plugeth/p2p/simulations/network_test.go

873 lines
23 KiB
Go
Raw Normal View History

// Copyright 2017 The go-ethereum Authors
// This file is part of the go-ethereum library.
//
// The go-ethereum library is free software: you can redistribute it and/or modify
// it under the terms of the GNU Lesser General Public License as published by
// the Free Software Foundation, either version 3 of the License, or
// (at your option) any later version.
//
// The go-ethereum library is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU Lesser General Public License for more details.
//
// You should have received a copy of the GNU Lesser General Public License
// along with the go-ethereum library. If not, see <http://www.gnu.org/licenses/>.
package simulations
import (
"bytes"
"context"
"encoding/json"
"fmt"
p2p, swarm: fix node up races by granular locking (#18976) * swarm/network: DRY out repeated giga comment I not necessarily agree with the way we wait for event propagation. But I truly disagree with having duplicated giga comments. * p2p/simulations: encapsulate Node.Up field so we avoid data races The Node.Up field was accessed concurrently without "proper" locking. There was a lock on Network and that was used sometimes to access the field. Other times the locking was missed and we had a data race. For example: https://github.com/ethereum/go-ethereum/pull/18464 The case above was solved, but there were still intermittent/hard to reproduce races. So let's solve the issue permanently. resolves: ethersphere/go-ethereum#1146 * p2p/simulations: fix unmarshal of simulations.Node Making Node.Up field private in 13292ee897e345045fbfab3bda23a77589a271c1 broke TestHTTPNetwork and TestHTTPSnapshot. Because the default UnmarshalJSON does not handle unexported fields. Important: The fix is partial and not proper to my taste. But I cut scope as I think the fix may require a change to the current serialization format. New ticket: https://github.com/ethersphere/go-ethereum/issues/1177 * p2p/simulations: Add a sanity test case for Node.Config UnmarshalJSON * p2p/simulations: revert back to defer Unlock() pattern for Network It's a good patten to call `defer Unlock()` right after `Lock()` so (new) error cases won't miss to unlock. Let's get back to that pattern. The patten was abandoned in 85a79b3ad3c5863f8612d25c246bcfad339f36b7, while fixing a data race. That data race does not exist anymore, since the Node.Up field got hidden behind its own lock. * p2p/simulations: consistent naming for test providers Node.UnmarshalJSON * p2p/simulations: remove JSON annotation from private fields of Node As unexported fields are not serialized. * p2p/simulations: fix deadlock in Network.GetRandomDownNode() Problem: GetRandomDownNode() locks -> getDownNodeIDs() -> GetNodes() tries to lock -> deadlock On Network type, unexported functions must assume that `net.lock` is already acquired and should not call exported functions which might try to lock again. * p2p/simulations: ensure method conformity for Network Connect* methods were moved to p2p/simulations.Network from swarm/network/simulation. However these new methods did not follow the pattern of Network methods, i.e., all exported method locks the whole Network either for read or write. * p2p/simulations: fix deadlock during network shutdown `TestDiscoveryPersistenceSimulationSimAdapter` often got into deadlock. The execution was stuck on two locks, i.e, `Kademlia.lock` and `p2p/simulations.Network.lock`. Usually the test got stuck once in each 20 executions with high confidence. `Kademlia` was stuck in `Kademlia.EachAddr()` and `Network` in `Network.Stop()`. Solution: in `Network.Stop()` `net.lock` must be released before calling `node.Stop()` as stopping a node (somehow - I did not find the exact code path) causes `Network.InitConn()` to be called from `Kademlia.SuggestPeer()` and that blocks on `net.lock`. Related ticket: https://github.com/ethersphere/go-ethereum/issues/1223 * swarm/state: simplify if statement in DBStore.Put() * p2p/simulations: remove faulty godoc from private function The comment started with the wrong method name. The method is simple and self explanatory. Also, it's private. => Let's just remove the comment.
2019-02-18 06:38:14 +00:00
"reflect"
"strconv"
"strings"
"testing"
"time"
"github.com/ethereum/go-ethereum/log"
"github.com/ethereum/go-ethereum/node"
all: new p2p node representation (#17643) Package p2p/enode provides a generalized representation of p2p nodes which can contain arbitrary information in key/value pairs. It is also the new home for the node database. The "v4" identity scheme is also moved here from p2p/enr to remove the dependency on Ethereum crypto from that package. Record signature handling is changed significantly. The identity scheme registry is removed and acceptable schemes must be passed to any method that needs identity. This means records must now be validated explicitly after decoding. The enode API is designed to make signature handling easy and safe: most APIs around the codebase work with enode.Node, which is a wrapper around a valid record. Going from enr.Record to enode.Node requires a valid signature. * p2p/discover: port to p2p/enode This ports the discovery code to the new node representation in p2p/enode. The wire protocol is unchanged, this can be considered a refactoring change. The Kademlia table can now deal with nodes using an arbitrary identity scheme. This requires a few incompatible API changes: - Table.Lookup is not available anymore. It used to take a public key as argument because v4 protocol requires one. Its replacement is LookupRandom. - Table.Resolve takes *enode.Node instead of NodeID. This is also for v4 protocol compatibility because nodes cannot be looked up by ID alone. - Types Node and NodeID are gone. Further commits in the series will be fixes all over the the codebase to deal with those removals. * p2p: port to p2p/enode and discovery changes This adapts package p2p to the changes in p2p/discover. All uses of discover.Node and discover.NodeID are replaced by their equivalents from p2p/enode. New API is added to retrieve the enode.Node instance of a peer. The behavior of Server.Self with discovery disabled is improved. It now tries much harder to report a working IP address, falling back to 127.0.0.1 if no suitable address can be determined through other means. These changes were needed for tests of other packages later in the series. * p2p/simulations, p2p/testing: port to p2p/enode No surprises here, mostly replacements of discover.Node, discover.NodeID with their new equivalents. The 'interesting' API changes are: - testing.ProtocolSession tracks complete nodes, not just their IDs. - adapters.NodeConfig has a new method to create a complete node. These changes were needed to make swarm tests work. Note that the NodeID change makes the code incompatible with old simulation snapshots. * whisper/whisperv5, whisper/whisperv6: port to p2p/enode This port was easy because whisper uses []byte for node IDs and URL strings in the API. * eth: port to p2p/enode Again, easy to port because eth uses strings for node IDs and doesn't care about node information in any way. * les: port to p2p/enode Apart from replacing discover.NodeID with enode.ID, most changes are in the server pool code. It now deals with complete nodes instead of (Pubkey, IP, Port) triples. The database format is unchanged for now, but we should probably change it to use the node database later. * node: port to p2p/enode This change simply replaces discover.Node and discover.NodeID with their new equivalents. * swarm/network: port to p2p/enode Swarm has its own node address representation, BzzAddr, containing both an overlay address (the hash of a secp256k1 public key) and an underlay address (enode:// URL). There are no changes to the BzzAddr format in this commit, but certain operations such as creating a BzzAddr from a node ID are now impossible because node IDs aren't public keys anymore. Most swarm-related changes in the series remove uses of NewAddrFromNodeID, replacing it with NewAddr which takes a complete node as argument. ToOverlayAddr is removed because we can just use the node ID directly.
2018-09-24 22:59:00 +00:00
"github.com/ethereum/go-ethereum/p2p/enode"
"github.com/ethereum/go-ethereum/p2p/simulations/adapters"
)
// Tests that a created snapshot with a minimal service only contains the expected connections
// and that a network when loaded with this snapshot only contains those same connections
func TestSnapshot(t *testing.T) {
// PART I
// create snapshot from ring network
// this is a minimal service, whose protocol will take exactly one message OR close of connection before quitting
adapter := adapters.NewSimAdapter(adapters.Services{
"noopwoop": func(ctx *adapters.ServiceContext) (node.Service, error) {
return NewNoopService(nil), nil
},
})
// create network
network := NewNetwork(adapter, &NetworkConfig{
DefaultService: "noopwoop",
})
// \todo consider making a member of network, set to true threadsafe when shutdown
runningOne := true
defer func() {
if runningOne {
network.Shutdown()
}
}()
// create and start nodes
nodeCount := 20
ids := make([]enode.ID, nodeCount)
for i := 0; i < nodeCount; i++ {
conf := adapters.RandomNodeConfig()
node, err := network.NewNodeWithConfig(conf)
if err != nil {
t.Fatalf("error creating node: %s", err)
}
if err := network.Start(node.ID()); err != nil {
t.Fatalf("error starting node: %s", err)
}
ids[i] = node.ID()
}
// subscribe to peer events
evC := make(chan *Event)
sub := network.Events().Subscribe(evC)
defer sub.Unsubscribe()
// connect nodes in a ring
// spawn separate thread to avoid deadlock in the event listeners
go func() {
for i, id := range ids {
peerID := ids[(i+1)%len(ids)]
if err := network.Connect(id, peerID); err != nil {
t.Fatal(err)
}
}
}()
// collect connection events up to expected number
ctx, cancel := context.WithTimeout(context.TODO(), time.Second)
defer cancel()
checkIds := make(map[enode.ID][]enode.ID)
connEventCount := nodeCount
OUTER:
for {
select {
case <-ctx.Done():
t.Fatal(ctx.Err())
case ev := <-evC:
if ev.Type == EventTypeConn && !ev.Control {
// fail on any disconnect
if !ev.Conn.Up {
t.Fatalf("unexpected disconnect: %v -> %v", ev.Conn.One, ev.Conn.Other)
}
checkIds[ev.Conn.One] = append(checkIds[ev.Conn.One], ev.Conn.Other)
checkIds[ev.Conn.Other] = append(checkIds[ev.Conn.Other], ev.Conn.One)
connEventCount--
log.Debug("ev", "count", connEventCount)
if connEventCount == 0 {
break OUTER
}
}
}
}
// create snapshot of current network
snap, err := network.Snapshot()
if err != nil {
t.Fatal(err)
}
j, err := json.Marshal(snap)
if err != nil {
t.Fatal(err)
}
log.Debug("snapshot taken", "nodes", len(snap.Nodes), "conns", len(snap.Conns), "json", string(j))
// verify that the snap element numbers check out
if len(checkIds) != len(snap.Conns) || len(checkIds) != len(snap.Nodes) {
t.Fatalf("snapshot wrong node,conn counts %d,%d != %d", len(snap.Nodes), len(snap.Conns), len(checkIds))
}
// shut down sim network
runningOne = false
sub.Unsubscribe()
network.Shutdown()
// check that we have all the expected connections in the snapshot
for nodid, nodConns := range checkIds {
for _, nodConn := range nodConns {
var match bool
for _, snapConn := range snap.Conns {
if snapConn.One == nodid && snapConn.Other == nodConn {
match = true
break
} else if snapConn.Other == nodid && snapConn.One == nodConn {
match = true
break
}
}
if !match {
t.Fatalf("snapshot missing conn %v -> %v", nodid, nodConn)
}
}
}
log.Info("snapshot checked")
// PART II
// load snapshot and verify that exactly same connections are formed
adapter = adapters.NewSimAdapter(adapters.Services{
