Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Fix TestLeftJoinUsingUnsharded and remove instability when running E2E locally #13973

Merged
merged 7 commits into from
Sep 18, 2023
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 12 additions & 4 deletions go/test/endtoend/cluster/cluster_process.go
Original file line number Diff line number Diff line change
Expand Up @@ -334,13 +334,12 @@ func (cluster *LocalProcessCluster) startKeyspace(keyspace Keyspace, shardNames
}
// Create the keyspace if it doesn't already exist.
_ = cluster.VtctlProcess.CreateKeyspace(keyspace.Name, keyspace.SidecarDBName)
var mysqlctlProcessList []*exec.Cmd
for _, shardName := range shardNames {
shard := &Shard{
Name: shardName,
}
log.Infof("Starting shard: %v", shardName)
mysqlctlProcessList = []*exec.Cmd{}
var mysqlctlProcessList []*exec.Cmd
for i := 0; i < totalTabletsRequired; i++ {
// instantiate vttablet object with reserved ports
tabletUID := cluster.GetAndReserveTabletUID()
Expand Down Expand Up @@ -1263,8 +1262,16 @@ func (cluster *LocalProcessCluster) VtprocessInstanceFromVttablet(tablet *Vttabl
}

// StartVttablet starts a new tablet
func (cluster *LocalProcessCluster) StartVttablet(tablet *Vttablet, servingStatus string,
supportBackup bool, cell string, keyspaceName string, hostname string, shardName string) error {
func (cluster *LocalProcessCluster) StartVttablet(
tablet *Vttablet,
explicitServingStatus bool,
servingStatus string,
supportBackup bool,
cell string,
keyspaceName string,
hostname string,
shardName string,
) error {
tablet.VttabletProcess = VttabletProcessInstance(
tablet.HTTPPort,
tablet.GrpcPort,
Expand All @@ -1282,6 +1289,7 @@ func (cluster *LocalProcessCluster) StartVttablet(tablet *Vttablet, servingStatu

tablet.VttabletProcess.SupportsBackup = supportBackup
tablet.VttabletProcess.ServingStatus = servingStatus
tablet.VttabletProcess.ExplicitServingStatus = explicitServingStatus
return tablet.VttabletProcess.Setup()
}

Expand Down
15 changes: 12 additions & 3 deletions go/test/endtoend/cluster/vttablet_process.go
Original file line number Diff line number Diff line change
Expand Up @@ -42,7 +42,7 @@ import (
"vitess.io/vitess/go/vt/sqlparser"
)

const vttabletStateTimeout = 30 * time.Second
const vttabletStateTimeout = 60 * time.Second

// VttabletProcess is a generic handle for a running vttablet .
// It can be spawned manually
Expand Down Expand Up @@ -71,6 +71,7 @@ type VttabletProcess struct {
QueryzURL string
StatusDetailsURL string
SupportsBackup bool
ExplicitServingStatus bool
ServingStatus string
DbPassword string
DbPort int
Expand All @@ -79,7 +80,7 @@ type VttabletProcess struct {
Charset string
ConsolidationsURL string

//Extra Args to be set before starting the vttablet process
// Extra Args to be set before starting the vttablet process
ExtraArgs []string

proc *exec.Cmd
Expand Down Expand Up @@ -150,7 +151,15 @@ func (vttablet *VttabletProcess) Setup() (err error) {
}()

if vttablet.ServingStatus != "" {
if err = vttablet.WaitForTabletStatus(vttablet.ServingStatus); err != nil {
// If the tablet has an explicit serving status we use the serving status
// otherwise we wait for any serving status to show up in the healthcheck.
var servingStatus []string
if vttablet.ExplicitServingStatus {
servingStatus = append(servingStatus, vttablet.ServingStatus)
} else {
servingStatus = append(servingStatus, "SERVING", "NOT_SERVING")
}
if err = vttablet.WaitForTabletStatuses(servingStatus); err != nil {
errFileContent, _ := os.ReadFile(fname)
if errFileContent != nil {
log.Infof("vttablet error:\n%s\n", string(errFileContent))
Expand Down
2 changes: 1 addition & 1 deletion go/test/endtoend/tabletmanager/custom_rule_topo_test.go
Original file line number Diff line number Diff line change
Expand Up @@ -71,7 +71,7 @@ func TestTopoCustomRule(t *testing.T) {
require.Nil(t, err, "error should be Nil")

