open-consul/consul/server.go

535 lines
14 KiB
Go
Raw Normal View History

2013-12-06 23:43:07 +00:00
package consul
import (
"crypto/tls"
2013-12-06 23:43:07 +00:00
"fmt"
"github.com/hashicorp/raft"
2014-04-19 20:31:56 +00:00
"github.com/hashicorp/raft-mdb"
2013-12-06 23:43:07 +00:00
"github.com/hashicorp/serf/serf"
"log"
2013-12-07 00:35:13 +00:00
"net"
"net/rpc"
2013-12-06 23:43:07 +00:00
"os"
"path/filepath"
"runtime"
"strconv"
2013-12-06 23:43:07 +00:00
"sync"
"time"
)
// These are the protocol versions that Consul can _understand_. These are
// Consul-level protocol versions, that are used to configure the Serf
// protocol versions.
const (
ProtocolVersionMin uint8 = 1
ProtocolVersionMax = 1
)
2013-12-06 23:43:07 +00:00
const (
serfLANSnapshot = "serf/local.snapshot"
serfWANSnapshot = "serf/remote.snapshot"
raftState = "raft/"
snapshotsRetained = 2
raftDBSize32bit uint64 = 128 * 1024 * 1024 // Limit Raft log to 128MB
raftDBSize64bit uint64 = 8 * 1024 * 1024 * 1024 // Limit Raft log to 8GB
2013-12-06 23:43:07 +00:00
)
// Server is Consul server which manages the service discovery,
// health checking, DC forwarding, Raft, and multiple Serf pools.
type Server struct {
config *Config
2013-12-09 20:09:57 +00:00
// Connection pool to other consul servers
connPool *ConnPool
2014-01-09 23:30:36 +00:00
// Endpoints holds our RPC endpoints
endpoints endpoints
// eventChLAN is used to receive events from the
// serf cluster in the datacenter
eventChLAN chan serf.Event
2013-12-06 23:43:07 +00:00
// eventChWAN is used to receive events from the
// serf cluster that spans datacenters
eventChWAN chan serf.Event
2013-12-06 23:43:07 +00:00
// fsm is the state machine used with Raft to provide
// strong consistency.
fsm *consulFSM
// Have we attempted to leave the cluster
left bool
2013-12-06 23:43:07 +00:00
// Logger uses the provided LogOutput
logger *log.Logger
// The raft instance is used among Consul nodes within the
// DC to protect operations that require strong consistency
raft *raft.Raft
raftLayer *RaftLayer
2013-12-09 23:29:01 +00:00
raftPeers raft.PeerStore
2014-04-19 20:31:56 +00:00
raftStore *raftmdb.MDBStore
raftTransport *raft.NetworkTransport
2013-12-06 23:43:07 +00:00
// reconcileCh is used to pass events from the serf handler
// into the leader manager, so that the strong state can be
// updated
reconcileCh chan serf.Member
2013-12-12 00:24:34 +00:00
// remoteConsuls is used to track the known consuls in
// remote data centers. Used to do DC forwarding.
remoteConsuls map[string][]net.Addr
remoteLock sync.RWMutex
2013-12-07 00:35:13 +00:00
// rpcListener is used to listen for incoming connections
rpcListener net.Listener
rpcServer *rpc.Server
// rpcTLS is the TLS config for incoming TLS requests
rpcTLS *tls.Config
// serfLAN is the Serf cluster maintained inside the DC
2013-12-06 23:43:07 +00:00
// which contains all the DC nodes
serfLAN *serf.Serf
2013-12-06 23:43:07 +00:00
// serfWAN is the Serf cluster maintained between DC's
2013-12-06 23:43:07 +00:00
// which SHOULD only consist of Consul servers
serfWAN *serf.Serf
2013-12-06 23:43:07 +00:00
shutdown bool
shutdownCh chan struct{}
shutdownLock sync.Mutex
}
2014-01-09 23:30:36 +00:00
// Holds the RPC endpoints
type endpoints struct {
2014-04-28 21:44:36 +00:00
Catalog *Catalog
Health *Health
Raft *Raft
Status *Status
KVS *KVS
Internal *Internal
2014-01-09 23:30:36 +00:00
}
2013-12-06 23:43:07 +00:00
// NewServer is used to construct a new Consul server from the
// configuration, potentially returning an error
func NewServer(config *Config) (*Server, error) {
// Check the protocol version
if err := config.CheckVersion(); err != nil {
return nil, err
}
2013-12-06 23:43:07 +00:00
// Check for a data directory!