"noopwoop": func(ctx *adapters.ServiceContext) (node.Service, error) {
return NewNoopService(nil), nil
},
})
network = NewNetwork(adapter, &NetworkConfig{
DefaultService: "noopwoop",
})
defer func() {
network.Shutdown()
}()
// subscribe to peer events
// every node up and conn up event will generate one additional control event
// therefore multiply the count by two
evC = make(chan *Event, (len(snap.Conns)*2)+(len(snap.Nodes)*2))
sub = network.Events().Subscribe(evC)
defer sub.Unsubscribe()
// load the snapshot
// spawn separate thread to avoid deadlock in the event listeners
err = network.Load(snap)
if err != nil {
t.Fatal(err)
}
// collect connection events up to expected number
ctx, cancel = context.WithTimeout(context.TODO(), time.Second*3)
defer cancel()
connEventCount = nodeCount
OuterTwo:
for {
select {
case <-ctx.Done():
t.Fatal(ctx.Err())
case ev := <-evC:
if ev.Type == EventTypeConn && !ev.Control {
// fail on any disconnect
if !ev.Conn.Up {
t.Fatalf("unexpected disconnect: %v -> %v", ev.Conn.One, ev.Conn.Other)
}
log.Debug("conn", "on", ev.Conn.One, "other", ev.Conn.Other)
checkIds[ev.Conn.One] = append(checkIds[ev.Conn.One], ev.Conn.Other)
checkIds[ev.Conn.Other] = append(checkIds[ev.Conn.Other], ev.Conn.One)
connEventCount--
log.Debug("ev", "count", connEventCount)
if connEventCount == 0 {
break OuterTwo
}
}
}
}
// check that we have all expected connections in the network
for _, snapConn := range snap.Conns {
var match bool
for nodid, nodConns := range checkIds {
for _, nodConn := range nodConns {
if snapConn.One == nodid && snapConn.Other == nodConn {
match = true
break
} else if snapConn.Other == nodid && snapConn.One == nodConn {
match = true
break
}
}
}
if !match {
t.Fatalf("network missing conn %v -> %v", snapConn.One, snapConn.Other)
}
}
// verify that network didn't generate any other additional connection events after the ones we have collected within a reasonable period of time
ctx, cancel = context.WithTimeout(context.TODO(), time.Second)
defer cancel()
select {
case <-ctx.Done():
case ev := <-evC:
if ev.Type == EventTypeConn {
t.Fatalf("Superfluous conn found %v -> %v", ev.Conn.One, ev.Conn.Other)
}
}
// This test validates if all connections from the snapshot
// are created in the network.
t.Run("conns after load", func(t *testing.T) {
// Create new network.
n := NewNetwork(
adapters.NewSimAdapter(adapters.Services{
"noopwoop": func(ctx *adapters.ServiceContext) (node.Service, error) {
return NewNoopService(nil), nil
},
}),
&NetworkConfig{
DefaultService: "noopwoop",
},
)
defer n.Shutdown()
// Load the same snapshot.
err := n.Load(snap)
if err != nil {
t.Fatal(err)
}
// Check every connection from the snapshot
// if it is in the network, too.
for _, c := range snap.Conns {
if n.GetConn(c.One, c.Other) == nil {
t.Errorf("missing connection: %s -> %s", c.One, c.Other)
}
}
})
}
// TestNetworkSimulation creates a multi-node simulation network with each node
// connected in a ring topology, checks that all nodes successfully handshake
// with each other and that a snapshot fully represents the desired topology
func TestNetworkSimulation(t *testing.T) {
// create simulation network with 20 testService nodes
adapter := adapters.NewSimAdapter(adapters.Services{
"test": newTestService,
})
network := NewNetwork(adapter, &NetworkConfig{
DefaultService: "test",
})
defer network.Shutdown()
nodeCount := 20
all: new p2p node representation (#17643) Package p2p/enode provides a generalized representation of p2p nodes which can contain arbitrary information in key/value pairs. It is also the new home for the node database. The "v4" identity scheme is also moved here from p2p/enr to remove the dependency on Ethereum crypto from that package. Record signature handling is changed significantly. The identity scheme registry is removed and acceptable schemes must be passed to any method that needs identity. This means records must now be validated explicitly after decoding. The enode API is designed to make signature handling easy and safe: most APIs around the codebase work with enode.Node, which is a wrapper around a valid record. Going from enr.Record to enode.Node requires a valid signature. * p2p/discover: port to p2p/enode This ports the discovery code to the new node representation in p2p/enode. The wire protocol is unchanged, this can be considered a refactoring change. The Kademlia table can now deal with nodes using an arbitrary identity scheme. This requires a few incompatible API changes: - Table.Lookup is not available anymore. It used to take a public key as argument because v4 protocol requires one. Its replacement is LookupRandom. - Table.Resolve takes *enode.Node instead of NodeID. This is also for v4 protocol compatibility because nodes cannot be looked up by ID alone. - Types Node and NodeID are gone. Further commits in the series will be fixes all over the the codebase to deal with those removals. * p2p: port to p2p/enode and discovery changes This adapts package p2p to the changes in p2p/discover. All uses of discover.Node and discover.NodeID are replaced by their equivalents from p2p/enode. New API is added to retrieve the enode.Node instance of a peer. The behavior of Server.Self with discovery disabled is improved. It now tries much harder to report a working IP address, falling back to 127.0.0.1 if no suitable address can be determined through other means. These changes were needed for tests of other packages later in the series. * p2p/simulations, p2p/testing: port to p2p/enode No surprises here, mostly replacements of discover.Node, discover.NodeID with their new equivalents. The 'interesting' API changes are: - testing.ProtocolSession tracks complete nodes, not just their IDs. - adapters.NodeConfig has a new method to create a complete node. These changes were needed to make swarm tests work. Note that the NodeID change makes the code incompatible with old simulation snapshots. * whisper/whisperv5, whisper/whisperv6: port to p2p/enode This port was easy because whisper uses []byte for node IDs and URL strings in the API. * eth: port to p2p/enode Again, easy to port because eth uses strings for node IDs and doesn't care about node information in any way. * les: port to p2p/enode Apart from replacing discover.NodeID with enode.ID, most changes are in the server pool code. It now deals with complete nodes instead of (Pubkey, IP, Port) triples. The database format is unchanged for now, but we should probably change it to use the node database later. * node: port to p2p/enode This change simply replaces discover.Node and discover.NodeID with their new equivalents. * swarm/network: port to p2p/enode Swarm has its own node address representation, BzzAddr, containing both an overlay address (the hash of a secp256k1 public key) and an underlay address (enode:// URL). There are no changes to the BzzAddr format in this commit, but certain operations such as creating a BzzAddr from a node ID are now impossible because node IDs aren't public keys anymore. Most swarm-related changes in the series remove uses of NewAddrFromNodeID, replacing it with NewAddr which takes a complete node as argument. ToOverlayAddr is removed because we can just use the node ID directly.
2018-09-24 22:59:00 +00:00
ids := make([]enode.ID, nodeCount)
for i := 0; i < nodeCount; i++ {
conf := adapters.RandomNodeConfig()
node, err := network.NewNodeWithConfig(conf)
if err != nil {
t.Fatalf("error creating node: %s", err)
}
if err := network.Start(node.ID()); err != nil {
t.Fatalf("error starting node: %s", err)
}
ids[i] = node.ID()
}
// perform a check which connects the nodes in a ring (so each node is
// connected to exactly two peers) and then checks that all nodes
// performed two handshakes by checking their peerCount
action := func(_ context.Context) error {
for i, id := range ids {
peerID := ids[(i+1)%len(ids)]
if err := network.Connect(id, peerID); err != nil {
return err
}
}
return nil
}
all: new p2p node representation (#17643) Package p2p/enode provides a generalized representation of p2p nodes which can contain arbitrary information in key/value pairs. It is also the new home for the node database. The "v4" identity scheme is also moved here from p2p/enr to remove the dependency on Ethereum crypto from that package. Record signature handling is changed significantly. The identity scheme registry is removed and acceptable schemes must be passed to any method that needs identity. This means records must now be validated explicitly after decoding. The enode API is designed to make signature handling easy and safe: most APIs around the codebase work with enode.Node, which is a wrapper around a valid record. Going from enr.Record to enode.Node requires a valid signature. * p2p/discover: port to p2p/enode This ports the discovery code to the new node representation in p2p/enode. The wire protocol is unchanged, this can be considered a refactoring change. The Kademlia table can now deal with nodes using an arbitrary identity scheme. This requires a few incompatible API changes: - Table.Lookup is not available anymore. It used to take a public key as argument because v4 protocol requires one. Its replacement is LookupRandom. - Table.Resolve takes *enode.Node instead of NodeID. This is also for v4 protocol compatibility because nodes cannot be looked up by ID alone. - Types Node and NodeID are gone. Further commits in the series will be fixes all over the the codebase to deal with those removals. * p2p: port to p2p/enode and discovery changes This adapts package p2p to the changes in p2p/discover. All uses of discover.Node and discover.NodeID are replaced by their equivalents from p2p/enode. New API is added to retrieve the enode.Node instance of a peer. The behavior of Server.Self with discovery disabled is improved. It now tries much harder to report a working IP address, falling back to 127.0.0.1 if no suitable address can be determined through other means. These changes were needed for tests of other packages later in the series. * p2p/simulations, p2p/testing: port to p2p/enode No surprises here, mostly replacements of discover.Node, discover.NodeID with their new equivalents. The 'interesting' API changes are: - testing.ProtocolSession tracks complete nodes, not just their IDs. - adapters.NodeConfig has a new method to create a complete node. These changes were needed to make swarm tests work. Note that the NodeID change makes the code incompatible with old simulation snapshots. * whisper/whisperv5, whisper/whisperv6: port to p2p/enode This port was easy because whisper uses []byte for node IDs and URL strings in the API. * eth: port to p2p/enode Again, easy to port because eth uses strings for node IDs and doesn't care about node information in any way. * les: port to p2p/enode Apart from replacing discover.NodeID with enode.ID, most changes are in the server pool code. It now deals with complete nodes instead of (Pubkey, IP, Port) triples. The database format is unchanged for now, but we should probably change it to use the node database later. * node: port to p2p/enode This change simply replaces discover.Node and discover.NodeID with their new equivalents. * swarm/network: port to p2p/enode Swarm has its own node address representation, BzzAddr, containing both an overlay address (the hash of a secp256k1 public key) and an underlay address (enode:// URL). There are no changes to the BzzAddr format in this commit, but certain operations such as creating a BzzAddr from a node ID are now impossible because node IDs aren't public keys anymore. Most swarm-related changes in the series remove uses of NewAddrFromNodeID, replacing it with NewAddr which takes a complete node as argument. ToOverlayAddr is removed because we can just use the node ID directly.
