I have three database server SQL1 SQL2, SQL3 form a group_replication cluster, because the network is not stable, SQL3 pulled out of the cluster, use start group_replication wanted him to rejoin the error:
This member has more executed the transactions than those present in the group
My understanding is: SQL3 exit cluster before a transaction carried out locally, has not been synchronized to SQL1, SQL2 exit, SQL3 SQL1 and SQL2 no transaction caused the SQL3 cannot join the cluster,
Online checked this question a lot of the solution is to reset the master. Particular way is:
turn off the replication,
The SQL1 or SQL2 data manual reduction to the SQL3, make three computer database content is consistent,
Then the three machines are reset to set up group_repliction master,
The question now is, when the above operation for large database, need data backup restore operation time is larger, and the data backup to reset before the master can't executed against the database write operations or content of the database will not consistent, so should not suitable for use in production this way,
So is there any way that can be without close SQL1 and SQL2 cluster add SQL3 to cluster? My idea is let SQL3 join where to set his local many transaction synchronization for SQL1 and SQL2, but in the official documentation check for a long time didn't find related content, please understand friend also glad, thank you!