Merge pull request #204 from github/reduce-minimum-max-lag

Reduce minimum maxLagMillisecondsThrottleThreshold to 100ms

Author: Shlomi Noach

doc/interactive-commands.md

@@ -18,7 +18,7 @@ Both interfaces may serve at the same time. Both respond to simple text command,
 - `status`: returns a detailed status summary of migration progress and configuration
 - `sup`: returns a brief status summary of migration progress
 - `chunk-size=<newsize>`: modify the `chunk-size`; applies on next running copy-iteration
-- `max-lag-millis=<max-lag>`: modify the maximum replication lag threshold (milliseconds, minimum value is `1000`, i.e. 1 second)
+- `max-lag-millis=<max-lag>`: modify the maximum replication lag threshold (milliseconds, minimum value is `100`, i.e. `0.1` second)
 - `max-load=<max-load-thresholds>`: modify the `max-load` config; applies on next running copy-iteration
   The `max-load` format must be: `some_status=<numeric-threshold>[,some_status=<numeric-threshold>...]`. For example: `Threads_running=50,threads_connected=1000`, and you would then write/echo `max-load=Threads_running=50,threads_connected=1000` to the socket.
 - `critical-load=<load>`: change critical load setting (exceeding given thresholds causes panic and abort)

go/base/context.go

@@ -41,6 +41,18 @@ var (
 	envVariableRegexp = regexp.MustCompile("[$][{](.*)[}]")
 )
 
+type ThrottleCheckResult struct {
+	ShouldThrottle bool
+	Reason         string
+}
+
+func NewThrottleCheckResult(throttle bool, reason string) *ThrottleCheckResult {
+	return &ThrottleCheckResult{
+		ShouldThrottle: throttle,
+		Reason:         reason,
+	}
+}
+
 // MigrationContext has the general, global state of migration. It is used by
 // all components throughout the migration process.
 type MigrationContext struct {
@@ -64,6 +76,7 @@ type MigrationContext struct {
 	CliUser     string
 	CliPassword string
 
+	HeartbeatIntervalMilliseconds       int64
 	defaultNumRetries                   int64
 	ChunkSize                           int64
 	niceRatio                           float64
@@ -95,32 +108,36 @@ type MigrationContext struct {
 	InitiallyDropGhostTable      bool
 	CutOverType                  CutOver
 
-	Hostname                  string
-	TableEngine               string
-	RowsEstimate              int64
-	RowsDeltaEstimate         int64
-	UsedRowsEstimateMethod    RowsEstimateMethod
-	HasSuperPrivilege         bool
-	OriginalBinlogFormat      string
-	OriginalBinlogRowImage    string
-	InspectorConnectionConfig *mysql.ConnectionConfig
-	ApplierConnectionConfig   *mysql.ConnectionConfig
-	StartTime                 time.Time
-	RowCopyStartTime          time.Time
-	RowCopyEndTime            time.Time
-	LockTablesStartTime       time.Time
-	RenameTablesStartTime     time.Time
-	RenameTablesEndTime       time.Time
-	pointOfInterestTime       time.Time
-	pointOfInterestTimeMutex  *sync.Mutex
-	CurrentLag                int64
-	TotalRowsCopied           int64
-	TotalDMLEventsApplied     int64
-	isThrottled               bool
-	throttleReason            string
-	throttleMutex             *sync.Mutex
-	IsPostponingCutOver       int64
-	CountingRowsFlag          int64
+	Hostname                               string
+	TableEngine                            string
+	RowsEstimate                           int64
+	RowsDeltaEstimate                      int64
+	UsedRowsEstimateMethod                 RowsEstimateMethod
+	HasSuperPrivilege                      bool
+	OriginalBinlogFormat                   string
+	OriginalBinlogRowImage                 string
+	InspectorConnectionConfig              *mysql.ConnectionConfig
+	ApplierConnectionConfig                *mysql.ConnectionConfig
+	StartTime                              time.Time
+	RowCopyStartTime                       time.Time
+	RowCopyEndTime                         time.Time
+	LockTablesStartTime                    time.Time
+	RenameTablesStartTime                  time.Time
+	RenameTablesEndTime                    time.Time
+	pointOfInterestTime                    time.Time
+	pointOfInterestTimeMutex               *sync.Mutex
+	CurrentLag                             int64
+	controlReplicasLagResult               mysql.ReplicationLagResult
+	TotalRowsCopied                        int64
+	TotalDMLEventsApplied                  int64
+	isThrottled                            bool
+	throttleReason                         string
+	throttleGeneralCheckResult             ThrottleCheckResult
+	throttleMutex                          *sync.Mutex
+	IsPostponingCutOver                    int64
+	CountingRowsFlag                       int64
+	AllEventsUpToLockProcessedInjectedFlag int64
+	CleanupImminentFlag                    int64
 