2018-09-24 22:59:00 +00:00
check := func(ctx context.Context, id enode.ID) (bool, error) {
// check we haven't run out of time
select {
case <-ctx.Done():
return false, ctx.Err()
default:
}
// get the node
node := network.GetNode(id)
if node == nil {
return false, fmt.Errorf("unknown node: %s", id)
}
// check it has exactly two peers
client, err := node.Client()
if err != nil {
return false, err
}
var peerCount int64
if err := client.CallContext(ctx, &peerCount, "test_peerCount"); err != nil {
return false, err
}
switch {
case peerCount < 2:
return false, nil
case peerCount == 2:
return true, nil
default:
return false, fmt.Errorf("unexpected peerCount: %d", peerCount)
}
}
timeout := 30 * time.Second
ctx, cancel := context.WithTimeout(context.Background(), timeout)
defer cancel()
// trigger a check every 100ms
all: new p2p node representation (#17643) Package p2p/enode provides a generalized representation of p2p nodes which can contain arbitrary information in key/value pairs. It is also the new home for the node database. The "v4" identity scheme is also moved here from p2p/enr to remove the dependency on Ethereum crypto from that package. Record signature handling is changed significantly. The identity scheme registry is removed and acceptable schemes must be passed to any method that needs identity. This means records must now be validated explicitly after decoding. The enode API is designed to make signature handling easy and safe: most APIs around the codebase work with enode.Node, which is a wrapper around a valid record. Going from enr.Record to enode.Node requires a valid signature. * p2p/discover: port to p2p/enode This ports the discovery code to the new node representation in p2p/enode. The wire protocol is unchanged, this can be considered a refactoring change. The Kademlia table can now deal with nodes using an arbitrary identity scheme. This requires a few incompatible API changes: - Table.Lookup is not available anymore. It used to take a public key as argument because v4 protocol requires one. Its replacement is LookupRandom. - Table.Resolve takes *enode.Node instead of NodeID. This is also for v4 protocol compatibility because nodes cannot be looked up by ID alone. - Types Node and NodeID are gone. Further commits in the series will be fixes all over the the codebase to deal with those removals. * p2p: port to p2p/enode and discovery changes This adapts package p2p to the changes in p2p/discover. All uses of discover.Node and discover.NodeID are replaced by their equivalents from p2p/enode. New API is added to retrieve the enode.Node instance of a peer. The behavior of Server.Self with discovery disabled is improved. It now tries much harder to report a working IP address, falling back to 127.0.0.1 if no suitable address can be determined through other means. These changes were needed for tests of other packages later in the series. * p2p/simulations, p2p/testing: port to p2p/enode No surprises here, mostly replacements of discover.Node, discover.NodeID with their new equivalents. The 'interesting' API changes are: - testing.ProtocolSession tracks complete nodes, not just their IDs. - adapters.NodeConfig has a new method to create a complete node. These changes were needed to make swarm tests work. Note that the NodeID change makes the code incompatible with old simulation snapshots. * whisper/whisperv5, whisper/whisperv6: port to p2p/enode This port was easy because whisper uses []byte for node IDs and URL strings in the API. * eth: port to p2p/enode Again, easy to port because eth uses strings for node IDs and doesn't care about node information in any way. * les: port to p2p/enode Apart from replacing discover.NodeID with enode.ID, most changes are in the server pool code. It now deals with complete nodes instead of (Pubkey, IP, Port) triples. The database format is unchanged for now, but we should probably change it to use the node database later. * node: port to p2p/enode This change simply replaces discover.Node and discover.NodeID with their new equivalents. * swarm/network: port to p2p/enode Swarm has its own node address representation, BzzAddr, containing both an overlay address (the hash of a secp256k1 public key) and an underlay address (enode:// URL). There are no changes to the BzzAddr format in this commit, but certain operations such as creating a BzzAddr from a node ID are now impossible because node IDs aren't public keys anymore. Most swarm-related changes in the series remove uses of NewAddrFromNodeID, replacing it with NewAddr which takes a complete node as argument. ToOverlayAddr is removed because we can just use the node ID directly.
2018-09-24 22:59:00 +00:00
trigger := make(chan enode.ID)
go triggerChecks(ctx, ids, trigger, 100*time.Millisecond)
result := NewSimulation(network).Run(ctx, &Step{
Action: action,
Trigger: trigger,
Expect: &Expectation{
Nodes: ids,
Check: check,
},
})
if result.Error != nil {
t.Fatalf("simulation failed: %s", result.Error)
}
// take a network snapshot and check it contains the correct topology
snap, err := network.Snapshot()
if err != nil {
t.Fatal(err)
}
if len(snap.Nodes) != nodeCount {
t.Fatalf("expected snapshot to contain %d nodes, got %d", nodeCount, len(snap.Nodes))
}
if len(snap.Conns) != nodeCount {
t.Fatalf("expected snapshot to contain %d connections, got %d", nodeCount, len(snap.Conns))
}
for i, id := range ids {
conn := snap.Conns[i]
if conn.One != id {
t.Fatalf("expected conn[%d].One to be %s, got %s", i, id, conn.One)
}
peerID := ids[(i+1)%len(ids)]
if conn.Other != peerID {
t.Fatalf("expected conn[%d].Other to be %s, got %s", i, peerID, conn.Other)
}
}
}
func createTestNodes(count int, network *Network) (nodes []*Node, err error) {
for i := 0; i < count; i++ {
nodeConf := adapters.RandomNodeConfig()
node, err := network.NewNodeWithConfig(nodeConf)
if err != nil {
return nil, err
}
if err := network.Start(node.ID()); err != nil {
return nil, err
}
nodes = append(nodes, node)
}
return nodes, nil
}
func createTestNodesWithProperty(property string, count int, network *Network) (propertyNodes []*Node, err error) {
for i := 0; i < count; i++ {
nodeConf := adapters.RandomNodeConfig()
nodeConf.Properties = append(nodeConf.Properties, property)
node, err := network.NewNodeWithConfig(nodeConf)
if err != nil {
return nil, err
}
if err := network.Start(node.ID()); err != nil {
return nil, err
}
propertyNodes = append(propertyNodes, node)
}
return propertyNodes, nil
}
// TestGetNodeIDs creates a set of nodes and attempts to retrieve their IDs,.
// It then tests again whilst excluding a node ID from being returned.
// If a node ID is not returned, or more node IDs than expected are returned, the test fails.
func TestGetNodeIDs(t *testing.T) {
adapter := adapters.NewSimAdapter(adapters.Services{
"test": newTestService,
})
network := NewNetwork(adapter, &NetworkConfig{
DefaultService: "test",
})
defer network.Shutdown()
numNodes := 5
nodes, err := createTestNodes(numNodes, network)
if err != nil {
t.Fatalf("Could not creat test nodes %v", err)
}
gotNodeIDs := network.GetNodeIDs()
if len(gotNodeIDs) != numNodes {
t.Fatalf("Expected %d nodes, got %d", numNodes, len(gotNodeIDs))
}
for _, node1 := range nodes {
match := false
for _, node2ID := range gotNodeIDs {
if bytes.Equal(node1.ID().Bytes(), node2ID.Bytes()) {
match = true
break
}
}
if !match {
t.Fatalf("A created node was not returned by GetNodes(), ID: %s", node1.ID().String())
}
}
excludeNodeID := nodes[3].ID()
gotNodeIDsExcl := network.GetNodeIDs(excludeNodeID)
if len(gotNodeIDsExcl) != numNodes-1 {
t.Fatalf("Expected one less node ID to be returned")
}
for _, nodeID := range gotNodeIDsExcl {
if bytes.Equal(excludeNodeID.Bytes(), nodeID.Bytes()) {
t.Fatalf("GetNodeIDs returned the node ID we excluded, ID: %s", nodeID.String())
}
}
}
// TestGetNodes creates a set of nodes and attempts to retrieve them again.
// It then tests again whilst excluding a node from being returned.
