swarm/network: WIP consider all nodes for healthy iteration (#19155)

* swarm/network: WIP consider all nodes for healthy iteration

* swarm/network/simulation: extend TestWaitTillHealthy to really check kads are healthy

* cmd/swarm/swarm-snapshot: fixed bugs in snapshot creation binary

* swarm/network/simulation: addressed PR comments

* swarm/network/simulation: defer sim.Clsoe()

* swarm/network/simulation: fixed wrong sim.Close()

* swarm/network/simulation: addressed PR comments

* cmd/swarm/swarm-snapshot: reducing default to 8 nodes, more to 4

* cmd/swarm/swarm-snapshot: extended timeout to 3 mins, or 256 nodes snapshot times out

* swarm/network/simulation: More PR comments

3 files changed, 10 insertions, 7 deletions

diff --git a/cmd/swarm/swarm-snapshot/create.go b/cmd/swarm/swarm-snapshot/create.go
index 127fde8ae..434561a49 100644
--- a/cmd/swarm/swarm-snapshot/create.go
+++ b/cmd/swarm/swarm-snapshot/create.go
@@ -59,13 +59,16 @@ func createSnapshot(filename string, nodes int, services []string) (err error) {
 	log.Debug("create snapshot", "filename", filename, "nodes", nodes, "services", services)
 
 	sim := simulation.New(map[string]simulation.ServiceFunc{
-		"bzz": func(ctx *adapters.ServiceContext, b *sync.Map) (node.Service, func(), error) {
+		"bzz": func(ctx *adapters.ServiceContext, bucket *sync.Map) (node.Service, func(), error) {
 			addr := network.NewAddr(ctx.Config.Node())
 			kad := network.NewKademlia(addr.Over(), network.NewKadParams())
 			hp := network.NewHiveParams()
 			hp.KeepAliveInterval = time.Duration(200) * time.Millisecond
 			hp.Discovery = true // discovery must be enabled when creating a snapshot
 
+			// store the kademlia in the bucket, needed later in the WaitTillHealthy function
+			bucket.Store(simulation.BucketKeyKademlia, kad)
+
 			config := &network.BzzConfig{
 				OverlayAddr:  addr.Over(),
 				UnderlayAddr: addr.Under(),
@@ -76,17 +79,17 @@ func createSnapshot(filename string, nodes int, services []string) (err error) {
 	})
 	defer sim.Close()
 
-	_, err = sim.AddNodes(nodes)
+	ids, err := sim.AddNodes(nodes)
 	if err != nil {
 		return fmt.Errorf("add nodes: %v", err)
 	}
 
-	err = sim.Net.ConnectNodesRing(nil)
+	err = sim.Net.ConnectNodesRing(ids)
 	if err != nil {
 		return fmt.Errorf("connect nodes: %v", err)
 	}
 
-	ctx, cancelSimRun := context.WithTimeout(context.Background(), 2*time.Minute)
+	ctx, cancelSimRun := context.WithTimeout(context.Background(), 3*time.Minute)
 	defer cancelSimRun()
 	if _, err := sim.WaitTillHealthy(ctx); err != nil {
 		return fmt.Errorf("wait for healthy kademlia: %v", err)
diff --git a/cmd/swarm/swarm-snapshot/create_test.go b/cmd/swarm/swarm-snapshot/create_test.go
index c9445168d..b2e30c201 100644
--- a/cmd/swarm/swarm-snapshot/create_test.go
+++ b/cmd/swarm/swarm-snapshot/create_test.go
@@ -48,7 +48,7 @@ func TestSnapshotCreate(t *testing.T) {
 		},
 		{
 			name:  "more nodes",
-			nodes: defaultNodes + 5,
+			nodes: defaultNodes + 4,
 		},
 		{
 			name:     "services",
@@ -81,7 +81,7 @@ func TestSnapshotCreate(t *testing.T) {
 			}
 			testCmd := runSnapshot(t, append(args, file.Name())...)
 
-			testCmd.ExpectExit()
+			testCmd.WaitExit()
 			if code := testCmd.ExitStatus(); code != 0 {
 				t.Fatalf("command exit code %v, expected 0", code)
 			}
diff --git a/cmd/swarm/swarm-snapshot/main.go b/cmd/swarm/swarm-snapshot/main.go
index 184727e4d..136295e51 100644
--- a/cmd/swarm/swarm-snapshot/main.go
+++ b/cmd/swarm/swarm-snapshot/main.go
@@ -27,7 +27,7 @@ import (
 var gitCommit string // Git SHA1 commit hash of the release (set via linker flags)
 
 // default value for "create" command --nodes flag
-const defaultNodes = 10
+const defaultNodes = 8
 
 func main() {
 	err := newApp().Run(os.Args)