How to add a custom metric provider#

This guide shows you how to expose your own metrics through the @@metrics endpoint.

A metric provider is an adapter on OFS.interfaces.IApplication that yields Metric instances. For the interface and the Metric fields, see Interfaces.

Implement the provider#

from zope.interface import implementer
from plone.observability.interfaces import IMetricProvider
from plone.observability.metric import Metric


@implementer(IMetricProvider)
class MyMetricProvider:
    name = "myapp"
    scope = "instance"

    def __init__(self, context):
        self.context = context

    def collect(self):
        yield Metric(
            name="myapp_queue_length",
            value=get_queue_length(),
            type="gauge",
            scope="instance",
            help="Number of items in the processing queue",
        )

Register the adapter#

<adapter
    factory=".metrics.MyMetricProvider"
    provides="plone.observability.interfaces.IMetricProvider"
    for="OFS.interfaces.IApplication"
    name="myapp"
    />

The next scrape of @@metrics includes the yielded metrics.

Keep labels low-cardinality#

Do not put unbounded values, such as a user id or a full URL, into metric names or labels. Each unique combination becomes a separate Prometheus time series. See About metrics and cardinality for why this matters and what to use instead.