// If a node is not returned, or more nodes than expected are returned, the test fails.
func TestGetNodes(t *testing.T) {
adapter := adapters.NewSimAdapter(adapters.Services{
"test": newTestService,
})
network := NewNetwork(adapter, &NetworkConfig{
DefaultService: "test",
})
defer network.Shutdown()
numNodes := 5
nodes, err := createTestNodes(numNodes, network)
if err != nil {
t.Fatalf("Could not creat test nodes %v", err)
}
gotNodes := network.GetNodes()
if len(gotNodes) != numNodes {
t.Fatalf("Expected %d nodes, got %d", numNodes, len(gotNodes))
}
for _, node1 := range nodes {
match := false
for _, node2 := range gotNodes {
if bytes.Equal(node1.ID().Bytes(), node2.ID().Bytes()) {
match = true
break
}
}
if !match {
t.Fatalf("A created node was not returned by GetNodes(), ID: %s", node1.ID().String())
}
}
excludeNodeID := nodes[3].ID()
gotNodesExcl := network.GetNodes(excludeNodeID)
if len(gotNodesExcl) != numNodes-1 {
t.Fatalf("Expected one less node to be returned")
}
for _, node := range gotNodesExcl {
if bytes.Equal(excludeNodeID.Bytes(), node.ID().Bytes()) {
t.Fatalf("GetNodes returned the node we excluded, ID: %s", node.ID().String())
}
}
}
// TestGetNodesByID creates a set of nodes and attempts to retrieve a subset of them by ID
// If a node is not returned, or more nodes than expected are returned, the test fails.
func TestGetNodesByID(t *testing.T) {
adapter := adapters.NewSimAdapter(adapters.Services{
"test": newTestService,
})
network := NewNetwork(adapter, &NetworkConfig{
DefaultService: "test",
})
defer network.Shutdown()
numNodes := 5
nodes, err := createTestNodes(numNodes, network)
if err != nil {
t.Fatalf("Could not create test nodes: %v", err)
}
numSubsetNodes := 2
subsetNodes := nodes[0:numSubsetNodes]
var subsetNodeIDs []enode.ID
for _, node := range subsetNodes {
subsetNodeIDs = append(subsetNodeIDs, node.ID())
}
gotNodesByID := network.GetNodesByID(subsetNodeIDs)
if len(gotNodesByID) != numSubsetNodes {
t.Fatalf("Expected %d nodes, got %d", numSubsetNodes, len(gotNodesByID))
}
for _, node1 := range subsetNodes {
match := false
for _, node2 := range gotNodesByID {
if bytes.Equal(node1.ID().Bytes(), node2.ID().Bytes()) {
match = true
break
}
}
if !match {
t.Fatalf("A created node was not returned by GetNodesByID(), ID: %s", node1.ID().String())
}
}
}
// TestGetNodesByProperty creates a subset of nodes with a property assigned.
// GetNodesByProperty is then checked for correctness by comparing the nodes returned to those initially created.
// If a node with a property is not found, or more nodes than expected are returned, the test fails.
func TestGetNodesByProperty(t *testing.T) {
adapter := adapters.NewSimAdapter(adapters.Services{
"test": newTestService,
})
network := NewNetwork(adapter, &NetworkConfig{
DefaultService: "test",
})
defer network.Shutdown()
numNodes := 3
_, err := createTestNodes(numNodes, network)
if err != nil {
t.Fatalf("Failed to create nodes: %v", err)
}
numPropertyNodes := 3
propertyTest := "test"
propertyNodes, err := createTestNodesWithProperty(propertyTest, numPropertyNodes, network)
if err != nil {
t.Fatalf("Failed to create nodes with property: %v", err)
}
gotNodesByProperty := network.GetNodesByProperty(propertyTest)
if len(gotNodesByProperty) != numPropertyNodes {
t.Fatalf("Expected %d nodes with a property, got %d", numPropertyNodes, len(gotNodesByProperty))
}
for _, node1 := range propertyNodes {
match := false
for _, node2 := range gotNodesByProperty {
if bytes.Equal(node1.ID().Bytes(), node2.ID().Bytes()) {
match = true
break
}
}
if !match {
t.Fatalf("A created node with property was not returned by GetNodesByProperty(), ID: %s", node1.ID().String())
}
}
}
// TestGetNodeIDsByProperty creates a subset of nodes with a property assigned.
// GetNodeIDsByProperty is then checked for correctness by comparing the node IDs returned to those initially created.
// If a node ID with a property is not found, or more nodes IDs than expected are returned, the test fails.
func TestGetNodeIDsByProperty(t *testing.T) {
adapter := adapters.NewSimAdapter(adapters.Services{
"test": newTestService,
})
network := NewNetwork(adapter, &NetworkConfig{
DefaultService: "test",
})
defer network.Shutdown()
numNodes := 3
_, err := createTestNodes(numNodes, network)
if err != nil {
t.Fatalf("Failed to create nodes: %v", err)
}
numPropertyNodes := 3
propertyTest := "test"
propertyNodes, err := createTestNodesWithProperty(propertyTest, numPropertyNodes, network)
if err != nil {
t.Fatalf("Failed to created nodes with property: %v", err)
}
gotNodeIDsByProperty := network.GetNodeIDsByProperty(propertyTest)
if len(gotNodeIDsByProperty) != numPropertyNodes {
t.Fatalf("Expected %d nodes with a property, got %d", numPropertyNodes, len(gotNodeIDsByProperty))
}
for _, node1 := range propertyNodes {
match := false
id1 := node1.ID()
for _, id2 := range gotNodeIDsByProperty {
if bytes.Equal(id1.Bytes(), id2.Bytes()) {
match = true
break
}
}
if !match {
t.Fatalf("Not all nodes IDs were returned by GetNodeIDsByProperty(), ID: %s", id1.String())
}
}
}
all: new p2p node representation (#17643) Package p2p/enode provides a generalized representation of p2p nodes which can contain arbitrary information in key/value pairs. It is also the new home for the node database. The "v4" identity scheme is also moved here from p2p/enr to remove the dependency on Ethereum crypto from that package. Record signature handling is changed significantly. The identity scheme registry is removed and acceptable schemes must be passed to any method that needs identity. This means records must now be validated explicitly after decoding. The enode API is designed to make signature handling easy and safe: most APIs around the codebase work with enode.Node, which is a wrapper around a valid record. Going from enr.Record to enode.Node requires a valid signature. * p2p/discover: port to p2p/enode This ports the discovery code to the new node representation in p2p/enode. The wire protocol is unchanged, this can be considered a refactoring change. The Kademlia table can now deal with nodes using an arbitrary identity scheme. This requires a few incompatible API changes: - Table.Lookup is not available anymore. It used to take a public key as argument because v4 protocol requires one. Its replacement is LookupRandom. - Table.Resolve takes *enode.Node instead of NodeID. This is also for v4 protocol compatibility because nodes cannot be looked up by ID alone. - Types Node and NodeID are gone. Further commits in the series will be fixes all over the the codebase to deal with those removals. * p2p: port to p2p/enode and discovery changes This adapts package p2p to the changes in p2p/discover. All uses of discover.Node and discover.NodeID are replaced by their equivalents from p2p/enode. New API is added to retrieve the enode.Node instance of a peer. The behavior of Server.Self with discovery disabled is improved. It now tries much harder to report a working IP address, falling back to 127.0.0.1 if no suitable address can be determined through other means. These changes were needed for tests of other packages later in the series. * p2p/simulations, p2p/testing: port to p2p/enode No surprises here, mostly replacements of discover.Node, discover.NodeID with their new equivalents. The 'interesting' API changes are: - testing.ProtocolSession tracks complete nodes, not just their IDs. - adapters.NodeConfig has a new method to create a complete node. These changes were needed to make swarm tests work. Note that the NodeID change makes the code incompatible with old simulation snapshots. * whisper/whisperv5, whisper/whisperv6: port to p2p/enode This port was easy because whisper uses []byte for node IDs and URL strings in the API. * eth: port to p2p/enode Again, easy to port because eth uses strings for node IDs and doesn't care about node information in any way. * les: port to p2p/enode Apart from replacing discover.NodeID with enode.ID, most changes are in the server pool code. It now deals with complete nodes instead of (Pubkey, IP, Port) triples. The database format is unchanged for now, but we should probably change it to use the node database later. * node: port to p2p/enode This change simply replaces discover.Node and discover.NodeID with their new equivalents. * swarm/network: port to p2p/enode Swarm has its own node address representation, BzzAddr, containing both an overlay address (the hash of a secp256k1 public key) and an underlay address (enode:// URL). There are no changes to the BzzAddr format in this commit, but certain operations such as creating a BzzAddr from a node ID are now impossible because node IDs aren't public keys anymore. Most swarm-related changes in the series remove uses of NewAddrFromNodeID, replacing it with NewAddr which takes a complete node as argument. ToOverlayAddr is removed because we can just use the node ID directly.