// Start Vttablet
err = clusterInstance.StartVttablet(rTablet, "SERVING", false, cell, keyspaceName, hostname, shardName)
err = clusterInstance.StartVttablet(rTablet, false, "SERVING", false, cell, keyspaceName, hostname, shardName)
require.Nil(t, err, "error should be Nil")

err = clusterInstance.VtctlclientProcess.ExecuteCommand("Validate")
Expand Down
2 changes: 1 addition & 1 deletion go/test/endtoend/tabletmanager/primary/tablet_test.go
Original file line number Diff line number Diff line change
Expand Up @@ -189,7 +189,7 @@ func TestPrimaryRestartSetsPTSTimestamp(t *testing.T) {
require.NoError(t, err)

// Start Vttablet
err = clusterInstance.StartVttablet(&replicaTablet, "SERVING", false, cell, keyspaceName, hostname, shardName)
err = clusterInstance.StartVttablet(&replicaTablet, false, "SERVING", false, cell, keyspaceName, hostname, shardName)
require.NoError(t, err)

// Make sure that the PTS did not change
Expand Down
8 changes: 4 additions & 4 deletions go/test/endtoend/tabletmanager/tablet_health_test.go
Original file line number Diff line number Diff line change
Expand Up @@ -73,7 +73,7 @@ func TestTabletReshuffle(t *testing.T) {

// SupportsBackup=False prevents vttablet from trying to restore
// Start vttablet process
err = clusterInstance.StartVttablet(rTablet, "SERVING", false, cell, keyspaceName, hostname, shardName)
err = clusterInstance.StartVttablet(rTablet, false, "SERVING", false, cell, keyspaceName, hostname, shardName)
require.NoError(t, err)

sql := "select value from t1"
Expand Down Expand Up @@ -106,7 +106,7 @@ func TestHealthCheck(t *testing.T) {
defer replicaConn.Close()

// start vttablet process, should be in SERVING state as we already have a primary
err = clusterInstance.StartVttablet(rTablet, "SERVING", false, cell, keyspaceName, hostname, shardName)
err = clusterInstance.StartVttablet(rTablet, true, "SERVING", false, cell, keyspaceName, hostname, shardName)
require.NoError(t, err)

conn, err := mysql.Connect(ctx, &primaryTabletParams)
Expand Down Expand Up @@ -227,7 +227,7 @@ func TestHealthCheckSchemaChangeSignal(t *testing.T) {
clusterInstance.VtTabletExtraArgs = oldArgs
}()
// start vttablet process, should be in SERVING state as we already have a primary.
err = clusterInstance.StartVttablet(tempTablet, "SERVING", false, cell, keyspaceName, hostname, shardName)
err = clusterInstance.StartVttablet(tempTablet, false, "SERVING", false, cell, keyspaceName, hostname, shardName)
require.NoError(t, err)

defer func() {
Expand Down Expand Up @@ -381,7 +381,7 @@ func TestHealthCheckDrainedStateDoesNotShutdownQueryService(t *testing.T) {
// - the second tablet will be set to 'drained' and we expect that
// - the query service won't be shutdown

//Wait if tablet is not in service state
// Wait if tablet is not in service state
defer cluster.PanicHandler(t)
clusterInstance.DisableVTOrcRecoveries(t)
defer clusterInstance.EnableVTOrcRecoveries(t)
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -39,7 +39,7 @@ func TestFallbackSecurityPolicy(t *testing.T) {

