After I was able to Deploy a Nanobus Web Server, I thought about using Nanbous to replace Nginx. I’ve moved on a little from Deploying Nginx + PHP + git-sync on Kubernetes. I still run Nginx + git-sync on Kubernetes. I’ve removed the need for PHP in my Nginx web servers anymore. I thought, do I still really need to run a full blown Nginx anymore on some of my static servers? Let’s find out!
Setting Up The Repo
I’m not going to bore you with the details here because I’ve covered the bulk of setting up a repo in previous articles such as the one noted above. I will at least cover here what I’m putting into the repo. I’m going to first create a super simple index.html file
<html>
<body>
Testing Content
</body>
</html>
I’ve added this to a content directory in my repo. In the root of my repo, I’ll add in a bus.yaml for the Nanobus web server:
id: static-content-server
version: 0.0.1
transports:
http:
uses: nanobus.transport.http.server/v1
with:
address: ':8000'
routers:
- uses: nanobus.transport.http.static/v1
with:
paths:
- file: /www/www/content/index.html
path: /
One import concept here is to make sure your file value points to where you actually plan to mount this repo. More on that in the next section. The directory structure for the repo looks like this:
.
./bus.yaml
./content
./content/index.html
Setting Up The Kubernetes Deployment
Since my previous configuration was all deployed via Kubernetes, I want to do the same with this. I put together a super simple YAML that has a service and deployment:
apiVersion: v1
kind: Service
metadata:
name: nano-web
labels:
tier: backend
spec:
selector:
app: nano-web
tier: backend
ports:
- name: http
port: 8000
---
apiVersion: apps/v1
kind: Deployment
metadata:
name: nano-web
labels:
tier: backend
spec:
replicas: 1
selector:
matchLabels:
app: nano-web
tier: backend
template:
metadata:
labels:
app: nano-web
tier: backend
spec:
securityContext:
fsGroup: 65533 # to make SSH key readable
volumes:
- name: dir
emptyDir: {}
- name: github-secret
secret:
secretName: github-creds
defaultMode: 288
containers:
- env:
- name: GIT_SYNC_REPO
value: [email protected]:<my-user>/<my-static-repo>.git
- name: GIT_SYNC_BRANCH
value: master
- name: GIT_SYNC_SSH
value: "true"
- name: GIT_SYNC_PERMISSIONS
value: "0777"
- name: GIT_SYNC_DEST
value: www
- name: GIT_SYNC_ROOT
value: /git
name: git-sync
image: k8s.gcr.io/git-sync:v3.1.1
securityContext:
runAsUser: 65533 # git-sync user
volumeMounts:
- name: git-secret
mountPath: /etc/git-secret
- name: dir
mountPath: /git
- name: nano-web
image: nanobus/nanobus
args:
- "run"
- "/www/www/bus.yaml"
ports:
- containerPort: 8000
volumeMounts:
- name: dir
mountPath: /www
It is also very important to again point out these items from the YAML:
...
- name: GIT_SYNC_DEST
value: www
...
args:
- "run"
- "/www/www/bus.yaml"
...
- name: dir
mountPath: /www
The GIT_SYNC_DEST says that we’ll mount the git repo into the www directory. The mountPath says that we should mount the git sync to /www. This means that our repo’s root will be /www/www. This means that our bus.yaml will be located in /www/www/bus.yaml and the index.html file will be at /www/www/content/index.html.
