Report user_id to New Relic for error tracking (Python 2.7) (#28)

* Report user_id to New Relic for error tracking

* Log user id reporting to NR

README.md

@@ -21,12 +21,16 @@ should subclass to implement your pure python jobs. It basically replicates
 the [delayed_job](https://github.com/collectiveidea/delayed_job) Ruby gem behavior.
 
 ## Installation
-### From Pypi:
+
+### From PyPI
+
     pip install rubydj-pyworker==0.2.0
 
-You can also pick up any version from [Pypi](https://pypi.org/project/rubydj-pyworker/)
+You can also pick up any version from [PyPI](https://pypi.org/project/rubydj-pyworker/)
 that supports Python 2.7 (< 1.0.0).
-### From Github branch:
+
+### From Github branch
+
     pip install git+https://github.com/rayyansys/pyworker.git@<branch_name>#egg=rubydj-pyworker
 
 ## Usage
@@ -97,14 +101,95 @@ configuration options:
 
 Youc an also provide a logger class (from `logging` module) to have full control on logging configuration:
 
-    import logging
+```python
+import logging
 
-    logging.basicConfig()
-    logger = logging.getLogger('pyworker')
-    logger.setLevel(logging.INFO)
+logging.basicConfig()
+logger = logging.getLogger('pyworker')
+logger.setLevel(logging.INFO)
 
-    w = Worker(dbstring, logger)
-    w.run()
+w = Worker(dbstring, logger)
+w.run()
+```
+
+## Monitoring
+
+Workers can be monitored using [New Relic](https://newrelic.com/). All you need
+to do is to create a free account there, then add the following to your
+environment variables:
+
+```bash
+NEW_RELIC_LICENSE_KEY=<your_newrelic_license_key>
+NEW_RELIC_APP_NAME=<your_newrelic_app_name>
+```
+
+All jobs will be reported under the `BackgroundTask` category. This includes
+standard metrics like throughput, response time (job duration), error rate, etc.
+Additional transaction custom attributes are also reported out of the box:
+
+1. `jobName`: the name of the job class
+1. `jobId`: the id of the delayed job in the database
+1. `jobQueue`: the queue name of the job
+1. `jobAttempts`: the number of attempts for the job
+1. `jobLatency`: the time between the job creation and the time it was picked up by the worker
+
+If you wish to automatically report additional attributes from the delayed job table
+(e.g. the priority of the job or any other custom fields), you can do so by
+providing a list of fields to the worker:
+
+```python
+worker = Worker(dbstring,
+    logger=logger,
+    extra_delayed_job_fields=['priority', 'custom_field'])
+```
+
+Columns of types `string`/`text`, `int`, `float` and `bool` are reported as is.
+`json`/`jsonb` types are expanded into separate attributes. All other fields are converted
+to JSON strings.
+
+You can also automatically prefix all attributes with a string before reporting:
+
+```python
+worker = Worker(dbstring,
+    logger=logger,
+    reported_attributes_prefix='myapp.')
+```
+
+If you wish to report additional custom attributes from your job, you can do so
+by calling the `reporter` object that is available in the job instance:
+
+```python
+class MyJob(Job):
+    def __init__(self, *args, **kwargs):
+        super(MyJob, self).__init__(*args, **kwargs)
+
+    def run(self):
+        ...
+        self.reporter.report(
+            my_custom_attribute='my_custom_value',
+            another_custom_attribute='another_custom_value',
+            ...)
+        ...
+```
+
+The prefix will be applied to all attributes reported from the job as well
+as the camel case conversion. If you wish to skip both, you can use the
+`report_raw` function instead of `report`.
+
+In all cases, two additional attributes are reported:
+
+1. `error`: a Boolean indicating whether the job has failed or not
+1. `jobFailure`: a Boolean indicating whether the job has permanently failed or not
+
+The `error` attribute is reported as is, i.e. without prefix or camel case conversion.
+
+Additionally, the end user id is reported if the custom attributes contain the id
+in any of the below formats:
+
+1. `user_id` (with and without prefix)
+1. `userId` (with and without prefix)
+
+This is useful in identifying impacted users count in case of job errors.
 
