Hello,

Investigating a backup failure here is what we found out :

SELECT * FROM information_schema.metadata_lock_info;
-----------------------------------------------------------------------------------+

-----------------------------------------------------------------------------------+

-----------------------------------------------------------------------------------+
2 rows in set (0.000 sec)

Thead Id 2

------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

A workaround is to disable event scheduler before taking backup

looking at the content of events we have purging process like
PROCEDURE `p_purge_syslog`(IN max_rows_deleted int unsigned)
COMMENT 'Remove rows older than 3 months'
BEGIN
IF (get_lock('p_purge_syslog', 0) = 1) THEN
DELETE FROM syslog.logs WHERE date < current_date - INTERVAL 3 MONTH LIMIT max_rows_deleted;
SELECT release_lock('p_purge_syslog');
END IF;

syslog.logs is engine innodb

PROCEDURE `p_purge_slow_query_log`(max_rows_deleted int unsigned)
COMMENT 'Remove rows older than 3 week'
BEGIN
IF (get_lock('p_purge_slow_query_log', 0) = 1) THEN

DELETE FROM global_query_review_history WHERE ts_min < current_date - INTERVAL 3 WEEK LIMIT max_rows_deleted;

SELECT release_lock('p_purge_slow_query_log');

END IF;

slow_query_log.global_query_review_history is innodb engine

Is that expected to have MDL lock on mysql.event when the scheduler is running or is this something inside the procedure code that is blocking backups ?