Send heartbeats immedately when pulling messages (#5)

This ensures there's some DB activity immediately after connecting to
stream WAL changes.  For some hosts, this is required.

pkg/replicator/pgreplicator/pg.go

@@ -66,7 +66,7 @@ func New(ctx context.Context, opts Opts) (PostgresReplicator, error) {
 	if opts.Log == nil {
 		opts.Log = slog.New(slog.NewJSONHandler(os.Stderr, &slog.HandlerOptions{
 			Level: slog.LevelInfo,
-		}))
+		})).With("host", opts.Config.Host)
 	}
 
 	cfg := opts.Config
@@ -256,6 +256,8 @@ func (p *pg) Pull(ctx context.Context, cc chan *changeset.Changeset) error {
 		return err
 	}
 
+	p.log.Debug("connected to replication slot")
+
 	// Postgres batches every individual insert, update, etc. within a BEGIN/COMMIT message.
 	// This is great for replication.  However, for Inngest events, we don't want superflous begin
 	// or commit messages as events.
@@ -268,64 +270,91 @@ func (p *pg) Pull(ctx context.Context, cc chan *changeset.Changeset) error {
 	go func() {
 		if p.version < pgconsts.MessagesVersion {
 			// doesn't support wal messages;  ignore.
+			p.log.Debug("heartbeat not supported", "pg_version", p.version)
 			return
 		}
 
+		// Send a hearbeat immediately.
+		if err := p.heartbeat(ctx); err != nil {
+			p.log.Warn("unable to emit immediate heartbeat", "error", err)
+		}
+
 		t := time.NewTicker(p.heartbeatTime)
-		for range t.C {
-			if p.queryConn.IsClosed() {
-				if err := p.createQueryConn(ctx); err != nil {
-					p.log.Error("error reconnecting for heartbeat", "error", err, "host", p.opts.Config.Host)
+		doneCheck := time.NewTicker(time.Second)
+		for {
+			select {
+			case <-ctx.Done():
+				return
+			case <-doneCheck.C:
+				// Check for the stopped signal internally every second. This lets us log
+				// the stopped message relatively close to the stop signal occurring.
+				if atomic.LoadInt32(&p.stopped) == 1 {
+					p.log.Debug("stopping heartbeat", "ctx_err", ctx.Err(), "stopped", atomic.LoadInt32(&p.stopped))
 					return
 				}
-			}
-
-			if atomic.LoadInt32(&p.stopped) == 1 || ctx.Err() != nil {
-				return
-			}
-
-			// Send a hearbeat every minute
-			p.queryLock.Lock()
-			_, err := p.queryConn.Exec(ctx, "SELECT pg_logical_emit_message(false, 'heartbeat', now()::varchar);")
-			p.queryLock.Unlock()
-
-			if isConnClosedErr(err) && p.queryConn.IsClosed() {
 				continue
-			}
+			case <-t.C:
+				if p.queryConn.IsClosed() {
+					if err := p.createQueryConn(ctx); err != nil {
+						p.log.Error("error reconnecting for heartbeat", "error", err)
+						return
+					}
+				}
+
+				// Send a hearbeat every minute
+				err := p.heartbeat(ctx)
+				if err != nil {
+					p.log.Warn("unable to emit heartbeat", "error", err)
+					continue
+				}
 
-			if err != nil {
-				p.log.Warn("unable to emit heartbeat", "error", err, "host", p.opts.Config.Host)
+				p.log.Debug("sent heartbeat", "error", err)
 			}
 		}
 	}()
 
 	for {
 		if ctx.Err() != nil || atomic.LoadInt32(&p.stopped) == 1 || p.conn.IsClosed() {
 			// Always call Close automatically.
+			p.log.Debug("stopping cdc connection", "conn_closed", p.conn.IsClosed())
 			p.Close(ctx)
 			return nil
 		}
 
 		changes, err := p.fetch(ctx)
 		if err != nil {
+			p.log.Warn("error pulling messages", "error", err)
 			return err
 		}
 		if changes == nil {
+			p.log.Debug("no messages pulled")
 			continue
 		}
 
 		if changes.Operation == changeset.OperationHeartbeat {
+			p.log.Debug("heartbeat pulled")
 			p.Commit(changes.Watermark)
 			if err := p.forceNextReport(ctx); err != nil {
 				p.log.Warn("unable to report lsn on heartbeat", "error", err, "host", p.opts.Config.Host)
 			}
 			continue
 		}
 
+		p.log.Debug("message pulled", "op", changes.Operation)
+
 		unwrapper.Process(changes)
 	}
 }
 
+func (p *pg) heartbeat(ctx context.Context) error {
+	// Send a hearbeat every minute
+	p.queryLock.Lock()
+	_, err := p.queryConn.Exec(ctx, "SELECT pg_logical_emit_message(false, 'heartbeat', now()::varchar);")
+	p.queryLock.Unlock()
+	return err
+
+}
+
 func (p *pg) fetch(ctx context.Context) (*changeset.Changeset, error) {
 	var err error
 
@@ -457,6 +486,9 @@ func (p *pg) report(ctx context.Context, forceReply bool) error {
 	if lsn == 0 {
 		return nil
 	}
+
+	p.log.Debug("reporting lsn to source", "lsn", p.LSN().String())
+
 	err := pglogrepl.SendStandbyStatusUpdate(ctx,
 		p.conn.PgConn(),
 		pglogrepl.StandbyStatusUpdate{
@@ -533,6 +565,6 @@ func standardizeErr(err error) (bool, error) {
 	return false, err
 }
 
-func isConnClosedErr(err error) bool {
+func IsConnClosedErr(err error) bool {
 	return err != nil && strings.Contains(err.Error(), "conn closed")
 }