Details
-
Bug
-
Status: Closed (View Workflow)
-
Major
-
Resolution: Fixed
-
2.7.10
-
None
Description
I'm using Aurora. When using two or more multi replica read db, a problem occurs when sending a kill query when query timeout occurs.
The problem occurs because of Aurora's cluster endpoint policy. Aurora provides a cluster endpoint for read db load balance, and the cluster endpoint's IP connected to DNS changes sequentially.
MariaDB Driver sends a kill query using domain. As a result, a situation occurs where a kill query may be sent to a db other than the db where the actual query is performed.
To solve this problem, I request that the kill query be sent to the IP rather than the domain.
I request changes as shown below.
copiedProtocol.setHostAddress(new HostAddress(this.getSocket().getInetAddress().getCanonicalHostName(), this.getHostAddress().port, this.getHostAddress().type));