A PetScan query is currently used to generate the list of untagged files. This was done to avoid introducing a database dependency, especially when used in inrcli. However, PetScan has recently been unreliable leading to a backlog of untagged files. The PetScan query could be replaced with a MediaWiki search query and further refinement with database queries. A pure database query is not possible due to the need to search page text, and the full query is too long to do purely in CirrusSearch.
A PetScan query is currently used to generate the list of untagged files. This was done to avoid introducing a database dependency, especially when used in inrcli. However, PetScan has recently been unreliable leading to a backlog of untagged files. The PetScan query could be replaced with a MediaWiki search query and further refinement with database queries. A pure database query is not possible due to the need to search page text, and the full query is too long to do purely in CirrusSearch.