Fedora, UUIDs, and user tracking

Posted Jan 15, 2019 18:20 UTC (Tue) by davidstrauss (guest, #85867)
Parent article: Fedora, UUIDs, and user tracking

> the "countme" flag could increment once per week, which would give some additional information about the longevity of systems—without providing much information that could be used for tracking

The addition of sequence numbers -- combined with the client IP addresses -- could allow identifying systems in a way similar to the approaches that de-mask NAT by relying on IP and TCP fields like sequence numbers. It would be better for clients to only classify themselves into demographic-style buckets: less than an hour, 1h-1d, 1d-1w, 1w-1m, 1m-1y, 1y+.

Fedora, UUIDs, and user tracking

Posted Jan 16, 2019 13:44 UTC (Wed) by sgallagh (guest, #80524) [Link]

This is something we've been talking about on the list. I think right now we're looking at "first week", "after the first week", "first week after a distro-upgrade", carrying the contents of /etc/os-release which should give us much of what we really need:
* Metrics on how many systems are short-lived (e.g. testing environments) vs long-lived and/or upgraded vs. freshly installed.
* Rough metrics on user counts that are a better approximation than the current guesses at the mirrorlist hits.

Fedora, UUIDs, and user tracking

Posted Jan 16, 2019 18:42 UTC (Wed) by mattdm (subscriber, #18) [Link]

I basically _was_ thinking week-sized buckets, but, sure, making a "log scale" bucket like you suggest would work too.