Make raft snapshot commit threshold configurable

agent/agent.go

@@ -763,6 +763,9 @@ func (a *Agent) consulConfig() (*consul.Config, error) {
 	if a.config.RaftProtocol != 0 {
 		base.RaftConfig.ProtocolVersion = raft.ProtocolVersion(a.config.RaftProtocol)
 	}
+	if a.config.RaftSnapshotThreshold != 0 {
+		base.RaftConfig.SnapshotThreshold = uint64(a.config.RaftSnapshotThreshold)
+	}
 	if a.config.ACLMasterToken != "" {
 		base.ACLMasterToken = a.config.ACLMasterToken
 	}

agent/config/builder.go

@@ -673,6 +673,7 @@ func (b *Builder) Build() (rt RuntimeConfig, err error) {
 		RPCProtocol:                 b.intVal(c.RPCProtocol),
 		RPCRateLimit:                rate.Limit(b.float64Val(c.Limits.RPCRate)),
 		RaftProtocol:                b.intVal(c.RaftProtocol),
+		RaftSnapshotThreshold:       b.intVal(c.RaftSnapshotThreshold),
 		ReconnectTimeoutLAN:         b.durationVal("reconnect_timeout", c.ReconnectTimeoutLAN),
 		ReconnectTimeoutWAN:         b.durationVal("reconnect_timeout_wan", c.ReconnectTimeoutWAN),
 		RejoinAfterLeave:            b.boolVal(c.RejoinAfterLeave),

agent/config/config.go

@@ -194,6 +194,7 @@ type Config struct {
 	Ports                       Ports                    `json:"ports,omitempty" hcl:"ports" mapstructure:"ports"`
 	RPCProtocol                 *int                     `json:"protocol,omitempty" hcl:"protocol" mapstructure:"protocol"`
 	RaftProtocol                *int                     `json:"raft_protocol,omitempty" hcl:"raft_protocol" mapstructure:"raft_protocol"`
+	RaftSnapshotThreshold       *int                     `json:"raft_snapshot_threshold,omitempty" hcl:"raft_snapshot_threshold" mapstructure:"raft_snapshot_threshold"`
 	ReconnectTimeoutLAN         *string                  `json:"reconnect_timeout,omitempty" hcl:"reconnect_timeout" mapstructure:"reconnect_timeout"`
 	ReconnectTimeoutWAN         *string                  `json:"reconnect_timeout_wan,omitempty" hcl:"reconnect_timeout_wan" mapstructure:"reconnect_timeout_wan"`
 	RejoinAfterLeave            *bool                    `json:"rejoin_after_leave,omitempty" hcl:"rejoin_after_leave" mapstructure:"rejoin_after_leave"`
@@ -264,6 +265,7 @@ type Consul struct {
 		ElectionTimeout    *string `json:"election_timeout,omitempty" hcl:"election_timeout" mapstructure:"election_timeout"`
 		HeartbeatTimeout   *string `json:"heartbeat_timeout,omitempty" hcl:"heartbeat_timeout" mapstructure:"heartbeat_timeout"`
 		LeaderLeaseTimeout *string `json:"leader_lease_timeout,omitempty" hcl:"leader_lease_timeout" mapstructure:"leader_lease_timeout"`
+		SnapshotThreshold  *int    `json:"snapshot_threshold,omitempty" hcl:"snapshot_threshold" mapstructure:"snapshot_threshold"`
 	} `json:"raft,omitempty" hcl:"raft" mapstructure:"raft"`
 
 	SerfLAN struct {

agent/config/runtime.go

@@ -899,6 +899,12 @@ type RuntimeConfig struct {
 	// hcl: raft_protocol = int
 	RaftProtocol int
 
