How to upgrade the application correctly

[zentavr]

Hi,

There are lines at the docker-entrypoint file:

docker-strapi/images/strapi-alpine/docker-entrypoint.sh

Lines 25 to 37 in 5b24cf8

	elif [ ! -d "node_modules" ] || [ ! "$(ls -qAL node_modules 2>/dev/null)" ]; then
	if [ -f "yarn.lock" ]; then
	echo "Node modules not installed. Installing using yarn ..."
	yarn install --prod --silent
	else
	echo "Node modules not installed. Installing using npm ..."
	npm install --only=prod --silent
	fi

If we upgrade strapio to the newer version, docker-entrypoint sees that the node_modules folder exists and wont install the upgrades if there are any. How to perform here correctly?

[naskio]

Why not just upgrading the docker image ?

[zentavr]

@naskio this is obvious actually. I'm just wondering how to upgrade all those modules (if needed) when I upgrade docker image?

[zentavr]

@naskio I'm currently building the Helm Chart which installs strapi.io (testing that with K8S in AWS EKS). We are going to have shared NFS folder between the pods, that's why I'm asking.

The code of the helm chart (WIP) is here

[cplepage]

Why not just upgrading the docker image ?

If we pull the newer image, but reuse the same volume for /srv/app that contains the .../public/uploads and ../src/api directories which are vitals for data persistence. Then with the line pointed out by @zentavr, even with a pull, the node_modules won't update and neither the strapi version.

Or maybe I am missing something ?

[zentavr]

Right now we are building our own image.
We use current image in development mode in order to create the collections (which creates the files in the src/api folder. Then we commit that into git and have ci/cd pipeline on top of that which runs in AWS Codepileline and builds and spins up the image from our personal repository.

probably we changed dockerfile and an entry point for that.