/bin/sh: [npm,: not found. ( Kubernetes )

[tron4x]

Hello,

have some problem running vitepress on kubernetes.

My dockerfile looks like this:

FROM node:lts-alpine AS base

FROM base AS buildtime
RUN apk add --no-cache libc6-compat git
RUN apk update && apk add bash && apk add npm

# install dependencies
FROM buildtime AS deps
WORKDIR /app
COPY package.json package-lock.json ./

RUN npm ci

FROM buildtime AS builder
RUN apk add --no-cache libc6-compat git
WORKDIR /app
COPY --from=deps /app/node_modules ./node_modules
COPY . .
RUN npm add -D vitepress
RUN npm fund

CMD ["npm", "run", "docs:dev"]

I have created a simple helm chart to deploy vitepress on kubernetes.
Can be found here: Helm Chart
Everything is starting with the exception of the pod.

In ArgoCD all resources are up running. The pod not.
Logs of pod:

/bin/sh: [npm,: not found

I don't understand why npm is not found even though I install it explicitly.
Everything works locally on my Mac.

I've been looking for solutions for a while, but unfortunately I haven't found anything yet.
Can you help me here please?

Thank you

[brc-dd]

Not sure, gpt solution:

...

FROM builder
CMD ["npm", "run", "docs:dev"]

Alternatively, ditch npm, just run ./node_modules/.bin/vitepress dev docs (or whatever your command is).

[tron4x]

It doesn't work.
Have simplify my dockerfile witn an older node version, but the result is the same.

FROM node:18-alpine3.20 AS base
RUN apk add --no-cache git

WORKDIR /app
COPY package.json package-lock.json ./
RUN node -v && npm -v

RUN npm ci

COPY . .

RUN npm install -D vitepress
RUN npm i vitepress

RUN ls -la /app/node_modules/.bin/vitepress

EXPOSE 5137

CMD ["./app/node_modules/.bin/vitepress"]

Error: ./app/node_modules/.bin/vitepress not found

With dot or without, it doesn't makes matter.

You could also use nginx as a web server and display the static pages but I think it looks better to use vitepress or npm.
Has anyone got this to work on Kubernetes ?

[brc-dd]

Has anyone got this to work on Kubernetes?

Most likely it is not related to vitepress. That binaries are missing somehow. Someone who has worked more on docker/k8s might be able to comment. Ask on StackOverflow or something. Try running any command from npm, it should fail similarly.

Also, I doubt that many people are using k8s with vitepress because you simply don't need it. You shouldn't be deploying the dev version anyway. It will be too slow. And vitepress build generates static site, which can be served with just nginx.

[tron4x]

You mean this here: Nginx
That's how it works.
The second Docker image I built also has all the nginx binaries.
The nginx conf file is provided in the pod via a config map.
I'll post the finished deployment here. I still have to test it a bit.

I was hoping there was another way.
I'll try again, maybe I'll think of something else.

Thanks for any tips though.

[tron4x]

Hi,

now its working on Kubernetes ( with nginx ). 😄
I spent a while on this until it worked properly.
This Dockerfile do the job:

FROM node:alpine3.20 AS base
RUN apk add --no-cache libc6-compat git

# Install build dependencies
FROM base AS deps
WORKDIR /app
COPY package.json package-lock.json ./
RUN npm ci

# Build static website
FROM base AS builder
WORKDIR /app
COPY --from=deps /app/node_modules ./node_modules
COPY . .
RUN npm add -D vitepress
RUN npm run docs:build

# Use nginx to serve the static website
FROM nginx:stable-alpine AS final
WORKDIR /usr/share/nginx/html
COPY --from=builder /app/.vitepress/dist ./
COPY default.conf /etc/nginx/conf.d/default.conf

# Expose necessary ports
EXPOSE 8080 8443

# Start nginx
CMD nginx -g 'daemon off;'

default.conf is from here
root location needs to be adjusted. In my case: /usr/share/nginx/html

Kubernetes resources are up and running:

NAME                             READY   STATUS    RESTARTS   AGE
pod/home-docs-5c68b994c9-smsgj   1/1     Running   0          15m

NAME                TYPE        CLUSTER-IP     EXTERNAL-IP   PORT(S)   AGE
service/home-docs   ClusterIP   10.43.70.150   <none>        80/TCP    6h28m

NAME                        READY   UP-TO-DATE   AVAILABLE   AGE
deployment.apps/home-docs   1/1     1            1           6h28m

NAME                                   DESIRED   CURRENT   READY   AGE
replicaset.apps/home-docs-5c68b994c9   1         1         1       15m
replicaset.apps/home-docs-7874977488   0         0         0       6h28m
replicaset.apps/home-docs-dd8bcfcc6    0         0         0       6h25m
replicaset.apps/home-docs-ddb8dffc8    0         0         0       6h24m

Logs from pod:

> k logs pod/home-docs-5c68b994c9-smsgj -n home

2024/08/25 08:33:15 [notice] 1#1: using the "epoll" event method
2024/08/25 08:33:15 [notice] 1#1: nginx/1.26.2
2024/08/25 08:33:15 [notice] 1#1: built by gcc 13.2.1 20240309 (Alpine 13.2.1_git20240309)
2024/08/25 08:33:15 [notice] 1#1: OS: Linux 6.1.0-22-amd64
2024/08/25 08:33:15 [notice] 1#1: getrlimit(RLIMIT_NOFILE): 1048576:1048576
2024/08/25 08:33:15 [notice] 1#1: start worker processes
2024/08/25 08:33:15 [notice] 1#1: start worker process 7
2024/08/25 08:33:15 [notice] 1#1: start worker process 8
2024/08/25 08:33:15 [notice] 1#1: start worker process 9
2024/08/25 08:33:15 [notice] 1#1: start worker process 10
2024/08/25 08:33:15 [notice] 1#1: start worker process 11
2024/08/25 08:33:15 [notice] 1#1: start worker process 12
2024/08/25 08:33:15 [notice] 1#1: start worker process 13
2024/08/25 08:33:15 [notice] 1#1: start worker process 14
2024/08/25 08:33:15 [notice] 1#1: start worker process 15
2024/08/25 08:33:15 [notice] 1#1: start worker process 16
2024/08/25 08:33:15 [notice] 1#1: start worker process 17
2024/08/25 08:33:15 [notice] 1#1: start worker process 18
2024/08/25 08:33:15 [notice] 1#1: start worker process 19
2024/08/25 08:33:15 [notice] 1#1: start worker process 20
2024/08/25 08:33:15 [notice] 1#1: start worker process 21
2024/08/25 08:33:15 [notice] 1#1: start worker process 22
78.46.98.221 - - [25/Aug/2024:08:33:16 +0000] "GET / HTTP/1.1" 200 10655 "-" "kube-probe/1.28" "-"
78.46.98.221 - - [25/Aug/2024:08:33:22 +0000] "GET / HTTP/1.1" 200 10655 "-" "kube-probe/1.28" "-"
78.46.98.221 - - [25/Aug/2024:08:33:22 +0000] "GET / HTTP/1.1" 200 10655 "-" "kube-probe/1.28" "-"

The trick here is to set the command for starting nginx correctly.
The usual would be: CMD ["nginx", "-g", "daemon off;"]
But that doesn't work! I still get: ["nginx" not found. on Kubernetes.

Passing CMD as a shell argument, however, works very well! 👍