Refactor of caching & data storage part 2 #118

server/handlers/api/addspot.py

@@ -8,7 +8,7 @@ import tornado
 from tornado import httputil
 from tornado.web import Application
 
-from core.config import ALLOW_SPOTTING, MAX_SPOT_AGE
+from core.config import ALLOW_SPOTTING
 from core.constants import UNKNOWN_BAND
 from core.lookup_helper import infer_band_from_freq
 from core.prometheus_metrics_handler import api_requests_counter

server/handlers/api/alerts.py

@@ -1,7 +1,6 @@
 import copy
 import logging
 from datetime import datetime
-from queue import Queue
 from typing import Any
 
 import pytz
@@ -11,12 +10,9 @@ from tornado import httputil
 from tornado.web import Application
 
 from core.prometheus_metrics_handler import api_requests_counter
-from core.utils import safe_json_dumps, empty_queue
+from core.utils import safe_json_dumps
 from data.lookup_credentials import extract_credentials
 
-SSE_HANDLER_MAX_QUEUE_SIZE = 100
-SSE_HANDLER_QUEUE_CHECK_INTERVAL = 5000
-
 
 class APIAlertsHandler(tornado.web.RequestHandler):
     """API request handler for /api/v1/alerts"""
@@ -73,16 +69,14 @@ class APIAlertsStreamHandler(tornado_eventsource.handler.EventSourceHandler):
     """API request handler for /api/v1/alerts/stream"""
 
     def __init__(self, application, request, **kwargs: Any):
-        self._sse_alert_queues = None
+        self._sse_alert_broadcaster = None
         self._web_server_metrics = None
         self._query_params = None
         self._credentials = None
-        self._alert_queue = None
-        self._heartbeat = None
         super().__init__(application, request, **kwargs)
 
-    def initialize(self, sse_alert_queues, web_server_metrics):
-        self._sse_alert_queues = sse_alert_queues
+    def initialize(self, _sse_alert_broadcaster, web_server_metrics):
+        self._sse_alert_broadcaster = _sse_alert_broadcaster
         self._web_server_metrics = web_server_metrics
 
     def custom_headers(self):
@@ -104,59 +98,32 @@ class APIAlertsStreamHandler(tornado_eventsource.handler.EventSourceHandler):
             self._query_params = {k: v[0].decode("utf-8") for k, v in self.request.arguments.items()}
             self._credentials = extract_credentials(self._query_params)
 
-            # Create a alert queue and add it to the web server's list. The web server will fill this when alerts arrive
-            self._alert_queue = Queue(maxsize=SSE_HANDLER_MAX_QUEUE_SIZE)
-            self._sse_alert_queues.append(self._alert_queue)
-
-            # Set up a timed callback to check if anything is in the queue
-            self._heartbeat = tornado.ioloop.PeriodicCallback(self._callback, SSE_HANDLER_QUEUE_CHECK_INTERVAL)
-            self._heartbeat.start()
-
             # Flush headers immediately so nginx doesn't time out waiting for a response
             self.write_message("keepalive", "")
 
+            # Register to handle new alerts arriving. The callback() method will get called with the new alert as an
+            # argument.
+            self._sse_alert_broadcaster.register(self)
+
         except Exception as e:
             logging.warning("Exception when serving SSE socket: %s", e, exc_info=True)
             self.close()
 
     def close(self):
-        """When the user closes the socket, empty our queue and remove it from the list so the server no longer fills it"""
+        """When the user closes the socket, deregister ourselves from the alert broadcaster"""
 
-        try:
-            if self._alert_queue in self._sse_alert_queues:
-                self._sse_alert_queues.remove(self._alert_queue)
-            empty_queue(self._alert_queue)
-        except:
-            pass
-        try:
-            self._heartbeat.stop()
-        except:
-            pass
-        self._alert_queue = None
+        self._sse_alert_broadcaster.unregister(self)
         super().close()
 
