Remove 'detail' dimension, fix a couple bugs

The 'detail' dimension in http_check is not actually an appropriate
dimension since it does not contain data which _identifies_ a metric.
This data is valuable, though, and will be captured and made available
in a future enhancement.

Bug fixes:
- Fix unix.py crash on EFI mountpoints which return 0 available inodes
- Rename 'service_name' to 'name' in nagios_wrapper for ServicesCheck
  compatibility
- Update nagios_wrapper YAML example for naming convention

Change-Id: I44877a856fac84f54ca764a79e75ace9f90d9302

conf.d/http_check.yaml.example

@@ -13,21 +13,10 @@ instances:
          # The (optional) match_pattern parameter will instruct the check
          # to match the HTTP response body against a regular-expression-
          # compatible pattern.  If the pattern matches the check will
-         # return 0 for OK.  Otherwise, it will return 1 for an error and
+         # return 0 for OK.  Otherwise, it will return 1 for an error
-         # include the response body as a 'detail' tag, as long as
-         # 'include_content' below is set to 'true'
 
 #        match_pattern: '.*OK.*OK.*OK.*OK.*OK'
 
-         # The (optional) include_content parameter will instruct the check
-         # to include the first 200 characters of the HTTP response body
-         # in notifications sent by this plugin. This is best used with
-         # "healthcheck"-type URLs, where the body contains a brief, human-
-         # readable summary of failure reasons in the case of errors. This
-         # defaults to false.
-
-#        include_content: true
-
          # The (optional) collect_response_time parameter will instruct the
          # check to create a metric 'network.http.response_time', tagged with
          # the url, reporting the response time in seconds.

conf.d/nagios_wrapper.yaml.example

@@ -5,19 +5,19 @@ init_config:
      # Where to store last-run timestamps for each check (required)
 #    temp_file_path:  /dev/shm/
 
-# List a check name under 'service_name' and the full command-line
+# List a check name under 'name' and the full command-line
 # under 'check_command'.  If the command exists in 'check_path' above,
 # it is not necessary to specify the full path.
 
 instances:
-#    - service_name:     load
+#    - name:             nagios.load
 #      check_command:    check_load -r -w 2,1.5,1 -c 10,5,4
 
-#    - service_name:     disk
+#    - name:             nagios.disk
 #      check_command:    check_disk -w 15\% -c 5\% -A -i /srv/node
 #      check_interval:   300
 
-#    - service_name:     swap
+#    - name:             nagios.swap
 #      check_command:    /usr/lib/nagios/plugins/check_swap -w 50\% -c 10\%
 #      check_interval:   120
 #      dimensions:       { 'group': 'memory' }

monagent/collector/checks/system/unix.py

@@ -114,6 +114,10 @@ class Disk(monagent.collector.checks.check.Check):
             except IndexError:
                 self.logger.exception("Cannot parse %s" % (parts,))
 
+            # Some partitions (EFI boot) may appear to have 0 available inodes
+            if parts[1] == 0:
+                continue
+
             if inodes:
                 usage_data['%s.disk_inode_utilization_perc' % parts[0]] = float(parts[2]) / parts[1] * 100
             else:

monagent/collector/checks_d/http_check.py

@@ -36,11 +36,10 @@ class HTTPCheck(ServicesCheck):
         pattern = instance.get('match_pattern', None)
         if url is None:
             raise Exception("Bad configuration. You must specify a url")
-        include_content = instance.get('include_content', False)
         ssl = instance.get('disable_ssl_validation', True)
         token = AgentCheck.keystone.get_token()
 
-        return url, username, password, timeout, include_content, headers, response_time, dimensions, ssl, pattern, use_keystone, token
+        return url, username, password, timeout, headers, response_time, dimensions, ssl, pattern, use_keystone, token
 
     def _create_status_event(self, status, msg, instance):
         """Does nothing: status events are not yet supported by Mon API.
@@ -49,7 +48,7 @@ class HTTPCheck(ServicesCheck):
         return
 
     def _check(self, instance):
-        addr, username, password, timeout, include_content, headers, response_time, dimensions, disable_ssl_validation, pattern, use_keystone, token = self._load_conf(
+        addr, username, password, timeout, headers, response_time, dimensions, disable_ssl_validation, pattern, use_keystone, token = self._load_conf(
             instance)
 
         content = ''
@@ -125,9 +124,7 @@ class HTTPCheck(ServicesCheck):
                 running_time = time.time() - start
                 self.gauge('http_response_time', running_time, dimensions=new_dimensions)
 
-            # Add a 'detail' tag if requested
+            # TODO(dschroeder): Save/send content data when supported by API
-            if include_content:
-                new_dimensions['detail'] = json.dumps(content)
 
             if int(resp.status) >= 400:
                 if use_keystone and int(resp.status) == 401:

monagent/collector/checks_d/nagios_wrapper.py

@@ -1,5 +1,5 @@
 #!/bin/env python
-"""Monitoring Agent wrapper for Nagios checks.
+"""Monasca Agent wrapper for Nagios checks.
 
 """
 
@@ -26,12 +26,11 @@ class WrapNagios(ServicesCheck):
     @staticmethod
     def _do_skip_check(instance, last_run_data):
         """Determine whether or not to skip a check depending on
-
         the checks's check_interval, if specified, and the last
         time the check was run
         """
-        if instance['service_name'] in last_run_data and 'check_interval' in instance:
+        if instance['name'] in last_run_data and 'check_interval' in instance:
-            if time.time() < last_run_data[instance['service_name']] + instance['check_interval']:
+            if time.time() < last_run_data[instance['name']] + instance['check_interval']:
                 return True
         else:
             return False
@@ -70,7 +69,7 @@ class WrapNagios(ServicesCheck):
             last_run_path +
             'nagios_wrapper_' +
             hashlib.md5(
-                instance['service_name']).hexdigest() +
+                instance['name']).hexdigest() +
             '.pck')
 
         # Load last-run data from shared memory file
@@ -92,22 +91,20 @@ class WrapNagios(ServicesCheck):
             output = proc.communicate()
             # The check detail is all the text before the pipe
             detail = output[0].split('|')[0]
-            if detail != '':
+            # TODO(dschroeder): Save/send 'detail' when supported by the API
-                # Serialize the output for JSON-friendliness and add to the dimensions
-                dimensions['detail'] = json.dumps(detail)
         except OSError:
             # Return an UNKNOWN code (3) if I have landed here
-            self.gauge(instance['service_name'], 3, dimensions=dimensions)
+            self.gauge(instance['name'], 3, dimensions=dimensions)
             self.log.info(instance['check_command'].split(" ")[0] + " is missing or unreadable")
             return
 
         status_code = proc.poll()
-        last_run_data[instance['service_name']] = time.time()
+        last_run_data[instance['name']] = time.time()
-        self.gauge(instance['service_name'], status_code, dimensions=dimensions)
+        self.gauge(instance['name'], status_code, dimensions=dimensions)
         # Return DOWN on critical, UP otherwise
         if status_code == "2":
-            return Status.DOWN, "DOWN: " + dimensions['detail']
+            return Status.DOWN, "DOWN: {}".format(detail)
-        return Status.UP, "UP: " + dimensions['detail']
+        return Status.UP, "UP: {}".format(detail)
 
         # Save last-run data
         file_w = open(last_run_file, "w")