stable containers don't work if downed once

[ArnaudCHEVALME]

Description:

When setting up the dev environment, the first launch of a stable container works perfectly. However, doing docker compose down stableXX and then docker compose up -d stableXX, there will be an error where the container says the server is not installed, it will try to install it and fail.

Steps to reproduce

Initial setup

git clone https://github.com/juliusknorr/nextcloud-docker-dev
cd nextcloud-docker-dev
./bootstrap.sh

cd workspace/server
git fetch origin stable31:stable31
git worktree add ../stable31 stable31
cd ../stable31
git submodule update --init

docker compose up -d nextcloud stable31

At this point, both nextcloud and stable31 works.

The breaking point

docker compose down
docker compose up -d nextcloud stable31

nextcloud.local works but stable31.local doesn't

You unfortunately have to remove the volumes to be able to relaunch the stable containers.

[juliusknorr]

Yes, right now stable containers use anonymous volumes which do get destroyed when running docker compose down but some other volumes are kept if -v is left out.

For now I'd recommend to always use down -v or just stop single containers. However we should migrate the volumes to named ones.

[pabzm]

I'm running into this problem if I want to down-and-re-up an instance while keeping another instance up, even with down -v:

docker compose up -d nextcloud stable31
…
docker compose down -v stable31
docker compose up -d stable31

⬆️ These lines leads to the error of this issue.

Is there any workaround for that situation?

[juliusknorr]

The problem is that we use a shared database container I think. As a workaround you can just stop the container instead of destroying it.