// Requesting an unregistered security_policy should fallback to deny-all.
clusterInstance.VtTabletExtraArgs = []string{"--security_policy", "bogus"}
err = clusterInstance.StartVttablet(mTablet, "SERVING", false, cell, keyspaceName, hostname, shardName)
err = clusterInstance.StartVttablet(mTablet, false, "SERVING", false, cell, keyspaceName, hostname, shardName)
require.NoError(t, err)

// It should deny ADMIN role.
Expand Down Expand Up @@ -94,7 +94,7 @@ func TestDenyAllSecurityPolicy(t *testing.T) {

// Requesting a deny-all security_policy.
clusterInstance.VtTabletExtraArgs = []string{"--security_policy", "deny-all"}
err = clusterInstance.StartVttablet(mTablet, "SERVING", false, cell, keyspaceName, hostname, shardName)
err = clusterInstance.StartVttablet(mTablet, false, "SERVING", false, cell, keyspaceName, hostname, shardName)
require.NoError(t, err)

// It should deny ADMIN role.
Expand Down Expand Up @@ -126,7 +126,7 @@ func TestReadOnlySecurityPolicy(t *testing.T) {

// Requesting a read-only security_policy.
clusterInstance.VtTabletExtraArgs = []string{"--security_policy", "read-only"}
err = clusterInstance.StartVttablet(mTablet, "SERVING", false, cell, keyspaceName, hostname, shardName)
err = clusterInstance.StartVttablet(mTablet, false, "SERVING", false, cell, keyspaceName, hostname, shardName)
require.NoError(t, err)

// It should deny ADMIN role.
Expand Down
4 changes: 2 additions & 2 deletions go/test/endtoend/tabletmanager/tablet_test.go
Original file line number Diff line number Diff line change
Expand Up @@ -43,7 +43,7 @@ func TestEnsureDB(t *testing.T) {

log.Info(fmt.Sprintf("Started vttablet %v", tablet))
// Start vttablet process as replica. It won't be able to serve because there's no db.
err = clusterInstance.StartVttablet(tablet, "NOT_SERVING", false, cell, "dbtest", hostname, "0")
err = clusterInstance.StartVttablet(tablet, false, "NOT_SERVING", false, cell, "dbtest", hostname, "0")
require.NoError(t, err)

// Make it the primary.
Expand Down Expand Up @@ -78,7 +78,7 @@ func TestResetReplicationParameters(t *testing.T) {

log.Info(fmt.Sprintf("Started vttablet %v", tablet))
// Start vttablet process as replica. It won't be able to serve because there's no db.
err = clusterInstance.StartVttablet(tablet, "NOT_SERVING", false, cell, "dbtest", hostname, "0")
err = clusterInstance.StartVttablet(tablet, false, "NOT_SERVING", false, cell, "dbtest", hostname, "0")
require.NoError(t, err)

// Set a replication source on the tablet and start replication
Expand Down
4 changes: 2 additions & 2 deletions go/test/endtoend/vtgate/queries/misc/misc_test.go
Original file line number Diff line number Diff line change
Expand Up @@ -301,6 +301,6 @@ func TestLeftJoinUsingUnsharded(t *testing.T) {
mcmp, closer := start(t)
defer closer()

utils.Exec(t, mcmp.VtConn, "insert into uks.unsharded(id1) values (1),(2),(3),(4),(5)")
utils.Exec(t, mcmp.VtConn, "select * from uks.unsharded as A left join uks.unsharded as B using(id1)")
utils.Exec(t, mcmp.VtConn, "insert /*vt+ QUERY_TIMEOUT_MS=1000 */ into uks.unsharded(id1) values (1),(2),(3),(4),(5)")
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Apparently the test setup defaults this to 100ms and test was failing after that time.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Correct, the query timeout is set to 100ms in the TestMain

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It's probably the select that is slow. the Insert is expected to be fast on the kind of small test data we use for tests. It's fine to leave the timeout for both queries, it won't hurt anything.

utils.Exec(t, mcmp.VtConn, "select /*vt+ QUERY_TIMEOUT_MS=1000 */ * from uks.unsharded as A left join uks.unsharded as B using(id1)")
}