-    def _callback(self):
-        """Callback to check if anything has arrived in the queue, and if so send it to the client"""
+    def callback(self, alert):
+        """Callback when a new alert arrives"""
 
         try:
-            if self._alert_queue:
-                if not self._alert_queue.empty():
-                    while not self._alert_queue.empty():
-                        alert = self._alert_queue.get()
-                        # If the new alert matches our param filters, send it to the client. If not, ignore it.
-                        if alert_allowed_by_query(alert, self._query_params):
-                            if self._credentials:
-                                alert = copy.deepcopy(alert)
-                                alert.infer_missing(self._credentials)
-                            self.write_message(msg=safe_json_dumps(alert))
-
-                else:
-                    # Send a keepalive comment if the queue was empty
-                    self.write_message("keepalive", "")
-
-                if self._alert_queue not in self._sse_alert_queues:
-                    logging.error("Web server cleared up a queue of an active connection!")
-                    self.close()
+            if alert_allowed_by_query(alert, self._query_params):
+                if self._credentials:
+                    alert = copy.deepcopy(alert)
+                    alert.infer_missing(self._credentials)
+                self.write_message(msg=safe_json_dumps(alert))
         except Exception as e:
             logging.warning("Exception in SSE callback, connection will be closed: %s", e, exc_info=True)
             self.close()
@@ -169,7 +136,7 @@ def get_alert_list_with_filters(all_alerts, query):
     # Create a shallow copy of the alert list ordered by start time, then filter the list to reduce it only to alerts
     # that match the filter parameters in the query string. Finally, apply a limit to the number of alerts returned.
     # The list of query string filters is defined in the API docs.
-    alert_ids = list(all_alerts.iterkeys())
+    alert_ids = all_alerts.keys()
     alerts = []
     for k in alert_ids:
         a = all_alerts.get(k)

server/handlers/api/dxstats.py

@@ -40,7 +40,7 @@ class APIDxStatsHandler(tornado.web.RequestHandler):
             one_hour_ago = (datetime.now(pytz.UTC) - timedelta(hours=1)).timestamp()
             counts = Counter()
 
-            for key in self._spots.iterkeys():
+            for key in self._spots.keys():
                 spot = self._spots.get(key)
                 if spot is None:
                     continue

server/handlers/api/spots.py

@@ -1,7 +1,6 @@
 import copy
 import logging
 from datetime import datetime, timedelta
-from queue import Queue
 from typing import Any
 
 import pytz
@@ -11,12 +10,9 @@ from tornado import httputil
 from tornado.web import Application
 
 from core.prometheus_metrics_handler import api_requests_counter
-from core.utils import safe_json_dumps, empty_queue
+from core.utils import safe_json_dumps
 from data.lookup_credentials import extract_credentials
 
-SSE_HANDLER_MAX_QUEUE_SIZE = 1000
-SSE_HANDLER_QUEUE_CHECK_INTERVAL = 5000
-
 
 class APISpotsHandler(tornado.web.RequestHandler):
     """API request handler for /api/v1/spots"""
@@ -73,16 +69,14 @@ class APISpotsStreamHandler(tornado_eventsource.handler.EventSourceHandler):
     """API request handler for /api/v1/spots/stream"""
 
     def __init__(self, application, request, **kwargs: Any):
-        self._sse_spot_queues = None
+        self._sse_spot_broadcaster = None
         self._web_server_metrics = None
         self._query_params = None
         self._credentials = None
-        self._spot_queue = None
-        self._heartbeat = None
         super().__init__(application, request, **kwargs)
 
-    def initialize(self, sse_spot_queues, web_server_metrics):
-        self._sse_spot_queues = sse_spot_queues
+    def initialize(self, sse_spot_broadcaster, web_server_metrics):
+        self._sse_spot_broadcaster = sse_spot_broadcaster
         self._web_server_metrics = web_server_metrics
 
     def custom_headers(self):
@@ -106,59 +100,33 @@ class APISpotsStreamHandler(tornado_eventsource.handler.EventSourceHandler):
             self._query_params = {k: v[0].decode("utf-8") for k, v in self.request.arguments.items()}
             self._credentials = extract_credentials(self._query_params)
 
-            # Create a spot queue and add it to the web server's list. The web server will fill this when spots arrive
-            self._spot_queue = Queue(maxsize=SSE_HANDLER_MAX_QUEUE_SIZE)
-            self._sse_spot_queues.append(self._spot_queue)
-
-            # Set up a timed callback to check if anything is in the queue
-            self._heartbeat = tornado.ioloop.PeriodicCallback(self._callback, SSE_HANDLER_QUEUE_CHECK_INTERVAL)
-            self._heartbeat.start()
-
             # Flush headers immediately so nginx doesn't time out waiting for a response
             self.write_message("keepalive", "")
 
+            # Register to handle new spots arriving. The callback() method will get called with the new spot as an
+            # argument.
+            self._sse_spot_broadcaster.register(self)
+
         except Exception as e:
             logging.warning("Exception when serving SSE socket: %s", e, exc_info=True)
             self.close()
 
     def close(self):
-        """When the user closes the socket, empty our queue and remove it from the list so the server no longer fills it"""
+        """When the user closes the socket, deregister ourselves from the spot broadcaster"""
 
