I just realized that we didn't deploy the fileserver in our deployment at all and instead enforce usage of s3 as we didn't see the need for it. We might have to clarify that in that stack though then

We would "donate" back to the community a docker stack template that can be used to set up the community edition. We have been running this on our Hetzner Docker Swarm for something of a year or so now? Not much in terms of HA for mongodb or Elastic Search, but we get "high enough" availability with this setup.

What I could do until we have the time to do this properly:

I could dump the whole thing but with all our credentials removed. As I said, we removed some things and some new components we are not using yet

also i wanna say that clearml is by far the least insane offering in the mlops space. and that's a good thing. no weird marriage between kubernetes and the platform just because. I applaud that design decision

yeah, I was just wondering why we did that and whether other people might want to do the same