raft.go

// Copyright 2015 CoreOS, Inc.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
//     http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.

package trr

import (
	"fmt"
	"log"
	"net/http"
	"net/url"
	"os"
	"strconv"
	"time"

	"github.com/coreos/etcd/Godeps/_workspace/src/golang.org/x/net/context"
	"github.com/coreos/etcd/etcdserver/stats"
	"github.com/coreos/etcd/pkg/types"
	"github.com/coreos/etcd/raft"
	"github.com/coreos/etcd/raft/raftpb"
	"github.com/coreos/etcd/rafthttp"
	"github.com/coreos/etcd/wal"
	"github.com/coreos/etcd/wal/walpb"
)

// A key-value stream backed by raft
type raftNode struct {
	proposeC    <-chan string            // proposed messages (k,v)
	confChangeC <-chan raftpb.ConfChange // proposed cluster config changes
	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
	lastIndex uint64   // index of log at start

	// raft backing for the commit/error channel
	node        raft.Node
	raftStorage *raft.MemoryStorage
	wal         *wal.WAL
	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
}

// 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, chan struct{}) {

	rc := &raftNode{
		proposeC:    proposeC,
		confChangeC: confChangeC,
		commitC:     make(chan *string),
		errorC:      make(chan error),
		id:          id,
		peers:       peers,
		join:        join,
		waldir:      fmt.Sprintf("trr-%d", id),
		raftStorage: raft.NewMemoryStorage(),
		stopc:       make(chan struct{}),
		httpstopc:   make(chan struct{}),
		httpdonec:   make(chan struct{}),
		// rest of structure populated after WAL replay
	}
	go rc.startRaft()
	return rc.commitC, rc.errorC, rc.httpstopc
}

// publishEntries writes committed log entries to commit channel and returns
// whether all entries could be published.
func (rc *raftNode) publishEntries(ents []raftpb.Entry) bool {
	for i := range ents {
		switch ents[i].Type {
		case raftpb.EntryNormal:
			if len(ents[i].Data) == 0 {
				// ignore empty messages
				break
			}
			s := string(ents[i].Data)
			select {
			case rc.commitC <- &s:
			case <-rc.stopc:
				return false
			}

		case raftpb.EntryConfChange:
			var cc raftpb.ConfChange
			cc.Unmarshal(ents[i].Data)
			rc.node.ApplyConfChange(cc)
			switch cc.Type {
			case raftpb.ConfChangeAddNode:
				if len(cc.Context) > 0 {
					rc.transport.AddPeer(types.ID(cc.NodeID), []string{string(cc.Context)})
				}
			case raftpb.ConfChangeRemoveNode:
				if cc.NodeID == uint64(rc.id) {
					log.Println("I've been removed from the cluster! Shutting down.")
					return false
				}
				rc.transport.RemovePeer(types.ID(cc.NodeID))
			}
		}

		// special nil commit to signal replay has finished
		if ents[i].Index == rc.lastIndex {
			select {
			case rc.commitC <- nil:
			case <-rc.stopc:
				return false
			}
		}
	}
	return true
}

// openWAL returns a WAL ready for reading.
func (rc *raftNode) openWAL() *wal.WAL {
	if wal.Exist(rc.waldir) == false {
		if err := os.Mkdir(rc.waldir, 0750); err != nil {
			log.Fatalf("trr: cannot create dir for wal (%v)", err)
		}

		w, err := wal.Create(rc.waldir, nil)
		if err != nil {
			log.Fatalf("trr: create wal error (%v)", err)
		}
		w.Close()
	}

	w, err := wal.Open(rc.waldir, walpb.Snapshot{})
	if err != nil {
		log.Fatalf("trr: error loading wal (%v)", err)
	}

	return w
}

// replayWAL replays WAL entries into the raft instance.
func (rc *raftNode) replayWAL() *wal.WAL {
	w := rc.openWAL()
	_, st, ents, err := w.ReadAll()
	if err != nil {
		log.Fatalf("trr: failed to read WAL (%v)", err)
	}
	// append to storage so raft starts at the right place in log
	rc.raftStorage.Append(ents)
	// send nil once lastIndex is published so client knows commit channel is current
	if len(ents) > 0 {
		rc.lastIndex = ents[len(ents)-1].Index
	} else {
		rc.commitC <- nil
	}
	rc.raftStorage.SetHardState(st)
	return w
}

func (rc *raftNode) writeError(err error) {
	rc.stopHTTP()
	close(rc.commitC)
	rc.errorC <- err
	close(rc.errorC)
	rc.node.Stop()
}

func (rc *raftNode) startRaft() {
	//log.Println("Raft start....")
	oldwal := wal.Exist(rc.waldir)
	rc.wal = rc.replayWAL()

