Merge pull request #49 from dmcnaught/master

Documentation for adding kube-prometheus to K8s on AWS w KOPS.

README.md

@@ -133,3 +133,6 @@ kube-system   kube-controller-manager-prometheus-discovery   10.2.30.2:10252
 kube-system   kube-scheduler-prometheus-discovery            10.2.30.4:10251    1h
 monitoring    etcd-k8s                                       172.17.4.51:2379   1h
 ```
+
+## Other Documentation
+[Install Docs for a cluster created with KOPS on AWS](docs/KOPSonAWS.md)
\ No newline at end of file

docs/KOPSonAWS.md

+# Adding kube-prometheus to [KOPS](https://github.com/kubernetes/kops) on AWS 1.5.x
+
+
+## Prerequisites
+
+A running Kubernetes cluster created with [KOPS](https://github.com/kubernetes/kops).
+ 
+These instructions have currently been tested with  **topology=public** on AWS with KOPS 1.5.1 and Kubernetes 1.5.x
+
+## Open AWS Security Groups:
+1. Open port 9100 on the masters security group to the nodes security group
+1. Open ports 10250-10252 on the masters security group to the nodes security group.
+
+Example script below requires $AWS\_DEFAULT_PROFILE and [$NAME](https://github.com/kubernetes/kops/blob/master/docs/aws.md#prepare-local-environment)
+
+```bash
+MASTER_SG=$(aws --profile ${AWS_DEFAULT_PROFILE} ec2 describe-security-groups --filters "Name=tag:Name,Values=masters.$NAME" --query "SecurityGroups[*].GroupId[]" --output=text)
+NODES_SG=$(aws --profile ${AWS_DEFAULT_PROFILE} ec2 describe-security-groups --filters "Name=tag:Name,Values=nodes.$NAME" --query "SecurityGroups[*].GroupId[]" --output=text)
+aws --profile ${AWS_DEFAULT_PROFILE} ec2 authorize-security-group-ingress --group-id $MASTER_SG --protocol tcp --port 9100 --source-group $NODES_SG
+aws --profile ${AWS_DEFAULT_PROFILE} ec2 authorize-security-group-ingress --group-id $MASTER_SG --protocol tcp --port 10250-10252 --source-group $NODES_SG
+```
+
+## Adding kube-prometheus
+Following the instructions in the [README](https://github.com/coreos/kube-prometheus/blob/master/README.md):
+
+Example:
+
+```bash
+git clone -b master https://github.com/coreos/kube-prometheus.git kube-prometheus-temp;
+cd kube-prometheus-temp
+./hack/cluster-monitoring/deploy
+kubectl -n kube-system create -f manifests/k8s/self-hosted/
+cd -
+rm -rf kube-prometheus-temp
+```
+This currently leads to a [certificate error on the kube-apiserver scrape](https://github.com/coreos/kube-prometheus/issues/35). Looks to be fixed by [PR33](https://github.com/coreos/kube-prometheus/pull/33)
+
+To create a config without this alert and error, you can deploy without the kube-apiserver spec:
+
+```bash
+kubectl -n kube-system create -f manifests/k8s/self-hosted/kube-controller-manager.yaml
+kubectl -n kube-system create -f manifests/k8s/self-hosted/kube-dns.yaml
+kubectl -n kube-system create -f manifests/k8s/self-hosted/kube-scheduler.yaml
+```
\ No newline at end of file