Deploying and Testing
Now that I’ve got everything in place, let’s wave the magic kubectl wand:
% kubectl apply -f ../static/nano-web.yaml
service/nano-web created
deployment.apps/nano-web created
We can check out the deployment:
% kubectl get pod
NAME READY STATUS RESTARTS AGE
nano-web-899b74798-4n5jt 1/2 Error 0 3s
% kubectl get pod
NAME READY STATUS RESTARTS AGE
nano-web-899b74798-4n5jt 2/2 Running 2 (1m ago) 1m
You will want to check the pod a few times because it will start in an Error state until the git sync finishes and makes the volume available. After we’ve got a Running status, let’s check the logs:
% kubectl logs -f nano-web-899b74798-4n5jt -c nano-web
2023-02-15T21:54:21.745Z INFO Initializing codec {"name": "cloudevents+json", "type": "cloudevents+json"}
2023-02-15T21:54:21.745Z INFO Initializing codec {"name": "text/plain", "type": "text/plain"}
2023-02-15T21:54:21.745Z INFO Initializing codec {"name": "text/html", "type": "text/html"}
2023-02-15T21:54:21.745Z INFO Initializing codec {"name": "bytes", "type": "bytes"}
2023-02-15T21:54:21.745Z INFO Initializing codec {"name": "json", "type": "json"}
2023-02-15T21:54:21.745Z INFO Initializing codec {"name": "msgpack", "type": "msgpack"}
2023-02-15T21:54:21.745Z INFO Initializing codec {"name": "cloudevents+avro", "type": "cloudevents+avro"}
2023-02-15T21:54:21.746Z INFO Initializing transport {"name": "http"}
2023-02-15T21:54:21.746Z INFO Serving static files {"path": "/", "dir": null, "file": "/www/www/content/index.html", "strip": null}
2023-02-15T21:54:21.753Z INFO HTTP server listening {"address": ":8000"}
This all looks good so let’s do the final test by checking it with our web browser. I haven’t exposed this yet outside of my cluster so we’ll port forward with a little kubectl port-forward svc/nano-web 8000:8000 and check the browser.
The Battle Continues
As I was writing this article, I noticed that my reference files from Deploying Nginx + PHP + git-sync on Kubernetes
were broken :facepalm:. This is a great opportunity to fix them! I updated my bus.yaml file to include the blog_posts directory:
id: static-content-server
version: 0.0.1
transports:
http:
uses: nanobus.transport.http.server/v1
with:
address: ':8000'
routers:
- uses: nanobus.transport.http.static/v1
with:
paths:
- file: /www/www/content/index.html
path: /
- dir: /www/www/content/my_blog_post_files/20191229
path: /my_blog_post_files/20191229
strip: /my_blog_post_files/20191229
Redeploying and checking the browser yields success! The YAML was returned.
This is a great next question. I had to redeploy to get this to work because it needed a new bus.yaml file. The whole point of this is to be able to update my web content by simply updating a git repo.
Testing For Success
I’ll start by first testing for something that does not exist
% curl http://localhost:8000/my_blog_post_files/20191229/does_not_exist.txt
404 page not found
This error is ok because it was totally expected. This file does not exist, yet. I’ll quickly create that file in my repo:
% echo "testing 1 2 3" > my_blog_post_files/20191229/does_not_exist.txt
Next, I do my git commit and git push to update the repo. After waiting a few seconds for the repo to sync, I’ll run another curl:
% curl http://localhost:8000/my_blog_post_files/20191229/does_not_exist.txt
testing 1 2 3
Success! Let’s see if I change the file and commit will it still update?
% echo "4 5 6 7 8" >> my_blog_post_files/20191229/does_not_exist.txt
% git commit -a
[master 4637b99] Altering test file
1 file changed, 1 insertion(+)
% git push origin master
Enumerating objects: 13, done.
Counting objects: 100% (13/13), done.
Delta compression using up to 8 threads
Compressing objects: 100% (6/6), done.
Writing objects: 100% (7/7), 593 bytes | 593.00 KiB/s, done.
Total 7 (delta 2), reused 0 (delta 0), pack-reused 0
remote: Resolving deltas: 100% (2/2), completed with 2 local objects.
To github.com:<some_user>/<some_repo>.git
9f5d696..4637b99 master -> master
% curl http://localhost:8000/my_blog_post_files/20191229/does_not_exist.txt
testing 1 2 3
4 5 6 7 8
I don’t just serve text documents with this nginx server so let’s see if we can also serve up images. I copied over a quick logo file and tested this: