add downtime calculation on alerts

- add timestamp on alert embeds
 - make Alert.post receive the worker instead of service info

elstat/adapters.py

@@ -114,7 +114,7 @@ class HttpAdapter(Adapter):
 
         if not succ:
             status_phrase = cls.get_phrase(resp.status)
-            err_str = f'HTTP Status - {resp.status} - {status_phrase}'
+            err_str = f'http status {resp.status} - {status_phrase}'
             return cls._construct(succ, latency, err_str)
 
         return cls._construct(succ, latency if succ else 0)

elstat/alerts.py

@@ -1,3 +1,4 @@
+import datetime
 import logging
 
 from aiohttp import ClientSession
@@ -12,20 +13,61 @@ class DiscordAlert:
         self.url = alert['url']
         self.session = ClientSession()
 
-    def _make_payload(self, service, status):
+    def _make_payload(self, worker, status: dict):
+        service = worker.service
+
         serv_name = service['name']
         is_up = status['status']
 
         err = status.get('error', 'No error provided')
         color = 0x00ff00 if is_up else 0xff0000
 
+        ts = status['timestamp'] / 1000
+        dt_iso = datetime.datetime.utcfromtimestamp(ts).isoformat()
         embed = {
             'title': serv_name,
-            'color': color
+            'color': color,
+            'timestamp': dt_iso
         }
 
         if not is_up:
-            embed['description'] = err
+            embed['description'] = f'error: {err}'
+        else:
+            # calculate downtime here
+
+            # first we need to find the work that had a success
+            # before the current one
+            conn = worker.manager.conn
+            cur = conn.cursor()
+
+            # find the last work that had a success
+            cur.execute(f"""
+            SELECT timestamp FROM {worker.name}
+            WHERE timestamp < ? AND status = true
+            ORDER BY timestamp DESC
+            LIMIT 1
+            """, (status['timestamp'], ))
+
+            row = cur.fetchone()
+            ts_success = row[0]
+
+            # now we fetch all the downtime after that ts_success
+            cur.execute(f"""
+            SELECT COUNT(*) FROM {worker.name}
+            WHERE timestamp > ? AND status = false
+            """, (ts_success,))
+
+            row = cur.fetchone()
+            count = row[0]
+
+            downtime_msec = count * worker.service['poll'] * 1000
+            downtime_sec = downtime_msec / 1000
+            downtime_min = round(downtime_sec / 60, 3)
+
+            embed['footer'] = {
+                'text': f'down for {downtime_min} minutes '
+                        f'({downtime_sec} seconds)'
+            }
 
         return {
             'content': '',

elstat/worker.py

@@ -50,6 +50,8 @@ class ServiceWorker:
         conn.commit()
 
         await self._dispatch_work(columns, timestamp, result)
+
+        result['timestamp'] = timestamp
         await self._check_alert(result)
 
     async def _dispatch_work(self, columns, timestamp: int, result: tuple):
@@ -96,7 +98,7 @@ class ServiceWorker:
                 self.log.error(f'alert not found: {alert!r}')
                 continue
 
-            await alert_obj.post(self.service, work)
+            await alert_obj.post(self, work)
 
     async def _work_loop(self):
         try: