-
Luis Soares authored
names in replicated statements BUG#50653: drop procedure implicitely treats db name in a case sensitive way This bug revealed itself while using case sensitive filesystems and exhibited two symptoms: 1. If setting lower_case_table_names=1 on the slave, but not on the master, this setting will not convert database name in replicated statements, ultimately breaking replication; 2. The same problem for symptom 1. surfaced in RBR, but this time for table names, as these would not be converted to lower case for row based replication events. Symptom 1. is addressed by conditionally converting to lower case, database name on Query_log_event constructor and Load_log_event::do_apply_event. Symptom 2. is addressed by conditionally converting to lower database name and table name when processing Table_map_log_event. A side note on BUG#50653. It's basically a duplicate of BUG#37656 but I included a test case for it in this patch.
Luis Soares authorednames in replicated statements BUG#50653: drop procedure implicitely treats db name in a case sensitive way This bug revealed itself while using case sensitive filesystems and exhibited two symptoms: 1. If setting lower_case_table_names=1 on the slave, but not on the master, this setting will not convert database name in replicated statements, ultimately breaking replication; 2. The same problem for symptom 1. surfaced in RBR, but this time for table names, as these would not be converted to lower case for row based replication events. Symptom 1. is addressed by conditionally converting to lower case, database name on Query_log_event constructor and Load_log_event::do_apply_event. Symptom 2. is addressed by conditionally converting to lower database name and table name when processing Table_map_log_event. A side note on BUG#50653. It's basically a duplicate of BUG#37656 but I included a test case for it in this patch.
Loading