一、官方介绍：

The value of this system variable can be changed while Group Replication is running, but the change only takes effect after you stop and restart Group Replication on the group member.

此系统变量的值可以在组复制运行时更改，但只有在停止并重新启动组成员上的组复制后，更改才会生效。

group_replication_recovery_compression_algorithms specifies the compression algorithms permitted for Group Replication distributed recovery connections for state transfer from a donor's binary log. The available algorithms are the same as for the protocol_compression_algorithms system variable. For more information, see Section 4.2.8, “Connection Compression Control”.

group_replication_recovery_compression_algorithms指定组复制分布式恢复连接允许的压缩算法，用于从施主的二进制日志进行状态传输。可用的算法与protocol_compression_agorithms系统变量相同。有关更多信息，请参阅第4.2.8节“连接压缩控制”。

This setting does not apply if the server has been set up to support cloning (see Section 18.5.4.2, “Cloning for Distributed Recovery”) and a remote cloning operation is used during distributed recovery. For this method of state transfer, the clone plugin's clone_enable_compression setting applies.

如果服务器已设置为支持克隆（请参阅第18.5.4.2节“分布式恢复的克隆”），并且在分布式恢复期间使用远程克隆操作，则此设置不适用。对于这种状态转移方法，将应用克隆插件的clone_enable_compression设置。

二、参数理解

默认情况下，连接是未压缩的，如果开启压缩，因为需要执行压缩和解压缩操作，所以会影响数据库的CPU负载。当mgr数据库节点之间带宽不充足时(例如不同机房、甚至异地机房节点），并且服务器cpu资源很充足的时候建议开启压缩 ，这样来提升异地机房节点分布式恢复的传输效率；

三、建议设置

当cpu资源充足，并且节点间带宽不足的时候，建议开启压缩

set  global group_replication_recovery_compression_algorithms=zlib

stop  group_replication;

start group_replication;