-        try:
-            if self._spot_queue in self._sse_spot_queues:
-                self._sse_spot_queues.remove(self._spot_queue)
-            empty_queue(self._spot_queue)
-        except:
-            pass
-        try:
-            self._heartbeat.stop()
-        except:
-            pass
-        self._spot_queue = None
+        self._sse_spot_broadcaster.unregister(self)
         super().close()
 
-    def _callback(self):
-        """Callback to check if anything has arrived in the queue, and if so send it to the client"""
+    def callback(self, spot):
+        """Callback when a new spot arrives"""
 
         try:
-            if self._spot_queue:
-                if not self._spot_queue.empty():
-                    while not self._spot_queue.empty():
-                        spot = self._spot_queue.get()
-                        # If the new spot matches our param filters, send it to the client. If not, ignore it.
-                        if spot_allowed_by_query(spot, self._query_params):
-                            if self._credentials:
-                                spot = copy.deepcopy(spot)
-                                spot.infer_missing(self._credentials)
-                            self.write_message(msg=safe_json_dumps(spot))
-
-                else:
-                    # Send a keepalive comment if the queue was empty
-                    self.write_message("keepalive", "")
-
-                if self._spot_queue not in self._sse_spot_queues:
-                    logging.error("Web server cleared up a queue of an active connection!")
-                    self.close()
+            # If the new spot matches our param filters, send it to the client. If not, ignore it.
+            if spot_allowed_by_query(spot, self._query_params):
+                if self._credentials:
+                    spot = copy.deepcopy(spot)
+                    spot.infer_missing(self._credentials)
+                self.write_message(msg=safe_json_dumps(spot))
         except Exception as e:
             logging.warning("Exception in SSE callback, connection will be closed: %s", e, exc_info=True)
             self.close()
@@ -171,7 +139,7 @@ def get_spot_list_with_filters(all_spots, query):
     # Create a shallow copy of the spot list, ordered by spot time, then filter the list to reduce it only to spots
     # that match the filter parameters in the query string. Finally, apply a limit to the number of spots returned.
     # The list of query string filters is defined in the API docs.
-    spot_ids = list(all_spots.iterkeys())
+    spot_ids = all_spots.keys()
     spots = []
     for k in spot_ids:
         s = all_spots.get(k)

server/sse_broadcaster.py

@@ -0,0 +1,36 @@
+import logging
+import threading
+
+from tornado.ioloop import IOLoop
+
+
+class SSEBroadcaster:
+    """Bridge between DataStore listener callbacks (which fire on provider threads) to Tornado's async SSE handlers
+    (which live on the IOLoop thread) to avoid any interdependency between them."""
+
+    def __init__(self):
+        self._handlers = set()
+        self._lock = threading.Lock()
+        self._loop = IOLoop.current()
+
+    def register(self, handler):
+        with self._lock:
+            self._handlers.add(handler)
+
+    def unregister(self, handler):
+        with self._lock:
+            self._handlers.discard(handler)
+
+    def publish(self, value):
+        self._loop.add_callback(self._fan_out, value)
+
+    def _fan_out(self, value):
+        with self._lock:
+            handlers = list(self._handlers)
+        for handler in handlers:
+            try:
+                handler.callback(value)
+            except Exception:
+                # Connection probably dropped, ignore and de-register the handler to stop getting future items.
+                logging.debug("Failed to push to an SSE client; dropping it")
+                self.unregister(handler)

server/webserver.py

@@ -6,7 +6,7 @@ import tornado
 from tornado.web import StaticFileHandler
 
 from core.config import ALLOW_SPOTTING, WEB_SERVER_PORT, API_ONLY_MODE, LOG_WEB_REQUESTS, BASE_URL
-from core.utils import empty_queue
+from core.data_store import DATA_STORE
 from server.handlers.api.addspot import APISpotHandler
 from server.handlers.api.alerts import APIAlertsHandler, APIAlertsStreamHandler
 from server.handlers.api.dxstats import APIDxStatsHandler
@@ -18,6 +18,7 @@ from server.handlers.api.status import APIStatusHandler
 from server.handlers.manifesthandler import ManifestHandler
 from server.handlers.metrics import PrometheusMetricsHandler
 from server.handlers.pagetemplate import PageTemplateHandler
+from server.sse_broadcaster import SSEBroadcaster
 
 _HERE = os.path.dirname(__file__ or "")
 
@@ -25,12 +26,12 @@ _HERE = os.path.dirname(__file__ or "")
 class WebServer:
     """Provides the public-facing web server."""
 
