Metric Submission: Custom Agent Check

Functions are used to submit metrics with a custom Agent check. Different functions are available depending on the metric type. Depending on the function used, the submission and actual metric type stored within Datadog might differ.

self.monotonic_count(name, value, tags=None, hostname=None, device_name=None)

self.count(name, value, tags=None, hostname=None, device_name=None)

self.gauge(name, value, tags=None, hostname=None, device_name=None)

self.rate(name, value, tags=None, hostname=None, device_name=None)

self.histogram(name, value, tags=None, hostname=None, device_name=None)

Tutorial

Follow the steps below to create a custom Agent check that sends all metric types periodically:

1. Create the directory metrics_example.d/ in the conf.d/ folder at the root of your Agent’s configuration directory.

2. In metrics_example.d/ folder, create an empty configuration file named metrics_example.yaml with the following content:

   instances: [{}]

3. Up one level from the conf.d/ folder, go to the checks.d/ folder. Create a custom check file named metrics_example.py with the content below:

   metrics_example.py

   Copy

       import random
   
       from datadog_checks.base import AgentCheck
   
       __version__ = "1.0.0"
   
       class MyClass(AgentCheck):
           def check(self, instance):
               self.count(
                   "example_metric.count",
                   2,
                   tags=["env:dev","metric_submission_type:count"],
               )
               self.count(
                   "example_metric.decrement",
                   -1,
                   tags=["env:dev","metric_submission_type:count"],
               )
               self.count(
                   "example_metric.increment",
                   1,
                   tags=["env:dev","metric_submission_type:count"],
               )
               self.rate(
                   "example_metric.rate",
                   1,
                   tags=["env:dev","metric_submission_type:rate"],
               )
               self.gauge(
                   "example_metric.gauge",
                   random.randint(0, 10),
                   tags=["env:dev","metric_submission_type:gauge"],
               )
               self.monotonic_count(
                   "example_metric.monotonic_count",
                   2,
                   tags=["env:dev","metric_submission_type:monotonic_count"],
               )
   
               # Calling the functions below twice simulates
               # several metrics submissions during one Agent run.
               self.histogram(
                   "example_metric.histogram",
                   random.randint(0, 10),
                   tags=["env:dev","metric_submission_type:histogram"],
               )
               self.histogram(
                   "example_metric.histogram",
                   random.randint(0, 10),
                   tags=["env:dev","metric_submission_type:histogram"],
               )
       

4. Restart the Agent.

5. Validate your custom check is running correctly with the Agent’s status subcommand. Look for metrics_example under the Checks section:

   =========
   Collector
   =========
   
     Running Checks
     ==============
   
       (...)
   
       metrics_example (1.0.0)
       -----------------------
         Instance ID: metrics_example:d884b5186b651429 [OK]
         Total Runs: 2
         Metric Samples: Last Run: 8, Total: 16
         Events: Last Run: 0, Total: 0
         Service Checks: Last Run: 0, Total: 0
         Average Execution Time : 2ms
   
       (...)

6. Verify your metrics are reporting to Datadog on your Metric Summary page:

Further Reading

Documentation, liens et articles supplémentaires utiles:

Write an Agent Custom CheckDOCUMENTATION