2018-09-24 22:59:00 +00:00
func triggerChecks(ctx context.Context, ids []enode.ID, trigger chan enode.ID, interval time.Duration) {
tick := time.NewTicker(interval)
defer tick.Stop()
for {
select {
case <-tick.C:
for _, id := range ids {
select {
case trigger <- id:
case <-ctx.Done():
return
}
}
case <-ctx.Done():
return
}
}
}
// \todo: refactor to implement shapshots
// and connect configuration methods once these are moved from
// swarm/network/simulations/connect.go
func BenchmarkMinimalService(b *testing.B) {
b.Run("ring/32", benchmarkMinimalServiceTmp)
}
func benchmarkMinimalServiceTmp(b *testing.B) {
// stop timer to discard setup time pollution
args := strings.Split(b.Name(), "/")
nodeCount, err := strconv.ParseInt(args[2], 10, 16)
if err != nil {
b.Fatal(err)
}
for i := 0; i < b.N; i++ {
// this is a minimal service, whose protocol will close a channel upon run of protocol
// making it possible to bench the time it takes for the service to start and protocol actually to be run
protoCMap := make(map[enode.ID]map[enode.ID]chan struct{})
adapter := adapters.NewSimAdapter(adapters.Services{
"noopwoop": func(ctx *adapters.ServiceContext) (node.Service, error) {
protoCMap[ctx.Config.ID] = make(map[enode.ID]chan struct{})
svc := NewNoopService(protoCMap[ctx.Config.ID])
return svc, nil
},
})
// create network
network := NewNetwork(adapter, &NetworkConfig{
DefaultService: "noopwoop",
})
defer network.Shutdown()
// create and start nodes
ids := make([]enode.ID, nodeCount)
for i := 0; i < int(nodeCount); i++ {
conf := adapters.RandomNodeConfig()
node, err := network.NewNodeWithConfig(conf)
if err != nil {
b.Fatalf("error creating node: %s", err)
}
if err := network.Start(node.ID()); err != nil {
b.Fatalf("error starting node: %s", err)
}
ids[i] = node.ID()
}
// ready, set, go
b.ResetTimer()
// connect nodes in a ring
for i, id := range ids {
peerID := ids[(i+1)%len(ids)]
if err := network.Connect(id, peerID); err != nil {
b.Fatal(err)
}
}
// wait for all protocols to signal to close down
ctx, cancel := context.WithTimeout(context.TODO(), time.Second)
defer cancel()
for nodid, peers := range protoCMap {
for peerid, peerC := range peers {
log.Debug("getting ", "node", nodid, "peer", peerid)
select {
case <-ctx.Done():
b.Fatal(ctx.Err())
case <-peerC:
}
}
}
}
}
p2p, swarm: fix node up races by granular locking (#18976) * swarm/network: DRY out repeated giga comment I not necessarily agree with the way we wait for event propagation. But I truly disagree with having duplicated giga comments. * p2p/simulations: encapsulate Node.Up field so we avoid data races The Node.Up field was accessed concurrently without "proper" locking. There was a lock on Network and that was used sometimes to access the field. Other times the locking was missed and we had a data race. For example: https://github.com/ethereum/go-ethereum/pull/18464 The case above was solved, but there were still intermittent/hard to reproduce races. So let's solve the issue permanently. resolves: ethersphere/go-ethereum#1146 * p2p/simulations: fix unmarshal of simulations.Node Making Node.Up field private in 13292ee897e345045fbfab3bda23a77589a271c1 broke TestHTTPNetwork and TestHTTPSnapshot. Because the default UnmarshalJSON does not handle unexported fields. Important: The fix is partial and not proper to my taste. But I cut scope as I think the fix may require a change to the current serialization format. New ticket: https://github.com/ethersphere/go-ethereum/issues/1177 * p2p/simulations: Add a sanity test case for Node.Config UnmarshalJSON * p2p/simulations: revert back to defer Unlock() pattern for Network It's a good patten to call `defer Unlock()` right after `Lock()` so (new) error cases won't miss to unlock. Let's get back to that pattern. The patten was abandoned in 85a79b3ad3c5863f8612d25c246bcfad339f36b7, while fixing a data race. That data race does not exist anymore, since the Node.Up field got hidden behind its own lock. * p2p/simulations: consistent naming for test providers Node.UnmarshalJSON * p2p/simulations: remove JSON annotation from private fields of Node As unexported fields are not serialized. * p2p/simulations: fix deadlock in Network.GetRandomDownNode() Problem: GetRandomDownNode() locks -> getDownNodeIDs() -> GetNodes() tries to lock -> deadlock On Network type, unexported functions must assume that `net.lock` is already acquired and should not call exported functions which might try to lock again. * p2p/simulations: ensure method conformity for Network Connect* methods were moved to p2p/simulations.Network from swarm/network/simulation. However these new methods did not follow the pattern of Network methods, i.e., all exported method locks the whole Network either for read or write. * p2p/simulations: fix deadlock during network shutdown `TestDiscoveryPersistenceSimulationSimAdapter` often got into deadlock. The execution was stuck on two locks, i.e, `Kademlia.lock` and `p2p/simulations.Network.lock`. Usually the test got stuck once in each 20 executions with high confidence. `Kademlia` was stuck in `Kademlia.EachAddr()` and `Network` in `Network.Stop()`. Solution: in `Network.Stop()` `net.lock` must be released before calling `node.Stop()` as stopping a node (somehow - I did not find the exact code path) causes `Network.InitConn()` to be called from `Kademlia.SuggestPeer()` and that blocks on `net.lock`. Related ticket: https://github.com/ethersphere/go-ethereum/issues/1223 * swarm/state: simplify if statement in DBStore.Put() * p2p/simulations: remove faulty godoc from private function The comment started with the wrong method name. The method is simple and self explanatory. Also, it's private. => Let's just remove the comment.
2019-02-18 06:38:14 +00:00
func TestNode_UnmarshalJSON(t *testing.T) {
t.Run("up_field", func(t *testing.T) {
runNodeUnmarshalJSON(t, casesNodeUnmarshalJSONUpField())
})
t.Run("config_field", func(t *testing.T) {
runNodeUnmarshalJSON(t, casesNodeUnmarshalJSONConfigField())
})
p2p, swarm: fix node up races by granular locking (#18976) * swarm/network: DRY out repeated giga comment I not necessarily agree with the way we wait for event propagation. But I truly disagree with having duplicated giga comments. * p2p/simulations: encapsulate Node.Up field so we avoid data races The Node.Up field was accessed concurrently without "proper" locking. There was a lock on Network and that was used sometimes to access the field. Other times the locking was missed and we had a data race. For example: https://github.com/ethereum/go-ethereum/pull/18464 The case above was solved, but there were still intermittent/hard to reproduce races. So let's solve the issue permanently. resolves: ethersphere/go-ethereum#1146 * p2p/simulations: fix unmarshal of simulations.Node Making Node.Up field private in 13292ee897e345045fbfab3bda23a77589a271c1 broke TestHTTPNetwork and TestHTTPSnapshot. Because the default UnmarshalJSON does not handle unexported fields. Important: The fix is partial and not proper to my taste. But I cut scope as I think the fix may require a change to the current serialization format. New ticket: https://github.com/ethersphere/go-ethereum/issues/1177 * p2p/simulations: Add a sanity test case for Node.Config UnmarshalJSON * p2p/simulations: revert back to defer Unlock() pattern for Network It's a good patten to call `defer Unlock()` right after `Lock()` so (new) error cases won't miss to unlock. Let's get back to that pattern. The patten was abandoned in 85a79b3ad3c5863f8612d25c246bcfad339f36b7, while fixing a data race. That data race does not exist anymore, since the Node.Up field got hidden behind its own lock. * p2p/simulations: consistent naming for test providers Node.UnmarshalJSON * p2p/simulations: remove JSON annotation from private fields of Node As unexported fields are not serialized. * p2p/simulations: fix deadlock in Network.GetRandomDownNode() Problem: GetRandomDownNode() locks -> getDownNodeIDs() -> GetNodes() tries to lock -> deadlock On Network type, unexported functions must assume that `net.lock` is already acquired and should not call exported functions which might try to lock again. * p2p/simulations: ensure method conformity for Network Connect* methods were moved to p2p/simulations.Network from swarm/network/simulation. However these new methods did not follow the pattern of Network methods, i.e., all exported method locks the whole Network either for read or write. * p2p/simulations: fix deadlock during network shutdown `TestDiscoveryPersistenceSimulationSimAdapter` often got into deadlock. The execution was stuck on two locks, i.e, `Kademlia.lock` and `p2p/simulations.Network.lock`. Usually the test got stuck once in each 20 executions with high confidence. `Kademlia` was stuck in `Kademlia.EachAddr()` and `Network` in `Network.Stop()`. Solution: in `Network.Stop()` `net.lock` must be released before calling `node.Stop()` as stopping a node (somehow - I did not find the exact code path) causes `Network.InitConn()` to be called from `Kademlia.SuggestPeer()` and that blocks on `net.lock`. Related ticket: https://github.com/ethersphere/go-ethereum/issues/1223 * swarm/state: simplify if statement in DBStore.Put() * p2p/simulations: remove faulty godoc from private function The comment started with the wrong method name. The method is simple and self explanatory. Also, it's private. => Let's just remove the comment.
2019-02-18 06:38:14 +00:00
}
func runNodeUnmarshalJSON(t *testing.T, tests []nodeUnmarshalTestCase) {
t.Helper()
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
var got *Node
if err := json.Unmarshal([]byte(tt.marshaled), &got); err != nil {
p2p, swarm: fix node up races by granular locking (#18976) * swarm/network: DRY out repeated giga comment I not necessarily agree with the way we wait for event propagation. But I truly disagree with having duplicated giga comments. * p2p/simulations: encapsulate Node.Up field so we avoid data races The Node.Up field was accessed concurrently without "proper" locking. There was a lock on Network and that was used sometimes to access the field. Other times the locking was missed and we had a data race. For example: https://github.com/ethereum/go-ethereum/pull/18464 The case above was solved, but there were still intermittent/hard to reproduce races. So let's solve the issue permanently. resolves: ethersphere/go-ethereum#1146 * p2p/simulations: fix unmarshal of simulations.Node Making Node.Up field private in 13292ee897e345045fbfab3bda23a77589a271c1 broke TestHTTPNetwork and TestHTTPSnapshot. Because the default UnmarshalJSON does not handle unexported fields. Important: The fix is partial and not proper to my taste. But I cut scope as I think the fix may require a change to the current serialization format. New ticket: https://github.com/ethersphere/go-ethereum/issues/1177 * p2p/simulations: Add a sanity test case for Node.Config UnmarshalJSON * p2p/simulations: revert back to defer Unlock() pattern for Network It's a good patten to call `defer Unlock()` right after `Lock()` so (new) error cases won't miss to unlock. Let's get back to that pattern. The patten was abandoned in 85a79b3ad3c5863f8612d25c246bcfad339f36b7, while fixing a data race. That data race does not exist anymore, since the Node.Up field got hidden behind its own lock. * p2p/simulations: consistent naming for test providers Node.UnmarshalJSON * p2p/simulations: remove JSON annotation from private fields of Node As unexported fields are not serialized. * p2p/simulations: fix deadlock in Network.GetRandomDownNode() Problem: GetRandomDownNode() locks -> getDownNodeIDs() -> GetNodes() tries to lock -> deadlock On Network type, unexported functions must assume that `net.lock` is already acquired and should not call exported functions which might try to lock again. * p2p/simulations: ensure method conformity for Network Connect* methods were moved to p2p/simulations.Network from swarm/network/simulation. However these new methods did not follow the pattern of Network methods, i.e., all exported method locks the whole Network either for read or write. * p2p/simulations: fix deadlock during network shutdown `TestDiscoveryPersistenceSimulationSimAdapter` often got into deadlock. The execution was stuck on two locks, i.e, `Kademlia.lock` and `p2p/simulations.Network.lock`. Usually the test got stuck once in each 20 executions with high confidence. `Kademlia` was stuck in `Kademlia.EachAddr()` and `Network` in `Network.Stop()`. Solution: in `Network.Stop()` `net.lock` must be released before calling `node.Stop()` as stopping a node (somehow - I did not find the exact code path) causes `Network.InitConn()` to be called from `Kademlia.SuggestPeer()` and that blocks on `net.lock`. Related ticket: https://github.com/ethersphere/go-ethereum/issues/1223 * swarm/state: simplify if statement in DBStore.Put() * p2p/simulations: remove faulty godoc from private function The comment started with the wrong method name. The method is simple and self explanatory. Also, it's private. => Let's just remove the comment.
