|
| 1 | +/* |
| 2 | +Copyright 2025 The Vitess Authors. |
| 3 | +
|
| 4 | +Licensed under the Apache License, Version 2.0 (the "License"); |
| 5 | +you may not use this file except in compliance with the License. |
| 6 | +You may obtain a copy of the License at |
| 7 | +
|
| 8 | + http://www.apache.org/licenses/LICENSE-2.0 |
| 9 | +
|
| 10 | +Unless required by applicable law or agreed to in writing, software |
| 11 | +distributed under the License is distributed on an "AS IS" BASIS, |
| 12 | +WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| 13 | +See the License for the specific language governing permissions and |
| 14 | +limitations under the License. |
| 15 | +*/ |
| 16 | + |
| 17 | +package clone |
| 18 | + |
| 19 | +import ( |
| 20 | + "os" |
| 21 | + "testing" |
| 22 | + "time" |
| 23 | + |
| 24 | + "github.com/stretchr/testify/assert" |
| 25 | + "github.com/stretchr/testify/require" |
| 26 | + |
| 27 | + "vitess.io/vitess/go/test/endtoend/cluster" |
| 28 | + "vitess.io/vitess/go/vt/log" |
| 29 | + vtutils "vitess.io/vitess/go/vt/utils" |
| 30 | +) |
| 31 | + |
| 32 | +func TestCloneBackup(t *testing.T) { |
| 33 | + t.Cleanup(func() { removeBackups(t) }) |
| 34 | + t.Cleanup(tearDown) |
| 35 | + |
| 36 | + // Initialize tablets first so we can connect to MySQL. |
| 37 | + for _, tablet := range []*cluster.Vttablet{primary, replica1} { |
| 38 | + err := localCluster.InitTablet(tablet, keyspaceName, shardName) |
| 39 | + require.NoError(t, err) |
| 40 | + err = tablet.VttabletProcess.Setup() |
| 41 | + require.NoError(t, err) |
| 42 | + } |
| 43 | + |
| 44 | + // Initialize shard primary. |
| 45 | + err := localCluster.VtctldClientProcess.InitShardPrimary(keyspaceName, shardName, cell, primary.TabletUID) |
| 46 | + require.NoError(t, err) |
| 47 | + |
| 48 | + // Now check if MySQL version supports clone (need vttablet running to query). |
| 49 | + if !mysqlVersionSupportsClone(t, primary) { |
| 50 | + t.Skip("Skipping clone test: MySQL version does not support CLONE (requires 8.0.17+)") |
| 51 | + } |
| 52 | + |
| 53 | + // Check if clone plugin is available. |
| 54 | + if !clonePluginAvailable(t, primary) { |
| 55 | + t.Skip("Skipping clone test: clone plugin not available") |
| 56 | + } |
| 57 | + |
| 58 | + // Set up clean test data (table may have data from previous tests). |
| 59 | + _, err = primary.VttabletProcess.QueryTablet(vtInsertTest, keyspaceName, true) |
| 60 | + require.NoError(t, err) |
| 61 | + _, err = primary.VttabletProcess.QueryTablet("TRUNCATE TABLE vt_insert_test", keyspaceName, true) |
| 62 | + require.NoError(t, err) |
| 63 | + _, err = primary.VttabletProcess.QueryTablet("insert into vt_insert_test (msg) values ('clone_test_1')", keyspaceName, true) |
| 64 | + require.NoError(t, err) |
| 65 | + _, err = primary.VttabletProcess.QueryTablet("insert into vt_insert_test (msg) values ('clone_test_2')", keyspaceName, true) |
| 66 | + require.NoError(t, err) |
| 67 | + |
| 68 | + // Verify data exists on primary. |
| 69 | + cluster.VerifyRowsInTablet(t, primary, keyspaceName, 2) |
| 70 | + |
| 71 | + // Wait for replica to catch up. |
| 72 | + time.Sleep(2 * time.Second) |
| 73 | + cluster.VerifyRowsInTablet(t, replica1, keyspaceName, 2) |
| 74 | + |
| 75 | + // Take a backup using clone from primary. |
| 76 | + log.Infof("Starting vtbackup with --clone-from-primary") |
| 77 | + err = vtbackupWithClone(t) |
| 78 | + require.NoError(t, err) |
| 79 | + |
| 80 | + // Verify a backup was created. |
| 81 | + backups := verifyBackupCount(t, shardKsName, 1) |
| 82 | + assert.NotEmpty(t, backups) |
| 83 | + |
| 84 | + // Insert more data AFTER the backup was taken. |
| 85 | + _, err = primary.VttabletProcess.QueryTablet("insert into vt_insert_test (msg) values ('after_backup')", keyspaceName, true) |
| 86 | + require.NoError(t, err) |