-    def __init__(self, data_store):
+    def __init__(self):
         """Constructor"""
 
-        self._data_store = data_store
-        self._sse_spot_queues = []
-        self._sse_alert_queues = []
+        self._data_store = DATA_STORE
+        self._spot_broadcaster = SSEBroadcaster()
+        self._alert_broadcaster = SSEBroadcaster()
         self._port = WEB_SERVER_PORT
         self._api_only_mode = API_ONLY_MODE
         self._shutdown_event = asyncio.Event()
@@ -42,6 +43,10 @@ class WebServer:
             "status": "Starting"
         }
 
+        # Listen for new spots and alerts being added to the cache, so we can notify SSE clients immediately
+        DATA_STORE.spots.add_listener(self._spot_broadcaster.publish)
+        DATA_STORE.alerts.add_listener(self._alert_broadcaster.publish)
+
     def start(self):
         """Start the web server"""
 
@@ -64,13 +69,13 @@ class WebServer:
             (r"/api/v1/spots", APISpotsHandler, {"spots": self._data_store.spots, **handler_opts}),
             (r"/api/v1/alerts", APIAlertsHandler, {"alerts": self._data_store.alerts, **handler_opts}),
             (r"/api/v1/spots/stream", APISpotsStreamHandler,
-             {"sse_spot_queues": self._sse_spot_queues, **handler_opts}),
+             {"sse_spot_broadcaster": self._spot_broadcaster, **handler_opts}),
             (r"/api/v1/alerts/stream", APIAlertsStreamHandler,
-             {"sse_alert_queues": self._sse_alert_queues, **handler_opts}),
+             {"sse_alert_broadcaster": self._alert_broadcaster, **handler_opts}),
             (r"/api/v1/solar", APISolarConditionsHandler, {"solar_conditions": self._data_store.solar, **handler_opts}),
             (r"/api/v1/dxstats", APIDxStatsHandler, {"spots": self._data_store.spots, **handler_opts}),
-            (r"/api/v1/options", APIOptionsHandler, {"status_data": self._data_store.status, **handler_opts}),
-            (r"/api/v1/status", APIStatusHandler, {"status_data": self._data_store.status, **handler_opts}),
+            (r"/api/v1/options", APIOptionsHandler, {"status_data": self._data_store.status_data, **handler_opts}),
+            (r"/api/v1/status", APIStatusHandler, {"status_data": self._data_store.status_data, **handler_opts}),
             (r"/api/v1/lookup/call", APILookupCallHandler, {**handler_opts}),
             (r"/api/v1/lookup/sigref", APILookupSIGRefHandler, {**handler_opts}),
             (r"/api/v1/lookup/grid", APILookupGridHandler, {**handler_opts}),
@@ -118,55 +123,6 @@ class WebServer:
         logging.info("You can access your copy of Spothole at " + BASE_URL)
         await self._shutdown_event.wait()
 
-    def notify_new_spot(self, spot):
-        """Internal method called when a new spot is added to the system. This is used to ping any SSE clients that are
-        awaiting a server-sent message with new spots."""
-
-        for queue in self._sse_spot_queues:
-            try:
-                queue.put(spot)
-            except:
-                # Cleanup thread was probably deleting the queue, that's fine
-                pass
-        pass
-
-    def notify_new_alert(self, alert):
-        """Internal method called when a new alert is added to the system. This is used to ping any SSE clients that are
-        awaiting a server-sent message with new spots."""
-
-        for queue in self._sse_alert_queues:
-            try:
-                queue.put(alert)
-            except:
-                # Cleanup thread was probably deleting the queue, that's fine
-                pass
-        pass
-
-    def clean_up_sse_queues(self):
-        """Clean up any SSE queues that are growing too large; probably their client disconnected and we didn't catch it
-        properly for some reason."""
-
-        for q in self._sse_spot_queues:
-            try:
-                if q.full():
-                    logging.warning(
-                        "A full SSE spot queue was found, presumably because the client disconnected strangely. It has been removed.")
-                    self._sse_spot_queues.remove(q)
-                    empty_queue(q)
-            except:
-                # Probably got deleted already on another thread
-                pass
-        for q in self._sse_alert_queues:
-            try:
-                if q.full():
-                    logging.warning(
-                        "A full SSE alert queue was found, presumably because the client disconnected strangely. It has been removed.")
-                    self._sse_alert_queues.remove(q)
-                    empty_queue(q)
-            except:
-                # Probably got deleted already on another thread
-                pass
-        pass
 
 def request_log(handler):
     """Custom log function to provide more data about requests when enabled, and to provide the ability to turn off