Logical_clock Concurrent Replication mysql5.7 Resolve replication Latency Issues

Tag: Delay indicates worker ati repos wait GIC cut read

slave-parallel-type=Logical_clock # #默认database, Logical_clock There may be bugs, 5.7.19 Fixed master-slave inconsistency bug

     * Replication:in Certain cases, the master could write to the       binary log a last_committed value which was smaller       than it should has been. This could cause the slave       into parallel transactions which should not having        been, leading to in Consistencies or other errors. (bug# 84471, Bug #253796595.7. 19 Version Repair

slave-parallel-Workers= 4 # Based on the concurrency setting, if concurrency is not required, set to 0 and do not set to 1. The performance of 1 is worse than 0. previous SQL single thread becomes 4 threadsWaiting for a event from coordinator

Mysql>show Processlist;+----+-------------+-----------+------+---------+--------+----------------------------------------------------- ---+------------------+| Id | User | Host | db | Command | Time | State | Info |+----+-------------+-----------+------+---------+--------+---------------------------------------------- ----------+------------------+| -|           System user | | NULL | Connect |606530| Slave have read all relay log; Waiting forMore Updates | NULL | | +|           System user | | NULL | Connect |606593| Waiting forAnEvent  fromCoordinator | NULL | | -|           System user | | NULL | Connect |838818| Waiting forAnEvent  fromCoordinator | NULL | | +|           System user | | NULL | Connect |841404| Waiting forAnEvent  fromCoordinator | NULL | | A|           System user | | NULL | Connect |841404| Waiting forAnEvent  fromCoordinator | NULL | | to|           System user | | NULL | Connect |862799| Waiting forMaster to sendEvent| NULL | | +| Root | localhost | NULL | Query |0| Starting | Show Processlist |+----+-------------+-----------+------+---------+--------+------------------------------------ --------------------+------------------+7Rowsinch Set(0.06Sec

It is not strictly required that all transactions that are concurrently in the prepare phase are executed concurrently in the repository, so that there may be different seq No in a set of committed transactions

You can find that more than the original binary log content last_committed and sequence_number,last_committed represent the time the transaction commits,

The number of the last transaction commit, if the transaction has the same last_committed, indicating that the transactions are within a group and can be played back in parallel.

#170623 22:27:43 server ID  end_log_pos 2781491 CRC32 0x1dcd41fa     GTID    last_committed=816      sequence_ number=817#170623 22:27:43 server ID  end_log_pos 2785022 CRC32 0x2d28e1b7 GTID    last_committed=816      sequence_number=818

master_info_repository=TABLE relay_log_info_repository=TABLE relay_log_recovery= on

Logical_clock Concurrent Replication mysql5.7 Resolve replication Latency Issues