Fix bug in dataservice use of obj store settings

[robertbartel]

Fixing usage of ServiceSetting object in service's main lifespan() func, as use was of earlier names for properties, which were changed in code review but not updated here; also bumping package version.

[robertbartel]

@christophertubbs, since Austin probably won't be able today, when you have 5 minutes, could you give this a quick review? It's very small, fixing some uses of attributes of ServiceSettings (that isn't change and can be found here) in the data service, which I think we just missed when we modified some the names during that PR.

[christophertubbs]

Any way I can reasonably complain about "object_store_exec_user_name" having exec in it? Does that mean "Object Store Execution User Name"? Also, is that user name used anywhere else? I would've assumed it should be something along the lines of access_key if it's going to be used for a access_key parameter.

I imagine this pickiness is out of scope though. Let me know if that's the case.

[robertbartel]

A fair question, @christophertubbs, but there is a reason behind it (albeit convoluted).

The Python attributes need to match the names of some configured Docker secrets in the main stack so that the ServiceSettings class can load those attributes' values from the secrets.

The main stack Docker secrets - object_store_exec_user_name/object_store_exec_user_passwd - are named that way to correspond to analogous secrets in the object_store Docker stack, with "object_store_" prefixed to provide context.

The object_store Docker stack secrets - exec_user_name/exec_user_passwd - are name as such to differentiate them from access_key/secret_key. The latter are the admin credentials for MinIO, and their names are dictated by the MinIO Docker image. exec_user_name/exec_user_passwd are for the user credentials that should be used during normal execution of DMOD.

[christophertubbs]

exec_user_name/exec_user_passwd are for the user credentials that should be used during normal execution of DMOD.

So... you're saying user_name and user_password (resulting in object_store_user_name and object_store_user_password per the above formulas) would fit? 😄

That makes sense, and that makes naming complaints even more out of scope than I thought. I'll throw a tantrum about that later, lol.

[christophertubbs]

lgtm!

But seriously, good fix.