lightning: support disable scheduler by key range (#34130)

close #35148

br/pkg/errors/errors.go

@@ -3,6 +3,9 @@
 package errors
 
 import (
+	"context"
+	stderrors "errors"
+
 	"github.com/pingcap/errors"
 )
 
@@ -15,6 +18,17 @@ func Is(err error, is *errors.Error) bool {
 	return errorFound != nil
 }
 
+// IsContextCanceled checks whether the is caused by context.Canceled.
+// errors.Cause does not work for the error wrapped by %w in fmt.Errorf.
+// So we need to call stderrors.Is to unwrap the error.
+func IsContextCanceled(err error) bool {
+	err = errors.Cause(err)
+	if err == context.Canceled || err == context.DeadlineExceeded {
+		return true
+	}
+	return stderrors.Is(err, context.Canceled) || stderrors.Is(err, context.DeadlineExceeded)
+}
+
 // BR errors.
 var (
 	ErrUnknown                   = errors.Normalize("internal error", errors.RFCCodeText("BR:Common:ErrUnknown"))

br/pkg/errors/errors_test.go

@@ -0,0 +1,24 @@
+// Copyright 2022 PingCAP, Inc. Licensed under Apache-2.0.
+
+package errors_test
+
+import (
+	"context"
+	"net/url"
+	"testing"
+
+	"github.com/pingcap/errors"
+	berrors "github.com/pingcap/tidb/br/pkg/errors"
+	"github.com/stretchr/testify/require"
+)
+
+func TestIsContextCanceled(t *testing.T) {
+	require.False(t, berrors.IsContextCanceled(nil))
+	require.False(t, berrors.IsContextCanceled(errors.New("connection closed")))
+	require.True(t, berrors.IsContextCanceled(context.Canceled))
+	require.True(t, berrors.IsContextCanceled(context.DeadlineExceeded))
+	require.True(t, berrors.IsContextCanceled(errors.Trace(context.Canceled)))
+	require.True(t, berrors.IsContextCanceled(errors.Trace(context.DeadlineExceeded)))
+	require.True(t, berrors.IsContextCanceled(&url.Error{Err: context.Canceled}))
+	require.True(t, berrors.IsContextCanceled(&url.Error{Err: context.DeadlineExceeded}))
+}

br/pkg/lightning/backend/local/local.go

@@ -1392,8 +1392,32 @@ func (local *local) ImportEngine(ctx context.Context, engineUUID uuid.UUID, regi
 		return err
 	}
 
+	if len(ranges) > 0 && local.pdCtl.CanPauseSchedulerByKeyRange() {
+		subCtx, cancel := context.WithCancel(ctx)
+		defer cancel()
+
+		var startKey, endKey []byte
+		if len(ranges[0].start) > 0 {
+			startKey = codec.EncodeBytes(nil, ranges[0].start)
+		}
+		if len(ranges[len(ranges)-1].end) > 0 {
+			endKey = codec.EncodeBytes(nil, ranges[len(ranges)-1].end)
+		}
+		done, err := local.pdCtl.PauseSchedulersByKeyRange(subCtx, startKey, endKey)
+		if err != nil {
+			return errors.Trace(err)
+		}
+		defer func() {
+			cancel()
+			<-done
+		}()
+	}
+
 	log.FromContext(ctx).Info("start import engine", zap.Stringer("uuid", engineUUID),
 		zap.Int("ranges", len(ranges)), zap.Int64("count", lfLength), zap.Int64("size", lfTotalSize))
+
+	failpoint.Inject("ReadyForImportEngine", func() {})
+
 	for {
 		unfinishedRanges := lf.unfinishedRanges(ranges)
 		if len(unfinishedRanges) == 0 {

br/pkg/lightning/backend/local/localhelper.go

@@ -19,7 +19,6 @@ import (
 	"context"
 	"database/sql"
 	"math"
-	"regexp"
 	"runtime"
 	"sort"
 	"strings"
@@ -332,14 +331,7 @@ func (local *local) SplitAndScatterRegionByRanges(
 	}
 
 	startTime := time.Now()
-	scatterCount := 0
-	for _, region := range scatterRegions {
-		local.waitForScatterRegion(ctx, region)
-		if time.Since(startTime) > split.ScatterWaitUpperInterval {
-			break
-		}
-		scatterCount++
-	}
+	scatterCount, err := local.waitForScatterRegions(ctx, scatterRegions)
 	if scatterCount == len(scatterRegions) {
 		log.FromContext(ctx).Info("waiting for scattering regions done",
 			zap.Int("skipped_keys", skippedKeys),
@@ -349,7 +341,8 @@ func (local *local) SplitAndScatterRegionByRanges(
 			zap.Int("skipped_keys", skippedKeys),
 			zap.Int("scatterCount", scatterCount),
 			zap.Int("regions", len(scatterRegions)),
-			zap.Duration("take", time.Since(startTime)))
+			zap.Duration("take", time.Since(startTime)),
+			zap.Error(err))
 	}
 	return nil
 }
@@ -447,28 +440,38 @@ func (local *local) waitForSplit(ctx context.Context, regionID uint64) {
 	}
 }
 
-func (local *local) waitForScatterRegion(ctx context.Context, regionInfo *split.RegionInfo) {
-	for i := 0; i < split.ScatterWaitMaxRetryTimes; i++ {
-		ok, err := local.checkScatterRegionFinishedOrReScatter(ctx, regionInfo)
-		if ok {
-			return
-		}
-		if err != nil {
-			if !common.IsRetryableError(err) {
-				log.FromContext(ctx).Warn("wait for scatter region encountered non-retryable error", logutil.Region(regionInfo.Region), zap.Error(err))
-				return
+func (local *local) waitForScatterRegions(ctx context.Context, regions []*split.RegionInfo) (scatterCount int, _ error) {
+	subCtx, cancel := context.WithTimeout(ctx, split.ScatterWaitUpperInterval)
+	defer cancel()
+
+	for len(regions) > 0 {
+		var retryRegions []*split.RegionInfo
+		for _, region := range regions {
+			scattered, err := local.checkRegionScatteredOrReScatter(subCtx, region)
+			if scattered {
+				scatterCount++
+				continue
 			}
-			log.FromContext(ctx).Warn("wait for scatter region encountered error, will retry again", logutil.Region(regionInfo.Region), zap.Error(err))
+			if err != nil {
+				if !common.IsRetryableError(err) {
+					log.FromContext(ctx).Warn("wait for scatter region encountered non-retryable error", logutil.Region(region.Region), zap.Error(err))
+					return scatterCount, err
+				}
+				log.FromContext(ctx).Warn("wait for scatter region encountered error, will retry again", logutil.Region(region.Region), zap.Error(err))
+			}
+			retryRegions = append(retryRegions, region)
 		}
+		regions = retryRegions
 		select {
 		case <-time.After(time.Second):
-		case <-ctx.Done():
+		case <-subCtx.Done():
 			return
 		}
 	}
+	return scatterCount, nil
 }
 
-func (local *local) checkScatterRegionFinishedOrReScatter(ctx context.Context, regionInfo *split.RegionInfo) (bool, error) {
+func (local *local) checkRegionScatteredOrReScatter(ctx context.Context, regionInfo *split.RegionInfo) (bool, error) {
 	resp, err := local.splitCli.GetOperator(ctx, regionInfo.Region.GetId())
 	if err != nil {
 		return false, err
@@ -478,13 +481,9 @@ func (local *local) checkScatterRegionFinishedOrReScatter(ctx context.Context, r
 		if respErr.GetType() == pdpb.ErrorType_REGION_NOT_FOUND {
 			return true, nil
 		}
-		// don't return error if region replicate not complete
-		// TODO: should add a new error type to avoid this check by string matching
-		matches, _ := regexp.MatchString("region \\d+ is not fully replicated", respErr.Message)
-		if matches {
-			return false, nil
-		}
-		return false, errors.Errorf("get operator error: %s", respErr.GetType())
+		return false, errors.Errorf(
+			"failed to get region operator, error type: %s, error message: %s",
+			respErr.GetType().String(), respErr.GetMessage())
 	}
 	// If the current operator of the region is not 'scatter-region', we could assume
 	// that 'scatter-operator' has finished.

br/pkg/lightning/common/retry.go

@@ -20,6 +20,7 @@ import (
 	"io"
 	"net"
 	"os"
+	"regexp"
 	"syscall"
 
