I have same error - did you manage to fix this?

I finally managed to get this running. Maybe this can help. I found that once the install was borked in portainer, there was no bringing it back when you are using stack...something is always remembered it seems, so here are a couple of things i discovered.

1. If install goes sideways, in portainer remove the 4 containers, go to volumes and delete any unused and then you also SSH in and go to /data where you should see a "compose" folder. Inside that will be numbered folder(s), how many will depend on how many times you tried to install it I think (I had 4 or 5). Use rm -r /data/compose - that will wipe out all traces.

2. in portainer, build a new stack with a different name, or at the very least stop the stack - this seems to be a necessary step - then redeploy stack.

3. I left everything the way the instructions said, but didn't want to use the default storage location, so instead of changing the UPLOAD_LOCATION variable from ./library (which is in /data/compose/##) I created a symbolic link from library to a share on my NAS after using mv to rename library to library_backup (use ln -s /mnt/<your share name> /data/compose/##/library. Doing this allowed me to keep everything exactly as the docs indicate, but use my own location for storage.

Hope something here helps you.