How does this work now?

[robinandeer]

I am (read: @ingkebil is) evaluating Ansible to get stuff automated at Clinical and thought this repo should be a good starting point. Have a few thoughts/questions from my brief overview:

1. Are the roles specific enough to be reusable? The nas.yml playbook simply runs the miniconda + "nas" role 😕 Maybe nothing overlaps between processing/nas but I still have some slight hope we might be able to collaborate more closely across SciLifeLab - would you be open to breaking some stuff up into more specific roles?
2. The group_vars content is encrypted of course but I don't seem to find any place where I can get an idea of what it controls... Is this intentional "security by obscurity" or can I ask for some hints?

[guillermo-carrasco]

HI @robinandeer ,

Absolutely, a closer collaboration is always welcome. However I am not gonna get involved in this, as it wouldn't make much sense since I am leaving soon. Maybe @senthil10 ? @Galithil ? In any case:

1. I think its quite reusable for both of us, if we agree on a common directory structure. It is basically creating a conda environment, installing TACA and adding cronjobs to transfer data and run TACA; something that you also need to do, right?
2. There is nothing about security through obscurity here :) in the group_vars there is only a username, that honestly I don't know why it is not only in roles/nas/vars/main.yaml, I was probably experimenting with variable files hierarchy or so, don't remember.

[johandahlberg]

A small note - something that I've noticed that is really useful is to use defaults instead of vars in the roles themselves, since this can then be overriden. It shows nicely what variables are expected and it has the lowest precedence order so it can be changed by e.g. group_vars later.

[robinandeer]

I totally get that @guillermo-carrasco, only when I did reach out yesterday I was told you are the only person with full insight into this repo, hehe

Thanks you for responding! I will look into how little we can modify to make it work for us. We still haven't implemented TACA for our data processing but this is still the plan

@johandahlberg: I remember something about that, checking it out, thanks!

[guillermo-carrasco]

I assume you are building it general enough so this is still possible

hmm... well, you'll have to judge that, but a lot of NGI specific stuff has been added lately 😓

[robinandeer]

Yeah, sorry... I edited that part out, didn't mean to bash. We could've been more active in the development for sure - we just need to get a pipeline up before summer