2019-02-18 06:38:14 +00:00
expectErrorMessageToContain(t, err, tt.wantErr)
got = nil
p2p, swarm: fix node up races by granular locking (#18976) * swarm/network: DRY out repeated giga comment I not necessarily agree with the way we wait for event propagation. But I truly disagree with having duplicated giga comments. * p2p/simulations: encapsulate Node.Up field so we avoid data races The Node.Up field was accessed concurrently without "proper" locking. There was a lock on Network and that was used sometimes to access the field. Other times the locking was missed and we had a data race. For example: https://github.com/ethereum/go-ethereum/pull/18464 The case above was solved, but there were still intermittent/hard to reproduce races. So let's solve the issue permanently. resolves: ethersphere/go-ethereum#1146 * p2p/simulations: fix unmarshal of simulations.Node Making Node.Up field private in 13292ee897e345045fbfab3bda23a77589a271c1 broke TestHTTPNetwork and TestHTTPSnapshot. Because the default UnmarshalJSON does not handle unexported fields. Important: The fix is partial and not proper to my taste. But I cut scope as I think the fix may require a change to the current serialization format. New ticket: https://github.com/ethersphere/go-ethereum/issues/1177 * p2p/simulations: Add a sanity test case for Node.Config UnmarshalJSON * p2p/simulations: revert back to defer Unlock() pattern for Network It's a good patten to call `defer Unlock()` right after `Lock()` so (new) error cases won't miss to unlock. Let's get back to that pattern. The patten was abandoned in 85a79b3ad3c5863f8612d25c246bcfad339f36b7, while fixing a data race. That data race does not exist anymore, since the Node.Up field got hidden behind its own lock. * p2p/simulations: consistent naming for test providers Node.UnmarshalJSON * p2p/simulations: remove JSON annotation from private fields of Node As unexported fields are not serialized. * p2p/simulations: fix deadlock in Network.GetRandomDownNode() Problem: GetRandomDownNode() locks -> getDownNodeIDs() -> GetNodes() tries to lock -> deadlock On Network type, unexported functions must assume that `net.lock` is already acquired and should not call exported functions which might try to lock again. * p2p/simulations: ensure method conformity for Network Connect* methods were moved to p2p/simulations.Network from swarm/network/simulation. However these new methods did not follow the pattern of Network methods, i.e., all exported method locks the whole Network either for read or write. * p2p/simulations: fix deadlock during network shutdown `TestDiscoveryPersistenceSimulationSimAdapter` often got into deadlock. The execution was stuck on two locks, i.e, `Kademlia.lock` and `p2p/simulations.Network.lock`. Usually the test got stuck once in each 20 executions with high confidence. `Kademlia` was stuck in `Kademlia.EachAddr()` and `Network` in `Network.Stop()`. Solution: in `Network.Stop()` `net.lock` must be released before calling `node.Stop()` as stopping a node (somehow - I did not find the exact code path) causes `Network.InitConn()` to be called from `Kademlia.SuggestPeer()` and that blocks on `net.lock`. Related ticket: https://github.com/ethersphere/go-ethereum/issues/1223 * swarm/state: simplify if statement in DBStore.Put() * p2p/simulations: remove faulty godoc from private function The comment started with the wrong method name. The method is simple and self explanatory. Also, it's private. => Let's just remove the comment.
2019-02-18 06:38:14 +00:00
}
expectNodeEquality(t, got, tt.want)
})
}
}
type nodeUnmarshalTestCase struct {
name string
marshaled string
want *Node
p2p, swarm: fix node up races by granular locking (#18976) * swarm/network: DRY out repeated giga comment I not necessarily agree with the way we wait for event propagation. But I truly disagree with having duplicated giga comments. * p2p/simulations: encapsulate Node.Up field so we avoid data races The Node.Up field was accessed concurrently without "proper" locking. There was a lock on Network and that was used sometimes to access the field. Other times the locking was missed and we had a data race. For example: https://github.com/ethereum/go-ethereum/pull/18464 The case above was solved, but there were still intermittent/hard to reproduce races. So let's solve the issue permanently. resolves: ethersphere/go-ethereum#1146 * p2p/simulations: fix unmarshal of simulations.Node Making Node.Up field private in 13292ee897e345045fbfab3bda23a77589a271c1 broke TestHTTPNetwork and TestHTTPSnapshot. Because the default UnmarshalJSON does not handle unexported fields. Important: The fix is partial and not proper to my taste. But I cut scope as I think the fix may require a change to the current serialization format. New ticket: https://github.com/ethersphere/go-ethereum/issues/1177 * p2p/simulations: Add a sanity test case for Node.Config UnmarshalJSON * p2p/simulations: revert back to defer Unlock() pattern for Network It's a good patten to call `defer Unlock()` right after `Lock()` so (new) error cases won't miss to unlock. Let's get back to that pattern. The patten was abandoned in 85a79b3ad3c5863f8612d25c246bcfad339f36b7, while fixing a data race. That data race does not exist anymore, since the Node.Up field got hidden behind its own lock. * p2p/simulations: consistent naming for test providers Node.UnmarshalJSON * p2p/simulations: remove JSON annotation from private fields of Node As unexported fields are not serialized. * p2p/simulations: fix deadlock in Network.GetRandomDownNode() Problem: GetRandomDownNode() locks -> getDownNodeIDs() -> GetNodes() tries to lock -> deadlock On Network type, unexported functions must assume that `net.lock` is already acquired and should not call exported functions which might try to lock again. * p2p/simulations: ensure method conformity for Network Connect* methods were moved to p2p/simulations.Network from swarm/network/simulation. However these new methods did not follow the pattern of Network methods, i.e., all exported method locks the whole Network either for read or write. * p2p/simulations: fix deadlock during network shutdown `TestDiscoveryPersistenceSimulationSimAdapter` often got into deadlock. The execution was stuck on two locks, i.e, `Kademlia.lock` and `p2p/simulations.Network.lock`. Usually the test got stuck once in each 20 executions with high confidence. `Kademlia` was stuck in `Kademlia.EachAddr()` and `Network` in `Network.Stop()`. Solution: in `Network.Stop()` `net.lock` must be released before calling `node.Stop()` as stopping a node (somehow - I did not find the exact code path) causes `Network.InitConn()` to be called from `Kademlia.SuggestPeer()` and that blocks on `net.lock`. Related ticket: https://github.com/ethersphere/go-ethereum/issues/1223 * swarm/state: simplify if statement in DBStore.Put() * p2p/simulations: remove faulty godoc from private function The comment started with the wrong method name. The method is simple and self explanatory. Also, it's private. => Let's just remove the comment.
2019-02-18 06:38:14 +00:00
wantErr string
}
func expectErrorMessageToContain(t *testing.T, got error, want string) {
t.Helper()
if got == nil && want == "" {
return
}
if got == nil && want != "" {
t.Errorf("error was expected, got: nil, want: %v", want)
return
}
if !strings.Contains(got.Error(), want) {
t.Errorf(
"unexpected error message, got %v, want: %v",
want,
got,
)
}
}
func expectNodeEquality(t *testing.T, got, want *Node) {
p2p, swarm: fix node up races by granular locking (#18976) * swarm/network: DRY out repeated giga comment I not necessarily agree with the way we wait for event propagation. But I truly disagree with having duplicated giga comments. * p2p/simulations: encapsulate Node.Up field so we avoid data races The Node.Up field was accessed concurrently without "proper" locking. There was a lock on Network and that was used sometimes to access the field. Other times the locking was missed and we had a data race. For example: https://github.com/ethereum/go-ethereum/pull/18464 The case above was solved, but there were still intermittent/hard to reproduce races. So let's solve the issue permanently. resolves: ethersphere/go-ethereum#1146 * p2p/simulations: fix unmarshal of simulations.Node Making Node.Up field private in 13292ee897e345045fbfab3bda23a77589a271c1 broke TestHTTPNetwork and TestHTTPSnapshot. Because the default UnmarshalJSON does not handle unexported fields. Important: The fix is partial and not proper to my taste. But I cut scope as I think the fix may require a change to the current serialization format. New ticket: https://github.com/ethersphere/go-ethereum/issues/1177 * p2p/simulations: Add a sanity test case for Node.Config UnmarshalJSON * p2p/simulations: revert back to defer Unlock() pattern for Network It's a good patten to call `defer Unlock()` right after `Lock()` so (new) error cases won't miss to unlock. Let's get back to that pattern. The patten was abandoned in 85a79b3ad3c5863f8612d25c246bcfad339f36b7, while fixing a data race. That data race does not exist anymore, since the Node.Up field got hidden behind its own lock. * p2p/simulations: consistent naming for test providers Node.UnmarshalJSON * p2p/simulations: remove JSON annotation from private fields of Node As unexported fields are not serialized. * p2p/simulations: fix deadlock in Network.GetRandomDownNode() Problem: GetRandomDownNode() locks -> getDownNodeIDs() -> GetNodes() tries to lock -> deadlock On Network type, unexported functions must assume that `net.lock` is already acquired and should not call exported functions which might try to lock again. * p2p/simulations: ensure method conformity for Network Connect* methods were moved to p2p/simulations.Network from swarm/network/simulation. However these new methods did not follow the pattern of Network methods, i.e., all exported method locks the whole Network either for read or write. * p2p/simulations: fix deadlock during network shutdown `TestDiscoveryPersistenceSimulationSimAdapter` often got into deadlock. The execution was stuck on two locks, i.e, `Kademlia.lock` and `p2p/simulations.Network.lock`. Usually the test got stuck once in each 20 executions with high confidence. `Kademlia` was stuck in `Kademlia.EachAddr()` and `Network` in `Network.Stop()`. Solution: in `Network.Stop()` `net.lock` must be released before calling `node.Stop()` as stopping a node (somehow - I did not find the exact code path) causes `Network.InitConn()` to be called from `Kademlia.SuggestPeer()` and that blocks on `net.lock`. Related ticket: https://github.com/ethersphere/go-ethereum/issues/1223 * swarm/state: simplify if statement in DBStore.Put() * p2p/simulations: remove faulty godoc from private function The comment started with the wrong method name. The method is simple and self explanatory. Also, it's private. => Let's just remove the comment.
