docs: Deprecate support for podnetwork etcd

joestringer · qmonnet · commit f99f10bd8843 · 2024-06-11T08:56:45.000Z
Running Etcd in podnetwork to distribute state between Cilium instances introduces a range of challenges to bootstrapping and ensuring reliable connectivity within the cluster. We've deprecated in-built support in the Helm charts for this sort of configuration for several releases, and documented suggested alternatives. If we deprecate this feature then we can simplify some of the operations inside the cilium-agent. For alternative installation steps, see https://docs.cilium.io/en/stable/installation/k8s-install-external-etcd/#admin-install-daemonset . Signed-off-by: Joe Stringer <joe@cilium.io>
diff --git a/Documentation/operations/upgrade.rst b/Documentation/operations/upgrade.rst
@@ -358,6 +358,11 @@ Annotations:
 * Certain invalid CiliumNetworkPolicies that have always been ignored will now be rejected by the apiserver.
   Specifically, policies with multiple L7 protocols on the same port, over 40 port rules, or over
   40 ICMP rules will now have server-side validation.
+* Cilium could previously be run in a configuration where the Etcd instances
+  that distribute Cilium state between nodes would be managed in pod network by
+  Cilium itself. This support was complicated and error prone, so the support
+  is now deprecated. The following guide provides alternatives for running
+  Cilium with Etcd: :ref:`k8s_install_etcd`.
 
 Removed Options
 ~~~~~~~~~~~~~~~
