Checks now have a new status: "paused". sendalerts management command will mark checks as paused if sending a notification throws exception. This should avoid potential infinite loops of sendalerts crashes/respawns.

.gitignore

@@ -1,5 +1,6 @@
 __pycache__/
 *.pyc
 .coverage
-local_settings.py
+hc.sqlite
+hc/local_settings.py
 static-collected

hc/api/management/commands/sendalerts.py

@@ -1,17 +1,63 @@
+import logging
 import sys
 import time
 
 from django.core.management.base import BaseCommand
+from django.db.models import Q
 from django.utils import timezone
 
 from hc.api.models import Check
 
+logger = logging.getLogger(__name__)
 
-def _log(message):
+
+def _stdout(message):
     sys.stdout.write(message)
     sys.stdout.flush()
 
 
+def handle_one():
+    """ Send an alert for a single check.
+
+    Return True if an appropriate check was selected and processed.
+    Return False if no checks need to be processed.
+
+    """
+
+    query = Check.objects.filter(user__isnull=False)
+
+    now = timezone.now()
+    going_down = Q(alert_after__lt=now, status="up")
+    going_up = Q(alert_after__gt=now, status="down")
+    query = query.filter(going_down | going_up)
+
+    try:
+        check = query[0]
+    except IndexError:
+        return False
+
+    check.status = check.get_status()
+
+    tmpl = "\nSending alert, status=%s, code=%s\n"
+    _stdout(tmpl % (check.status, check.code))
+
+    try:
+        check.send_alert()
+    except:
+        # Catch EVERYTHING. If we crash here, what can happen is:
+        # - the sendalerts command will crash
+        # - supervisor will respawn sendalerts command
+        # - sendalerts will try same thing again, resulting in infinite loop
+        # So instead we catch and log all exceptions, and mark
+        # the checks as paused so they are not retried.
+        logger.error("Could not alert %s" % check.code, exc_info=True)
+        check.status = "paused"
+    finally:
+        check.save()
+
+    return True
+
+
 class Command(BaseCommand):
     help = 'Sends UP/DOWN email alerts'
 
@@ -19,36 +65,12 @@ class Command(BaseCommand):
 
         ticks = 0
         while True:
-            # Gone down?
+            success = True
-            query = Check.objects
+            while success:
-            query = query.filter(alert_after__lt=timezone.now())
+                success = handle_one()
-            query = query.filter(user__isnull=False)
+                ticks = 0 if success else ticks + 1
-            query = query.filter(status="up")
-            for check in query:
-                check.status = "down"
-
-                _log("\nSending notification(s) about going down for %s\n" % check.code)
-                check.send_alert()
-                ticks = 0
-
-                # Save status after the notification is sent
-                check.save()
-
-            # Gone up?
-            query = Check.objects
-            query = query.filter(alert_after__gt=timezone.now())
-            query = query.filter(user__isnull=False)
-            query = query.filter(status="down")
-            for check in query:
-                check.status = "up"
-
-                _log("\nSending notification(s) about going up for %s\n" % check.code)
-                check.send_alert()
-                ticks = 0
-
-                # Save status after the notification is sent
-                check.save()
 
             time.sleep(1)
-            ticks = (ticks + 1) % 80
+            _stdout(".")
-            _log("." + ("\n" if ticks == 0 else ""))
+            if ticks % 60 == 0:
+                _stdout("\n")

hc/api/migrations/0016_auto_20151030_1107.py

@@ -0,0 +1,19 @@
+# -*- coding: utf-8 -*-
+from __future__ import unicode_literals
+
+from django.db import models, migrations
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('api', '0015_auto_20151022_1008'),
+    ]
+
+    operations = [
+        migrations.AlterField(
+            model_name='check',
+            name='status',
+            field=models.CharField(default='new', max_length=6, choices=[('up', 'Up'), ('down', 'Down'), ('new', 'New'), ('paused', 'Paused')]),
+        ),
+    ]

hc/api/models.py

@@ -15,8 +15,12 @@ import requests
 
 from hc.lib import emails
 
-
+STATUSES = (
-STATUSES = (("up", "Up"), ("down", "Down"), ("new", "New"))
+    ("up", "Up"),
+    ("down", "Down"),
+    ("new", "New"),
+    ("paused", "Paused")
+)
 DEFAULT_TIMEOUT = td(days=1)
 DEFAULT_GRACE = td(hours=1)
 CHANNEL_KINDS = (("email", "Email"), ("webhook", "Webhook"),
@@ -60,8 +64,8 @@ class Check(models.Model):
             channel.notify(self)
 
     def get_status(self):
-        if self.status == "new":
+        if self.status in ("new", "paused"):
-            return "new"
+            return self.status
 
         now = timezone.now()
 

static/css/base.css

@@ -42,7 +42,7 @@ body {
     font-size: small;
 }
 
-.glyphicon.up, .glyphicon.new, .glyphicon.grace, .glyphicon.down {
+.glyphicon.up, .glyphicon.new, .glyphicon.paused, .glyphicon.grace, .glyphicon.down {
     font-size: 22px;
 }
 
@@ -50,7 +50,7 @@ body {
     color: #5cb85c;
 }
 
-.glyphicon.new {
+.glyphicon.new, .glyphicon.paused {
     color: #AAA;
 }
 

templates/emails/alert-body-html.html

@@ -19,6 +19,7 @@
     }
 
     .new { background: #AAA; }
+    .paused { background: #AAA; }
     .up { background: #5cb85c; }
     .grace { background: #f0ad4e; }
     .down { background: #d9534f; }
@@ -55,6 +56,8 @@
                 <span class="badge grace">LATE</span>
             {% elif check.get_status == "down" %}
                 <span class="badge down">DOWN</span>
+            {% elif check.get_status == "paused" %}
+                <span class="badge paused">PAUSED</span>
             {% endif %}
         </td>
         <td>

templates/front/my_checks_desktop.html

@@ -26,6 +26,8 @@
                 <span class="glyphicon glyphicon-exclamation-sign grace"></span>
             {% elif check.get_status == "down" %}
                 <span class="glyphicon glyphicon-exclamation-sign down"></span>
+            {% elif check.get_status == "paused" %}
+                <span class="glyphicon glyphicon-minus-sign paused"></span>
             {% endif %}
         </td>
         <td class="name-cell">

templates/front/my_checks_mobile.html

@@ -31,6 +31,8 @@
                         <span class="label label-warning">LATE</span>
                     {% elif check.get_status == "down" %}
                         <span class="label label-danger">DOWN</span>
+                    {% elif check.get_status == "paused" %}
+                        <span class="label label-default">PAUSED</span>
                     {% endif %}
                 </td>
             </tr>