionality). |
siso_search.default.index_facet_fields |
Solr string fields are lowercased, therefore this configuration is used to specify an additional Solr _id field.ID fields are not lowercased, which results in facets that can preserve their original case.Field names must be specified without the _s suffix. |