if config.DataDir == "" {
return nil, fmt.Errorf("Config must provide a DataDir")
}
// Ensure we have a log output
if config.LogOutput == nil {
config.LogOutput = os.Stderr
}
// Create the tlsConfig for outgoing connections
var tlsConfig *tls.Config
var err error
if config.VerifyOutgoing {
if tlsConfig, err = config.OutgoingTLSConfig(); err != nil {
return nil, err
}
}
// Get the incoming tls config
incomingTLS, err := config.IncomingTLSConfig()
if err != nil {
return nil, err
}
2013-12-06 23:43:07 +00:00
// Create a logger
logger := log.New(config.LogOutput, "", log.LstdFlags)
// Create server
s := &Server{
2013-12-12 00:24:34 +00:00
config: config,
connPool: NewPool(time.Minute, tlsConfig),
2013-12-12 00:24:34 +00:00
eventChLAN: make(chan serf.Event, 256),
eventChWAN: make(chan serf.Event, 256),
logger: logger,
reconcileCh: make(chan serf.Member, 32),
2013-12-12 00:24:34 +00:00
remoteConsuls: make(map[string][]net.Addr),
rpcServer: rpc.NewServer(),
rpcTLS: incomingTLS,
2013-12-12 00:24:34 +00:00
shutdownCh: make(chan struct{}),
2013-12-06 23:43:07 +00:00
}
// Initialize the RPC layer
if err := s.setupRPC(tlsConfig); err != nil {
s.Shutdown()
return nil, fmt.Errorf("Failed to start RPC layer: %v", err)
}
2013-12-09 23:29:01 +00:00
// Initialize the Raft server
if err := s.setupRaft(); err != nil {
s.Shutdown()
return nil, fmt.Errorf("Failed to start Raft: %v", err)
}
2013-12-06 23:43:07 +00:00
// Start the Serf listeners to prevent a deadlock
go s.lanEventHandler()
go s.wanEventHandler()
2013-12-06 23:43:07 +00:00
// Initialize the lan Serf
2013-12-07 01:18:09 +00:00
s.serfLAN, err = s.setupSerf(config.SerfLANConfig,
s.eventChLAN, serfLANSnapshot, false)
2013-12-06 23:43:07 +00:00
if err != nil {
s.Shutdown()
return nil, fmt.Errorf("Failed to start lan serf: %v", err)
2013-12-06 23:43:07 +00:00
}
// Initialize the wan Serf
2013-12-07 01:18:09 +00:00
s.serfWAN, err = s.setupSerf(config.SerfWANConfig,
s.eventChWAN, serfWANSnapshot, true)
2013-12-06 23:43:07 +00:00
if err != nil {
s.Shutdown()
return nil, fmt.Errorf("Failed to start wan serf: %v", err)
2013-12-06 23:43:07 +00:00
}
return s, nil
}
// setupSerf is used to setup and initialize a Serf
func (s *Server) setupSerf(conf *serf.Config, ch chan serf.Event, path string, wan bool) (*serf.Serf, error) {
addr := s.rpcListener.Addr().(*net.TCPAddr)
2014-01-30 21:13:29 +00:00
conf.Init()
if wan {
conf.NodeName = fmt.Sprintf("%s.%s", s.config.NodeName, s.config.Datacenter)
} else {
conf.NodeName = s.config.NodeName
}
2014-01-30 21:13:29 +00:00
conf.Tags["role"] = "consul"
conf.Tags["dc"] = s.config.Datacenter
conf.Tags["vsn"] = fmt.Sprintf("%d", s.config.ProtocolVersion)
conf.Tags["vsn_min"] = fmt.Sprintf("%d", ProtocolVersionMin)
conf.Tags["vsn_max"] = fmt.Sprintf("%d", ProtocolVersionMax)
2014-01-30 21:13:29 +00:00
conf.Tags["port"] = fmt.Sprintf("%d", addr.Port)
2014-01-20 23:39:07 +00:00
if s.config.Bootstrap {
2014-01-30 21:13:29 +00:00
conf.Tags["bootstrap"] = "1"
2014-01-20 23:39:07 +00:00
}
2013-12-06 23:43:07 +00:00
conf.MemberlistConfig.LogOutput = s.config.LogOutput
conf.LogOutput = s.config.LogOutput
conf.EventCh = ch
conf.SnapshotPath = filepath.Join(s.config.DataDir, path)
conf.ProtocolVersion = protocolVersionMap[s.config.ProtocolVersion]
// Until Consul supports this fully, we disable automatic resolution.
