I'm interested in learning more about the replication code, specifically
more about the part where the replication on a slave is happening in a
multi-threaded environment (released in 5.6) is implemented.
Where and how does the code blocks parallel execution once the query is
decided as it to be on the same database. Is there a lock per database? Or
is there a unique thread per database?
In any case, can you point me more specifically to the code? I've been
trying to look over it and I am a bit stuck around rpl_rli, rpl_slave,
|• Replication code||Alexandru Palade||16 Mar|