Get your application events and errors to Trakerr via the Trakerr API.

You can send both errors and non-errors (plain log statements, for example) to Trakerr with this API.

The 3-minute integration guide is primarily oriented around sending errors or warnings and does not allow you to specify additional parameters. The logging handler also has the ability to specify the log level that then controls whether infos or debugs are also sent.

For sending additional parameters, Option-4 in the detailed integration guide describes how you could send non-errors along with additional parameters.

The SDK takes performance impact seriously and all communication between the SDK <=> Trakerr avoids blocking the calling function. The SDK also applies asynchronous patterns where applicable.

A Trakerr Event can consist of various parameters as described here in AppEvent. Some of these parameters are populated by default and others are optional and can be supplied by you.

Since some of these parameters are common across all event's, the API has the option of setting these on the TrakerrClient instance (described towards the bottom) and offers a factory API for creating AppEvent's.

• Log Level This enum specifies the logging level to be used for this event ('debug','info','warning','error' or 'fatal')
• Event Type This defines the type of event or logger name. This is automatically set for errors.
• Classification This is a user settable property that controls how the events are grouped. Defaults to 'Issue'. Set this to a different value to group this event in a different group.

• Event User This is the user that is associated with this event. This can be any user data or could be encrypted if privacy is required.
• Event Session This is any session specific information associated with this event.
• Cross App Correlation ID This is an additional ID that can be used for cross-application correlation of the same event.

• Operation Time This property in milliseconds measures the operation time for this specific event.

In addition to the above, you can use custom properties and segments to send custom event, performance data. These can then be visualized in Trakerr's dashboards.

Python 2.7.9+ and 3.4+

If you're already using the python logger in some capacity, you can integrate with Trakerr quickly. First, issue a pip install to get the latest version:

To install from master, simply use:

pip install git+https://github.com/trakerr-com/trakerr-python.git

(you may need to run pip with root permission: sudo -H pip install git+https://github.com/trakerr-com/trakerr-python.git)

or upgrade an installation with a command from the advanced pip commands section.

Once installation is complete, add this import from wherever you instantiate your loggers.

import logging #Your previous logging import here
from trakerr import TrakerrHandler #New import.

And then you'll need to create a handler and a logger object and attach them before you can use them later on.

#Your original logger.
logger = logging.getLogger("Logger name")

#Instantiate Trakerr's logger handler. By default the handler will only log WARNING and above.
th = TrakerrHandler("<api-key>", "App Version number here", "Deployment stage here")

#Attach our handler to the root logger.
logging.getLogger('').addHandler(th)

#Alternatively attach our handler to your logger for a single logger instance.
logger.addHandler(th)

Follow the shutdown section when you are closing your app to ensure a graceful shutdown.

The handler's full constructor signature is as follows:

def __init__(self, api_key=None, app_version="1.0", context_deployment_stage="deployment",
                 client=None, level=logging.WARNING):

The parameters are as follows

• api_key: should be your api key from trakerr.
• app_version: should be your code's app version.
• deployment_stage: is the stage of the codebase you are pulling events from (production, test, development, ect).
• client: allows you to pass in a preconfigured TrakerrClient instance, which allows you to provide a client with custom values. If you provide a client, the values of the first three parameters do not matter.
• level: Allows you to set the level of events that this handler can emit. By default, TrakerrHandler will send warnings and above to trakerr. This way, if you're using more than one handler, you can set the logger to log all events, and set trakerr to log only warnings and above, while sending infos and above with another handler.

You should be able to now send basic events and information to trakerr. If you are only using trakerr to log, or want to send more in depth events, read on below.

When cleaning up your application, or implementing a shutdown, be sure to call:

TrakerrClient.shutdown()

This will close the perfomange monitoring thread gracefully. You only need to call this once, no matter how many loggers you have. Add it to any cleanup function you have for your program.

By using the pip command above, you can also use trakerr in a multitude of different ways.

Follow the instructions in the three minute integration guide to attach trakerr to a logger you've already written.