 	OriginalTableColumns             *sql.ColumnList
 	OriginalTableUniqueKeys          [](*sql.UniqueKey)
@@ -326,9 +343,19 @@ func (this *MigrationContext) TimeSincePointOfInterest() time.Duration {
 	return time.Since(this.pointOfInterestTime)
 }
 
+func (this *MigrationContext) SetHeartbeatIntervalMilliseconds(heartbeatIntervalMilliseconds int64) {
+	if heartbeatIntervalMilliseconds < 100 {
+		heartbeatIntervalMilliseconds = 100
+	}
+	if heartbeatIntervalMilliseconds > 1000 {
+		heartbeatIntervalMilliseconds = 1000
+	}
+	this.HeartbeatIntervalMilliseconds = heartbeatIntervalMilliseconds
+}
+
 func (this *MigrationContext) SetMaxLagMillisecondsThrottleThreshold(maxLagMillisecondsThrottleThreshold int64) {
-	if maxLagMillisecondsThrottleThreshold < 1000 {
-		maxLagMillisecondsThrottleThreshold = 1000
+	if maxLagMillisecondsThrottleThreshold < 100 {
+		maxLagMillisecondsThrottleThreshold = 100
 	}
 	atomic.StoreInt64(&this.MaxLagMillisecondsThrottleThreshold, maxLagMillisecondsThrottleThreshold)
 }
@@ -343,6 +370,20 @@ func (this *MigrationContext) SetChunkSize(chunkSize int64) {
 	atomic.StoreInt64(&this.ChunkSize, chunkSize)
 }
 
+func (this *MigrationContext) SetThrottleGeneralCheckResult(checkResult *ThrottleCheckResult) *ThrottleCheckResult {
+	this.throttleMutex.Lock()
+	defer this.throttleMutex.Unlock()
+	this.throttleGeneralCheckResult = *checkResult
+	return checkResult
+}
+
+func (this *MigrationContext) GetThrottleGeneralCheckResult() *ThrottleCheckResult {
+	this.throttleMutex.Lock()
+	defer this.throttleMutex.Unlock()
+	result := this.throttleGeneralCheckResult
+	return &result
+}
+
 func (this *MigrationContext) SetThrottled(throttle bool, reason string) {
 	this.throttleMutex.Lock()
 	defer this.throttleMutex.Unlock()
@@ -454,6 +495,20 @@ func (this *MigrationContext) ReadCriticalLoad(criticalLoadList string) error {
 	return nil
 }
 
+func (this *MigrationContext) GetControlReplicasLagResult() mysql.ReplicationLagResult {
+	this.throttleMutex.Lock()
+	defer this.throttleMutex.Unlock()
+
+	lagResult := this.controlReplicasLagResult
+	return lagResult
+}
+
+func (this *MigrationContext) SetControlReplicasLagResult(lagResult *mysql.ReplicationLagResult) {
+	this.throttleMutex.Lock()
+	defer this.throttleMutex.Unlock()
+	this.controlReplicasLagResult = *lagResult
+}
+
 func (this *MigrationContext) GetThrottleControlReplicaKeys() *mysql.InstanceKeyMap {
 	this.throttleMutex.Lock()
 	defer this.throttleMutex.Unlock()

