guarentee batch doesn't starve

add more comments (will need to cleanup later)

net_sink.go

@@ -278,12 +278,15 @@ func (s *netSink) run() {
 
 	var reconnectFailed bool // true if last reconnect failed
 
-	batchSize := GetSettings().BatchSize
+	settings := GetSettings()
+
+	batchSize := settings.BatchSize
 	isBatchEnabled := batchSize > 0
 	batch := make([]bytes.Buffer, 0, batchSize+cap(s.outc)) // expand allocation to allow for draining outc. todo change to an array to optimize
 	sendBatch := false
 
 	t := time.NewTicker(flushInterval)
+	bt := time.NewTicker(time.Duration(settings.FlushIntervalS) * time.Second) // todo maybe use a new configuration for batch flush interval
 	defer t.Stop()
 
 	// flush all/any buffered stats (on specified ticker) and send loop
@@ -322,11 +325,11 @@ func (s *netSink) run() {
 		}
 
 		// send batched outc data anytime indicated or the batch is full
-		if sendBatch || len(batch) == batchSize {
+		if sendBatch || len(batch) >= batchSize {
 			var err error
 			batch, err = s.sendBatch(batch)
 			if err != nil {
-				sendBatch = true // guarantee we conitnue to complete sending the batch after retrying and before procecessing anything else
+				sendBatch = true // guarentee for all cases that we complete sending the batch after retrying and before procecessing anything else
 				continue         // cut the iteration to process retries (nothing can be sent anyway since writeToConn failure with set s.conn to nil)
 			}
 			sendBatch = false
@@ -356,6 +359,7 @@ func (s *netSink) run() {
 				if err := s.send(buf); err == nil {
 					putBuffer(buf)
 				}
+				// if an error occurs we send the metric to the retry channel and make s.conn nil, breaking this loop and preventing further batch processing in this stack
 			}
 
 			close(done)
@@ -372,6 +376,11 @@ func (s *netSink) run() {
 			}
 		}
 
+		select {
+		case <-bt.C:
+			sendBatch = true
+		default:
+		}
 	}
 }