raftexample: implement Raft snapshot

contrib/raftexample/httpapi.go

@@ -18,7 +18,6 @@ import (
 	"io/ioutil"
 	"log"
 	"net/http"
-	"os"
 	"strconv"
 
 	"github.com/coreos/etcd/raft/raftpb"
@@ -102,25 +101,22 @@ func (h *httpKVAPI) ServeHTTP(w http.ResponseWriter, r *http.Request) {
 }
 
 // serveHttpKVAPI starts a key-value server with a GET/PUT API and listens.
-func serveHttpKVAPI(port int, proposeC chan<- string, confChangeC chan<- raftpb.ConfChange,
-	commitC <-chan *string, errorC <-chan error) {
-
-	// exit when raft goes down
-	go func() {
-		if err, ok := <-errorC; ok {
-			log.Fatal(err)
-		}
-		os.Exit(0)
-	}()
-
+func serveHttpKVAPI(kv *kvstore, port int, confChangeC chan<- raftpb.ConfChange, errorC <-chan error) {
 	srv := http.Server{
 		Addr: ":" + strconv.Itoa(port),
 		Handler: &httpKVAPI{
-			store:       newKVStore(proposeC, commitC, errorC),
+			store:       kv,
 			confChangeC: confChangeC,
 		},
 	}
-	if err := srv.ListenAndServe(); err != nil {
+	go func() {
+		if err := srv.ListenAndServe(); err != nil {
+			log.Fatal(err)
+		}
+	}()
+
+	// exit when raft goes down
+	if err, ok := <-errorC; ok {
 		log.Fatal(err)
 	}
 }

contrib/raftexample/kvstore.go

@@ -20,13 +20,16 @@ import (
 	"encoding/json"
 	"log"
 	"sync"
+
+	"github.com/coreos/etcd/snap"
 )
 
 // a key-value store backed by raft
 type kvstore struct {
-	proposeC chan<- string // channel for proposing updates
-	mu       sync.RWMutex
-	kvStore  map[string]string // current committed key-value pairs
+	proposeC    chan<- string // channel for proposing updates
+	mu          sync.RWMutex
+	kvStore     map[string]string // current committed key-value pairs
+	snapshotter *snap.Snapshotter
 }
 
 type kv struct {
@@ -34,8 +37,8 @@ type kv struct {
 	Val string
 }
 
-func newKVStore(proposeC chan<- string, commitC <-chan *string, errorC <-chan error) *kvstore {
-	s := &kvstore{proposeC: proposeC, kvStore: make(map[string]string)}
+func newKVStore(snapshotter *snap.Snapshotter, proposeC chan<- string, commitC <-chan *string, errorC <-chan error) *kvstore {
+	s := &kvstore{proposeC: proposeC, kvStore: make(map[string]string), snapshotter: snapshotter}
 	// replay log into key-value map
 	s.readCommits(commitC, errorC)
 	// read commits from raft into kvStore map until error
@@ -62,7 +65,18 @@ func (s *kvstore) readCommits(commitC <-chan *string, errorC <-chan error) {
 	for data := range commitC {
 		if data == nil {
 			// done replaying log; new data incoming
-			return
+			// OR signaled to load snapshot
+			snapshot, err := s.snapshotter.Load()
+			if err == snap.ErrNoSnapshot {
+				return
+			}
+			if err != nil && err != snap.ErrNoSnapshot {
+				log.Panic(err)
+			}
+			log.Printf("loading snapshot at term %d and index %d", snapshot.Metadata.Term, snapshot.Metadata.Index)
+			if err := s.recoverFromSnapshot(snapshot.Data); err != nil {
+				log.Panic(err)
+			}
 		}
 
 		var dataKv kv

contrib/raftexample/main.go

@@ -34,8 +34,12 @@ func main() {
 	defer close(confChangeC)
 
 	// raft provides a commit stream for the proposals from the http api
-	commitC, errorC := newRaftNode(*id, strings.Split(*cluster, ","), *join, proposeC, confChangeC)
+	var kvs *kvstore
+	getSnapshot := func() ([]byte, error) { return kvs.getSnapshot() }
+	commitC, errorC, snapshotterReady := newRaftNode(*id, strings.Split(*cluster, ","), *join, getSnapshot, proposeC, confChangeC)
+
+	kvs = newKVStore(<-snapshotterReady, proposeC, commitC, errorC)
 
