Merge branch 'main' into discovery_metrics

das/state.go

@@ -96,6 +96,10 @@ func (s *coordinatorState) handleResult(res result) {
 		// if job was already in retry and failed again, carry over attempt count
 		lastRetry, ok := s.inRetry[h]
 		if ok {
+			if res.job.jobType != retryJob {
+				// retry job has been already created by another worker (recent or catchup)
+				continue
+			}
 			delete(s.inRetry, h)
 		}
 

share/availability/discovery/discovery_test.go

@@ -19,6 +19,8 @@ import (
 func TestDiscovery(t *testing.T) {
 	const nodes = 10 // higher number brings higher coverage
 
+	discoveryRetryTimeout = time.Millisecond * 100 // defined in discovery.go
+
 	ctx, cancel := context.WithTimeout(context.Background(), time.Second*30)
 	t.Cleanup(cancel)
 
@@ -28,7 +30,6 @@ func TestDiscovery(t *testing.T) {
 		WithPeersLimit(nodes),
 		WithAdvertiseInterval(-1),
 	)
-	discoveryRetryTimeout = time.Millisecond * 100 // defined in discovery.go
 
 	type peerUpdate struct {
 		peerID  peer.ID
@@ -42,7 +43,6 @@ func TestDiscovery(t *testing.T) {
 	discs := make([]*Discovery, nodes)
 	for i := range discs {
 		discs[i] = tn.discovery(WithPeersLimit(0), WithAdvertiseInterval(time.Millisecond*100))
-		discoveryRetryTimeout = -1 // defined in discovery.go
 
 		select {
 		case res := <-updateCh:

share/p2p/shrexeds/client.go

@@ -57,7 +57,7 @@ func (c *Client) RequestEDS(
 	log.Debugw("client: eds request to peer failed", "peer", peer, "hash", dataHash.String(), "error", err)
 	if errors.Is(err, context.DeadlineExceeded) || errors.Is(err, context.Canceled) {
 		c.metrics.ObserveRequests(1, p2p.StatusTimeout)
-		return nil, ctx.Err()
+		return nil, err
 	}
 	// some net.Errors also mean the context deadline was exceeded, but yamux/mocknet do not
 	// unwrap to a ctx err