InfrastructureAWSGoDatabaseMonitoringRaspberryPIPythonTerraformGitAIDIYData Analytics
Catalogue
Running Hubot on GKE + Kubernetes Within the Free Tier — Running from the CLI
Posted 2017-05-10Infrastructure9 minutes read (About 1329 words)

Running Hubot on GKE + Kubernetes Within the Free Tier — Running from the CLI

🌐 日本語で読む

Overview

  • We’ll build a Slack-integrated Hubot on GKE using the free tier.
  • As a bonus, we’ll also add JIRA integration.

Installing and Initializing the Google Cloud SDK

Download the SDK by referring to the Quickstart for macOS.

  • Check whether you’re on Mac OS X (x86_64) or (x86) with the following command:
1
2
3
macOS%$ uname -m

x86_64

Installing kubectl

1
macOS%$ gcloud components update kubectl

gcloud Default Configuration

Below, we configure the project, region, and zone we created.
This means you won’t have to specify the region and so on each time you run a gcloud command from now on.

  • Created project ID: hubot-167007
  • We set the region to a US region so we can use the free tier with us-west.
1
2
3
4
macOS%$ gcloud auth login
macOS%$ gcloud config set project hubot-167007
macOS%$ gcloud config set compute/region us-west1
macOS%$ gcloud config set compute/zone us-west1-b

Please refer to the Google Cloud Platform Free Tier.

Creating the Cluster

  • To use the free tier, we set f1-micro with 30GB.
  • However, 3 nodes are required at creation time.
  • After creation completes, we resize down to 1 node.
1
2
3
4
5
6
7
8
9
10
macOS%$ gcloud container clusters create hubot-cluster-free \
      --machine-type f1-micro \
      --disk-size=30 \
      --num-nodes=3

Creating cluster hubot-cluster-free...done.
Created [https://container.googleapis.com/v1/projects/hubot-167007/zones/us-west1-b/clusters/hubot-cluster-free].
kubeconfig entry generated for hubot-cluster-free.
NAME                ZONE        MASTER_VERSION  MASTER_IP       MACHINE_TYPE  NODE_VERSION  NUM_NODES  STATUS
hubot-cluster-free  us-west1-b  1.5.7           35.xxx.xxx.xxx  f1-micro      1.5.7         3          RUNNING
  • Looking at the console, you can confirm that it’s being created.

Imgur

  • You can verify it with the following command:
1
2
3
4
5
macOS% $ kubectl get nodes
NAME                                                STATUS    AGE       VERSION
gke-hubot-cluster-free-default-pool-a3b110d2-9k6s   Ready     59s       v1.5.7
gke-hubot-cluster-free-default-pool-a3b110d2-lqxg   Ready     1m        v1.5.7
gke-hubot-cluster-free-default-pool-a3b110d2-xqs8   Ready     1m        v1.5.7
  • Resize down to 1 node
1
2
3
4
5
6
7
8
macOS%$ gcloud container clusters resize hubot-cluster-free --size=1

Pool [default-pool] for [hubot-cluster-free] will be resized to 1.

Do you want to continue (Y/n)?  y

Resizing hubot-cluster-free...done.
Updated [https://container.googleapis.com/v1/projects/hubot-167007/zones/us-west1-b/clusters/hubot-cluster-free].

Imgur

If we can resize it, I wish they’d just let us create it with 1 node from the start (>_<)

Doing it from the console still doesn’t work, after all (T_T)

Imgur

Obtaining Credentials

  • Obtain the container cluster’s credentials so you can use kubectl to create containers on the container cluster.
1
2
3
4
macOS%$ gcloud container clusters get-credentials hubot-cluster-free

Fetching cluster endpoint and auth data.
kubeconfig entry generated for hubot-cluster-free.
  • Display container cluster information
1
macOS%$ gcloud container clusters describe hubot-cluster-free

Starting Docker Locally

[https://github.com/kenzo0107/hubot-slack-on-docker:embed:cite]

1
2
3
macOS%$ git clone https://github.com/kenzo0107/hubot-slack-on-docker
macOS%$ cd hubot-slack-on-docker
macOS%$ docker-compose up -d
1
2
3
4
macOS%$ docker ps

CONTAINER ID        IMAGE                      COMMAND                  CREATED             STATUS              PORTS                              NAMES
12f77feb09b4        hubotslackondocker_hubot   "/bin/sh -c 'bash ..."   24 minutes ago      Up 24 minutes       6379/tcp, 0.0.0.0:8080->8080/tcp   hubotslackondocker_hubot_1

Verifying Hubot Works

If Hubot shows up in Slack and replies with Hi when you say hello to it, you’ve succeeded.

Committing an Image from the CONTAINER ID

1
2
3
4
5
macOS%$ docker commit 12f77feb09b4 gcr.io/hubot-167007/hubot:latest

macOS%$ docker images
REPOSITORY                       TAG                 IMAGE ID            CREATED             SIZE
gcr.io/hubot-167007/hubot        latest              2f7336b3a3ce        3 seconds ago       484 MB

Pushing to the GKE Registry

Reference: Pushing to Container Registry

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
macOS%$ gcloud docker -- push gcr.io/hubot-167007/hubot:latest

The push refers to a repository [gcr.io/hubot-167007/hubot]
0569b419082b: Pushed
a7637cfcdfba: Pushed
9f0bdbb7b1fa: Pushed
f1d85eafc75a: Pushed
c2c2b58591f2: Pushed
51c94eacef50: Pushed
69e7fcf7ba41: Pushed
293d09ca6a9d: Pushed
247e72dfcaf5: Pushed
8c2bc9bf1f19: Pushed
40907ce0d959: Pushed
bfba578a7fbe: Pushed
561cbcaac156: Pushed
293a1e72e88b: Pushed
ae09eb3da3dc: Pushed
c06c14d7f919: Pushed
e14577d2cac5: Layer already exists
e8829d5bbd2c: Layer already exists
674ce3c5d814: Layer already exists
308b39a73046: Layer already exists
638903ee8579: Layer already exists

latest: digest: sha256:0c3b29d18b64c1f8ecc1a1bf67462c84d5915a4a708fe87df714d09198eb5fa1 size: 4704
  • If the latest tag collides, the tag of the previous image gets taken away. It’s a waste of storage, so delete it.

