I think the biggest reason, is that

- we don't have anyone that is familiar enough with mongodb to admin it without issues

- mongodb licensing is a bit uncomfortable, while from what I've read using mongo isn't really an issue, but apparently running it as a service might require a comercial license for us

The rest are more like questions

- I assume it's scalable base on a backing database, but is it somewhat tested for usage in a scaled deployment? Assuming mongo supports clutering and Grain is stateless?

- Is logging easily parseible and uniform? Like journald could capture it easily and I could use something like vector to capture and transform it?

- Is state easily shared either by database or with a small number of files?

- Would you consider implementing either docker secret store, secret files, or HCP vault integration for passwords/connection strings?