Deployment of influxdb cluster in kubernetes

The main goals of the project:
 - create influxdb cluster in kubernetes.
 - configure influxdb cluster.
 - Stress Test influxdb cluster.
 - Tune influxdb configs to achieve current usecases.

Prerequisites

• [docker-machine] (https://docs.docker.com/engine/installation/mac/)
• Local Machine: [kubernetes with OSX] (https://github.com/TheNewNormal/kube-solo-osx) or [kubernetes with vagrant] (http://kubernetes.io/v1.1/docs/getting-started-guides/vagrant.html)
• AWS: [kubernetes with coreos] (https://coreos.com/kubernetes/docs/latest/kubernetes-on-aws.html)

Influxdb version

0.10.0-1

Deploy influxdb cluster to kubernetes cluster

• Create "infra" namespace

   	$ kubectl create -f ./k8s/infra-namespace.yaml
  

• Set default namespace to "infra"

   	$ export CONTEXT=$(kubectl config view | grep current-context | awk '{print $2}')
   	$ kubectl config set-context $(CONTEXT) --namespace=<insert-namespace-name-here>
  

• Deploy influxdb service and cluster

  • A prebuilt docker image is deployed at: https://hub.docker.com/r/supershal/influxdb/
  1. deploy influxdb service

   	$ kubectl create -f ./k8s/influxdb-svc.yaml
  

  2. deploy influxdb raft (jury) node

   	$ kubectl create -f ./k8s/influxdb-raft-rc.yaml
  

  3. scale raft node to 3. no more than 3 allowed.

   	$ kubectl scale --replicas=3 rc/influxdb-raft-rc
  

  4. deploy influxdb data nodes

   	$ kubectl create -f ./k8s/influxdb-peers-rc.yaml
  

  5. scale data node to any number. we set it up to 4

   	$ kubectl scale --replicas=4 rc/influxdb-peers-rc
  

  6. deploy test data producers

   	$ kubectl create -f ./k8s/telegraf-rc.yaml
  

• Deploy grafana service

  1. deploy grafana service

   	$ kubectl create -f ./k8s/influxdb-grafana-svc.yaml
  

  2. deploy grafana node

   	$ kubectl create -f ./k8s/influxdb-grafana-rc.yaml
  

Verify cluster setup

• ssh to k8s minion

   	$ vagrant ssh minion-1
   	OR
   	$ ssh vagrant@minion-1 # make sure you have set up vagrant config in your ./ssh/config. use vagrant ssh-config command.
  

• locate one of the influxdb container and exec

   	$ sudo docker ps
   	$ sudo docker exec -it /bin/bash <container_id> /bin/bash
  

• launch influxdb cli and check cluster members

   	$ influx
   	$ show servers
  

  It should give all members of the cluser something like.

show servers name: data_nodes

---

id http_addr tcp_addr 4 10.244.1.3:8086 10.244.1.3:8088 5 10.244.2.200:8086 10.244.2.200:8088 6 10.244.1.185:8086 10.244.1.185:8088 7 10.244.2.183:8086 10.244.2.183:8088

name: meta_nodes

id http_addr tcp_addr 1 10.244.1.2:8091 10.244.1.2:8088 2 10.244.2.199:8091 10.244.2.199:8088 3 10.244.1.184:8091 10.244.1.184:8088

## Terminate cluster
- delete influxdb service
	```
	 $ kubectl delete svc influxdb-svc
	``` 
- delete influxdb rc
	```
 	 $ kubectl delete rc -l app=influxdb
	```
- delete grafana service
	```
	 $ kubectl delete svc influxdb-grafana-svc
	``` 
- delete grafana rc
	```
 	 $ kubectl delete rc influxdb-grafana-rc
	```
- delete telegraf(data producers) rc
	```
 	 $ kubectl delete rc telegraf-rc
	```

## Build influxdb docker image
If you are not making any code changes and/or building docker image for influxdb, you can skip this section.
1. Install GO latest version and set up go workspace. [Instructions]
2. Install godep

	```
		$ go get -u github.com/tools/godep
	```
3. Restore dependency

	```
		$ godep restore
	```
4. Install and test influxdbconfig program locally.
	- Build go binary for testing.

      ```
		$ go build -o influxdblocal ./influxdb/main.go
      ```
    - Setup kubectl proxy to proxy influxdb api server
   
      ```
      	$ kubectl proxy --port=9090 &
      ```
    - test locally. 
   
      ```
      	$ LOCAL_PROXY="http://localhost:9090" INFLUXDB_POD_SELECTORS="app=influxdb" NAMESPACE="infra" influxdblocal test
      ```
      It will spit out the the influxdb cluster config parameters to console.
4. build influxdbconfig executable from the go program and create docker image for influxdbconfig + influxdb
	- Create docker-machine vm and set docker daemon

	 ```
	 	$ docker-machine create --driver=virtualbox default
	 	$ eval "$(docker-machine env default)"
	 ```
	- build the image

	 ```
		$ ./influxdb/build.sh
	 ```
	This will create image in your docker-machine VM.
5. Upload image to docker hub or directy to k8s-minion.
  - (Faster) Copy image directly to k8s minion. Use this method if you frequently building the image.
    * One time setup for ssh to minion.
    	Go to your kubernetes installation and locate Vagrantfile.
  	
  		```
  			$ vagrant ssh-config
  		```
  		Copy output of above command to your ~/.ssh/config file.
    * copy the image from docker-machine VM to K8S minion.
    
    	```
    		$ docker save supershal/influxdb:stresstest | ssh vagrant@minion-1 sudo docker load
    	```
  - (Slower) push to docker hub. 
  
  	 ```
  	 	$ docker push supershal/influxdb:stresstest
  	 ```

## Deploy influxdb cluster to Aws kubernetes cluster
 In progress

## Test cases
 In progress

## Stress Test cases
 In progress

## Useful links

- influxdb clusering:
https://docs.influxdata.com/influxdb/v0.9/guides/clustering/

- Access k8s cluster using apis:
https://github.com/kubernetes/kubernetes/blob/master/docs/user-guide/accessing-the-cluster.md#accessing-the-cluster-api

- K8s: developer guide: 
https://github.com/kubernetes/kubernetes/blob/master/docs/devel/README.md

- setup developer environment for k8s:
https://github.com/kubernetes/kubernetes/blob/master/docs/devel/development.md

- swagger doc for k8s:
http://kubernetes.io/third_party/swagger-ui/

- client library for k8s:
https://github.com/kubernetes/kubernetes/blob/master/docs/devel/client-libraries.md

- browse Golang doc and sourcecode for k8s (or any golang project)
   * install Dash https://kapeli.com/dash 
   * load golang docs for any go project. checkout docs for Dash