Connection leak when executing query with invalid parameters

Happens when running in sequential HA mode with Client Prepared Statement in a transaction.

Create prepared statement and bind parameter to non-serializable object:

try(PreparedStatement statement = connection.prepareStatement("SELECT * FROM test WHERE column = ?"))

{

    statement.setObject(1, new Object());

    statement.execute();

}

Executing such query and writing parameters into the OutputStream (i.e. when calling "ComQuery.sendSubCmd"), results in "java.io.NotSerializableException: java.lang.Object".

Stack trace:

java.io.NotSerializableException: java.lang.Object

	at java.io.ObjectOutputStream.writeObject0(ObjectOutputStream.java:1184)

	at java.io.ObjectOutputStream.writeObject(ObjectOutputStream.java:348)

	at org.mariadb.jdbc.internal.com.send.parameters.SerializableParameter.writeObjectToBytes(SerializableParameter.java:91)

	at org.mariadb.jdbc.internal.com.send.parameters.SerializableParameter.writeTo(SerializableParameter.java:80)

	at org.mariadb.jdbc.internal.com.send.ComQuery.sendSubCmd(ComQuery.java:85)

	at org.mariadb.jdbc.internal.protocol.AbstractQueryProtocol.executeQuery(AbstractQueryProtocol.java:285)

	at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)

	at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)

	at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)

	at java.lang.reflect.Method.invoke(Method.java:498)

	at org.mariadb.jdbc.internal.failover.AbstractMastersListener.invoke(AbstractMastersListener.java:398)

	at org.mariadb.jdbc.internal.failover.FailoverProxy.executeInvocation(FailoverProxy.java:278)

	at org.mariadb.jdbc.internal.failover.FailoverProxy.invoke(FailoverProxy.java:270)

	at com.sun.proxy.$Proxy8.executeQuery(Unknown Source)

	at org.mariadb.jdbc.ClientSidePreparedStatement.executeInternal(ClientSidePreparedStatement.java:221)

	at org.mariadb.jdbc.ClientSidePreparedStatement.execute(ClientSidePreparedStatement.java:157)

	at org.mariadb.jdbc.ClientSidePreparedStatement.executeUpdate(ClientSidePreparedStatement.java:192)

What is happening:

1. "AbstractQueryProtocol.handleIoException" method will mark the protocol as not connected, via: "connected = false;"
2. FailoverProxy.executeInvocation catches the exception and enters the block:

   if (hasToHandleFailover(queryException)) {

   return handleFailOver(queryException, method, args, protocol);

   }

3. MastersFailoverLustener.primaryFail will reconnect with a new protocol, but because the protocol returns false on "currentProtocol.isConnected()", existing protocol is not closed.

   @Override

   public HandleErrorResult primaryFail(Method method, Object[] args, boolean killCmd) {

   boolean alreadyClosed = !currentProtocol.isConnected();

   boolean inTransaction = currentProtocol != null && currentProtocol.inTransaction();

   if (currentProtocol.isConnected()) {

   currentProtocol.close();

   }

   try {

   reconnectFailedConnection(new SearchFilter(true, false));

4. During reconnection in "reconnectFailedConnection" method, "MastersFailoverListener.foundActiveMaster" method is called, which override current protocol's reference, without closing existing one, as "currentProtocol.isClosed()" returns true

   if (currentProtocol != null && !currentProtocol.isClosed()) {

   currentProtocol.close();

   }

   currentProtocol = protocol;

Results in hanging open connection to database every time you run into above scenario. Calling "close" on the Connection, does not close the old protocol.

I would expect that "AbstractQueryProtocol.handleIoException" method to not assume that IOExceptions is due to connection being broken and actually close the protocol and not just flag it as "closed".