* Enhance network policy to isolate synced workload pods and improve cross-cluster pod isolation handling
* Add test for label update on synced Pod to ensure isolation label persistence
* Derive host pod CIDRs dynamically for the isolation NetworkPolicy
Compute the egress-exclude CIDRs from the --cluster-cidr flag or the live
Node PodCIDR(s) via FindPodCIDRs, instead of a hardcoded guess, so cross-cluster
pod isolation is enforced against the host's real pod network. Adds unit,
integration, and e2e coverage.
* update comment
* Sort CIDR list in FindPodCIDRs function to ensure consistent order for egress rules
* Use `t.Context()` instead of `context.Background()`
Co-authored-by: Kevin McDermott <bigkevmcd@gmail.com>
* Use Ginkgo provided context
* Refactor FindPodCIDRs to use sets for CIDR collection and simplify logic
* fix lint
---------
Co-authored-by: Kevin McDermott <bigkevmcd@gmail.com>
* Add tests asserting GetPods is scoped to the node's own Pods
Adds a unit test (fake host + virtual clients, Pods across multiple nodes plus
a dangling one) and a multi-node e2e test (a Deployment with one nginx Pod per
node via required anti-affinity; restart every k3k-kubelet agent Pod; assert no
workload Pod is deleted from the host or virtual cluster). Both pin the intended
behavior and fail against the current code; the fix follows in the next commit.
* Scope Provider.GetPods to the node's own Pods to prevent cross-node dangling-pod deletion
GetPods() listed host Pods cluster-wide (by the k3k.io/clusterName label only).
The vendored virtual-kubelet library's deleteDanglingPods reconciliation deletes
any Pod returned here that is missing from this instance's virtual Pod lister,
which is scoped to spec.nodeName == agentHostname. On a multi-node host cluster,
every restarting k3k-kubelet instance therefore saw Pods owned by other nodes as
'dangling' and deleted them from the host (and, in turn, the virtual cluster).
Scope GetPods by the *virtual* Pod's spec.nodeName -- the same ownership signal
the framework uses -- excluding Pods owned by other nodes while still returning
own-node Pods and genuinely dangling ones. The host Pod's physical node is not a
reliable owner (it is scheduled with only a soft, sometimes-absent affinity), so
it must not be used. Virtual Pods are read live to avoid a startup cache-sync
race. Makes the previous commit's tests pass.
* Refactor GetPods to scope to the cluster namespace and update tests accordingly
* Add AgentNameLabel to track Pods synced by the k3k-kubelet agent and update tests accordingly
* Add failing test for updatePod
* Add HCP (Hosted Control Plane) support
Introduce hosted control plane mode for k3k virtual clusters, including
API types, controller logic, server endpoint handling, CLI flags,
CRD updates, kubeconfig generation, and examples.
Co-Authored-By: RuFlo <ruv@ruv.net>
* removed hcpRegitration command
added HCP conformance tests
warning for hcp
fix multi-VM HCP conformance test networking
Both QEMU workers booted with `-net user` and ended up registering the
same InternalIP (10.0.2.15) because each VM gets its own isolated NAT
slirp. Flannel propagated this to `public-ip` on both nodes, so VXLAN
could not tunnel between workers and any cross-node pod traffic broke
(89 failed / 335 passed of 424 conformance specs).
Replace user-mode networking with a Linux bridge (k3kbr0,
192.168.100.0/24) and one TAP device per VM, so the two workers share
an L2 segment with unique routable IPs. NAT outbound from the bridge
keeps internet access working for image pulls.
Also set unique hostnames via cloud-init (worker-1/worker-2) and drop
the `--node-name` flag from INSTALL_K3S_EXEC, since k3s now picks the
correct node name from the OS hostname on its own.
Bump hydrophone back to `--parallel 4` to match the single-VM job
(parallelism was reduced earlier when the failure was thought to be
resource-related).
added HCP print command
updated crds
adding e2e tests
Refactor selectNonLoopbackSAN function to accept SANs directly and update related logic in ensureHCPRegistration
* Update agent flag validation and enhance ingress host check with a warning log
Refactor descriptions for cluster provisioning mode and role in CRDs and documentation
Refactor logging in ServerURL function to use controller-runtime logger
Rename selectNonLoopbackSAN to findNonLoopbackSAN for clarity and update references
Refactor ServerURL function and related code to remove unused parameters and improve clarity
Remove unused imports from kubeconfig.go to improve code clarity
* suggested changes
* fix comment
* fix test
---------
Co-authored-by: jpgouin <jeanphilippe.gouin@suse.com>
Co-authored-by: RuFlo <ruv@ruv.net>
* Adding new k3s function to get the config
* Adding a new way to get the bootstrap data directly from the server Pod
* Gate the bootstrap fetch if external datastore is in use
---------
Signed-off-by: galal-hussein <hussein.galal.ahmed.11@gmail.com>
* Refactor kubeconfig generation to remove unused port parameter and update related functions
* Refactor kubeconfig generation to streamline error handling and remove unused imports
* Refactor kubeconfig URL generation functions and deprecate old implementation
* restore old behavior
* Add 'k3kcli kubeconfig get' command and update documentation
* Refactor URL generation by removing deprecated getURLFromService function and updating tests to use new implementation
* Fix expected URL for LoadBalancer test case to include hostname
* Refactor kubeconfig test documentation to clarify URL generation behavior for ClusterIP, NodePort, and LoadBalancer service types
* Refactor kubeconfig URL generation functions to improve clarity and maintainability
* Remove deprecated 'k3kcli kubeconfig get' command and update related documentation
* Set logger to discard in NewRootCmd for improved logging control
* Refactor getURLFromService to streamline ingress key retrieval
* removed all webhook references
* fix tests
* removal of webhook
* updated doc
* add downward API test
* small refactor of virtual-kubelet
* split tests
* updates to services
- added loadBalancerConfig
- removed service-port
- added logic to not expose services
* Refactor cluster tests to improve readability and maintainability
- Simplified service port expectations by directly accessing elements instead of using `ContainElement`.
- Enhanced clarity of test assertions for `k3s-server-port` and `k3s-etcd-port` attributes.
- Removed redundant code for checking service ports.
* fix ports for ingress expose, update kubeconfig generate
Hussein GalalandGitHub
