Expose swarm raft tuning parameters in engine config

Signed-off-by: David Chung <david.chung@docker.com>

cmd/dockerd/daemon.go

@@ -257,6 +257,8 @@ func (cli *DaemonCli) start(opts *daemonOptions) (err error) {
 		PluginBackend:          d.PluginManager(),
 		NetworkSubnetsProvider: d,
 		DefaultAdvertiseAddr:   cli.Config.SwarmDefaultAdvertiseAddr,
+		RaftHeartbeatTick:      cli.Config.SwarmRaftHeartbeatTick,
+		RaftElectionTick:       cli.Config.SwarmRaftElectionTick,
 		RuntimeRoot:            cli.getSwarmRunRoot(),
 		WatchStream:            watchStream,
 	})

daemon/cluster/cluster.go

@@ -96,6 +96,13 @@ type Config struct {
 
 	// WatchStream is a channel to pass watch API notifications to daemon
 	WatchStream chan *swarmapi.WatchMessage
+
+	// RaftHeartbeatTick is the number of ticks for heartbeat of quorum members
+	RaftHeartbeatTick uint32
+
+	// RaftElectionTick is the number of ticks to elapse before followers propose a new round of leader election
+	// This value should be 10x that of RaftHeartbeatTick
+	RaftElectionTick uint32
 }
 
 // Cluster provides capabilities to participate in a cluster as a worker or a
@@ -134,6 +141,14 @@ func New(config Config) (*Cluster, error) {
 	if config.RuntimeRoot == "" {
 		config.RuntimeRoot = root
 	}
+	if config.RaftHeartbeatTick == 0 {
+		config.RaftHeartbeatTick = 1
+	}
+	if config.RaftElectionTick == 0 {
+		// 10X heartbeat tick is the recommended ratio according to etcd docs.
+		config.RaftElectionTick = 10 * config.RaftHeartbeatTick
+	}
+
 	if err := os.MkdirAll(config.RuntimeRoot, 0700); err != nil {
 		return nil, err
 	}

daemon/cluster/noderunner.go

@@ -124,11 +124,11 @@ func (n *nodeRunner) start(conf nodeStartConfig) error {
 			n.cluster.config.Backend,
 			n.cluster.config.PluginBackend,
 			n.cluster.config.ImageBackend),
-		HeartbeatTick: 1,
+		HeartbeatTick: n.cluster.config.RaftHeartbeatTick,
 		// Recommended value in etcd/raft is 10 x (HeartbeatTick).
 		// Lower values were seen to have caused instability because of
 		// frequent leader elections when running on flakey networks.
-		ElectionTick:     10,
+		ElectionTick:     n.cluster.config.RaftElectionTick,
 		UnlockKey:        conf.lockKey,
 		AutoLockManagers: conf.autolock,
 		PluginGetter:     n.cluster.config.Backend.PluginGetter(),

daemon/config/config.go

@@ -158,7 +158,18 @@ type CommonConfig struct {
 	// given to the /swarm/init endpoint and no advertise address is
 	// specified.
 	SwarmDefaultAdvertiseAddr string `json:"swarm-default-advertise-addr"`
-	MetricsAddress            string `json:"metrics-addr"`
+
+	// SwarmRaftHeartbeatTick is the number of ticks in time for swarm mode raft quorum heartbeat
+	// Typical value is 1
+	SwarmRaftHeartbeatTick uint32 `json:"swarm-raft-heartbeat-tick"`
+
+	// SwarmRaftElectionTick is the number of ticks to elapse before followers in the quorum can propose
+	// a new round of leader election.  Default, recommended value is at least 10X that of Heartbeat tick.
+	// Higher values can make the quorum less sensitive to transient faults in the environment, but this also
+	// means it takes longer for the managers to detect a down leader.
+	SwarmRaftElectionTick uint32 `json:"swarm-raft-election-tick"`
+
+	MetricsAddress string `json:"metrics-addr"`
 
 	LogConfig
 	BridgeConfig // bridgeConfig holds bridge network specific configuration.