#38 - Adding support to milliseconds and different units for latency and changing http status to a range, instead of a single value.

README.md

@@ -23,7 +23,7 @@ endpoint:
   timeout: 1 # seconds
   expectation:
     - type: HTTP_STATUS
-      status: 200
+      status_range: 200-300
     - type: LATENCY
       threshold: 1
     - type: REGEX
@@ -38,6 +38,7 @@ cachet:
     - UPDATE_STATUS
   public_incidents: true
 frequency: 30
+latency_unit: ms
 ```
 
 - **endpoint**, the configuration about the URL that will be monitored.
@@ -45,7 +46,7 @@ frequency: 30
     - **method**, the HTTP method that will be used by the monitor.
     - **timeout**, how long we'll wait to consider the request failed. The unit of it is seconds.
     - **expectation**, the list of expectations set for the URL.
-        - **HTTP_STATUS**, we will verify if the response status code matches what we expect.
+        - **HTTP_STATUS**, we will verify if the response status code falls into the expected range. Please keep in mind the range is inclusive on the first number and exclusive on the second number. If just one value is specified, it will default to only the given value, for example `200` will be converted to `200-201`. 
         - **LATENCY**, we measure how long the request took to get a response and fail if it's above the threshold. The unit is in seconds.
         - **REGEX**, we verify if the response body matches the given regex.
 - **cachet**, this is the settings for our cachet server.
@@ -58,6 +59,7 @@ frequency: 30
         - **UPDATE_STATUS**, updates the component status
     - **public_incidents**, boolean to decide if created incidents should be visible to everyone or only to logged in users. Important only if `CREATE_INCIDENT` or `UPDATE_STATUS` are set.
 - **frequency**, how often we'll send a request to the given URL. The unit is in seconds.
+- **latency_unit**, the latency unit used when reporting the metrics. It will automatically convert to the specified unit. It's not mandatory and it will default to **seconds**. Available units: `ms`, `s`, `m`, `h`.
 
 ## Setting up
 

cachet_url_monitor/configuration.py

@@ -10,6 +10,7 @@ import requests
 from yaml import dump
 from yaml import load
 
+import latency_unit
 import status as st
 
 # This is the mandatory fields that must be in the configuration file in this
@@ -89,12 +90,15 @@ class Configuration(object):
         self.component_id = os.environ.get('CACHET_COMPONENT_ID') or self.data['cachet']['component_id']
         self.metric_id = os.environ.get('CACHET_METRIC_ID') or self.data['cachet'].get('metric_id')
         self.default_metric_value = self.get_default_metric_value()
+        # The latency_unit configuration is not mandatory and we fallback to seconds, by default.
+        self.latency_unit = os.environ.get('LATENCY_UNIT') or self.data['cachet'].get('latency_unit') or 's'
 
         # We need the current status so we monitor the status changes. This is necessary for creating incidents.
         self.status = get_current_status(self.api_url, self.component_id, self.headers)
 
         # Get remaining settings
-        self.public_incidents = int(os.environ.get('CACHET_PUBLIC_INCIDENTS') or self.data['cachet']['public_incidents'])
+        self.public_incidents = int(
+            os.environ.get('CACHET_PUBLIC_INCIDENTS') or self.data['cachet']['public_incidents'])
 
         self.logger.info('Monitoring URL: %s %s' % (self.endpoint_method, self.endpoint_url))
         self.expectations = [Expectaction.create(expectation) for expectation in self.data['endpoint']['expectation']]
@@ -207,7 +211,9 @@ class Configuration(object):
         In case of failed connection trial pushes the default metric value.
         """
         if 'metric_id' in self.data['cachet'] and hasattr(self, 'request'):
-            value = self.default_metric_value if self.status != 1 else self.request.elapsed.total_seconds()
+            # We convert the elapsed time from the request, in seconds, to the configured unit.
+            value = self.default_metric_value if self.status != 1 else latency_unit.convert_to_unit(self.latency_unit,
+                                                                                                    self.request.elapsed.total_seconds())
             params = {'id': self.metric_id, 'value': value,
                       'timestamp': self.current_timestamp}
             metrics_request = requests.post('%s/metrics/%d/points' % (self.api_url, self.metric_id), params=params,
@@ -226,7 +232,8 @@ class Configuration(object):
         """
         if hasattr(self, 'incident_id') and self.status == st.COMPONENT_STATUS_OPERATIONAL:
             # If the incident already exists, it means it was unhealthy but now it's healthy again.
-            params = {'status': 4, 'visible': self.public_incidents, 'component_id': self.component_id, 'component_status': self.status,
+            params = {'status': 4, 'visible': self.public_incidents, 'component_id': self.component_id,
+                      'component_status': self.status,
                       'notify': True}
 
             incident_request = requests.put('%s/incidents/%d' % (self.api_url, self.incident_id), params=params,
@@ -287,10 +294,19 @@ class Expectaction(object):
 
 class HttpStatus(Expectaction):
     def __init__(self, configuration):
-        self.status = configuration['status']
+        self.status_range = self.parse_range(configuration['status_range'])
+
+    def parse_range(self, range_string):
+        statuses = range_string.split("-")
+        if len(statuses) == 1:
+            # When there was no range given, we should treat the first number as a single status check.
+            return (int(statuses[0]), int(statuses[0]) + 1)
+        else:
+            # We shouldn't look into more than one value, as this is a range value.