 	// the key-value http handler will propose updates to raft
-	serveHttpKVAPI(*kvport, proposeC, confChangeC, commitC, errorC)
+	serveHttpKVAPI(kvs, *kvport, confChangeC, errorC)
 }

contrib/raftexample/raft.go

@@ -43,12 +43,13 @@ type raftNode struct {
 	commitC     chan<- *string           // entries committed to log (k,v)
 	errorC      chan<- error             // errors from raft session
 
-	id        int      // client ID for raft session
-	peers     []string // raft peer URLs
-	join      bool     // node is joining an existing cluster
-	waldir    string   // path to WAL directory
-	snapdir   string   // path to snapshot directory
-	lastIndex uint64   // index of log at start
+	id          int      // client ID for raft session
+	peers       []string // raft peer URLs
+	join        bool     // node is joining an existing cluster
+	waldir      string   // path to WAL directory
+	snapdir     string   // path to snapshot directory
+	getSnapshot func() ([]byte, error)
+	lastIndex   uint64 // index of log at start
 
 	confState     raftpb.ConfState
 	snapshotIndex uint64
@@ -58,20 +59,26 @@ type raftNode struct {
 	node        raft.Node
 	raftStorage *raft.MemoryStorage
 	wal         *wal.WAL
-	snapshotter *snap.Snapshotter
-	transport   *rafthttp.Transport
-	stopc       chan struct{} // signals proposal channel closed
-	httpstopc   chan struct{} // signals http server to shutdown
-	httpdonec   chan struct{} // signals http server shutdown complete
+
+	snapshotter      *snap.Snapshotter
+	snapshotterReady chan *snap.Snapshotter // signals when snapshotter is ready
+
+	snapCount uint64
+	transport *rafthttp.Transport
+	stopc     chan struct{} // signals proposal channel closed
+	httpstopc chan struct{} // signals http server to shutdown
+	httpdonec chan struct{} // signals http server shutdown complete
 }
 
+var defaultSnapCount uint64 = 10000
+
 // newRaftNode initiates a raft instance and returns a committed log entry
 // channel and error channel. Proposals for log updates are sent over the
 // provided the proposal channel. All log entries are replayed over the
 // commit channel, followed by a nil message (to indicate the channel is
 // current), then new log entries. To shutdown, close proposeC and read errorC.
-func newRaftNode(id int, peers []string, join bool, proposeC <-chan string,
-	confChangeC <-chan raftpb.ConfChange) (<-chan *string, <-chan error) {
+func newRaftNode(id int, peers []string, join bool, getSnapshot func() ([]byte, error), proposeC <-chan string,
+	confChangeC <-chan raftpb.ConfChange) (<-chan *string, <-chan error, <-chan *snap.Snapshotter) {
 
 	commitC := make(chan *string)
 	errorC := make(chan error)
@@ -86,14 +93,18 @@ func newRaftNode(id int, peers []string, join bool, proposeC <-chan string,
 		join:        join,
 		waldir:      fmt.Sprintf("raftexample-%d", id),
 		snapdir:     fmt.Sprintf("raftexample-%d-snap", id),
+		getSnapshot: getSnapshot,
 		raftStorage: raft.NewMemoryStorage(),
+		snapCount:   defaultSnapCount,
 		stopc:       make(chan struct{}),
 		httpstopc:   make(chan struct{}),
 		httpdonec:   make(chan struct{}),
+
+		snapshotterReady: make(chan *snap.Snapshotter, 1),
 		// rest of structure populated after WAL replay
 	}
 	go rc.startRaft()
-	return commitC, errorC
+	return commitC, errorC, rc.snapshotterReady
 }
 
 func (rc *raftNode) saveSnap(snap raftpb.Snapshot) error {
@@ -229,9 +240,11 @@ func (rc *raftNode) startRaft() {
 			log.Fatalf("raftexample: cannot create dir for snapshot (%v)", err)
 		}
 	}
+	rc.snapshotter = snap.New(rc.snapdir)
+	rc.snapshotterReady <- rc.snapshotter
+
 	oldwal := wal.Exist(rc.waldir)
 	rc.wal = rc.replayWAL()
-	rc.snapshotter = snap.New(rc.snapdir)
 
 	rpeers := make([]raft.Peer, len(rc.peers))
 	for i := range rpeers {
@@ -293,6 +306,56 @@ func (rc *raftNode) stopHTTP() {
 	<-rc.httpdonec
 }
 
