Sorry , comment of #2 is incorrect.
It's ok to call after_flush after signal_update because the LOCK_log was hold by the leader thread.
(It's a problem for me because I have split the LOCK_log for binlog write/dump).
Sorry , comment of #2 is incorrect.
It's ok to call after_flush after signal_update because the LOCK_log was hold by the leader thread.
(It's a problem for me because I have split the LOCK_log for binlog write/dump).