Building search indexes on a mirror server usually takes a lot of time with the current indexer on an average mirror server.
A solution would be to load a backup of Solr data instead. These search indexes must correspond to the exported MB DB or any more recent state of the MB DB.
Only then, replication can be run along live indexing to keep both mirrored MB DB and local search indexes synchronous.
Since Solr almost instantly updates its data when MB DB changes, a dump of Solr data taken right after MB DB full export would be satisfactory, that's why it should be triggered from MusicBrainz cron rather than independently.