"Historgam" fails with Python 2.7.6

Hello!

I’m running prometheus_client==0.0.19 with Flask. As far as I understand the source code of the client, this registers every “Metric” in default prometheus_client.core.REGISTRY. So in my app I do the following:

def set_monitoring(self):
    self.server_summary = Histogram('main_flask_server', 'All requests timing', ['instance', 'endpoint'])
    self.server_errors = Histogram('main_flask_server_errors', 'Bad requests', ['instance', 'endpoint'])

...

app. set_monitoring()

...

app.server_summary.labels('i1', key).observe(dt_ms)

...

push_to_gateway(gateway=gateway, job='flask_meme', registry=REGISTRY)

On my local machine (MacOS Sierra, Python 2.7.12) everything is fine and works perfectly. But on the Ubuntu server with Python 2.7.6 I got the exception on startup. Traceback:

unable to load app 0 (mountpoint='') (callable not found or import error)
Traceback (most recent call last):
  File "/home/vladimir/membot/imager/server.py", line 104, in uwsgi_app
    return get_app(conf=conf)
  File "/home/vladimir/membot/imager/server.py", line 95, in get_app
    flask_app.set_monitoring()
  File "/home/vladimir/membot/imager/server.py", line 46, in set_monitoring
    self.server_summary = Histogram('main_flask_server', 'All requests timing', ['instance', 'endpoint'])
  File "/usr/local/lib/python2.7/dist-packages/prometheus_client/core.py", line 556, in init
    registry.register(collector)
  File "/usr/local/lib/python2.7/dist-packages/prometheus_client/core.py", line 54, in register
    'in CollectorRegistry: ' + name)
ValueError: Timeseries already present in CollectorRegistry: main_flask_server_bucket

What am I doing wrong? Why Timeseries is already present in CollectorRegistry?

[patrickleet]

the whole community can benefit from the answers provided.

except those who look on github

[brian-brazil]

Random github issues are not a good place for support requests, I’m locking this now.