add an easy way, good integration with Prometheus and Gratana to monitor multi-instance applications

I'm trying to implement a very simple rest API with just one endpoint using shelf initializing multiple instances and I want to be able to integrate with Prometheus and Gratana to display a graph of requests per second and/or total requests per day, I made this implementation using the stream_isolate package and prometheus_client , I would like to know if there is a simpler and cleaner way to do this
import 'dart:async';
import 'dart:convert';

import 'dart:isolate';
import 'package:eloquent/eloquent.dart';
import 'package:shelf/shelf.dart';
import 'package:shelf_router/shelf_router.dart';
import 'package:shelf/shelf_io.dart' as io;
import 'package:stack_trace/stack_trace.dart';
import 'package:new_sali_backend/src/db/db_layer.dart';
import 'package:new_sali_backend/src/modules/protocolo/repositories/processo_repository.dart';
import 'package:new_sali_core/src/utils/core_utils.dart';
import 'package:new_sali_core/src/models/status_message.dart';
import 'package:prometheus_client/prometheus_client.dart';
import 'package:prometheus_client/runtime_metrics.dart' as runtime_metrics;
import 'package:prometheus_client_shelf/shelf_metrics.dart' as shelf_metrics;
import 'package:prometheus_client/format.dart' as format;
import 'package:args/args.dart' show ArgParser;
import 'shelf_cors_headers_base.dart';
import 'stream_isolate.dart';

// to test
// xargs -I % -P 8 curl "http:/192.168.66.123:3161/api/v1/protocolo/processos/public/site/2023/10" < <(printf '%s\n' {1..400})

const defaultHeaders = {'Content-Type': 'application/json;charset=utf-8'};

Response responseError(String message,
    {dynamic exception, dynamic stackTrace, int statusCode = 400}) {
  final v = jsonEncode({
    'is_error': true,
    'status_code': statusCode,
    'message': message,
    'exception': exception?.toString(),
    'stackTrace': stackTrace?.toString()
  });
  return Response(statusCode, body: v, headers: defaultHeaders);
}

final basePath = '/api/v1';
final streamIsolates = <Map<int, BidirectionalStreamIsolate>>[];
void main(List<String> args) async {
  final parser = new ArgParser()
    ..addOption('address', abbr: 'a', defaultsTo: '0.0.0.0')
    ..addOption('port', abbr: 'p', defaultsTo: '3161')
    ..addOption('isolates', abbr: 'i', defaultsTo: '3');

  final argsParsed = parser.parse(args);

  final arguments = [argsParsed['address'], int.parse(argsParsed['port'])];

  final numberOfIsolates = int.parse(argsParsed['isolates']);
  for (var i = 0; i < numberOfIsolates - 1; i++) {
    final streamIsolate = await StreamIsolate.spawnBidirectional(isolateMain,
        debugName: i.toString(), argument: [i, ...arguments]);
    streamIsolates.add({i: streamIsolate});
    streamIsolate.stream.listen((event) => receiveAndPass(event, i));
  }
}

/// receive msg from isolate and send to all isolates
void receiveAndPass(event, int idx) {
  streamIsolates.forEach((item) {
    item.values.first.send(event);
  });
}

Stream isolateMain(Stream inc, dynamic args) {
  final arguments = args as List;
  int id = arguments[0];
  String address = arguments[1];
  int port = arguments[2];

  final streamController = StreamController.broadcast();

  final reg = CollectorRegistry(); //CollectorRegistry.defaultRegistry;
  // Register default runtime metrics
  runtime_metrics.register(reg);
  // Register http requests total
  final http_requests_total = Counter(
      name: 'http_requests_total', help: 'Total number of http api requests');
  http_requests_total.register(reg);
  // listen msg from main
  inc.listen((msg) {
    http_requests_total.inc();
  });

  _startServer([id, streamController, reg, address, port]);
  return streamController.stream;
}

void _startServer(List args) async {

  final streamController = args[1] as StreamController;
  final reg = args[2] as CollectorRegistry;
  String address = args[3];
  int port = args[4];

  final app = Router();
  routes(app, reg);

  final handler = Pipeline()
      .addMiddleware(corsHeaders())
      .addMiddleware(shelf_metrics.register(reg))
      .addMiddleware((innerHandler) {
        return (request) async {
          // Every time http_request is called, increase the counter by one
          final resp = await innerHandler(request);
          if (!request.url.path.contains('metrics')) {
            //send msg to main
            streamController.add('+1');
          }
          return resp;
        };
      })
      .addMiddleware(logRequestsCustom())
      .addHandler(app);

  final server = await io.serve(handler, address, port, shared: true);
  server.defaultResponseHeaders.remove('X-Frame-Options', 'SAMEORIGIN');

  print('Serving at http://${server.address.host}:${server.port}');
}

void routes(Router app, CollectorRegistry reg) {
  // Register a handler to expose the metrics in the Prometheus text format
  app.get('/metrics', (Request request) async {
    final buffer = StringBuffer();
    final metrics = await reg.collectMetricFamilySamples();
    format.write004(buffer, metrics);
    return Response.ok(
      buffer.toString(),
      headers: {'Content-Type': format.contentType},
    );
  });

  app.get('$basePath/protocolo/processos/public/site/<ano>/<codigo>',
      (Request request, String ano, String codigo) async {
    //final key = request.headers['Authorization'];
    Connection? conn;
    try {
      final codProcesso = int.tryParse(codigo);
      if (codProcesso == null) {
        return responseError('codProcesso invalido');
      }
      final anoExercicio = ano;
      conn = await DBLayer().connect();
      final procRepo = ProcessoRepository(conn);
      final proc =
          await procRepo.getProcessoByCodigoPublic(codProcesso, anoExercicio);
      await conn.disconnect();
      return Response.ok(
        jsonEncode(proc, toEncodable: SaliCoreUtils.customJsonEncode),
        headers: defaultHeaders,
      );
    } catch (e, s) {
      await conn?.disconnect();
      print('public_backend@getProcessoByCodigoPublic $e $s');
      return responseError(StatusMessage.ERROR_GENERIC);
    }
  });
}

Middleware logRequestsCustom(
        {void Function(String message, bool isError)? logger}) =>
    (innerHandler) {
      final theLogger = logger ?? _defaultLogger;
      return (request) {
        var startTime = DateTime.now();
        var watch = Stopwatch()..start();
        return Future.sync(() => innerHandler(request)).then((response) {
          var msg = _message(startTime, response.statusCode,
              request.requestedUri, request.method, watch.elapsed);
          theLogger(msg, false);
          return response;
        }, onError: (Object error, StackTrace stackTrace) {
          if (error is HijackException) throw error;
          var msg = _errorMessage(startTime, request.requestedUri,
              request.method, watch.elapsed, error, stackTrace);
          theLogger(msg, true);
          // ignore: only_throw_errors
          throw error;
        });
      };
    };

String _formatQuery(String query) {
  return query == '' ? '' : '?$query';
}

String _message(DateTime requestTime, int statusCode, Uri requestedUri,
    String method, Duration elapsedTime) {
  return '${requestTime.toIso8601String()} '
      '${elapsedTime.toString().padLeft(15)} '
      '${method.padRight(7)} [$statusCode] ' // 7 - longest standard HTTP method
      '${requestedUri.path}${_formatQuery(requestedUri.query)}'
      '  isolate: ${Isolate.current.debugName}';
}

String _errorMessage(DateTime requestTime, Uri requestedUri, String method,
    Duration elapsedTime, Object error, StackTrace? stack) {
  var chain = Chain.current();
  if (stack != null) {
    chain = Chain.forTrace(stack)
        .foldFrames((frame) => frame.isCore || frame.package == 'shelf')
        .terse;
  }

  var msg = '$requestTime\t$elapsedTime\t$method\t${requestedUri.path}'
      '${_formatQuery(requestedUri.query)}\n$error';

  return '$msg\n$chain';
}

void _defaultLogger(String msg, bool isError) {
  if (isError) {
    print('[ERROR] $msg');
  } else {
    print(msg);
  }
}
dart-lang / shelf

add an easy way, good integration with Prometheus and Gratana to monitor multi-instance applications #390
