Merge pull request #129 from hyperledger/fix-127

Do not keep dispatching after block listener falls behind

Author: matthew1001

internal/confirmations/confirmed_block_listener.go

@@ -194,12 +194,12 @@ func (cbl *confirmedBlockListener) processBlockNotification(block *apitypes.Bloc
 	var dispatchHead *apitypes.BlockInfo
 	if len(cbl.newHeadToAdd) > 0 {
 		// we've snuck in multiple notifications while the dispatcher is busy... don't add indefinitely to this list
-		if len(cbl.newHeadToAdd) > 10 /* not considered worth adding/explaining a tuning property for this */ {
+		if len(cbl.newHeadToAdd) >= 10 /* not considered worth adding/explaining a tuning property for this */ {
 			log.L(cbl.ctx).Infof("Block listener fell behind head of chain")
-			cbl.newHeadToAdd = nil
-		} else {
-			dispatchHead = cbl.newHeadToAdd[len(cbl.newHeadToAdd)-1]
+			// Nothing more we can do in this function until it catches up - we just have to discard the notification
+			return
 		}
+		dispatchHead = cbl.newHeadToAdd[len(cbl.newHeadToAdd)-1]
 	}
 	if dispatchHead == nil && len(cbl.blocksSinceCheckpoint) > 0 {
 		dispatchHead = cbl.blocksSinceCheckpoint[len(cbl.blocksSinceCheckpoint)-1]

internal/confirmations/confirmed_block_listener_test.go

@@ -15,6 +15,7 @@
 package confirmations
 
 import (
+	"context"
 	"fmt"
 	"strings"
 	"sync/atomic"
@@ -338,38 +339,39 @@ func TestCBLDispatcherFallsBehindHead(t *testing.T) {
 	mbiHash := mca.On("BlockInfoByHash", mock.Anything, mock.Anything)
 	mbiHash.Run(func(args mock.Arguments) { mockBlockHashReturn(mbiHash, args, blocks) })
 
-	// We'll fall back to this because we don't keep up
-	mbiNum := mca.On("BlockInfoByNumber", mock.Anything, mock.Anything)
-	mbiNum.Run(func(args mock.Arguments) { mockBlockNumberReturn(mbiNum, args, blocks) })
-
 	bcm.requiredConfirmations = 5
+
+	// Start a CBL, but then cancel the dispatcher
 	cbl, err := bcm.startConfirmedBlockListener(bcm.ctx, id, "", nil, esDispatch)
 	assert.NoError(t, err)
+	cbl.cancelFunc()
+	<-cbl.dispatcherDone
+	<-cbl.processorDone
+
+	// Restart just the processor
+	cbl.ctx, cbl.cancelFunc = context.WithCancel(bcm.ctx)
+	cbl.processorDone = make(chan struct{})
+	go cbl.notificationProcessor()
 
-	// Notify all the blocks before we process any
+	// Notify all the blocks
 	for i := 0; i < len(blocks); i++ {
 		bcm.propagateBlockHashToCBLs(&ffcapi.BlockHashEvent{
 			BlockHashes: []string{blocks[i].BlockHash},
 		})
 	}
 
-	for i := 0; i < len(blocks)-bcm.requiredConfirmations; i++ {
-		// The dispatches should have been added, until it got too far ahead
-		// and then set to nil.
-		for cbl.newHeadToAdd != nil {
-			time.Sleep(1 * time.Millisecond)
-		}
-		b := <-esDispatch
-		assert.Equal(t, b.BlockEvent.BlockNumber.Uint64(), blocks[i].BlockNumber.Uint64())
-		assert.Equal(t, b.BlockEvent.BlockInfo, blocks[i].BlockInfo)
-	}
+	// Wait until there's a first tap to the dispatcher, meaning
+	// dispatches should have been added to the newHeadToAdd list
+	<-cbl.dispatcherTap
 
-	time.Sleep(1 * time.Millisecond)
-	assert.Len(t, cbl.blocksSinceCheckpoint, bcm.requiredConfirmations)
-	select {
-	case <-esDispatch:
-		assert.Fail(t, "should not have received block in confirmation window")
-	default: // good - we should have the confirmations sat there, but no dispatch
+	// ... until it got too far ahead and then set to nil.
+	checkHeadLen := func() int {
+		cbl.stateLock.Lock()
+		defer cbl.stateLock.Unlock()
+		return len(cbl.newHeadToAdd)
+	}
+	for checkHeadLen() < 10 {
+		time.Sleep(1 * time.Millisecond)
 	}
 
 	bcm.Stop()