I have 4 MariaDB replication slaves sync'ed with single master via SSL. Slaves keep up-to-date, until somehow they permanently lag behind a certain amount of updates. Has anyone seen this before? Here's what I mean. On the master:
SHOW MASTER STATUS \G
displays its current log position as
881134, then on each slave:
SHOW SLAVE STATUS \G
displays the master's
Exec_Master_Log_Pos: 862183. As you can see, the 2 numbers are quite different. It never catches up on its own, no matter how long I wait.
If I issue a few more updates to the master, eventually the slaves will catch up to where I'm expecting them to, but they still remain several updates behind the master.
If I restart the slave MariaDBs, they catch up immediately and stay caught up for a few days until it happens again.
Things I have checked: No errors reported by the replication. There is no CPU usage and no slow queries holding things up. There is plenty of free memory and disk space on all MariaDB hosts. No major network latency or outage. Nothing in the MariaDB logs. All slaves have separate
server_id settings. All are running
INNODB. All system clocks are synchronized within a second of each other.