+	// RaftSnapshotThreshold sets the minimum threshold of raft commits after which
+	// a snapshot is created. Defaults to 8192
+	//
+	// hcl: raft_snapshot_threshold = int
+	RaftSnapshotThreshold int
+
 	// ReconnectTimeoutLAN specifies the amount of time to wait to reconnect with
 	// another agent before deciding it's permanently gone. This can be used to
 	// control the time it takes to reap failed nodes from the cluster.

agent/config/runtime_test.go

@@ -2421,6 +2421,7 @@ func TestFullConfig(t *testing.T) {
 			},
 			"protocol": 30793,
 			"raft_protocol": 19016,
+            "raft_snapshot_threshold": 16384,
 			"reconnect_timeout": "23739s",
 			"reconnect_timeout_wan": "26694s",
 			"recursors": [ "63.38.39.58", "92.49.18.18" ],
@@ -2852,6 +2853,7 @@ func TestFullConfig(t *testing.T) {
 			}
 			protocol = 30793
 			raft_protocol = 19016
+            raft_snapshot_threshold = 16384
 			reconnect_timeout = "23739s"
 			reconnect_timeout_wan = "26694s"
 			recursors = [ "63.38.39.58", "92.49.18.18" ]
@@ -3409,6 +3411,7 @@ func TestFullConfig(t *testing.T) {
 		RPCRateLimit:              12029.43,
 		RPCMaxBurst:               44848,
 		RaftProtocol:              19016,
+		RaftSnapshotThreshold:     16384,
 		ReconnectTimeoutLAN:       23739 * time.Second,
 		ReconnectTimeoutWAN:       26694 * time.Second,
 		RejoinAfterLeave:          true,
@@ -4089,6 +4092,7 @@ func TestSanitize(t *testing.T) {
     "RPCProtocol": 0,
     "RPCRateLimit": 0,
     "RaftProtocol": 0,
+    "RaftSnapshotThreshold": 0,
     "ReconnectTimeoutLAN": "0s",
     "ReconnectTimeoutWAN": "0s",
     "RejoinAfterLeave": false,

agent/consul/config.go

@@ -451,6 +451,9 @@ func DefaultConfig() *Config {
 	// Check every 5 seconds to see if there are enough new entries for a snapshot
 	conf.RaftConfig.SnapshotInterval = 5 * time.Second
 
+	// Snapshots are created every 8192 entries by default, can be overridden
+	conf.RaftConfig.SnapshotThreshold = 8192
+
 	return conf
 }
 

website/source/docs/agent/options.html.md

@@ -359,6 +359,9 @@ will exit with an error at startup.
   [Raft Protocol Version Compatibility](/docs/upgrade-specific.html#raft-protocol-version-compatibility)
   for more details.
 
+* <a name="_raft_snapshot_threshold"></a><a href="#_raft_snapshot_threshold">`-raft-snapshot-threshold`</a> - This
+  control the minimum number of raft commit entries between snapshots that are saved to disk.
+
 * <a name="_recursor"></a><a href="#_recursor">`-recursor`</a> - Specifies the address of an upstream DNS
   server. This option may be provided multiple times, and is functionally
   equivalent to the [`recursors` configuration option](#recursors).
@@ -935,6 +938,9 @@ Consul will not enable TLS for the HTTP API unless the `https` port has been ass
 * <a name="raft_protocol"></a><a href="#raft_protocol">`raft_protocol`</a> Equivalent to the
   [`-raft-protocol` command-line flag](#_raft_protocol).
 
+* <a name="raft_snapshot_threshold"></a><a href="#raft_snapshot_threshold">`raft_snapshot_threshold`</a> Equivalent to the
+    [`-raft-snapshot-threshold` command-line flag](#_raft_snapshot_threshold).
+
 * <a name="reap"></a><a href="#reap">`reap`</a> This controls Consul's automatic reaping of child processes,
   which is useful if Consul is running as PID 1 in a Docker container. If this isn't specified, then Consul will
   automatically reap child processes if it detects it is running as PID 1. If this is set to true or false, then