Description:
MySQL master:
set expire_logs_days = 30
its binlog :
mysql_bin.000003 #30 days ago
mysql_bin.000004 #new binlog
MySQL master have many days no data to write .
MySQL slave replication;
replication I/O position "mysql_bin.000003, position 18365907"
master -> slave replication status is ok.
There will be a problem in the following scene：
①master -> slave ，slave crash
②master , begin to write data
mysql_error.log:
2018-03-20T16:30:28.165334+08:00 1038542 [Warning] Timeout waiting for reply of binlog (file: mysql_bin.000004, pos: 1085), semi-sync up to file mysql_bin.000003, position 18365907.
2018-03-20T16:30:28.165363+08:00 1038542 [Note] Semi-sync replication switched OFF.
master do this:
signal_update();
rotate_and_purge(RP_LOCK_LOG_IS_ALREADY_LOCKED);
"mysql_bin.000003" was removed more than 30 days later
③slave started，replication reoprt a error:
Last_IO_Error: Got fatal error 1236 from master when reading data from binary log: 'Could not find first log file name in binary log index file'
How to repeat:
As shown above
Suggested fix:
Suggest on master set a flag ,when semi-sync binlog to slave fail，don't to purge binlog at semi-sync fail binlog .
Description: MySQL master: set expire_logs_days = 30 its binlog : mysql_bin.000003 #30 days ago mysql_bin.000004 #new binlog MySQL master have many days no data to write . MySQL slave replication; replication I/O position "mysql_bin.000003, position 18365907" master -> slave replication status is ok. There will be a problem in the following scene： ①master -> slave ，slave crash ②master , begin to write data mysql_error.log: 2018-03-20T16:30:28.165334+08:00 1038542 [Warning] Timeout waiting for reply of binlog (file: mysql_bin.000004, pos: 1085), semi-sync up to file mysql_bin.000003, position 18365907. 2018-03-20T16:30:28.165363+08:00 1038542 [Note] Semi-sync replication switched OFF. master do this: signal_update(); rotate_and_purge(RP_LOCK_LOG_IS_ALREADY_LOCKED); "mysql_bin.000003" was removed more than 30 days later ③slave started，replication reoprt a error: Last_IO_Error: Got fatal error 1236 from master when reading data from binary log: 'Could not find first log file name in binary log index file' How to repeat: As shown above Suggested fix: Suggest on master set a flag ,when semi-sync binlog to slave fail，don't to purge binlog at semi-sync fail binlog .