2019-02-18 06:38:14 +00:00
t.Helper()
if !reflect.DeepEqual(got, want) {
t.Errorf("Node.UnmarshalJSON() = %v, want %v", got, want)
}
}
func casesNodeUnmarshalJSONUpField() []nodeUnmarshalTestCase {
return []nodeUnmarshalTestCase{
{
name: "empty json",
marshaled: "{}",
want: newNode(nil, nil, false),
p2p, swarm: fix node up races by granular locking (#18976) * swarm/network: DRY out repeated giga comment I not necessarily agree with the way we wait for event propagation. But I truly disagree with having duplicated giga comments. * p2p/simulations: encapsulate Node.Up field so we avoid data races The Node.Up field was accessed concurrently without "proper" locking. There was a lock on Network and that was used sometimes to access the field. Other times the locking was missed and we had a data race. For example: https://github.com/ethereum/go-ethereum/pull/18464 The case above was solved, but there were still intermittent/hard to reproduce races. So let's solve the issue permanently. resolves: ethersphere/go-ethereum#1146 * p2p/simulations: fix unmarshal of simulations.Node Making Node.Up field private in 13292ee897e345045fbfab3bda23a77589a271c1 broke TestHTTPNetwork and TestHTTPSnapshot. Because the default UnmarshalJSON does not handle unexported fields. Important: The fix is partial and not proper to my taste. But I cut scope as I think the fix may require a change to the current serialization format. New ticket: https://github.com/ethersphere/go-ethereum/issues/1177 * p2p/simulations: Add a sanity test case for Node.Config UnmarshalJSON * p2p/simulations: revert back to defer Unlock() pattern for Network It's a good patten to call `defer Unlock()` right after `Lock()` so (new) error cases won't miss to unlock. Let's get back to that pattern. The patten was abandoned in 85a79b3ad3c5863f8612d25c246bcfad339f36b7, while fixing a data race. That data race does not exist anymore, since the Node.Up field got hidden behind its own lock. * p2p/simulations: consistent naming for test providers Node.UnmarshalJSON * p2p/simulations: remove JSON annotation from private fields of Node As unexported fields are not serialized. * p2p/simulations: fix deadlock in Network.GetRandomDownNode() Problem: GetRandomDownNode() locks -> getDownNodeIDs() -> GetNodes() tries to lock -> deadlock On Network type, unexported functions must assume that `net.lock` is already acquired and should not call exported functions which might try to lock again. * p2p/simulations: ensure method conformity for Network Connect* methods were moved to p2p/simulations.Network from swarm/network/simulation. However these new methods did not follow the pattern of Network methods, i.e., all exported method locks the whole Network either for read or write. * p2p/simulations: fix deadlock during network shutdown `TestDiscoveryPersistenceSimulationSimAdapter` often got into deadlock. The execution was stuck on two locks, i.e, `Kademlia.lock` and `p2p/simulations.Network.lock`. Usually the test got stuck once in each 20 executions with high confidence. `Kademlia` was stuck in `Kademlia.EachAddr()` and `Network` in `Network.Stop()`. Solution: in `Network.Stop()` `net.lock` must be released before calling `node.Stop()` as stopping a node (somehow - I did not find the exact code path) causes `Network.InitConn()` to be called from `Kademlia.SuggestPeer()` and that blocks on `net.lock`. Related ticket: https://github.com/ethersphere/go-ethereum/issues/1223 * swarm/state: simplify if statement in DBStore.Put() * p2p/simulations: remove faulty godoc from private function The comment started with the wrong method name. The method is simple and self explanatory. Also, it's private. => Let's just remove the comment.
2019-02-18 06:38:14 +00:00
},
{
name: "a stopped node",
marshaled: "{\"up\": false}",
want: newNode(nil, nil, false),
p2p, swarm: fix node up races by granular locking (#18976) * swarm/network: DRY out repeated giga comment I not necessarily agree with the way we wait for event propagation. But I truly disagree with having duplicated giga comments. * p2p/simulations: encapsulate Node.Up field so we avoid data races The Node.Up field was accessed concurrently without "proper" locking. There was a lock on Network and that was used sometimes to access the field. Other times the locking was missed and we had a data race. For example: https://github.com/ethereum/go-ethereum/pull/18464 The case above was solved, but there were still intermittent/hard to reproduce races. So let's solve the issue permanently. resolves: ethersphere/go-ethereum#1146 * p2p/simulations: fix unmarshal of simulations.Node Making Node.Up field private in 13292ee897e345045fbfab3bda23a77589a271c1 broke TestHTTPNetwork and TestHTTPSnapshot. Because the default UnmarshalJSON does not handle unexported fields. Important: The fix is partial and not proper to my taste. But I cut scope as I think the fix may require a change to the current serialization format. New ticket: https://github.com/ethersphere/go-ethereum/issues/1177 * p2p/simulations: Add a sanity test case for Node.Config UnmarshalJSON * p2p/simulations: revert back to defer Unlock() pattern for Network It's a good patten to call `defer Unlock()` right after `Lock()` so (new) error cases won't miss to unlock. Let's get back to that pattern. The patten was abandoned in 85a79b3ad3c5863f8612d25c246bcfad339f36b7, while fixing a data race. That data race does not exist anymore, since the Node.Up field got hidden behind its own lock. * p2p/simulations: consistent naming for test providers Node.UnmarshalJSON * p2p/simulations: remove JSON annotation from private fields of Node As unexported fields are not serialized. * p2p/simulations: fix deadlock in Network.GetRandomDownNode() Problem: GetRandomDownNode() locks -> getDownNodeIDs() -> GetNodes() tries to lock -> deadlock On Network type, unexported functions must assume that `net.lock` is already acquired and should not call exported functions which might try to lock again. * p2p/simulations: ensure method conformity for Network Connect* methods were moved to p2p/simulations.Network from swarm/network/simulation. However these new methods did not follow the pattern of Network methods, i.e., all exported method locks the whole Network either for read or write. * p2p/simulations: fix deadlock during network shutdown `TestDiscoveryPersistenceSimulationSimAdapter` often got into deadlock. The execution was stuck on two locks, i.e, `Kademlia.lock` and `p2p/simulations.Network.lock`. Usually the test got stuck once in each 20 executions with high confidence. `Kademlia` was stuck in `Kademlia.EachAddr()` and `Network` in `Network.Stop()`. Solution: in `Network.Stop()` `net.lock` must be released before calling `node.Stop()` as stopping a node (somehow - I did not find the exact code path) causes `Network.InitConn()` to be called from `Kademlia.SuggestPeer()` and that blocks on `net.lock`. Related ticket: https://github.com/ethersphere/go-ethereum/issues/1223 * swarm/state: simplify if statement in DBStore.Put() * p2p/simulations: remove faulty godoc from private function The comment started with the wrong method name. The method is simple and self explanatory. Also, it's private. => Let's just remove the comment.
2019-02-18 06:38:14 +00:00
},
{
name: "a running node",
marshaled: "{\"up\": true}",
want: newNode(nil, nil, true),
p2p, swarm: fix node up races by granular locking (#18976) * swarm/network: DRY out repeated giga comment I not necessarily agree with the way we wait for event propagation. But I truly disagree with having duplicated giga comments. * p2p/simulations: encapsulate Node.Up field so we avoid data races The Node.Up field was accessed concurrently without "proper" locking. There was a lock on Network and that was used sometimes to access the field. Other times the locking was missed and we had a data race. For example: https://github.com/ethereum/go-ethereum/pull/18464 The case above was solved, but there were still intermittent/hard to reproduce races. So let's solve the issue permanently. resolves: ethersphere/go-ethereum#1146 * p2p/simulations: fix unmarshal of simulations.Node Making Node.Up field private in 13292ee897e345045fbfab3bda23a77589a271c1 broke TestHTTPNetwork and TestHTTPSnapshot. Because the default UnmarshalJSON does not handle unexported fields. Important: The fix is partial and not proper to my taste. But I cut scope as I think the fix may require a change to the current serialization format. New ticket: https://github.com/ethersphere/go-ethereum/issues/1177 * p2p/simulations: Add a sanity test case for Node.Config UnmarshalJSON * p2p/simulations: revert back to defer Unlock() pattern for Network It's a good patten to call `defer Unlock()` right after `Lock()` so (new) error cases won't miss to unlock. Let's get back to that pattern. The patten was abandoned in 85a79b3ad3c5863f8612d25c246bcfad339f36b7, while fixing a data race. That data race does not exist anymore, since the Node.Up field got hidden behind its own lock. * p2p/simulations: consistent naming for test providers Node.UnmarshalJSON * p2p/simulations: remove JSON annotation from private fields of Node As unexported fields are not serialized. * p2p/simulations: fix deadlock in Network.GetRandomDownNode() Problem: GetRandomDownNode() locks -> getDownNodeIDs() -> GetNodes() tries to lock -> deadlock On Network type, unexported functions must assume that `net.lock` is already acquired and should not call exported functions which might try to lock again. * p2p/simulations: ensure method conformity for Network Connect* methods were moved to p2p/simulations.Network from swarm/network/simulation. However these new methods did not follow the pattern of Network methods, i.e., all exported method locks the whole Network either for read or write. * p2p/simulations: fix deadlock during network shutdown `TestDiscoveryPersistenceSimulationSimAdapter` often got into deadlock. The execution was stuck on two locks, i.e, `Kademlia.lock` and `p2p/simulations.Network.lock`. Usually the test got stuck once in each 20 executions with high confidence. `Kademlia` was stuck in `Kademlia.EachAddr()` and `Network` in `Network.Stop()`. Solution: in `Network.Stop()` `net.lock` must be released before calling `node.Stop()` as stopping a node (somehow - I did not find the exact code path) causes `Network.InitConn()` to be called from `Kademlia.SuggestPeer()` and that blocks on `net.lock`. Related ticket: https://github.com/ethersphere/go-ethereum/issues/1223 * swarm/state: simplify if statement in DBStore.Put() * p2p/simulations: remove faulty godoc from private function The comment started with the wrong method name. The method is simple and self explanatory. Also, it's private. => Let's just remove the comment.
