- update to 1.66.4:
* Fixed: Restored UDP connectivity through Mullvad exit nodes
* Stateful filtering is now off by default
- update to 1.66.3:
* Login URLs did not always appear in the console when running tailscale up
* Starting with v1.66, the Kubernetes operator must always run the same or later version
as the proxies it manages.
* Expose cloud services on cluster network to the tailnet, using Kubernetes ExternalName Services
* Expose tailnet services that use Tailscale HTTPS to cluster workloads
* Cluster workloads can now refer to Tailscale Ingress resources by their MagicDNS names
* Configure environment variables for Tailscale Kubernetes operator proxies using ProxyClass CRD
* Expose tailscaled metrics endpoint for Tailscale Kubernetes operator proxies through ProxyClass CRD
* Configure labels for the Kubernetes operator Pods with Helm chart values
* Configure affinity rules for Kubernetes operator proxy Pods with ProxyClass
* Kubernetes operator proxy init container no longer attempts to enable IPv6 forwarding on systems
that don't have IPv6 module loaded
* Tailscale containers running on Kubernetes no longer error if an empty Kubernetes Secret is
pre-created for the tailscaled state
* Improved the ambiguous error messages when Tailscale running on Kubernetes does not have the right
permissions to perform actions against the tailscaled state Secret