frankwiles / django-app-metrics Goto Github PK

It is possible to emulate gauges in Mixpanel by posting the gauge value as a property of a metric and then using the aggregation functions (SUM or AVG) in the segmentation view to graph it.

No big deal to implement that in django-app-metrics and I'd be glad to do it, just wanted to ask if you think the idea is worth pursuing. Looking forward to feedback!

There are some updates I'd like to make e.g. #42, which would require dropping support for unsupported versions of Django.

@frankwiles would you be okay with making a major version update that breaks support for legacy Django? I'll help with PRs.

Installing celery (not especially a trivial task) if you don't need it seems especially onerous. I haven't examined the code but would conditional imports not remove the need for this?

(On a side note - there's often a simpler solution than Celery for many things it's used for - is this aspect of the code pluggable? A simple management command run via cron would be a good alternative for some)

The emails that I receive would make more sense to me if they showed the following:

• today so far
• yesterday
• today - 2
• ...
• today - 7
• this week
• this week - 1
  Etc

Would this be useful to others? I'd be happy to code it up if so.

Create a way to store up metrics in shared memory and periodically flush them to the backend removing the need/overhead of celery for extremely high traffic sites.

Does this app allow me to do this or should i just store metric("john_created_object") then rely on the backend to split by "_" and make sense of it?

Hi,

Is there an intention to add Mongodb integration?

Using django-app-metrics in a larger projects that is not otherwise using django-mailer means that the tests fail (they can't even be imported) because of a single use of mailer. Maybe the test can be written to avoid the use of mailer.send_all() or maybe it should be conditional upon the ability to import mailer. As it currently stands, things blow up.

The tests fail for me (Bare virtualenv, Django 1.4.1, python manage.py test) with connection errors in celery/kombu/amqplib. Is it necessary to have a fully configured celery running?

It would be very helpful to have a short description about the test requirements, alternatively removing as many external dependencies as possible would also be nice :-)

Let me know If I can support you with this!

Test output: https://gist.github.com/3417395

My question is wider than this project context. I integrated my django with Sentry (https://getsentry.com/welcome/), I provided a SECRET_KEY variable in my settings.py. Also in this project "tests/settings.py" have same variable.

When I import "APP_METRICS_BACKEND" from "app_metrics.tests.settings" to config djang-app-metrics. Django faced by this exception: "django.core.exceptions.ImproperlyConfigured: The SECRET_KEY setting must not be empty.".

/a/venvs/venv/lib/python2.6/site-packages/app_metrics/models.pyc in unicode(self)
87 def unicode(self):
88 return "'%s' for week %d of %d" % (self.metric.name,
---> 89 self.created.strftime("%m"))
90

TypeError: %d format: a number is required, not str

I'm using redis-py 2.7.1 (instead of 2.6.2 as listed in requirements), and redis 2.6.7, and I'm getting the following exception:

Task app_metrics.tasks.redis_metric_task[d83b61b0-e35a-4170-902b-97d6e7c5a2f9] raised exception: AttributeError("'StrictRedis' object has no attribute 'incrby'",)
Traceback (most recent call last):
  File "/Users/brad/.virtualenvs/wfp-app-metrics/lib/python2.7/site-packages/celery/task/trace.py", line 224, in trace_task
    R = retval = fun(*args, **kwargs)
  File "/Users/brad/.virtualenvs/wfp-app-metrics/lib/python2.7/site-packages/celery/task/trace.py", line 406, in __protected_call__
    return self.run(*args, **kwargs)
  File "/Users/brad/django/workforpie/app_metrics/tasks.py", line 161, in redis_metric_task
    r.incrby(day_key, num)
AttributeError: 'StrictRedis' object has no attribute 'incrby'

This breaks in tasks/redis_metric_task.py where you increment keys for day, week, month, year:

r.incrby(day_key, num)

It appears that redis-py no longer includes an incrby method. There's an incrbyfloat method for float values, but I'm not sure that's what you want either.

Any thoughts on a preferred way to make this work with the current redis-py?