Deploying Application code in Persisted Volume in Kubernetes

Question

I am new to Kuberentes and am trying to deploying a simple webapp in it, currently I am building an image with all the dependencies and my application bundled in , uploading the image to repo and then deploying it in the pod, but the issue with this approach is that if I just want to edit my application code (let's say to add a print statement) I would need again build the image , upload it to repo and deploy it in K8.

Can I use the PVC to host my application code and refer it in my init container , is this a standard practice ?

Answer 1

Technically, you can. Practically, as far as I know, it is never a standard practice.

• Majority of PVC solutions are ReadWriteOnce. So it would mean that you can't use scaling feature - replicas. You can't duplicate code easily across pods.

• PVC is usually used for dynamic data: files or database - you separate code and configuration from data using PVC

• Some images detect when there is nothing in PVC so app can be reinitialised from scratch, for example: prometheus, mysql and so on

• PVC could be slower than data in Docker image (clustering mechanisms behind PVC)

• If you have your code in PVC, you can't use anything app code related in Docker build phase (some testing, libraries resolution, etc.)

• Pipeline may move away from standards (apps are usually built and deployed in a pipeline in Kubernetes)

• and other consistency issues could occur

In case you need to edit code in development environment, other methods are used:

• Locally, you can mount your local app folder to folder on Docker container, for instance it's volumes feature in Docker

• Use minikube to recreate all containers locally for convenient development

• Use tools to attach to containers such as https://code.visualstudio.com/docs/remote/attach-container

There should be a lot more methods, but you haven't mentioned the reason why building image with code is an issue.