Details
-
Bug
-
Status: Done
-
Medium
-
Resolution: Fixed
-
8.0.x
-
None
-
None
-
Yes
Description
While testing PXC-8.0.23, it is seen that the cluster break with a sequence of SQLs when binlog encryption is enabled. To further investigate, the same sequence of SQLs were executed against a PS server in a normal master-slave setup and the issue was successfully reproduced.
The problem was successfully verified on PS-8.0.25. Due to this the PXC versions are also impacted. Upstream versions does not have the issue. Verfied on both upstream 8.0.23,8.0.24 versions.
To repeat the issue. Follow below steps:
Step 1: Start a 2 node server using MTR. Make sure to enable binlog encryption ./mtr rpl_encryption --start --mysqld=--early-plugin-load=keyring_file.so --mysqld=--binlog-encryption=ON --mysqld=--keyring_file_data=keyring Step 2: Execute on running server on port 13002 CHANGE MASTER TO master_host='127.0.0.1',master_port=13000,master_user='root'; START SLAVE; SHOW replica status\G Step 3: On master node (port: 13000) drop database test; source ~/my_reduced.sql; SHOW replica status\G Last_SQL_Error: Could not execute Delete_rows event on table test.tt_8; Corrupted replication event was detected, Error_code: 1610; Got error 1610 - 'Unknown error 1610' from storage engine, Error_code: 1030; handler error No Error!; the event's master log master-bin.000001, end_log_pos 418338
