This example demonstrates the use of XRd vRouter as a redundant Cloud Router.
Two Alpine Linux containers are connected to each other via a redundant pair of XRd vRouter instances.
The Bootstrap configuration must be applied before running this example.
Then to run this example, execute:
terraform -chdir=infra init
terraform -chdir=infra apply
terraform -chdir=workload init
terraform -chdir=workload apply
This should take around two minutes to complete. You may then configure kubectl so that you can connect to the cluster:
aws eks update-kubeconfig --name $(terraform -chdir=workload output -raw cluster_name)
Traffic from the CNF to the Peer container is routed via the active XRd vRouter instance. To simulate failure of the active XRd vRouter instance:
# Ping from the CNF to the Peer should succeed after initialization:
kubectl exec deploy/cnf-vrid1 -- ping -c5 10.0.10.12
kubectl exec deploy/cnf-vrid2 -- ping -c5 10.0.10.12
# Simulate a failure of the active XRd vRouter instance:
kubectl rollout restart sts/xrd1
# Traffic is restored after up to ~30 seconds:
kubectl exec deploy/cnf-vrid1 -- ping 10.0.10.12
kubectl exec deploy/cnf-vrid2 -- ping -c5 10.0.10.12To destroy this example, execute:
terraform -chdir=workload destroy
terraform -chdir=infra destroy
Configuration inputs are documented on the infra configuration and workload configuration pages.
Configuration outputs are documented on the workload configuration page.
It is possible to use a local copy of the HA app Helm chart when applying the workload configuration, by providing a local path for the workload configuration variable ha_app_chart_name, and null for the workload configuration variable ha_app_chart_repository.
All on-disk dependencies of the local Helm chart must be present. If not, then you may see the following terraform apply error:
│ Error: found in Chart.yaml, but missing in charts/ directory: xrd-vrouter
│
│ with helm_release.xrd1,
│ on main.tf line 42, in resource "helm_release" "xrd1":
│ 42: resource "helm_release" "xrd1" {
The fix is to run helm dependency update.