*: make TestConfigTestSuite stable (tikv#7398)

close tikv#7395

Signed-off-by: lhy1024 <[email protected]>

client/retry/backoff.go

@@ -0,0 +1,88 @@
+// Copyright 2023 TiKV Project Authors.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package retry
+
+import (
+	"context"
+	"time"
+
+	"github.com/pingcap/failpoint"
+)
+
+// BackOffer is a backoff policy for retrying operations.
+type BackOffer struct {
+	max  time.Duration
+	next time.Duration
+	base time.Duration
+}
+
+// Exec is a helper function to exec backoff.
+func (bo *BackOffer) Exec(
+	ctx context.Context,
+	fn func() error,
+) error {
+	if err := fn(); err != nil {
+		after := time.NewTimer(bo.nextInterval())
+		defer after.Stop()
+		select {
+		case <-ctx.Done():
+		case <-after.C:
+			failpoint.Inject("backOffExecute", func() {
+				testBackOffExecuteFlag = true
+			})
+		}
+		return err
+	}
+	// reset backoff when fn() succeed.
+	bo.resetBackoff()
+	return nil
+}
+
+// InitialBackOffer make the initial state for retrying.
+func InitialBackOffer(base, max time.Duration) BackOffer {
+	return BackOffer{
+		max:  max,
+		base: base,
+		next: base,
+	}
+}
+
+// nextInterval for now use the `exponentialInterval`.
+func (bo *BackOffer) nextInterval() time.Duration {
+	return bo.exponentialInterval()
+}
+
+// exponentialInterval returns the exponential backoff duration.
+func (bo *BackOffer) exponentialInterval() time.Duration {
+	backoffInterval := bo.next
+	bo.next *= 2
+	if bo.next > bo.max {
+		bo.next = bo.max
+	}
+	return backoffInterval
+}
+
+// resetBackoff resets the backoff to initial state.
+func (bo *BackOffer) resetBackoff() {
+	bo.next = bo.base
+}
+
+// Only used for test.
+var testBackOffExecuteFlag = false
+
+// TestBackOffExecute Only used for test.
+func TestBackOffExecute() bool {
+	return testBackOffExecuteFlag
+}

pkg/schedule/operator/operator_test.go

@@ -17,7 +17,6 @@ package operator
 import (
 	"context"
 	"encoding/json"
-	"fmt"
 	"sync/atomic"
 	"testing"
 	"time"
@@ -514,7 +513,7 @@ func (suite *operatorTestSuite) TestOpStepTimeout() {
 		},
 	}
 	for i, v := range testData {
-		fmt.Printf("case:%d\n", i)
+		suite.T().Logf("case: %d", i)
 		for _, step := range v.step {
 			suite.Equal(v.expect, step.Timeout(v.regionSize))
 		}

tests/integrations/mcs/tso/server_test.go

@@ -89,7 +89,7 @@ func (suite *tsoServerTestSuite) TearDownSuite() {
 func (suite *tsoServerTestSuite) TestTSOServerStartAndStopNormally() {
 	defer func() {
 		if r := recover(); r != nil {
-			fmt.Println("Recovered from an unexpected panic", r)
+			suite.T().Log("Recovered from an unexpected panic", r)
 			suite.T().Errorf("Expected no panic, but something bad occurred with")
 		}
 	}()

tests/pdctl/config/config_test.go

@@ -832,7 +832,6 @@ func (suite *configTestSuite) checkPDServerConfig(cluster *tests.TestCluster) {
 		LastHeartbeat: time.Now().UnixNano(),
 	}
 	tests.MustPutStore(re, cluster, store)
-	defer cluster.Destroy()
 
 	output, err := pdctl.ExecuteCommand(cmd, "-u", pdAddr, "config", "show", "server")
 	re.NoError(err)
@@ -844,7 +843,9 @@ func (suite *configTestSuite) checkPDServerConfig(cluster *tests.TestCluster) {
 	re.Equal("table", conf.KeyType)
 	re.Equal(typeutil.StringSlice([]string{}), conf.RuntimeServices)
 	re.Equal("", conf.MetricStorage)
-	re.Equal("auto", conf.DashboardAddress)
+	if conf.DashboardAddress != "auto" { // dashboard has been assigned
+		re.Equal(leaderServer.GetAddr(), conf.DashboardAddress)
+	}
 	re.Equal(int(3), conf.FlowRoundByDigit)
 }
 

tests/server/api/rule_test.go

@@ -1092,7 +1092,6 @@ func (suite *regionRuleTestSuite) checkRegionPlacementRule(cluster *tests.TestCl
 	var label labeler.LabelRule
 	escapedID := url.PathEscape("keyspaces/0")
 	u = fmt.Sprintf("%s/config/region-label/rule/%s", urlPrefix, escapedID)
-	fmt.Println("u====", u)
 	err = tu.ReadGetJSON(re, testDialClient, u, &label)
 	suite.NoError(err)
 	suite.Equal(label.ID, "keyspaces/0")

tests/server/cluster/cluster_test.go

@@ -1288,7 +1288,7 @@ func TestTransferLeaderForScheduler(t *testing.T) {
 	re.NoError(err)
 	tc.WaitLeader()
 	// start
-	leaderServer := tc.GetServer(tc.GetLeader())
+	leaderServer := tc.GetLeaderServer()
 	re.NoError(leaderServer.BootstrapCluster())
 	rc := leaderServer.GetServer().GetRaftCluster()
 	re.NotNil(rc)
@@ -1327,7 +1327,7 @@ func TestTransferLeaderForScheduler(t *testing.T) {
 	tc.ResignLeader()
 	rc.Stop()
 	tc.WaitLeader()
-	leaderServer = tc.GetServer(tc.GetLeader())
+	leaderServer = tc.GetLeaderServer()
 	rc1 := leaderServer.GetServer().GetRaftCluster()
 	rc1.Start(leaderServer.GetServer())
 	re.NoError(err)
@@ -1347,7 +1347,7 @@ func TestTransferLeaderForScheduler(t *testing.T) {
 	tc.ResignLeader()
 	rc1.Stop()
 	tc.WaitLeader()
-	leaderServer = tc.GetServer(tc.GetLeader())
+	leaderServer = tc.GetLeaderServer()
 	rc = leaderServer.GetServer().GetRaftCluster()
 	rc.Start(leaderServer.GetServer())
 	re.NotNil(rc)

tests/server/member/member_test.go

@@ -260,7 +260,7 @@ func TestPDLeaderLostWhileEtcdLeaderIntact(t *testing.T) {
 	re.NoError(err)
 
 	leader1 := cluster.WaitLeader()
-	memberID := cluster.GetServer(leader1).GetLeader().GetMemberId()
+	memberID := cluster.GetLeaderServer().GetLeader().GetMemberId()
 
 	re.NoError(failpoint.Enable("github.com/tikv/pd/server/leaderLoopCheckAgain", fmt.Sprintf("return(\"%d\")", memberID)))
 	re.NoError(failpoint.Enable("github.com/tikv/pd/server/exitCampaignLeader", fmt.Sprintf("return(\"%d\")", memberID)))

tests/server/region_syncer/region_syncer_test.go

@@ -259,7 +259,7 @@ func TestPrepareCheckerWithTransferLeader(t *testing.T) {
 	err = cluster.RunInitialServers()
 	re.NoError(err)
 	cluster.WaitLeader()
-	leaderServer := cluster.GetServer(cluster.GetLeader())
+	leaderServer := cluster.GetLeaderServer()
 	re.NoError(leaderServer.BootstrapCluster())
 	rc := leaderServer.GetServer().GetRaftCluster()
 	re.NotNil(rc)