The feed query is making a subquery that gets all IDs for a given collection, and then sorting all the updated_at values, which is slow if the collection is large. This patch moves the timestamp into the documents table and creates an index for (collection, digested_at), to speed up feed queries.
The feed query is making a subquery that gets all IDs for a given collection, and then sorting all the
updated_at
values, which is slow if the collection is large. This patch moves the timestamp into the documents table and creates an index for(collection, digested_at)
, to speed up feed queries.