hacking metrics into kubeadm

gistfile1.txt

10226  # 1) import components-base into the kubeadm package
10227  # 2) add a few metrics as is done in  vi pkg/kubelet/metrics/metrics.go
10228  # 3) run "make kubeadm"
10229  # 4) run "kubeadm init" on your mac
10230  # 5) like this .//_output/local/go/bin/kubeadm init


diff --git a/cmd/kubeadm/app/kubeadm.go b/cmd/kubeadm/app/kubeadm.go
index 1842cfd8cb4..0ab66a1cadb 100644
--- a/cmd/kubeadm/app/kubeadm.go
+++ b/cmd/kubeadm/app/kubeadm.go
@@ -26,6 +26,24 @@ import (
 
 	cliflag "k8s.io/component-base/cli/flag"
 	"k8s.io/kubernetes/cmd/kubeadm/app/cmd"
+
+        "k8s.io/component-base/metrics"
+//        "k8s.io/component-base/metrics/legacyregistry"
+
+)
+
+
+var (
+        // NodeName is a Gauge that tracks the ode's name. The count is always 1.
+        _ = metrics.NewGaugeVec(
+                &metrics.GaugeOpts{
+                        Subsystem:      "kubeadm",
+                        Name:           "a",
+                        Help:           "The node's name. The count is always 1.",
+                        StabilityLevel: "1",
+                },
+                []string{"a"},
+        )
 )
 
 // Run creates and executes new kubeadm command

What does this put into the image?

Maybe (?) -- need to verify:

• kubelet
• api-server
• scheduler
• controller-manager

To build the kind node-image from source:

• kind build node-image --kube-root=/tmp/kubernetes

Instead of using KinD, could also get a linux laptop and:

• git clone kubernetes
• hack/local-up-cluster.sh

get a linux vm with 4 cores, 16 GB of RAM and use it for development

• can ssh into it

Follow along with KinD CI here