You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository has been archived by the owner on Oct 21, 2022. It is now read-only.
Filter for in- and outgoing network traffic as configurable Karydia feature
User Story
As Kubernetes cluster owner I want to prevent applications and users from reaching remote hosts or from being reached by remote hosts in order to mitigating DDoS attacks, avoiding SPAM, blocking access to or from services for specific geographic regions and so on.
Does this REST based API already exist or does it need to be designed? If it exists, it would be good to see documentation with the endpoints, the input/output formats, authentication, whether there is pagination.
Possible list of tasks:
Update CRD KarydiaConfig (scope: cluster) with additional parameters:
Whenever the KarydiaConfig changes, an update is scheduled immediately
Tests
Mock REST API server providing the CIDR list
Manual functional tests: check that the policy is enforced correctly
Connectivity from a pod with networkHost=true (or from the host)
Connectivity from a pod with networkHost=false
Automatic tests: Karydia has e2e tests (tests/e2e directory) but it does not seem to be integrated in a CI pipeline (.ci directory empty). We can add more e2e tests for this feature.
Description
Filter for in- and outgoing network traffic as configurable Karydia feature
User Story
As Kubernetes cluster owner I want to prevent applications and users from reaching remote hosts or from being reached by remote hosts in order to mitigating DDoS attacks, avoiding SPAM, blocking access to or from services for specific geographic regions and so on.
Implementation Idea
This kind of filtering is discussed in the Kubernetes community already in recent blog postings (see Performance Benchmark Analysis of Egress Filtering on Linux and BPF Isn't Just About Speed.
The idea is that one or more Reputation Block Lists are received via HTTPS and a REST based API and are transformed in Cilium or other technology based network filters.
The text was updated successfully, but these errors were encountered: