scheduler: do not balance the empty regions (#3344)

* do not balance the empty regions

Signed-off-by: Ryan Leung <rleungx@gmail.com>

* add a threshold for balance empty region

Signed-off-by: rleungx <rleungx@gmail.com>

* address the comment

Signed-off-by: Ryan Leung <rleungx@gmail.com>

* Update server/schedule/opt/healthy.go

Signed-off-by: Ryan Leung <rleungx@gmail.com>

Co-authored-by: Ti Chi Robot <71242396+ti-chi-bot@users.noreply.github.com>

pkg/mock/mockcluster/mockcluster.go

@@ -15,6 +15,7 @@ package mockcluster
 
 import (
 	"fmt"
+	"strconv"
 	"time"
 
 	"github.com/gogo/protobuf/proto"
@@ -585,7 +586,11 @@ func (mc *Cluster) CheckLabelProperty(typ string, labels []*metapb.StoreLabel) b
 
 // PutRegionStores mocks method.
 func (mc *Cluster) PutRegionStores(id uint64, stores ...uint64) {
-	meta := &metapb.Region{Id: id}
+	meta := &metapb.Region{
+		Id:       id,
+		StartKey: []byte(strconv.FormatUint(id, 10)),
+		EndKey:   []byte(strconv.FormatUint(id+1, 10)),
+	}
 	for _, s := range stores {
 		meta.Peers = append(meta.Peers, &metapb.Peer{StoreId: s})
 	}

server/schedule/opt/healthy.go

@@ -13,7 +13,12 @@
 
 package opt
 
-import "github.com/tikv/pd/server/core"
+import (
+	"github.com/tikv/pd/server/core"
+)
+
+// BalanceEmptyRegionThreshold is a threshold which allow balance the empty region if the region number is less than this threshold.
+var balanceEmptyRegionThreshold = 50
 
 // IsRegionHealthy checks if a region is healthy for scheduling. It requires the
 // region does not have any down or pending peers. And when placement rules
@@ -31,6 +36,11 @@ func IsHealthyAllowPending(cluster Cluster, region *core.RegionInfo) bool {
 	return len(region.GetDownPeers()) == 0
 }
 
+// IsEmptyRegionAllowBalance checks if a region is an empty region and can be balanced.
+func IsEmptyRegionAllowBalance(cluster Cluster, region *core.RegionInfo) bool {
+	return region.GetApproximateSize() > core.EmptyRegionApproximateSize || cluster.GetRegionCount() < balanceEmptyRegionThreshold
+}
+
 // HealthRegion returns a function that checks if a region is healthy for
 // scheduling. It requires the region does not have any down or pending peers,
 // and does not have any learner peers when placement rules is disabled.
@@ -45,6 +55,11 @@ func HealthAllowPending(cluster Cluster) func(*core.RegionInfo) bool {
 	return func(region *core.RegionInfo) bool { return IsHealthyAllowPending(cluster, region) }
 }
 
+// AllowBalanceEmptyRegion returns a function that checks if a region is an empty region and can be balanced.
+func AllowBalanceEmptyRegion(cluster Cluster) func(*core.RegionInfo) bool {
+	return func(region *core.RegionInfo) bool { return IsEmptyRegionAllowBalance(cluster, region) }
+}
+
 // IsRegionReplicated checks if a region is fully replicated. When placement
 // rules is enabled, its peers should fit corresponding rules. When placement
 // rules is disabled, it should have enough replicas and no any learner peer.

server/schedulers/balance_region.go

@@ -152,18 +152,18 @@ func (s *balanceRegionScheduler) Schedule(cluster opt.Cluster) []*operator.Opera
 		for i := 0; i < balanceRegionRetryLimit; i++ {
 			// Priority pick the region that has a pending peer.
 			// Pending region may means the disk is overload, remove the pending region firstly.
-			region := cluster.RandPendingRegion(sourceID, s.conf.Ranges, opt.HealthAllowPending(cluster), opt.ReplicatedRegion(cluster))
+			region := cluster.RandPendingRegion(sourceID, s.conf.Ranges, opt.HealthAllowPending(cluster), opt.ReplicatedRegion(cluster), opt.AllowBalanceEmptyRegion(cluster))
 			if region == nil {
 				// Then pick the region that has a follower in the source store.
-				region = cluster.RandFollowerRegion(sourceID, s.conf.Ranges, opt.HealthRegion(cluster), opt.ReplicatedRegion(cluster))
+				region = cluster.RandFollowerRegion(sourceID, s.conf.Ranges, opt.HealthRegion(cluster), opt.ReplicatedRegion(cluster), opt.AllowBalanceEmptyRegion(cluster))
 			}
 			if region == nil {
 				// Then pick the region has the leader in the source store.
-				region = cluster.RandLeaderRegion(sourceID, s.conf.Ranges, opt.HealthRegion(cluster), opt.ReplicatedRegion(cluster))
+				region = cluster.RandLeaderRegion(sourceID, s.conf.Ranges, opt.HealthRegion(cluster), opt.ReplicatedRegion(cluster), opt.AllowBalanceEmptyRegion(cluster))
 			}
 			if region == nil {
 				// Finally pick learner.
-				region = cluster.RandLearnerRegion(sourceID, s.conf.Ranges, opt.HealthRegion(cluster), opt.ReplicatedRegion(cluster))
+				region = cluster.RandLearnerRegion(sourceID, s.conf.Ranges, opt.HealthRegion(cluster), opt.ReplicatedRegion(cluster), opt.AllowBalanceEmptyRegion(cluster))
 			}
 			if region == nil {
 				schedulerCounter.WithLabelValues(s.GetName(), "no-region").Inc()

server/schedulers/balance_test.go

@@ -932,6 +932,43 @@ func (s *testBalanceRegionSchedulerSuite) TestShouldNotBalance(c *C) {
 	}
 }
 
+func (s *testBalanceRegionSchedulerSuite) TestEmptyRegion(c *C) {
+	opt := config.NewTestOptions()
+	tc := mockcluster.NewCluster(opt)
+	tc.DisableFeature(versioninfo.JointConsensus)
+	oc := schedule.NewOperatorController(s.ctx, nil, nil)
+	sb, err := schedule.CreateScheduler(BalanceRegionType, oc, core.NewStorage(kv.NewMemoryKV()), schedule.ConfigSliceDecoder(BalanceRegionType, []string{"", ""}))
+	c.Assert(err, IsNil)
+	tc.AddRegionStore(1, 10)
+	tc.AddRegionStore(2, 9)
+	tc.AddRegionStore(3, 10)
+	tc.AddRegionStore(4, 10)
+	region := core.NewRegionInfo(
+		&metapb.Region{
+			Id:       5,
+			StartKey: []byte("a"),
+			EndKey:   []byte("b"),
+			Peers: []*metapb.Peer{
+				{Id: 6, StoreId: 1},
+				{Id: 7, StoreId: 3},
+				{Id: 8, StoreId: 4},
+			},
+		},
+		&metapb.Peer{Id: 7, StoreId: 3},
+		core.SetApproximateSize(1),
+		core.SetApproximateKeys(1),
+	)
+	tc.PutRegion(region)
+	operators := sb.Schedule(tc)
+	c.Assert(operators, NotNil)
+
+	for i := uint64(10); i < 60; i++ {
+		tc.PutRegionStores(i, 1, 3, 4)
+	}
+	operators = sb.Schedule(tc)
+	c.Assert(operators, IsNil)
+}
+
 var _ = Suite(&testRandomMergeSchedulerSuite{})
 
 type testRandomMergeSchedulerSuite struct{}