// When enabled, the Serf gossip may just turn off if we are the minority
// node which is rather unexpected.
conf.EnableNameConflictResolution = false
2013-12-19 22:18:55 +00:00
if err := ensurePath(conf.SnapshotPath, false); err != nil {
2013-12-06 23:43:07 +00:00
return nil, err
}
return serf.Create(conf)
}
// setupRaft is used to setup and initialize Raft
func (s *Server) setupRaft() error {
// If we are in bootstrap mode, enable a single node cluster
if s.config.Bootstrap {
s.config.RaftConfig.EnableSingleNode = true
}
2013-12-06 23:43:07 +00:00
// Create the base path
path := filepath.Join(s.config.DataDir, raftState)
2013-12-19 22:18:55 +00:00
if err := ensurePath(path, true); err != nil {
2013-12-06 23:43:07 +00:00
return err
}
2013-12-11 01:00:48 +00:00
// Create the FSM
var err error
2014-02-03 23:21:56 +00:00
s.fsm, err = NewFSM(s.config.LogOutput)
2013-12-11 01:00:48 +00:00
if err != nil {
return err
}
// Set the maximum raft size based on 32/64bit. Since we are
// doing an mmap underneath, we need to limit our use of virtual
// address space on 32bit, but don't have to care on 64bit.
dbSize := raftDBSize32bit
if runtime.GOARCH == "amd64" {
dbSize = raftDBSize64bit
}
2013-12-19 00:23:17 +00:00
// Create the MDB store for logs and stable storage
store, err := raftmdb.NewMDBStoreWithSize(path, dbSize)
2013-12-06 23:43:07 +00:00
if err != nil {
return err
}
s.raftStore = store
2013-12-06 23:43:07 +00:00
// Create the snapshot store
snapshots, err := raft.NewFileSnapshotStore(path, snapshotsRetained, s.config.LogOutput)
2013-12-06 23:43:07 +00:00
if err != nil {
store.Close()
return err
}
// Create a transport layer
2014-02-22 19:13:59 +00:00
trans := raft.NewNetworkTransport(s.raftLayer, 3, 10*time.Second, s.config.LogOutput)
s.raftTransport = trans
2013-12-06 23:43:07 +00:00
// Setup the peer store
2013-12-09 23:29:01 +00:00
s.raftPeers = raft.NewJSONPeers(path, trans)
2013-12-06 23:43:07 +00:00
// Ensure local host is always included if we are in bootstrap mode
if s.config.Bootstrap {
peers, err := s.raftPeers.Peers()
if err != nil {
store.Close()
return err
}
if !raft.PeerContained(peers, trans.LocalAddr()) {
s.raftPeers.SetPeers(raft.AddUniquePeer(peers, trans.LocalAddr()))
}
}
2013-12-23 23:30:45 +00:00
// Make sure we set the LogOutput
s.config.RaftConfig.LogOutput = s.config.LogOutput
2013-12-06 23:43:07 +00:00
// Setup the Raft store
2013-12-09 23:29:01 +00:00
s.raft, err = raft.NewRaft(s.config.RaftConfig, s.fsm, store, store,
snapshots, s.raftPeers, trans)
2013-12-06 23:43:07 +00:00
if err != nil {
