From d644709b22bd1872e78a0938dec7eb77d5ed03f3 Mon Sep 17 00:00:00 2001
From: 6543 <6543@obermui.de>
Date: Sat, 8 May 2021 20:27:00 +0200
Subject: [PATCH] Exponential Backoff for ByteFIFO (#15724) (#15793)

This PR is another in the vein of queue improvements. It suggests an
exponential backoff for bytefifo queues to reduce the load from queue
polling. This will mostly be useful for redis queues.

Signed-off-by: Andrew Thornton <art27@cantab.net>

Co-authored-by: Lauris BH <lauris@nix.lv>

Co-authored-by: zeripath <art27@cantab.net>
Co-authored-by: Lauris BH <lauris@nix.lv>
---
 modules/queue/queue_bytefifo.go | 88 ++++++++++++++++++++++-----------
 1 file changed, 59 insertions(+), 29 deletions(-)

diff --git a/modules/queue/queue_bytefifo.go b/modules/queue/queue_bytefifo.go
index fe5178ff2d..6f92e3cb6a 100644
--- a/modules/queue/queue_bytefifo.go
+++ b/modules/queue/queue_bytefifo.go
@@ -114,43 +114,73 @@ func (q *ByteFIFOQueue) Run(atShutdown, atTerminate func(context.Context, func()
 }
 
 func (q *ByteFIFOQueue) readToChan() {
+	// handle quick cancels
+	select {
+	case <-q.closed:
+		// tell the pool to shutdown.
+		q.cancel()
+		return
+	default:
+	}
+
+	backOffTime := time.Millisecond * 100
+	maxBackOffTime := time.Second * 3
 	for {
-		select {
-		case <-q.closed:
-			// tell the pool to shutdown.
-			q.cancel()
-			return
-		default:
-			q.lock.Lock()
-			bs, err := q.byteFIFO.Pop()
-			if err != nil {
-				q.lock.Unlock()
-				log.Error("%s: %s Error on Pop: %v", q.typ, q.name, err)
-				time.Sleep(time.Millisecond * 100)
-				continue
-			}
+		success, resetBackoff := q.doPop()
+		if resetBackoff {
+			backOffTime = 100 * time.Millisecond
+		}
 
-			if len(bs) == 0 {
-				q.lock.Unlock()
-				time.Sleep(time.Millisecond * 100)
-				continue
+		if success {
+			select {
+			case <-q.closed:
+				// tell the pool to shutdown.
+				q.cancel()
+				return
+			default:
 			}
-
-			data, err := unmarshalAs(bs, q.exemplar)
-			if err != nil {
-				log.Error("%s: %s Failed to unmarshal with error: %v", q.typ, q.name, err)
-				q.lock.Unlock()
-				time.Sleep(time.Millisecond * 100)
-				continue
+		} else {
+			select {
+			case <-q.closed:
+				// tell the pool to shutdown.
+				q.cancel()
+				return
+			case <-time.After(backOffTime):
+			}
+			backOffTime += backOffTime / 2
+			if backOffTime > maxBackOffTime {
+				backOffTime = maxBackOffTime
 			}
-
-			log.Trace("%s %s: Task found: %#v", q.typ, q.name, data)
-			q.WorkerPool.Push(data)
-			q.lock.Unlock()
 		}
 	}
 }
 
+func (q *ByteFIFOQueue) doPop() (success, resetBackoff bool) {
+	q.lock.Lock()
+	defer q.lock.Unlock()
+	bs, err := q.byteFIFO.Pop()
+	if err != nil {
+		log.Error("%s: %s Error on Pop: %v", q.typ, q.name, err)
+		return
+	}
+	if len(bs) == 0 {
+		return
+	}
+
+	resetBackoff = true
+
+	data, err := unmarshalAs(bs, q.exemplar)
+	if err != nil {
+		log.Error("%s: %s Failed to unmarshal with error: %v", q.typ, q.name, err)
+		return
+	}
+
+	log.Trace("%s %s: Task found: %#v", q.typ, q.name, data)
+	q.WorkerPool.Push(data)
+	success = true
+	return
+}
+
 // Shutdown processing from this queue
 func (q *ByteFIFOQueue) Shutdown() {
 	log.Trace("%s: %s Shutting down", q.typ, q.name)