Enhance replicate feature

[PotentialIngenuity]

Is your feature request related to a problem? Please describe.
I performed a replicate with a tag filter and found that it deleted every remote/repository/distribution that didnt match the filter. This is not how I would expect the feature to work.

We have several repos we mirror from public repos that get updated once a week. We also have internal repos that get pushed to irregularly. We would like to replicate these repos separately but we currently cannot since anything outside of the filter gets deleted.

Describe the solution you'd like
I think it should never touch anything outside of the filter. Only change what the user is targeting. The removal step could still be run for the filtered content just don't remove anything outside of the filter.

[mdellweg]

This is how the replicate feature was designed and I'm unsure we can actually change that. The idea is that a replica is an hands off box that just fully mirrors a selected subset of upstream distributions.
I think, you can however limit its reach to a domain (given all your plugins already support that).

In case you need much more control, consider writing an ansible playbook a bit like this: mdellweg/squeezer@f426142

[dkliban]

It would be better if the replicate task would create on the replicate Pulp, the pulp labels that are specified in the UpstreamPulp config. THen when the repositories, remotes, and distributions are created, these labels are applied to them. In subsequent runs of the replicate task, the replicate will only consider local repositories with the labels. All other repositories will not be affected by the replicate task. This will allow users to run replicate on a subset of repositories.

[PotentialIngenuity]

This feature has become more prevalent as I have added other plugins to our environment. Those being containers and python. Each of those occasionally have errors and cause the whole replication task to be marked as a failure. Its makes error handling hard trying to distinguish which plugin actually failed. I was able to create a process that scrapes the logs for the error and this has made it wayy quicker to diagnose the problem.

I would still like to be able to replicate individual repos based on a tag or even plugin type. We have a CI/CD pipeline that builds our rpms and pushes them through our environments. Including the public rpms, containers and python packages is unnecessary and makes the jobs take too long.