1. Proposed title of this feature request

Default Alerting for Network Observability

2. What is the nature and description of the request?

With metrics being collected by the openstack-network-exporter, there are some metrics that always indicate a problem: buffers overflowing, packets being dropped, that sort of thing.  We should incorporate a base set of alarms that will trigger when these metrics are non-zero.  This spans both kernel data path and userspace data path.

3. Why does the customer need this? (List the business requirements here)

Having metrics observability is great, but the added value of showing we understand knowable bad metrics patterns and we will alert the customer to them shows a holistic approach to observability design.  And getting these alerts for free could be a great value add for customers.

4. List any affected packages or components.

openstack-network-exporter

thanos

References:

Alerts in Specification