go/cmd/gh-ost/main.go

@@ -81,6 +81,7 @@ func main() {
 	replicationLagQuery := flag.String("replication-lag-query", "", "Query that detects replication lag in seconds. Result can be a floating point (by default gh-ost issues SHOW SLAVE STATUS and reads Seconds_behind_master). If you're using pt-heartbeat, query would be something like: SELECT ROUND(UNIX_TIMESTAMP() - MAX(UNIX_TIMESTAMP(ts))) AS delay FROM my_schema.heartbeat")
 	throttleControlReplicas := flag.String("throttle-control-replicas", "", "List of replicas on which to check for lag; comma delimited. Example: myhost1.com:3306,myhost2.com,myhost3.com:3307")
 	throttleQuery := flag.String("throttle-query", "", "when given, issued (every second) to check if operation should throttle. Expecting to return zero for no-throttle, >0 for throttle. Query is issued on the migrated server. Make sure this query is lightweight")
+	heartbeatIntervalMillis := flag.Int64("heartbeat-interval-millis", 500, "how frequently would gh-ost inject a heartbeat value")
 	flag.StringVar(&migrationContext.ThrottleFlagFile, "throttle-flag-file", "", "operation pauses when this file exists; hint: use a file that is specific to the table being altered")
 	flag.StringVar(&migrationContext.ThrottleAdditionalFlagFile, "throttle-additional-flag-file", "/tmp/gh-ost.throttle", "operation pauses when this file exists; hint: keep default, use for throttling multiple gh-ost operations")
 	flag.StringVar(&migrationContext.PostponeCutOverFlagFile, "postpone-cut-over-flag-file", "", "while this file exists, migration will postpone the final stage of swapping tables, and will keep on syncing the ghost table. Cut-over/swapping would be ready to perform the moment the file is deleted.")
@@ -184,6 +185,7 @@ func main() {
 	if migrationContext.ServeSocketFile == "" {
 		migrationContext.ServeSocketFile = fmt.Sprintf("/tmp/gh-ost.%s.%s.sock", migrationContext.DatabaseName, migrationContext.OriginalTableName)
 	}
+	migrationContext.SetHeartbeatIntervalMilliseconds(*heartbeatIntervalMillis)
 	migrationContext.SetNiceRatio(*niceRatio)
 	migrationContext.SetChunkSize(*chunkSize)
 	migrationContext.SetMaxLagMillisecondsThrottleThreshold(*maxLagMillis)

go/logic/applier.go

@@ -258,7 +258,7 @@ func (this *Applier) WriteChangelogState(value string) (string, error) {
 
 // InitiateHeartbeat creates a heartbeat cycle, writing to the changelog table.
 // This is done asynchronously
-func (this *Applier) InitiateHeartbeat(heartbeatIntervalMilliseconds int64) {
+func (this *Applier) InitiateHeartbeat() {
 	var numSuccessiveFailures int64
 	injectHeartbeat := func() error {
 		if _, err := this.WriteChangelog("heartbeat", time.Now().Format(time.RFC3339Nano)); err != nil {
@@ -273,10 +273,10 @@ func (this *Applier) InitiateHeartbeat(heartbeatIntervalMilliseconds int64) {
 	}
 	injectHeartbeat()
 
-	heartbeatTick := time.Tick(time.Duration(heartbeatIntervalMilliseconds) * time.Millisecond)
+	heartbeatTick := time.Tick(time.Duration(this.migrationContext.HeartbeatIntervalMilliseconds) * time.Millisecond)
 	for range heartbeatTick {
 		// Generally speaking, we would issue a goroutine, but I'd actually rather
-		// have this blocked rather than spam the master in the event something
+		// have this block the loop rather than spam the master in the event something
 		// goes wrong
 		if err := injectHeartbeat(); err != nil {
 			return