Imgur

Creating the Deployment

1
2
3
4
5
6
7
8
9
10
macOS%$ kubectl run pod-hubot \
      --image=gcr.io/hubot-167007/hubot:latest \
      --env="HUBOT_SLACK_TOKEN=xoxb-xxxxxxxxxxxx-xxxxxxxxxxxxxxxxxxxxxxxx" \
      --env="HUBOT_SLACK_TEAM=xxxxxx.slack.com" \
      --env="HUBOT_SLACK_BOTNAME=hubot" \
      --env="HUBOT_JIRA_URL=https://<jira_server_domain_or_ip>" \
      --port=8080 \
      --restart='Always'

deployment "pod-hubot" created
  • Check the deployments status
1
2
3
macOS%$ kubectl get deployments
NAME        DESIRED   CURRENT   UP-TO-DATE   AVAILABLE   AGE
pod-hubot   1         1         1            0           10s
  • Check the Pod status
1
2
3
4
macOS%$ kubectl get pods

NAME                         READY     STATUS             RESTARTS   AGE
pod-hubot-1713414922-b2dkq   0/1       ImagePullBackOff   0          23s
  • Log in to the Pod
1
$ kubectl exec -it pod-hubot-1713414922-b2dkq /bin/bash
  • Check the service status
1
2
3
4
5
macOS%$ kubectl get service


NAME         CLUSTER-IP    EXTERNAL-IP   PORT(S)   AGE
kubernetes   10.23.240.1   <none>        443/TCP   22m

EXTERNAL-IP: <none> … this means there’s no IP open to the outside.
A Private IP has been assigned, but there’s no Public IP, so it can’t be accessed from external networks.

Exposing the Container

  • Attach a load balancer to the Service to expose it.
  • Note: adding a load balancer makes the billing jump by an order of magnitude.
    (Around 2,000 yen/month. Just to be safe, I found out via the budget alert I had set up.)
1
2
macOS%$ kubectl expose deployment pod-hubot --type="LoadBalancer"
service "pod-hubot" exposed
  • Check the Service

It shows EXTERNAL-IP: <pending>, which tells us it’s still being created.

1
2
3
4
5
macOS%$ kubectl get service

NAME         CLUSTER-IP      EXTERNAL-IP   PORT(S)          AGE
kubernetes   10.23.240.1     <none>        443/TCP          25m
pod-hubot    10.23.244.214   <pending>     8080:30453/TCP   8s
  • Check the Service again

We can see it has been successfully assigned.

1
2
3
4
macOS%$ kubectl get service
NAME         CLUSTER-IP      EXTERNAL-IP     PORT(S)          AGE
kubernetes   10.23.240.1     <none>          443/TCP          27m
pod-hubot    10.23.244.214   104.xxx.x.xxx   8080:30453/TCP   1m

Testing

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
macOS%$ curl \
-X POST \
-H "Content-Type: application/json" \
-d \
'{
 "webhookEvent":"jira:issue_updated",
 "comment":{
   "author":{
     "name":"himuko"
    },
    "body":"[~kenzo.tanaka] 東京03 秋山 ケンコバ 劇団ひとり"
 },
 "issue": {
   "key":"key",
   "fields":{
     "summary":"summary"
    }
  }
}' \
http://104.xxx.x.xxx:8080/hubot/jira-comment-dm

Imgur

Cleanup

Run the following if you want to clean things up.

  • Delete the service
1
2
macOS%$ kubectl delete service pod-hubot
service "pod-hubot" deleted
  • Delete the pod
1
2
macOS%$ kubectl delete pod pod-hubot-729436916-htw3r
service "pod-hubot" deleted
  • Delete the deployments
1
macOS%$ kubectl delete deployments pod-hubot
  • Delete the container clusters
    If you delete the container cluster, the deployments, service, and pod associated with it are also deleted.
1
macOS%$ gcloud container clusters delete hubot-cluster-free

That’s all.

Overall Impressions

GKE has a lot of concepts, and there’s plenty to learn all at once — deployment, pod, service, kubernetes, and so on — but it’s fun to learn while actually running things.

I was able to set up almost everything on my Mac at hand!
Since it can all be done locally, macOS%$ wasn’t really necessary…

The service we created this time leaves port 8080 wide open to the outside.

Next time, I’ll cover restricting the source of access for port access and updating containers.

[http://kenzo0107.hatenablog.com/entry/2017/05/16/222815:embed:cite]

Running Hubot on GKE + Kubernetes Within the Free Tier — Running from the CLI

https://kenzo0107.github.io/en/2017/05/10/gke-kubernetes-hubot-cli/

Author

Kenzo Tanaka

Posted on

2017-05-10

Licensed under

#

Like this article? Support the author with

×