2019-02-18 06:38:14 +00:00
},
{
name: "invalid JSON value on valid key",
marshaled: "{\"up\": foo}",
wantErr: "invalid character",
},
{
name: "invalid JSON key and value",
marshaled: "{foo: bar}",
wantErr: "invalid character",
},
{
name: "bool value expected but got something else (string)",
marshaled: "{\"up\": \"true\"}",
wantErr: "cannot unmarshal string into Go struct",
},
}
}
func casesNodeUnmarshalJSONConfigField() []nodeUnmarshalTestCase {
// Don't do a big fuss around testing, as adapters.NodeConfig should
// handle it's own serialization. Just do a sanity check.
return []nodeUnmarshalTestCase{
{
name: "Config field is omitted",
marshaled: "{}",
want: newNode(nil, nil, false),
p2p, swarm: fix node up races by granular locking (#18976) * swarm/network: DRY out repeated giga comment I not necessarily agree with the way we wait for event propagation. But I truly disagree with having duplicated giga comments. * p2p/simulations: encapsulate Node.Up field so we avoid data races The Node.Up field was accessed concurrently without "proper" locking. There was a lock on Network and that was used sometimes to access the field. Other times the locking was missed and we had a data race. For example: https://github.com/ethereum/go-ethereum/pull/18464 The case above was solved, but there were still intermittent/hard to reproduce races. So let's solve the issue permanently. resolves: ethersphere/go-ethereum#1146 * p2p/simulations: fix unmarshal of simulations.Node Making Node.Up field private in 13292ee897e345045fbfab3bda23a77589a271c1 broke TestHTTPNetwork and TestHTTPSnapshot. Because the default UnmarshalJSON does not handle unexported fields. Important: The fix is partial and not proper to my taste. But I cut scope as I think the fix may require a change to the current serialization format. New ticket: https://github.com/ethersphere/go-ethereum/issues/1177 * p2p/simulations: Add a sanity test case for Node.Config UnmarshalJSON * p2p/simulations: revert back to defer Unlock() pattern for Network It's a good patten to call `defer Unlock()` right after `Lock()` so (new) error cases won't miss to unlock. Let's get back to that pattern. The patten was abandoned in 85a79b3ad3c5863f8612d25c246bcfad339f36b7, while fixing a data race. That data race does not exist anymore, since the Node.Up field got hidden behind its own lock. * p2p/simulations: consistent naming for test providers Node.UnmarshalJSON * p2p/simulations: remove JSON annotation from private fields of Node As unexported fields are not serialized. * p2p/simulations: fix deadlock in Network.GetRandomDownNode() Problem: GetRandomDownNode() locks -> getDownNodeIDs() -> GetNodes() tries to lock -> deadlock On Network type, unexported functions must assume that `net.lock` is already acquired and should not call exported functions which might try to lock again. * p2p/simulations: ensure method conformity for Network Connect* methods were moved to p2p/simulations.Network from swarm/network/simulation. However these new methods did not follow the pattern of Network methods, i.e., all exported method locks the whole Network either for read or write. * p2p/simulations: fix deadlock during network shutdown `TestDiscoveryPersistenceSimulationSimAdapter` often got into deadlock. The execution was stuck on two locks, i.e, `Kademlia.lock` and `p2p/simulations.Network.lock`. Usually the test got stuck once in each 20 executions with high confidence. `Kademlia` was stuck in `Kademlia.EachAddr()` and `Network` in `Network.Stop()`. Solution: in `Network.Stop()` `net.lock` must be released before calling `node.Stop()` as stopping a node (somehow - I did not find the exact code path) causes `Network.InitConn()` to be called from `Kademlia.SuggestPeer()` and that blocks on `net.lock`. Related ticket: https://github.com/ethersphere/go-ethereum/issues/1223 * swarm/state: simplify if statement in DBStore.Put() * p2p/simulations: remove faulty godoc from private function The comment started with the wrong method name. The method is simple and self explanatory. Also, it's private. => Let's just remove the comment.
2019-02-18 06:38:14 +00:00
},
{
name: "Config field is nil",
marshaled: "{\"config\": null}",
want: newNode(nil, nil, false),
p2p, swarm: fix node up races by granular locking (#18976) * swarm/network: DRY out repeated giga comment I not necessarily agree with the way we wait for event propagation. But I truly disagree with having duplicated giga comments. * p2p/simulations: encapsulate Node.Up field so we avoid data races The Node.Up field was accessed concurrently without "proper" locking. There was a lock on Network and that was used sometimes to access the field. Other times the locking was missed and we had a data race. For example: https://github.com/ethereum/go-ethereum/pull/18464 The case above was solved, but there were still intermittent/hard to reproduce races. So let's solve the issue permanently. resolves: ethersphere/go-ethereum#1146 * p2p/simulations: fix unmarshal of simulations.Node Making Node.Up field private in 13292ee897e345045fbfab3bda23a77589a271c1 broke TestHTTPNetwork and TestHTTPSnapshot. Because the default UnmarshalJSON does not handle unexported fields. Important: The fix is partial and not proper to my taste. But I cut scope as I think the fix may require a change to the current serialization format. New ticket: https://github.com/ethersphere/go-ethereum/issues/1177 * p2p/simulations: Add a sanity test case for Node.Config UnmarshalJSON * p2p/simulations: revert back to defer Unlock() pattern for Network It's a good patten to call `defer Unlock()` right after `Lock()` so (new) error cases won't miss to unlock. Let's get back to that pattern. The patten was abandoned in 85a79b3ad3c5863f8612d25c246bcfad339f36b7, while fixing a data race. That data race does not exist anymore, since the Node.Up field got hidden behind its own lock. * p2p/simulations: consistent naming for test providers Node.UnmarshalJSON * p2p/simulations: remove JSON annotation from private fields of Node As unexported fields are not serialized. * p2p/simulations: fix deadlock in Network.GetRandomDownNode() Problem: GetRandomDownNode() locks -> getDownNodeIDs() -> GetNodes() tries to lock -> deadlock On Network type, unexported functions must assume that `net.lock` is already acquired and should not call exported functions which might try to lock again. * p2p/simulations: ensure method conformity for Network Connect* methods were moved to p2p/simulations.Network from swarm/network/simulation. However these new methods did not follow the pattern of Network methods, i.e., all exported method locks the whole Network either for read or write. * p2p/simulations: fix deadlock during network shutdown `TestDiscoveryPersistenceSimulationSimAdapter` often got into deadlock. The execution was stuck on two locks, i.e, `Kademlia.lock` and `p2p/simulations.Network.lock`. Usually the test got stuck once in each 20 executions with high confidence. `Kademlia` was stuck in `Kademlia.EachAddr()` and `Network` in `Network.Stop()`. Solution: in `Network.Stop()` `net.lock` must be released before calling `node.Stop()` as stopping a node (somehow - I did not find the exact code path) causes `Network.InitConn()` to be called from `Kademlia.SuggestPeer()` and that blocks on `net.lock`. Related ticket: https://github.com/ethersphere/go-ethereum/issues/1223 * swarm/state: simplify if statement in DBStore.Put() * p2p/simulations: remove faulty godoc from private function The comment started with the wrong method name. The method is simple and self explanatory. Also, it's private. => Let's just remove the comment.
2019-02-18 06:38:14 +00:00
},
{
name: "a non default Config field",
marshaled: "{\"config\":{\"name\":\"node_ecdd0\",\"port\":44665}}",
want: newNode(nil, &adapters.NodeConfig{Name: "node_ecdd0", Port: 44665}, false),
p2p, swarm: fix node up races by granular locking (#18976) * swarm/network: DRY out repeated giga comment I not necessarily agree with the way we wait for event propagation. But I truly disagree with having duplicated giga comments. * p2p/simulations: encapsulate Node.Up field so we avoid data races The Node.Up field was accessed concurrently without "proper" locking. There was a lock on Network and that was used sometimes to access the field. Other times the locking was missed and we had a data race. For example: https://github.com/ethereum/go-ethereum/pull/18464 The case above was solved, but there were still intermittent/hard to reproduce races. So let's solve the issue permanently. resolves: ethersphere/go-ethereum#1146 * p2p/simulations: fix unmarshal of simulations.Node Making Node.Up field private in 13292ee897e345045fbfab3bda23a77589a271c1 broke TestHTTPNetwork and TestHTTPSnapshot. Because the default UnmarshalJSON does not handle unexported fields. Important: The fix is partial and not proper to my taste. But I cut scope as I think the fix may require a change to the current serialization format. New ticket: https://github.com/ethersphere/go-ethereum/issues/1177 * p2p/simulations: Add a sanity test case for Node.Config UnmarshalJSON * p2p/simulations: revert back to defer Unlock() pattern for Network It's a good patten to call `defer Unlock()` right after `Lock()` so (new) error cases won't miss to unlock. Let's get back to that pattern. The patten was abandoned in 85a79b3ad3c5863f8612d25c246bcfad339f36b7, while fixing a data race. That data race does not exist anymore, since the Node.Up field got hidden behind its own lock. * p2p/simulations: consistent naming for test providers Node.UnmarshalJSON * p2p/simulations: remove JSON annotation from private fields of Node As unexported fields are not serialized. * p2p/simulations: fix deadlock in Network.GetRandomDownNode() Problem: GetRandomDownNode() locks -> getDownNodeIDs() -> GetNodes() tries to lock -> deadlock On Network type, unexported functions must assume that `net.lock` is already acquired and should not call exported functions which might try to lock again. * p2p/simulations: ensure method conformity for Network Connect* methods were moved to p2p/simulations.Network from swarm/network/simulation. However these new methods did not follow the pattern of Network methods, i.e., all exported method locks the whole Network either for read or write. * p2p/simulations: fix deadlock during network shutdown `TestDiscoveryPersistenceSimulationSimAdapter` often got into deadlock. The execution was stuck on two locks, i.e, `Kademlia.lock` and `p2p/simulations.Network.lock`. Usually the test got stuck once in each 20 executions with high confidence. `Kademlia` was stuck in `Kademlia.EachAddr()` and `Network` in `Network.Stop()`. Solution: in `Network.Stop()` `net.lock` must be released before calling `node.Stop()` as stopping a node (somehow - I did not find the exact code path) causes `Network.InitConn()` to be called from `Kademlia.SuggestPeer()` and that blocks on `net.lock`. Related ticket: https://github.com/ethersphere/go-ethereum/issues/1223 * swarm/state: simplify if statement in DBStore.Put() * p2p/simulations: remove faulty godoc from private function The comment started with the wrong method name. The method is simple and self explanatory. Also, it's private. => Let's just remove the comment.
2019-02-18 06:38:14 +00:00
},
}
}