We're running AWS RDS(mysql) with 2 replicas and need more replicas.
However, we've failed making the 3rd replica that is synchronized with master.
When I create new 3rd replica, its replica lag keeps increased forever.
What we've tried were,
- making same instance type of all replicas with master one (r3.4xlarge)
- tweak innodb_flush_log_at_trx_commit, sync_binlog to 0 on replica side.
- tried creating new replica when write operation is the lowest (870 write IOPS, 191.89 inserts/s, 355.97 updates/s, 25.45 deletes/s, 139477.05 reads/s)
show processlist tells
| 2 | system user | | NULL | Connect | 4816 | Waiting for master to send event | NULL |
| 3 | system user | | NULL | Connect | 7823 | System lock | NULL |
or DML queries are processed.
It seems like replica is processing bin log so slowly.
However other replicas (1st are 2nd working okay)
Can anyone have experienced similar issue or know what's the problem?
Thank you