store.Close()
trans.Close()
return err
}
2014-01-09 23:49:09 +00:00
// Start monitoring leadership
go s.monitorLeadership()
2013-12-06 23:43:07 +00:00
return nil
}
2013-12-07 00:35:13 +00:00
// setupRPC is used to setup the RPC listener
func (s *Server) setupRPC(tlsConfig *tls.Config) error {
2014-01-09 23:30:36 +00:00
// Create endpoints
s.endpoints.Status = &Status{s}
s.endpoints.Raft = &Raft{s}
s.endpoints.Catalog = &Catalog{s}
s.endpoints.Health = &Health{s}
2014-03-31 21:15:49 +00:00
s.endpoints.KVS = &KVS{s}
2014-04-28 21:44:36 +00:00
s.endpoints.Internal = &Internal{s}
2014-01-09 23:30:36 +00:00
2013-12-09 22:49:07 +00:00
// Register the handlers
2014-01-09 23:30:36 +00:00
s.rpcServer.Register(s.endpoints.Status)
s.rpcServer.Register(s.endpoints.Raft)
s.rpcServer.Register(s.endpoints.Catalog)
s.rpcServer.Register(s.endpoints.Health)
2014-03-31 21:15:49 +00:00
s.rpcServer.Register(s.endpoints.KVS)
2014-04-28 21:44:36 +00:00
s.rpcServer.Register(s.endpoints.Internal)
2013-12-09 22:49:07 +00:00
2014-01-01 00:45:13 +00:00
list, err := net.ListenTCP("tcp", s.config.RPCAddr)
2013-12-07 00:35:13 +00:00
if err != nil {
return err
}
s.rpcListener = list
2013-12-31 22:00:25 +00:00
var advertise net.Addr
if s.config.RPCAdvertise != nil {
advertise = s.config.RPCAdvertise
} else {
advertise = s.rpcListener.Addr()
}
// Verify that we have a usable advertise address
addr, ok := advertise.(*net.TCPAddr)
if !ok {
list.Close()
return fmt.Errorf("RPC advertise address is not a TCP Address: %v", addr)
}
if addr.IP.IsUnspecified() {
list.Close()
return fmt.Errorf("RPC advertise address is not advertisable: %v", addr)
}
s.raftLayer = NewRaftLayer(advertise, tlsConfig)
2013-12-07 00:35:13 +00:00
go s.listen()
return nil
}
2013-12-06 23:43:07 +00:00
// Shutdown is used to shutdown the server
func (s *Server) Shutdown() error {
2014-01-10 19:06:11 +00:00
s.logger.Printf("[INFO] consul: shutting down server")
2013-12-06 23:43:07 +00:00
s.shutdownLock.Lock()
defer s.shutdownLock.Unlock()
if s.shutdown {
return nil
}
s.shutdown = true
close(s.shutdownCh)
if s.serfLAN != nil {
s.serfLAN.Shutdown()
2013-12-06 23:43:07 +00:00
}
if s.serfWAN != nil {
s.serfWAN.Shutdown()
2013-12-06 23:43:07 +00:00
}
if s.raft != nil {
s.raftTransport.Close()
s.raftLayer.Close()
future := s.raft.Shutdown()
if err := future.Error(); err != nil {
2013-12-24 00:20:51 +00:00
s.logger.Printf("[WARN] consul: Error shutting down raft: %s", err)
}
2013-12-06 23:43:07 +00:00
s.raftStore.Close()
// Clear the peer set on a graceful leave to avoid
// triggering elections on a rejoin.