 	"github.com/go-sql-driver/mysql"
@@ -30,6 +31,8 @@ import (
 	"google.golang.org/grpc/status"
 )
 
+var regionNotFullyReplicatedRe = regexp.MustCompile(`region \d+ is not fully replicated`)
+
 // IsRetryableError returns whether the error is transient (e.g. network
 // connection dropped) or irrecoverable (e.g. user pressing Ctrl+C). This
 // function returns `false` (irrecoverable) if `err == nil`.
@@ -88,6 +91,9 @@ func isSingleRetryableError(err error) bool {
 		}
 		return false
 	default:
+		if regionNotFullyReplicatedRe.MatchString(err.Error()) {
+			return true
+		}
 		switch status.Code(err) {
 		case codes.DeadlineExceeded, codes.NotFound, codes.AlreadyExists, codes.PermissionDenied, codes.ResourceExhausted, codes.Aborted, codes.OutOfRange, codes.Unavailable, codes.DataLoss:
 			return true

br/pkg/lightning/common/retry_test.go

@@ -94,4 +94,6 @@ func TestIsRetryableError(t *testing.T) {
 	require.False(t, IsRetryableError(multierr.Combine(context.Canceled, context.Canceled)))
 	require.True(t, IsRetryableError(multierr.Combine(&net.DNSError{IsTimeout: true}, &net.DNSError{IsTimeout: true})))
 	require.False(t, IsRetryableError(multierr.Combine(context.Canceled, &net.DNSError{IsTimeout: true})))
+
+	require.True(t, IsRetryableError(errors.Errorf("region %d is not fully replicated", 1234)))
 }

br/pkg/lightning/lightning.go

@@ -209,6 +209,14 @@ func (l *Lightning) goServe(statusAddr string, realAddrWriter io.Writer) error {
 	mux.HandleFunc("/debug/pprof/symbol", pprof.Symbol)
 	mux.HandleFunc("/debug/pprof/trace", pprof.Trace)
 
+	// Enable failpoint http API for testing.
+	failpoint.Inject("EnableTestAPI", func() {
+		mux.HandleFunc("/fail/", func(w http.ResponseWriter, r *http.Request) {
+			r.URL.Path = strings.TrimPrefix(r.URL.Path, "/fail")
+			new(failpoint.HttpHandler).ServeHTTP(w, r)
+		})
+	})
+
 	handleTasks := http.StripPrefix("/tasks", http.HandlerFunc(l.handleTask))
 	mux.Handle("/tasks", httpHandleWrapper(handleTasks.ServeHTTP))
 	mux.Handle("/tasks/", httpHandleWrapper(handleTasks.ServeHTTP))

br/pkg/lightning/restore/meta_manager.go

@@ -555,6 +555,8 @@ type taskMetaMgr interface {
 	// need to update or any new tasks. There is at most one lightning who can execute the action function at the same time.
 	// Note that action may be executed multiple times due to transaction retry, caller should make sure it's idempotent.
 	CheckTasksExclusively(ctx context.Context, action func(tasks []taskMeta) ([]taskMeta, error)) error
+	// CanPauseSchedulerByKeyRange returns whether the scheduler can pause by the key range.
+	CanPauseSchedulerByKeyRange() bool
 	CheckAndPausePdSchedulers(ctx context.Context) (pdutil.UndoFunc, error)
 	// CheckAndFinishRestore check task meta and return whether to switch cluster to normal state and clean up the metadata
 	// Return values: first boolean indicates whether switch back tidb cluster to normal state (restore schedulers, switch tikv to normal)
@@ -867,6 +869,10 @@ func (m *dbTaskMetaMgr) CheckAndPausePdSchedulers(ctx context.Context) (pdutil.U
 	}, nil
 }
 
+func (m *dbTaskMetaMgr) CanPauseSchedulerByKeyRange() bool {
+	return m.pd.CanPauseSchedulerByKeyRange()
+}
+
 // CheckAndFinishRestore check task meta and return whether to switch cluster to normal state and clean up the metadata
 // Return values: first boolean indicates whether switch back tidb cluster to normal state (restore schedulers, switch tikv to normal)
 // the second boolean indicates whether to clean up the metadata in tidb
@@ -1058,6 +1064,10 @@ func (m noopTaskMetaMgr) CheckAndPausePdSchedulers(ctx context.Context) (pdutil.
 	}, nil
 }
 