+            return (int(statuses[0]), int(statuses[1]))
 
     def get_status(self, response):
-        if response.status_code == self.status:
+        if response.status_code >= self.status_range[0] and response.status_code < self.status_range[1]:
             return st.COMPONENT_STATUS_OPERATIONAL
         else:
             return st.COMPONENT_STATUS_PARTIAL_OUTAGE
@@ -299,7 +315,7 @@ class HttpStatus(Expectaction):
         return 'Unexpected HTTP status (%s)' % (response.status_code,)
 
     def __str__(self):
-        return repr('HTTP status: %s' % (self.status,))
+        return repr('HTTP status range: %s' % (self.status_range,))
 
 
 class Latency(Expectaction):

cachet_url_monitor/latency_unit.py

@@ -0,0 +1,16 @@
+#!/usr/bin/env python
+
+seconds_per_unit = {"ms": 1000, "milliseconds": 1000, "s": 1, "seconds": 1, "m": float(1) / 60,
+                    "minutes": float(1) / 60, "h": float(1) / 3600, "hours": float(1) / 3600}
+
+
+def convert_to_unit(time_unit, value):
+    """
+    Will convert the given value from seconds to the given time_unit.
+
+    :param time_unit: The time unit to which the value will be converted to, from seconds.
+    This is a string parameter. The unit must be in the short form.
+    :param value: The given value that will be converted. This value must be in seconds.
+    :return: The converted value.
+    """
+    return value * seconds_per_unit[time_unit]

cachet_url_monitor/status.py

@@ -1,4 +1,9 @@
 #!/usr/bin/env python
+"""
+This file defines all the different status different values.
+These are all constants and are coupled to cachet's API configuration.
+"""
+
 
 COMPONENT_STATUS_OPERATIONAL = 1
 COMPONENT_STATUS_PERFORMANCE_ISSUES = 2

config.yml

@@ -4,7 +4,7 @@ endpoint:
   timeout: 0.01
   expectation:
     - type: HTTP_STATUS
-      status: 200
+      status_range: 200-300
     - type: LATENCY
       threshold: 1
     - type: REGEX
@@ -19,3 +19,4 @@ cachet:
     - UPDATE_STATUS
   public_incidents: true
 frequency: 30
+latency_unit: ms

tests/test_configuration.py

@@ -2,10 +2,11 @@
 import sys
 import unittest
 
-import cachet_url_monitor.status
 import mock
 from requests import ConnectionError, HTTPError, Timeout
 
+import cachet_url_monitor.status
+
 sys.modules['requests'] = mock.Mock()
 sys.modules['logging'] = mock.Mock()
 from cachet_url_monitor.configuration import Configuration
@@ -38,7 +39,7 @@ class ConfigurationTest(unittest.TestCase):
         sys.modules['requests'].HTTPError = HTTPError
 
     def test_init(self):
-        self.assertEqual(len(self.configuration.data), 3, 'Configuration data size is incorrect')
+        self.assertEqual(len(self.configuration.data), 4, 'Number of root elements in config.yml is incorrect')
         self.assertEquals(len(self.configuration.expectations), 3, 'Number of expectations read from file is incorrect')
         self.assertDictEqual(self.configuration.headers, {'X-Cachet-Token': 'token2'}, 'Header was not set correctly')
         self.assertEquals(self.configuration.api_url, 'https://demo.cachethq.io/api/v1',

tests/test_expectation.py

@@ -1,8 +1,10 @@
 #!/usr/bin/env python
+import re
 import unittest
 
 import mock
-import re
+import pytest
+
 from cachet_url_monitor.configuration import HttpStatus, Regex
 from cachet_url_monitor.configuration import Latency
 
@@ -51,10 +53,21 @@ class LatencyTest(unittest.TestCase):
 
 class HttpStatusTest(unittest.TestCase):
     def setUp(self):
-        self.expectation = HttpStatus({'type': 'HTTP_STATUS', 'status': 200})
+        self.expectation = HttpStatus({'type': 'HTTP_STATUS', 'status_range': "200-300"})
 
     def test_init(self):
-        assert self.expectation.status == 200
+        assert self.expectation.status_range == (200, 300)
+
+    def test_init_with_one_status(self):
+        """With only one value, we still expect a valid tuple"""
+        self.expectation = HttpStatus({'type': 'HTTP_STATUS', 'status_range': "200"})
+
+        assert self.expectation.status_range == (200, 201)
+
+    def test_init_with_invalid_number(self):
+        """Invalid values should just fail with a ValueError, as we can't convert it to int."""
+        with pytest.raises(ValueError) as excinfo:
+            self.expectation = HttpStatus({'type': 'HTTP_STATUS', 'status_range': "foo"})
 
     def test_get_status_healthy(self):
         request = mock.Mock()

tests/test_latency_unit.py

@@ -0,0 +1,18 @@
+#!/usr/bin/env python
+import unittest
+
+from cachet_url_monitor.latency_unit import convert_to_unit
+
+
+class ConfigurationTest(unittest.TestCase):
+    def test_convert_to_unit_ms(self):
+        assert convert_to_unit("ms", 1) == 1000
+
+    def test_convert_to_unit_s(self):
+        assert convert_to_unit("s", 20) == 20
+
+    def test_convert_to_unit_m(self):
+        assert convert_to_unit("m", 3) == float(3) / 60
+
+    def test_convert_to_unit_h(self):
+        assert convert_to_unit("h", 7200) == 2