gcs-copy is a simple binary that copies files from one location in a gcs bucket to another It facilitates providing a public static version of your chartmuseum server via direct public gcs bucket access
You will need to generate a service account JSON key for your cluster and store that as a secret in Kubernetes Then you can provision gcs-copy as a cronjob with something like this:
apiVersion: batch/v1beta1
kind: CronJob
metadata:
name: gcs-copy
spec:
schedule: "*/1 * * * *"
jobTemplate:
spec:
template:
spec:
containers:
- name: gcs-copy
image: i0nw/gcs-copy:0.0.1
command: ["gcs-copy"]
env:
- name: BUCKET_NAME
value: "jenkinsx-chartmuseum"
- name: COPY_FROM
value: "charts/index-cache.yaml"
- name: COPY_TO
value: "index.yaml"
- name: GOOGLE_APPLICATION_CREDENTIALS
value: "/gcs-service-key/i0ntest.key.json"
volumeMounts:
- name: gcs-service-key
mountPath: /gcs-service-key
readOnly: true
volumes:
- name: gcs-service-key
secret:
secretName: gcs-jenkinsx-chartmuseum
restartPolicy: OnFailure
You will need to adjust your chartmuseum server settings to prefix the "charts" directory and use gcs as storage.
Now whenever you hit the api to upload a chart simply run helm repo update afterwards to force a recreation of the index-cache.yaml
You now have a highly available (serverless) public endpoint to serve your helm charts from!
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent