librdkafka

You can choose to use the C bindings using confluent-kafka:

$ pip install kser[confluent]

kafka-python

You can choose the pure python library kafka-python:

$ pip install kser[pykafka]

http

There is also a light version using HTTP ideal to produce messages (not recommended for consumption)

$ pip install kser[http]

dummy

There is a dummy version for tests, no additional import required, just use / patch with the dummy module.

encrypt data

It is possible to encrypt messages in the Kafka queue using libsodium.

$ pip install kser[crypto]

prometheus exporter

You can export metrics using the prometheus format:

$ pip install kser[prometheus]

A few set of environment variables allow to manage the exporter:

The exporter has only 2 metrics defined by default, it’s just a sample. A good way to implement your own is to override the triggers methods (prefixed with ‘_’) like the following example:

from kser import KSER_METRICS_ENABLED
from prometheus_client import Counter
from kser.entry import Entrypoint

MY_METRIC = Counter('kser_my_metric', 'a usefull metric')


class MyEntrypoint(Entrypoint):
    def _run(self):
        if KSER_METRICS_ENABLED == "yes":
            MY_METRIC.inc()

        return self.run()

See also:

prometheus_client documentation

Documentation of the python Prometheus client

opentracing support

You can use opentracing with Jaeger to follow operation and task:

$ pip install kser[opentracing]

Configuration can be set using environment variable:

See also:

Opentracing support with Jaeger

Kser documentation