Grant Gainey

Draft until I figure out what is up w/ permissions - can't reproduce locally?!?

### Some notes on performance: Some unscientific performance results follow. The repositories are a selection of "real" repos, of small (zoo), medium (baseos-9) and large (appstream-9, epel) amounts of content....

Drafting temporarily while I think on some suggestions.

Note: belongs to RPM since the "interesting" flags/howto are pulp_rpm-specific.

@dralley What problem do we think needs investigation/solving here? Is it "make sure uploading same NEVRA, diff checksum, into a repo replaces the existing one with the incoming one"?

> So, no "perfect" solutions. Thoughts? I'm always hesitant to rely on build_time, because "when it was built" can have very little to do with "which one do we want"...

Contact with the API workers is impacted by things like idle-connection-timeouts as well - if the call takes "too long", nginx/apache can decide to drop the connection. This leans me...

If/when we decide to implement, [this](https://github.com/pulp/pulp_rpm/blob/main/pulp_rpm/app/tasks/synchronizing.py#L1257) is prob the place. Other thoughts: * exclude must be incompatible with any of the "mirror" options * exclude by Name? By full NEVRA?...

The typical workflow in Pulp is exactly this - you sync the upstream to provide the set of content that's available to the Pulp Admin, and then curate that content...

> It would not however stop cold if you did an on-demand sync first, because that skips the package download process Ah! That's an outstanding observation, good point Daniel.