Connecting to MySQL Server 8.0.3 using Connector/J resulted in an SQLException: Unknown system variable 'query_cache_size', because the query cache has been removed from the server. All query cache server variables and functionality has now been removed from Connector/J 5.1. (Bug #26399958)

Updated the timezone mappings using the latest IANA and CLDR time zone databases. (Bug #25946965)

A new option for the loadBalancingStrategy connection property called serverAffinity has been added. The servers listed in the new connection property serverAffinityOrder (which should be a subset of the servers in the host list of the connection URL) are contacted in the order they are listed until a server is available or until the list of servers is exhausted, at which point a random load-balancing strategy is used with the hosts not listed by serverAffinityOrder. See descriptions for loadBalancingStrategy and serverAffinityOrder in Configuration Properties for Connector/J for details. (Bug #20182108)

Secure JDBC connections did not fall back to the default truststore when a custom one was not provided. (Bug #26243128)

Connector/J failed a number of regression tests in the testsuite related to geographic information system (GIS) functions because of changes to GIS support by the MySQL server. The fix corrects the tests. (Bug #26239946, Bug #26140577)

Attempts to connect to a server started with collation utf8mb4_de_pb_0900_ai_ci resulted in null pointer exceptions. (Bug #26090721)

In com/mysql/jdbc/ServerPreparedStatement.java, the arguments resultSetType and resultSetConcurrency for a call of Connection.preparedStatement() were swapped. (Bug #25874048, Bug #85885)

A NullPointerException was returned when getDate(), getTime(), or getTimestamp() was called with a null Calendar. This fix makes Connector/J throw an SQLException in the case. (Bug #25650305)

Some JDBC proxied objects were missing the proper handlings of the equals() methods, thus even comparison of one of these proxied objects to its own self with equals() yielded false. This patch introduces proper handlings for the equals() method in all the relevant proxies. (Bug #21931572, Bug #78313)

A server-side prepared statement was not closed when the same statement was being prepared again while the original statement was being cached. This was caused by the silent replacement of the cache entry of the old statement by the new. When this happened repeatedly, it caused eventually the complaint that max_prepared_stmt_count was exceeded. This fix makes sure that when a cache entry for a statement replaces an older one, the older statement is immediately closed. (Bug #20066806, Bug #74932)