Moving collections between Funnelback servers

Background

This article gives guidelines about moving Funnelback collections between servers. This is only supported if the source and target servers are running the same Funnelback version.

Copy the collection configuration folder

  • Retrieve the complete collection configuration folder from $SEARCH_HOME/conf/<COLLECTION-ID>/ on the source server

  • Copy the complete folder onto the target server, at the same location $SEARCH_HOME/conf/<COLLECTION-ID>/

Re-create the collection on the target server

To re-create the collection from the existing configuration, run the following command:

Linux:

$ $SEARCH_HOME/linbin/ActivePerl/bin/perl $SEARCH_HOME/bin/create-collection.pl $SEARCH_HOME/conf/<COLLECTION-ID>/collection.cfg

Windows:

$ %SEARCH_HOME%\wbin\ActivePerl\bin\perl %SEARCH_HOME%\bin\create-collection.pl %SEARCH_HOME%\conf\<COLLECTION-ID>\collection.cfg

This will re-create all the folders associated with the collection, such as $SEARCH_HOME/data/<COLLECTION-ID>/, $SEARCH_HOME/admin/reports/<COLLECTION-ID>/

Copy additional data from the source server

At this point the collection is created but doesn’t contain any data or search indexes. If there is no data to transfer from the source server an update can be triggered on the newly created collection to start crawling content.

Additional data can be copied from the source server:

  • Search indexes and data files: To be able to search the collection immediately without having to update i

  • Analytics logs: To make sure historical logs are preserved for future analytics updates

  • Update history database: To preserve the update history reports in the administration interface

  • Analytics database: To view analytics reports without having to rebuild them from the logs

  • WCAG database: To preserve existing WCAG reports

Copy search indexes, data and analytics logs

The search indexes and data are located under $SEARCH_HOME/data/<COLLECTION-ID>/. This folder contains both the live indexes & data (used for querying) and the offline ones (from the previous update). Either both views or a single can be copied depending on your needs.

The live/ and offline/ folders are actually symbolic links to the one/ and two/ folders. These links are swapped during an update to make the new index live.

Do not copy live/ and offline/ directly as the copy will create actual folders that will prevent Funnelback from swapping the views of the collection during the next update.

Instead, make sure that the sub-folders of live/ and offline/ are copied individually (e.g. idx/, log/, data/) or only copy the one and two folders then recreate the symbolic links.

To restore the full collection, copy each sub-folder of the live/ and offline/ folders (e.g. live/idx/ from the source server should be copied to live/idx/ in the target server).

This will restore the search indexes, crawled data, update logs, analytics archived logs.

Copy update history, analytics & WCAG databases

The database are located under $SEARCH_HOME/admin/reports/<COLLECTION-ID>/. The content of this folder can be copied to the target folder to restore the databases.

The update history database may already exist in the target server as it will have been re-created with the collection. It can be safely overwritten.

Troubleshooting

If the live/ and offline/ symbolic links have been transformed to actual folders, Funnelback may get stuck repeatedly attempting to swap the views for the collection. As a result the collection cannot be updated nor deleted via the administration interface.

To resolve this situation:

  • Delete the collection from the administration interface

  • Ensure the collection has been correctly deleted: The $SEARCH_HOME/conf/<COLLECTION-ID>/ and $SEARCH_HOME/data/<COLLECTION-ID>/ folders should have been removed. If they are still present, delete them manually.

  • Re-create the collection by following the instructions of this article