	rpeers := make([]raft.Peer, len(rc.peers))
	for i := range rpeers {
		rpeers[i] = raft.Peer{ID: uint64(i + 1)}
	}
	c := &raft.Config{
		ID:              uint64(rc.id),
		ElectionTick:    10,
		HeartbeatTick:   1,
		Storage:         rc.raftStorage,
		MaxSizePerMsg:   1024 * 1024,
		MaxInflightMsgs: 256,
	}

	if oldwal {
		rc.node = raft.RestartNode(c)
	} else {
		startPeers := rpeers
		if rc.join {
			startPeers = nil
		}
		rc.node = raft.StartNode(c, startPeers)
	}

	ss := &stats.ServerStats{}
	ss.Initialize()

	rc.transport = &rafthttp.Transport{
		ID:          types.ID(rc.id),
		ClusterID:   0x1000,
		Raft:        rc,
		ServerStats: ss,
		LeaderStats: stats.NewLeaderStats(strconv.Itoa(rc.id)),
		ErrorC:      make(chan error),
	}

	rc.transport.Start()
	log.Println(" peer size:", len(rc.peers))
	for i := range rc.peers {
		if i+1 != rc.id {
			log.Println("Transport: ID:", i, " addr:", rc.peers[i])
			rc.transport.AddPeer(types.ID(i+1), []string{rc.peers[i]})
		}
	}

	log.Println("Prepare ----->")
	go rc.serveRaft()
	go rc.serveChannels()
}

// stop closes http, closes all channels, and stops raft.
func (rc *raftNode) stop() {
	rc.stopHTTP()
	close(rc.commitC)
	close(rc.errorC)
	rc.node.Stop()
}

func (rc *raftNode) stopHTTP() {
	log.Println("raft stop............")
	rc.transport.Stop()
	close(rc.httpstopc)
	<-rc.httpdonec
}

func (rc *raftNode) serveChannels() {
	defer rc.wal.Close()

	ticker := time.NewTicker(100 * time.Millisecond)
	defer ticker.Stop()

	// send proposals over raft
	go func() {
		var confChangeCount uint64 = 0

		for rc.proposeC != nil && rc.confChangeC != nil {
			select {
			case prop, ok := <-rc.proposeC:
				//log.Println("Go propose:", prop)
				if !ok {
					rc.proposeC = nil
				} else {
					// blocks until accepted by raft state machine
					rc.node.Propose(context.TODO(), []byte(prop))
				}

			case cc, ok := <-rc.confChangeC:
				if !ok {
					rc.confChangeC = nil
				} else {
					confChangeCount += 1
					cc.ID = confChangeCount
					rc.node.ProposeConfChange(context.TODO(), cc)
				}
			}
		}
		// client closed channel; shutdown raft if not already
		close(rc.stopc)
	}()

	// event loop on raft state machine updates
	for {
		select {
		case <-ticker.C:
			rc.node.Tick()

		// store raft entries to wal, then publish over commit channel
		case rd := <-rc.node.Ready():
			rc.wal.Save(rd.HardState, rd.Entries)
			rc.raftStorage.Append(rd.Entries)
			//log.Println("raft send:=>", rd.Messages)
			rc.transport.Send(rd.Messages)
			if ok := rc.publishEntries(rd.CommittedEntries); !ok {
				rc.stop()
				return
			}
			rc.node.Advance()

		case err := <-rc.transport.ErrorC:
			rc.writeError(err)
			return

		case <-rc.stopc:
			rc.stop()
			return
		}
	}
}

func (rc *raftNode) serveRaft() {
	url, err := url.Parse(rc.peers[rc.id-1])
	//log.Println("URL-->", url, err)
	if err != nil {
		log.Fatalf("trr: Failed parsing URL (%v)", err)
	}

	ln, err := newStoppableListener(url.Host, rc.httpstopc)
	if err != nil {
		log.Fatalf("trr: Failed to listen rafthttp (%v)", err)
	}

	err = (&http.Server{Handler: rc.transport.Handler()}).Serve(ln)
	select {
	case <-rc.httpstopc:
	default:
		log.Fatalf("trr: Failed to serve rafthttp (%v)", err)
	}
	close(rc.httpdonec)
}

func (rc *raftNode) Process(ctx context.Context, m raftpb.Message) error {
	//log.Println("[Process]------->")
	return rc.node.Step(ctx, m)
}
func (rc *raftNode) IsIDRemoved(id uint64) bool                           { return false }
func (rc *raftNode) ReportUnreachable(id uint64)                          {}
func (rc *raftNode) ReportSnapshot(id uint64, status raft.SnapshotStatus) {}