new change

Author: moizpgedge

clustertest/add_remove_host_test.go

@@ -154,7 +154,7 @@ func TestForcedHostRemovalWithDatabase(t *testing.T) {
 	cluster.Host("host-3").Stop(t)
 
 	// Allow cluster some time to detect the container termination
-	time.Sleep(5 * time.Second)
+	time.Sleep(15 * time.Second)
 
 	tLog(t, "removing host-3 from cluster")
 	resp, err := cluster.Client().RemoveHost(ctx, &api.RemoveHostPayload{
@@ -175,7 +175,7 @@ func TestForcedHostRemovalWithDatabase(t *testing.T) {
 	require.NoError(t, err, "database not healthy with 2 nodes")
 
 	tLog(t, "verifying cluster has 2 hosts")
-	err = waitForHostCount(ctx, cluster.Client(), 2, 30*time.Second)
+	err = waitForHostCount(ctx, cluster.Client(), 2, 60*time.Second)
 	require.NoError(t, err, "cluster should have 2 hosts")
 
 	// Ensure host-3 is not in the list

e2e/database_test.go

@@ -405,7 +405,7 @@ func (f *DatabaseFixture) WaitForReplication(ctx context.Context, t testing.TB,
 			f.WithConnection(ctx, primaryOpts, t, func(conn *pgx.Conn) {
 				var synced bool
 
-				row := conn.QueryRow(ctx, "CALL spock.wait_for_sync_event(true, $1, $2::pg_lsn, 30);", peerNode, lsn)
+				row := conn.QueryRow(ctx, "CALL spock.wait_for_sync_event(true, $1, $2::pg_lsn, 60);", peerNode, lsn)
 				require.NoError(t, row.Scan(&synced))
 				assert.True(t, synced)
 			})

e2e/db_create_test.go

@@ -145,12 +145,12 @@ func testCreateDB(t *testing.T, nodeCount int, deployReplicas bool) {
 				row := conn.QueryRow(ctx, "SELECT pg_current_wal_lsn() AS commit_lsn")
 				require.NoError(t, row.Scan(&commitLSN))
 
-				// Wait for up to 10 seconds and verify commit_lsn has replayed
+				// Wait for up to 20 seconds and verify commit_lsn has replayed
 				// This should prevent flaky tests
 				var replayLSN string
 				var hasReplayed bool
 
-				for i := 0; i < 10; i++ {
+				for i := 0; i < 20; i++ {
 					rows, err := conn.Query(ctx, "SELECT replay_lsn, (replay_lsn >= $1::pg_lsn) AS has_replayed FROM pg_stat_replication", commitLSN)
 					require.NoError(t, err)
 					defer rows.Close()

e2e/db_update_add_node_test.go

@@ -175,7 +175,7 @@ func verifyWALReplay(ctx context.Context, t *testing.T, conn *pgx.Conn) {
 	var commitLSN string
 	require.NoError(t, conn.QueryRow(ctx, "SELECT pg_current_wal_lsn()").Scan(&commitLSN))
 
-	for i := 0; i < 10; i++ {
+	for i := 0; i < 20; i++ {
 		rows, err := conn.Query(ctx, "SELECT (replay_lsn >= $1::pg_lsn) AS has_replayed FROM pg_stat_replication", commitLSN)
 		require.NoError(t, err)
 		defer rows.Close()

e2e/service_provisioning_test.go

@@ -645,6 +645,22 @@ func TestProvisionMCPServiceRecovery(t *testing.T) {
 	require.NoError(t, err, "GetDatabase should succeed after update")
 	assert.Equal(t, "available", db.State, "Database should be available after recovery update")
 
+	// Service instance creation can lag briefly behind the database
+	// reaching "available", so poll instead of asserting immediately.
+	if len(db.ServiceInstances) == 0 {
+		t.Log("Service instance not yet present, waiting for it to appear...")
+
+		instanceDeadline := time.Now().Add(1 * time.Minute)
+		for time.Now().Before(instanceDeadline) && len(db.ServiceInstances) == 0 {
+			time.Sleep(2 * time.Second)
+
+			db, err = fixture.Client.GetDatabase(ctx, &controlplane.GetDatabasePayload{
+				DatabaseID: dbID,
+			})
+			require.NoError(t, err, "Failed to refresh database")
+		}
+	}
+
 	// Service instance should now exist
 	require.NotNil(t, db.ServiceInstances, "ServiceInstances should not be nil after recovery")
 	require.Len(t, db.ServiceInstances, 1, "Expected 1 service instance after recovery")