Hello! Good afternoon!
Don wanted me to try and replicate an issue on 6.8
So, I cloned the develop branch and attempted to spin up the docker-compose-dev for 6.8
There were some issues which I don't think the team could possibly do anything about like MinIO dying and such, which I used some workarounds for
But, then it started pulling Payara 7 and Postgres 17 for 6.8
I think this is one of the quirks of using tags like :latest or :unstable
But, do we want to think about container reproducibility? I just wanted to collect some thoughts
We're talking about this file at this version, right? https://github.com/IQSS/dataverse/blob/v6.8/docker-compose-dev.yml
image: postgres:${POSTGRES_VERSION}
POSTGRES_VERSION=17 is expected. It's from https://github.com/IQSS/dataverse/blob/v6.8/.env
We've been using PG17 in dev for a long time.
Payara should be 6, though, not 7. Specifically <payara.version>6.2025.3</payara.version> according to https://github.com/IQSS/dataverse/blob/v6.8/modules/dataverse-parent/pom.xml#L152
My workflow would be something like this:
I think this is what caused a Payara 7 pull
"APP_IMAGE=gdcc/dataverse:unstable"
ah, from .env, yes
I guess you have to know to build the base image if you're going back in time like that.
Should we document it, at least? :thinking:
It makes sense to. I kept running into a brick wall.
Also the note about MinIO too, if its not already documented!
docker pull fails on everything
Hmm. Ok. Where would you look for those tidbits?
Within the container guide, maybe a section named backwards compatability or legacy installs?
I had to use this for minIO:
docker pull pgsty/minio:latest
docker tag pgsty/minio:latest minio/minio:latest
Sure, sounds find. Do you want to make a PR?
Sure! I'll make a PR
Thanks!
I just added this to the agenda for Thursday. Please see #containers > meetings.
Please note that APP_IMAGE is overwritten by whatever Maven has in app.image. If you are spinning up using docker directly, you'd fall into that trap.
It's really hard to maintain compatibility between both worlds ![]()
Sorry we didn't get to this today.
no worries, today's discussions were more important in my opinion
Last updated: Oct 02 2026 at 18:51 UTC