Fix data race while accessing connection in partitionProducer

[wolfstudy]

Signed-off-by: xiaolongran [email protected]

Motivation

In #700, we use a separate go rutine to handle the logic of reconnect, so here you may encounter the same data race problem as #535

Modifications

Now, the conn field is read and written atomically; avoiding race conditions.

[wolfstudy]

cc @dferstay PTAL thanks

[freeznet]

maybe clean up the cached connection when the cnx is closed? otherwise _getConn() will return a closed connection.

[wolfstudy]

Looks like #703 is working on this

[wolfstudy]

@freeznet Looking at the implementation of the current Close method, when triggering the logic of Close, the state of the current connection will first be set to connectionClosed. When GetConnection() is executed, if the state of the connection is detected to be closed, the connection will be removed from the cache.

func (c *connection) Close() {
	c.closeOnce.Do(func() {
		c.Lock()
		cnx := c.cnx
		c.Unlock()
		c.changeState(connectionClosed)

if conn.closed() {
			delete(p.connections, key)
			p.log.Infof("Removed connection from pool key=%s logical_addr=%+v physical_addr=%+v",
				key, conn.logicalAddr, conn.physicalAddr)
			conn = nil // set to nil so we create a new one
		}

[leizhiyuan]

LGTM