CoreCommunity ExtensionsIncubatorDistributionsTYPO3 4.5 ProjectsTYPO3 4.6 ProjectsTYPO3 4.7 ProjectsTYPO3 6.0 ProjectsTYPO3 6.1 ProjectsTYPO3 6.2 Projects (+)
History

v1.3.0 to v2.0.0 Upgrade Guide

This upgrade guide also covers upgrades for dkd's early access version from 1.6.0 to 2.5.0.

Upgrading

  • Update the extension using the Extension Manager.
  • Update schema.xml and solrconfig.xml with the ones you find shipped with the extension. They are located in resources/solr/.
  • The new Solr configuration is a multicore configuration, previous versions of the extension shipped with singlecore and multicore configurations.
  • The path where the access filter plugin is located has changed in solrconfig.xml. Until now it was typo3/ in the solr home directory. The new location is typo3lib/, just rename the directory. For existing installation the solr home directory should be /opt/solr-tomcat/solr/.
  • There is a new plugin that replaces the old access filter plugin. The new plugin, at some point, will do more than just access checks. The new plugin is available at http://www.typo3-solr.com/fileadmin/files/solr/solr-typo3-plugin-1.2.0.jar . Remove the old plugin.
  • Restart Tomcat
  • Finally, go to the reports module and make sure everything is ok under the solr section.
  • Existing installations need to re-index their sites due to the changes in schema.xml and the page indexer

Important Notes

  • TypoScript option allowEmptyQuery has been moved into query configuration: plugin.tx_solr.search.query.allowEmptyQuery. Also it now does what it implies; allow or disallow empty user queries.
  • To achieve the behavior of the old allowEmptyQuery option use initializeWithEmptyQuery and showResultsOfInitialEmptyQuery instead now.
  • The site field now uses the first domain record of a site instead of the TYPO3_SITE_URL constant. If you have facets or filters configured on that field you may have to check them.
  • When using the Index Queue to index custom records, the default indexer now takes care of adding the most common fields to Solr documents by default already. Those fields are id (document ID), type (table name), appKey, site, siteHash, uid, pid, created (crdate), changed (tstamp), access, endtime. If you added any of those in your Index Queue field mappings yourself until now, you need to remove them since they are handled automatically now.
  • Handling and configuring the Index Queue Worker task is now easier as it is more obvious now that only one task is needed for each site. However some code obviously has been changed for that. As the scheduler stores the task objects in the DB (serialized) those try to invoke now missing methods when deserialized, leading to a fatal error. Solution is to remove the Index Queue Worker tasks from table tx_scheduler_task and re-create it in the scheduler BE module.
  • plugin.tx_solr.templateFile has changed to plugin.tx_solr.templateFiles
  • plugin.tx_solr.addDefaultCss and plugin.tx_solr.addDefaultJs have been removed. Use plugin.tx_solr.cssFiles and plugin.tx_solr.javascriptFiles arrays instead
  • plugin.tx_solr.search.highlighting has changed to plugin.tx_solr.search.results.resultsHighlighting
  • change your setting for plugin.tx_solr.search.commonTerms to plugin.tx_solr.search.frequentSearches

New Installations

For new installations the install script in EXT:solr/resources/shell/ will do all the necessary steps for you.