if s.left {
s.raftPeers.SetPeers(nil)
}
2013-12-06 23:43:07 +00:00
}
2013-12-07 00:35:13 +00:00
if s.rpcListener != nil {
s.rpcListener.Close()
}
2013-12-09 20:09:57 +00:00
// Close the connection pool
s.connPool.Shutdown()
// Close the fsm
if s.fsm != nil {
s.fsm.Close()
}
2013-12-06 23:43:07 +00:00
return nil
}
2013-12-07 01:18:09 +00:00
2013-12-09 20:10:27 +00:00
// Leave is used to prepare for a graceful shutdown of the server
func (s *Server) Leave() error {
2014-01-10 19:06:11 +00:00
s.logger.Printf("[INFO] consul: server starting leave")
s.left = true
2013-12-09 20:10:27 +00:00
// Leave the WAN pool
if s.serfWAN != nil {
if err := s.serfWAN.Leave(); err != nil {
2014-01-10 19:06:11 +00:00
s.logger.Printf("[ERR] consul: failed to leave WAN Serf cluster: %v", err)
2013-12-09 20:10:27 +00:00
}
}
// Leave the LAN pool
if s.serfLAN != nil {
if err := s.serfLAN.Leave(); err != nil {
2014-01-10 19:06:11 +00:00
s.logger.Printf("[ERR] consul: failed to leave LAN Serf cluster: %v", err)
2013-12-09 20:10:27 +00:00
}
}
// Leave the Raft cluster
if s.raft != nil {
// Check if we have other raft nodes
peers, _ := s.raftPeers.Peers()
if len(peers) <= 1 {
2014-01-10 19:06:11 +00:00
s.logger.Printf("[WARN] consul: not leaving Raft cluster, no peers")
goto AFTER_LEAVE
}
2013-12-09 20:10:27 +00:00
// Get the leader
leader := s.raft.Leader()
if leader == nil {
2014-01-10 19:06:11 +00:00
s.logger.Printf("[ERR] consul: failed to leave Raft cluster: no leader")
2013-12-09 20:10:27 +00:00
goto AFTER_LEAVE
}
// Request that we are removed
ch := make(chan error, 1)
go func() {
var out struct{}
peer := s.raftTransport.LocalAddr().String()
err := s.connPool.RPC(leader, "Raft.RemovePeer", peer, &out)
ch <- err
2013-12-09 20:10:27 +00:00
}()
// Wait for the commit
select {
case err := <-ch:
// Ignore if we have already been deregistered by the leader
if err != nil && err.Error() != raft.ErrUnknownPeer.Error() {
2014-01-10 19:06:11 +00:00
s.logger.Printf("[ERR] consul: failed to leave Raft cluster: %v", err)
2013-12-09 20:10:27 +00:00
}
case <-time.After(3 * time.Second):
2014-01-10 19:06:11 +00:00
s.logger.Printf("[ERR] consul: timed out leaving Raft cluster")
2013-12-09 20:10:27 +00:00
}
}
AFTER_LEAVE:
return nil
}
2013-12-07 01:18:09 +00:00
// JoinLAN is used to have Consul join the inner-DC pool
// The target address should be another node inside the DC
// listening on the Serf LAN address
func (s *Server) JoinLAN(addrs []string) (int, error) {
return s.serfLAN.Join(addrs, true)
2013-12-07 01:18:09 +00:00
}
// JoinWAN is used to have Consul join the cross-WAN Consul ring
// The target address should be another node listening on the
// Serf WAN address
func (s *Server) JoinWAN(addrs []string) (int, error) {
return s.serfWAN.Join(addrs, true)
2013-12-07 01:18:09 +00:00
}
// LANMembers is used to return the members of the LAN cluster
func (s *Server) LANMembers() []serf.Member {
return s.serfLAN.Members()
}
// WANMembers is used to return the members of the LAN cluster
func (s *Server) WANMembers() []serf.Member {
return s.serfWAN.Members()
}
// RemoveFailedNode is used to remove a failed node from the cluster
func (s *Server) RemoveFailedNode(node string) error {
if err := s.serfLAN.RemoveFailedNode(node); err != nil {
return err
}
if err := s.serfWAN.RemoveFailedNode(node); err != nil {
return err
}
return nil
}
// IsLeader checks if this server is the cluster leader
func (s *Server) IsLeader() bool {
return s.raft.State() == raft.Leader
}
// RPC is used to make a local RPC call
func (s *Server) RPC(method string, args interface{}, reply interface{}) error {
addr := s.rpcListener.Addr()
return s.connPool.RPC(addr, method, args, reply)
}
// Stats is used to return statistics for debugging and insight
// for various sub-systems
func (s *Server) Stats() map[string]map[string]string {
toString := func(v uint64) string {
return strconv.FormatUint(v, 10)
}
stats := map[string]map[string]string{
"consul": map[string]string{
2014-02-24 02:08:58 +00:00
"server": "true",
"leader": fmt.Sprintf("%v", s.IsLeader()),
"bootstrap": fmt.Sprintf("%v", s.config.Bootstrap),
"known_datacenters": toString(uint64(len(s.remoteConsuls))),
},
2014-02-24 02:08:58 +00:00
"raft": s.raft.Stats(),
"serf_lan": s.serfLAN.Stats(),
"serf_wan": s.serfWAN.Stats(),
"runtime": runtimeStats(),
}
return stats
}