 ## Limitations
 
@@ -133,21 +218,14 @@ Do your changes, then send a pull request.
 ## Publish
 
 ### Using Python
+
 1. Increment the version number in `setup.py`
 1. Install twine: `pip install twine`. You may need to upgrade pip first: `pip install --upgrade pip`
 1. Create the distribution files: `python setup.py sdist bdist_wheel`
-1. Optionally upload to [Test PyPi](https://test.pypi.org/) as a dry run: `twine upload -r testpypi dist/*`. You will need a separate account there
-1. Upload to [PyPi](https://pypi.org/): `twine upload dist/*`
-
-### Using Docker
-Increment the version as in the first step above then:
-
-```bash
-docker build . -t pyworker:0.1.0
-docker run -it --rm pyworker:0.1.0
-```
+1. Optionally upload to [Test PyPI](https://test.pypi.org/) as a dry run: `twine upload -r testpypi dist/*`. You will need a separate account there
+1. Upload to [PyPI](https://pypi.org/): `twine upload dist/*`
 
-Enter your PyPi username and password when prompted.
+Enter your PyPI username and password when prompted.
 
 ## License
 

pyworker/reporter.py

@@ -72,6 +72,14 @@ def _convert_value(value):
     def _report_newrelic(self, attributes):
         if self._logger:
             self._logger.debug('Reporter: reporting to NewRelic: %s' % attributes)
+        # report user id if available in attributes in any form
+        possible_keys = [self._prefix + 'userId', self._prefix + 'user_id', 'userId', 'user_id']
+        possible_values = [attributes.get(key) for key in possible_keys if key in attributes]
+        if possible_values:
+            user_id = str(possible_values[0])
+            newrelic.agent.set_user_id(user_id)
+            if self._logger:
+                self._logger.debug('Reporter: reporting to NewRelic user_id: %s' % user_id)
         # convert attributes dict to list of tuples
         attributes = list(attributes.items())
         newrelic.agent.add_custom_attributes(attributes)

tests/test_reporter.py

@@ -188,7 +188,7 @@ def test_reporter_convert_value_converts_value_to_json_if_not_supported(self):
     #********** ._report_newrelic tests **********#
 
     @patch('pyworker.reporter.newrelic.agent')
-    def test_reporter_report_newrelic_calls_newrelic_record_exception(self, newrelic_agent):
+    def test_reporter_report_newrelic_calls_newrelic_add_custom_attributes(self, newrelic_agent):
         reporter = Reporter()
         reporter._report_newrelic({
             'test_key1': 'test_value',
@@ -199,3 +199,57 @@ def test_reporter_report_newrelic_calls_newrelic_record_exception(self, newrelic
             ('test_key1', 'test_value'),
             ('test_key2', 2)
         ])
+
+    @patch('pyworker.reporter.newrelic.agent')
+    def test_reporter_report_newrelic_does_not_call_newrelic_set_user_id_if_user_id_not_in_attributes(self, newrelic_agent):
+        reporter = Reporter()
+        reporter._report_newrelic({
+            'test_key1': 'test_value',
+            'test_key2': 2
+        })
+
+        newrelic_agent.set_user_id.assert_not_called()
+
+    @patch('pyworker.reporter.newrelic.agent')
+    def test_reporter_report_newrelic_calls_newrelic_set_user_id_if_user_id_in_attributes(self, newrelic_agent):
+        reporter = Reporter()
+        reporter._report_newrelic({
+            'test_key1': 'test_value',
+            'test_key2': 2,
+            'user_id': 123
+        })
+
+        newrelic_agent.set_user_id.assert_called_once_with('123')
+
+    @patch('pyworker.reporter.newrelic.agent')
+    def test_reporter_report_newrelic_calls_newrelic_set_user_id_if_userId_in_attributes(self, newrelic_agent):
+        reporter = Reporter()
+        reporter._report_newrelic({
+            'test_key1': 'test_value',
+            'test_key2': 2,
+            'userId': 123
+        })
+
+        newrelic_agent.set_user_id.assert_called_once_with('123')
+
+    @patch('pyworker.reporter.newrelic.agent')
+    def test_reporter_report_newrelic_calls_newrelic_set_user_id_if_prefixed_user_id_in_attributes(self, newrelic_agent):
+        reporter = Reporter(attribute_prefix='test_')
+        reporter._report_newrelic({
+            'test_key1': 'test_value',
+            'test_key2': 2,
+            'test_user_id': 123
+        })
+
+        newrelic_agent.set_user_id.assert_called_once_with('123')
+
+    @patch('pyworker.reporter.newrelic.agent')
+    def test_reporter_report_newrelic_calls_newrelic_set_user_id_if_prefixed_userId_in_attributes(self, newrelic_agent):
+        reporter = Reporter(attribute_prefix='test_')
+        reporter._report_newrelic({
+            'test_key1': 'test_value',
+            'test_key2': 2,
+            'test_userId': 123
+        })
+
+        newrelic_agent.set_user_id.assert_called_once_with('123')