Along with your imports, add:

from trakerr import Trakerr

And then you can get a python logger object that ties in to trakerr to handle the various levels of errors, giving the appropriate data as strings.

def main(argv=None):
    if argv is None:
        argv = sys.argv
    
    logger = Trakerr.get_logger("<api-key>", "App Version number here", "Logger Name")

    try:
        raise FloatingPointError()
    except:
       logger.exception("Bad math.")

    return 0

You can quickly send a simple event with partial custom data from the log function of the TrakerrClient. Add these imports:

from trakerr import TrakerrClient

you can then send call log simply to send a quick error to Trakerr.

client = TrakerrClient("<api-key>", "App Version number")

client.log({"user":"[email protected]", "session":"25", "eventtype":"user logon issue",
            "eventmessage":"User refreshed the page."}, "info", "logon script", False)

The full signature of log is as follows:

def log(self, arg_dict, log_level="error", classification="issue", exc_info=True):

If exc_info is True, it will poll sys.exc_info() for the latest error stack information. If this is false, then the event will not generate a stack trace. If exc_info is is a tuple of exc_info, then the event will be created using that exc_info.

arg_dict is a dictionary which makes it simple to pass in basic AppEvent information without using the more extensive methods below. The items arg_dict looks for are as follows:

• "eventtype":(maps to string) The name of the event. This will autmatically be filled if nothing is passed in when you are sending and event with a stacktrace.
• "eventmessage":(maps to string) The message of the event. This will autmatically be filled if nothing is passed in when you are sending and event with a stacktrace.
• "user":(maps to string) User that triggered the event
• "session":(maps to string) Session that triggered the event
• "time":(maps to string) Time that the operation took (usually in miliseconds)
• "url":(maps to string) URL of the page the error occurred on
• "corrid":(maps to string) The correlation id
• "device":(maps to string) The machine name or type you are targeting

You can of course leave out anything you don't need, or pass in an empty dictionary to arg_dict if you don't wish to give any data

You can send custom data as part of your error event if you need to. This circumvents the python handler. Add these imports:

from trakerr import TrakerrClient
from trakerr_client.models import CustomData, CustomStringData

You'll then need to initialize custom properties once you create the event. Note that you can use the second call commented out to instead create an app event without a stacktrace.

def main(argv=None):
    if argv is None:
        argv = sys.argv

    client = TrakerrClient("<api-key>", "App Version number")

    try:
        raise IndexError("Bad Math")
    except:
        appevent = client.create_new_app_event("FATAL", exc_info=True)

        #Populate any field with your own data, or send your own custom data
        appevent.context_app_browser = "Chrome"
        appevent.context_app_browser_version = "67.x"

        #Can support multiple ways to input data
        appevent.custom_properties = CustomData("Custom Data holder!")
        appevent.custom_properties.string_data = CustomStringData("Custom String Data 1",
                                                                  "Custom String Data 2")
        appevent.custom_properties.string_data.custom_data3 = "More Custom Data!"
        appevent.event_user = "[email protected]"
        appevent.event_session = "6"

        appevent.context_operation_time_millis = 1000
        appevent.context_device = "pc"
        appevent.context_app_sku = "mobile"
        appevent.context_tags = ["client, frontend"]

        #Send it to trakerr
        client.send_event_async(appevent)

    return 0

create_new_app_event's full signature is as follows:

create_new_app_event(self, log_level="error", classification="issue", event_type="unknown",
                     event_message="unknown", exc_info=False):

exc_info can be set to True to get the latest exception traces from sys.exc_info or simply passed in.

If you are using Django, we recommend that you look at django user agents as a simple and quick way of getting the browser's name and version rather than parsing the user agent yourself. The library also allows you to check the client browser in the template, allowing you to modify the front end to the error accordingly. Please note that this library is not maintained by or related to Trakerr in any way.

TrakerrClient's constructor initalizes the default values to all of TrakerrClient's properties.

def __init__(self, api_key, context_app_version="1.0",
             context_deployment_stage="development", application_sku="", tags=[],
             threads=4, connnections=4):

The threads parameter specify the number of max_workers in the thread pool. This only matters if you are using send_event_async in python 3.2+. The connections parameter specificies the number of connections in the connection pool. If there are more threads than connections, the connection pool will block the derivitive async calls until it can serve a connection.

The TrakerrClient class however has a lot of exposed properties. The benefit to setting these immediately after after you create the TrakerrClient is that AppEvent will default it's values with the TrakerrClient instance that created it. This way if there is a value that all your AppEvents uses, and the constructor default value currently doesn't suit you; it may be easier to change it in the TrakerrClient instance as it will become the default value for all AppEvents created after. A lot of these are populated by default value by the constructor, but you can populate them with whatever string data you want. The following table provides an in depth look at each of those.

If you're populating an app event directly, you'll want to take a look at the AppEvent properties as they contain properties unique to each AppEvent which do not have defaults you may set in the client.

You can run the following command to update an exsisting installation to the latest commit on master:

pip install git+https://github.com/trakerr-com/trakerr-python.git --upgrade

You can install from a branch for development or testing a new feature (Not recommended for production use):

pip install git+https://github.com/trakerr-com/trakerr-python.git@<branch_name_here>

• AppEvent

RM