| 87 | + cluster.VerifyRowsInTablet(t, primary, keyspaceName, 3) |
| 88 | + |
| 89 | + // Now bring up replica2 and restore from the backup we just created. |
| 90 | + // This verifies the clone-based backup actually contains the data. |
| 91 | + log.Infof("Restoring replica2 from backup to verify clone worked") |
| 92 | + err = localCluster.InitTablet(replica2, keyspaceName, shardName) |
| 93 | + require.NoError(t, err) |
| 94 | + restore(t, replica2, "replica", "SERVING") |
| 95 | + |
| 96 | + // Give replica2 time to catch up via replication. |
| 97 | + time.Sleep(5 * time.Second) |
| 98 | + |
| 99 | + // Verify replica2 has ALL the data (2 rows from before backup + 1 from after). |
| 100 | + // The 2 pre-backup rows prove the clone-based backup worked. |
| 101 | + // The 3rd row proves replication is working after restore. |
| 102 | + cluster.VerifyRowsInTablet(t, replica2, keyspaceName, 3) |
| 103 | + log.Infof("Clone backup verification successful: replica2 has all data") |
| 104 | +} |
| 105 | + |
| 106 | +func vtbackupWithClone(t *testing.T) error { |
| 107 | + mysqlSocket, err := os.CreateTemp("", "vtbackup_clone_test_mysql.sock") |
| 108 | + require.NoError(t, err) |
| 109 | + defer os.Remove(mysqlSocket.Name()) |
| 110 | + |
| 111 | + extraArgs := []string{ |
| 112 | + "--allow_first_backup", |
| 113 | + "--db-credentials-file", dbCredentialFile, |
| 114 | + "--mysql-clone-enabled", |
| 115 | + vtutils.GetFlagVariantForTests("--mysql-socket"), mysqlSocket.Name(), |
| 116 | + // Clone from primary instead of restoring from backup. |
| 117 | + "--restore-with-clone", |
| 118 | + "--clone-from-primary", |
| 119 | + // Clone credentials - use vt_clone user which is created with @'%' host |
| 120 | + // and BACKUP_ADMIN privilege in init_db.sql (no password). |
| 121 | + "--db-clone-user", "vt_clone", |
| 122 | + "--db-clone-password", "", |
| 123 | + "--db-clone-use-ssl=false", |
| 124 | + } |
| 125 | + |
| 126 | + log.Infof("Starting vtbackup with clone args: %v", extraArgs) |
| 127 | + return localCluster.StartVtbackup(newInitDBFile, false, keyspaceName, shardName, cell, extraArgs...) |
| 128 | +} |
| 129 | + |
| 130 | +func verifyBackupCount(t *testing.T, shardKsName string, expected int) []string { |
| 131 | + backups, err := localCluster.VtctldClientProcess.ExecuteCommandWithOutput("GetBackups", shardKsName) |
| 132 | + require.NoError(t, err) |
| 133 | + |
| 134 | + var result []string |
| 135 | + for _, line := range splitLines(backups) { |
| 136 | + if line != "" { |
| 137 | + result = append(result, line) |
| 138 | + } |
| 139 | + } |
| 140 | + assert.Equalf(t, expected, len(result), "expected %d backups, got %d", expected, len(result)) |
| 141 | + return result |
| 142 | +} |
| 143 | + |
| 144 | +func restore(t *testing.T, tablet *cluster.Vttablet, tabletType string, waitForState string) { |
| 145 | + // Start tablet with restore enabled. MySQL is already running from TestMain. |
| 146 | + log.Infof("restoring tablet %s", time.Now()) |
| 147 | + tablet.VttabletProcess.ExtraArgs = []string{"--db-credentials-file", dbCredentialFile} |
| 148 | + tablet.VttabletProcess.TabletType = tabletType |
| 149 | + tablet.VttabletProcess.ServingStatus = waitForState |
| 150 | + tablet.VttabletProcess.SupportsBackup = true |
| 151 | + err := tablet.VttabletProcess.Setup() |
| 152 | + require.NoError(t, err) |
| 153 | +} |
| 154 | + |
| 155 | +func tearDown() { |
| 156 | + for _, tablet := range []*cluster.Vttablet{primary, replica1, replica2} { |
| 157 | + if tablet != nil && tablet.VttabletProcess != nil { |
| 158 | + _ = tablet.VttabletProcess.TearDown() |
| 159 | + } |
| 160 | + if tablet != nil { |
| 161 | + _ = localCluster.VtctldClientProcess.ExecuteCommand("DeleteTablets", "--allow-primary", tablet.Alias) |
| 162 | + } |
| 163 | + } |
| 164 | +} |
0 commit comments