+func (rc *raftNode) publishSnapshot(snapshotToSave raftpb.Snapshot) {
+	if raft.IsEmptySnap(snapshotToSave) {
+		return
+	}
+
+	log.Printf("publishing snapshot at index %d", rc.snapshotIndex)
+	defer log.Printf("finished publishing snapshot at index %d", rc.snapshotIndex)
+
+	if snapshotToSave.Metadata.Index <= rc.appliedIndex {
+		log.Fatalf("snapshot index [%d] should > progress.appliedIndex [%d] + 1", snapshotToSave.Metadata.Index, rc.appliedIndex)
+	}
+	rc.commitC <- nil // trigger kvstore to load snapshot
+
+	rc.confState = snapshotToSave.Metadata.ConfState
+	rc.snapshotIndex = snapshotToSave.Metadata.Index
+	rc.appliedIndex = snapshotToSave.Metadata.Index
+}
+
+var snapshotCatchUpEntriesN uint64 = 10000
+
+func (rc *raftNode) maybeTriggerSnapshot() {
+	if rc.appliedIndex-rc.snapshotIndex <= rc.snapCount {
+		return
+	}
+
+	log.Printf("start snapshot [applied index: %d | last snapshot index: %d]", rc.appliedIndex, rc.snapshotIndex)
+	data, err := rc.getSnapshot()
+	if err != nil {
+		log.Panic(err)
+	}
+	snap, err := rc.raftStorage.CreateSnapshot(rc.appliedIndex, &rc.confState, data)
+	if err != nil {
+		panic(err)
+	}
+	if err := rc.saveSnap(snap); err != nil {
+		panic(err)
+	}
+
+	compactIndex := uint64(1)
+	if rc.appliedIndex > snapshotCatchUpEntriesN {
+		compactIndex = rc.appliedIndex - snapshotCatchUpEntriesN
+	}
+	if err := rc.raftStorage.Compact(compactIndex); err != nil {
+		panic(err)
+	}
+
+	log.Printf("compacted log at index %d", compactIndex)
+	rc.snapshotIndex = rc.appliedIndex
+}
+
 func (rc *raftNode) serveChannels() {
 	snap, err := rc.raftStorage.Snapshot()
 	if err != nil {
@@ -347,15 +410,15 @@ func (rc *raftNode) serveChannels() {
 			if !raft.IsEmptySnap(rd.Snapshot) {
 				rc.saveSnap(rd.Snapshot)
 				rc.raftStorage.ApplySnapshot(rd.Snapshot)
+				rc.publishSnapshot(rd.Snapshot)
 			}
 			rc.raftStorage.Append(rd.Entries)
 			rc.transport.Send(rd.Messages)
-			// TODO: apply snapshot
 			if ok := rc.publishEntries(rc.entriesToApply(rd.CommittedEntries)); !ok {
 				rc.stop()
 				return
 			}
-			// TODO: trigger snapshot
+			rc.maybeTriggerSnapshot()
 			rc.node.Advance()
 
 		case err := <-rc.transport.ErrorC:

contrib/raftexample/raftexample_test.go

@@ -50,7 +50,7 @@ func newCluster(n int) *cluster {
 		os.RemoveAll(fmt.Sprintf("raftexample-%d-snap", i+1))
 		clus.proposeC[i] = make(chan string, 1)
 		clus.confChangeC[i] = make(chan raftpb.ConfChange, 1)
-		clus.commitC[i], clus.errorC[i] = newRaftNode(i+1, clus.peers, false, clus.proposeC[i], clus.confChangeC[i])
+		clus.commitC[i], clus.errorC[i], _ = newRaftNode(i+1, clus.peers, false, nil, clus.proposeC[i], clus.confChangeC[i])
 	}
 
 	return clus