query: fix error "leak"

Long-running queries can build up large error sets that we never actually use.
This is exacerbated by libp2p/go-libp2p-swarm#115.

fixes libp2p/go-libp2p-swarm#119

query.go

@@ -2,9 +2,9 @@ package dht
 
 import (
 	"context"
+	"errors"
 	"sync"
 
-	u "github.com/ipfs/go-ipfs-util"
 	logging "github.com/ipfs/go-log"
 	todoctr "github.com/ipfs/go-todocounter"
 	process "github.com/jbenet/goprocess"
@@ -18,6 +18,9 @@ import (
 	notif "github.com/libp2p/go-libp2p-routing/notifications"
 )
 
+// ErrNoPeersQueried is returned when we failed to connect to any peers.
+var ErrNoPeersQueried = errors.New("failed to query any peers")
+
 var maxQueryConcurrency = AlphaValue
 
 type dhtQuery struct {
@@ -77,7 +80,6 @@ type dhtQueryRunner struct {
 	peersRemaining todoctr.Counter  // peersToQuery + currently processing
 
 	result *dhtQueryResult // query result
-	errs   u.MultiErr      // result errors. maybe should be a map[peer.ID]error
 
 	rateLimit chan struct{} // processing semaphore
 	log       logging.EventLogger
@@ -158,12 +160,10 @@ func (r *dhtQueryRunner) Run(ctx context.Context, peers []peer.ID) (*dhtQueryRes
 		r.RLock()
 		defer r.RUnlock()
 
-		err = routing.ErrNotFound
-
-		// if every query to every peer failed, something must be very wrong.
-		if len(r.errs) > 0 && len(r.errs) == r.peersSeen.Size() {
-			logger.Debugf("query errs: %s", r.errs)
-			err = r.errs[0]
+		if r.peersQueried.Size() == 0 {
+			err = ErrNoPeersQueried
+		} else {
+			err = routing.ErrNotFound
 		}
 
 	case <-r.proc.Closed():
@@ -257,10 +257,6 @@ func (r *dhtQueryRunner) dialPeer(ctx context.Context, p peer.ID) error {
 			ID:    p,
 		})
 
-		r.Lock()
-		r.errs = append(r.errs, err)
-		r.Unlock()
-
 		// This peer is dropping out of the race.
 		r.peersRemaining.Decrement(1)
 		return err
@@ -289,10 +285,6 @@ func (r *dhtQueryRunner) queryPeer(proc process.Process, p peer.ID) {
 
 	if err != nil {
 		logger.Debugf("ERROR worker for: %v %v", p, err)
-		r.Lock()
-		r.errs = append(r.errs, err)
-		r.Unlock()
-
 	} else if res.success {
 		logger.Debugf("SUCCESS worker for: %v %s", p, res)
 		r.Lock()