physical: deflake TestCreateTenantFromReplicationUsingID

We occasionally see slow quiesces on the
`TestCreateTenantFromReplicationUsingID` test. This appears to be caused
by shutting down the server while the source tenant is still being
created. To avoid this, we wait until the SQL server on the source
tenant is ready before teardown.

Fixes: #155345

Release note: None

Author: kev-cao

pkg/crosscluster/physical/stream_ingestion_planning_test.go

@@ -13,6 +13,7 @@ import (
 	"github.com/cockroachdb/cockroach/pkg/crosscluster/replicationtestutils"
 	"github.com/cockroachdb/cockroach/pkg/jobs"
 	"github.com/cockroachdb/cockroach/pkg/security/username"
+	"github.com/cockroachdb/cockroach/pkg/testutils"
 	"github.com/cockroachdb/cockroach/pkg/testutils/serverutils"
 	"github.com/cockroachdb/cockroach/pkg/testutils/sqlutils"
 	"github.com/cockroachdb/cockroach/pkg/util/leaktest"
@@ -67,4 +68,16 @@ func TestCreateTenantFromReplicationUsingID(t *testing.T) {
 		t.Logf("starting replication [50]->[51]")
 		sqlB.Exec(t, "CREATE VIRTUAL CLUSTER [51] FROM REPLICATION OF 'cluster-50' ON $1", serverAURL.String())
 	})
+
+	// We wait for tenant 50's SQL server to be up and running before ending the
+	// test. This should help reduce flakes where the tenant is being setup when
+	// the stopper begins quiescing.
+	testutils.SucceedsSoon(t, func() error {
+		tenantConn, err := serverA.ApplicationLayer().SQLConnE(serverutils.DBName("cluster:cluster-50"))
+		if err != nil {
+			return err
+		}
+		defer tenantConn.Close()
+		return tenantConn.PingContext(ctx)
+	})
 }