Unique UUIDs from multiple processes on same machine

Question

To tag the data we create, I'm considering using uuids. Security is not an issue, so I was going to use version 1 (date-and-mac-address-based). The only concern is that each user may be creating multiple data files at once from different processes with multiple threads. Assuming Python's uuid library is thread-safe (though it doesn't look it), that still leaves the multiple process issue. I'm considering suffixing the uuid with a dash and the process number.

Since our group has little experience with uuids, are there any issues I need to keep in mind? How is the multiple process issue usually handled?

Answer 1

Just use uuid4 for completely random UUIDs. There is no need to worry about collisions.

edit in response to comment: In my experience, redundant data leads to inconsistencies sooner or later. There is a reason that avoiding redundancy is a dogma of relational database design.

So don't use the UUID as a "redundancy backup" for the actual "originating computer" and "timestamp" data. Either use it as a pure unique ID carrying no other information, or don't use it at all.