+func (m noopTaskMetaMgr) CanPauseSchedulerByKeyRange() bool {
+	return false
+}
+
 func (m noopTaskMetaMgr) CheckTaskExist(ctx context.Context) (bool, error) {
 	return true, nil
 }
@@ -1168,6 +1178,10 @@ func (m *singleTaskMetaMgr) CheckAndPausePdSchedulers(ctx context.Context) (pdut
 	return m.pd.RemoveSchedulers(ctx)
 }
 
+func (m *singleTaskMetaMgr) CanPauseSchedulerByKeyRange() bool {
+	return m.pd.CanPauseSchedulerByKeyRange()
+}
+
 func (m *singleTaskMetaMgr) CheckTaskExist(ctx context.Context) (bool, error) {
 	return m.initialized, nil
 }

br/pkg/lightning/restore/restore.go

@@ -1370,42 +1370,51 @@ func (rc *Controller) restoreTables(ctx context.Context) (finalErr error) {
 	// make split region and ingest sst more stable
 	// because importer backend is mostly use for v3.x cluster which doesn't support these api,
 	// so we also don't do this for import backend
-	finishSchedulers := func() {}
+	finishSchedulers := func() {
+		if rc.taskMgr != nil {
+			rc.taskMgr.Close()
+		}
+	}
 	// if one lightning failed abnormally, and can't determine whether it needs to switch back,
 	// we do not do switch back automatically
 	switchBack := false
 	cleanup := false
 	postProgress := func() error { return nil }
 	if rc.cfg.TikvImporter.Backend == config.BackendLocal {
+		var restoreFn pdutil.UndoFunc
 
-		logTask.Info("removing PD leader&region schedulers")
+		if !rc.taskMgr.CanPauseSchedulerByKeyRange() {
+			logTask.Info("removing PD leader&region schedulers")
 
-		restoreFn, err := rc.taskMgr.CheckAndPausePdSchedulers(ctx)
-		if err != nil {
-			return errors.Trace(err)
+			var err error
+			restoreFn, err = rc.taskMgr.CheckAndPausePdSchedulers(ctx)
+			if err != nil {
+				return errors.Trace(err)
+			}
 		}
 
 		finishSchedulers = func() {
-			if restoreFn != nil {
-				taskFinished := finalErr == nil
-				// use context.Background to make sure this restore function can still be executed even if ctx is canceled
-				restoreCtx := context.Background()
-				needSwitchBack, needCleanup, err := rc.taskMgr.CheckAndFinishRestore(restoreCtx, taskFinished)
-				if err != nil {
-					logTask.Warn("check restore pd schedulers failed", zap.Error(err))
-					return
-				}
-				switchBack = needSwitchBack
-				if needSwitchBack {
-					logTask.Info("add back PD leader&region schedulers")
-					if restoreE := restoreFn(restoreCtx); restoreE != nil {
-						logTask.Warn("failed to restore removed schedulers, you may need to restore them manually", zap.Error(restoreE))
-					}
+			taskFinished := finalErr == nil
+			// use context.Background to make sure this restore function can still be executed even if ctx is canceled
+			restoreCtx := context.Background()
+			needSwitchBack, needCleanup, err := rc.taskMgr.CheckAndFinishRestore(restoreCtx, taskFinished)
+			if err != nil {
+				logTask.Warn("check restore pd schedulers failed", zap.Error(err))
+				return
+			}
+			switchBack = needSwitchBack
+			cleanup = needCleanup
+
+			if needSwitchBack && restoreFn != nil {
+				logTask.Info("add back PD leader&region schedulers")
+				if restoreE := restoreFn(restoreCtx); restoreE != nil {
+					logTask.Warn("failed to restore removed schedulers, you may need to restore them manually", zap.Error(restoreE))
 				}
-				cleanup = needCleanup
 			}
 
-			rc.taskMgr.Close()
+			if rc.taskMgr != nil {
+				rc.taskMgr.Close()
+			}
 		}
 	}