#3000 Actually, really deal with people who don't have bz email addresses but need them
Closed: Fixed None Opened 12 years ago by toshio.

We have two scripts that deal with the confluence of fas and bz. One syncs package ownership information from pkgdb+fas => bugzilla. This script operates through the json interfaces to both pkgdb and fas. This runs on bapp01. The other syncs packager permissions from fas => bugzilla. This runs on fas01. It talks to the fas db directly.

To enable contributors to have new account permissions as rapidly as possible, these scripts are run hourly. However, if the scripts fail to set permissions/ownership for a package (the most common being because some of the accounts being changed do not have a valid bugzilla email address) email is sent hourly to tell people about the problems. Currently, the fas => bugzilla sync is going to the contirubtors in question. The pkgdb+fas=>bugzilla sync is sending email only to me (where it's being ignored when I don't have time to email nag mails to the contributors in question).

What can we do about this? We can make a persistent record (either separate ones for fas01 and bapp01 or we can run the fas+pkgdb script on fas01 even though it doesn't need to) that shows who has been emailed in the past and not nag them hourly. If we do this, we need to also have some method of cleaning these accounts up. Here's a sample proposal:

For pkgdb+fas=>bugzilla sync, if after two weeks, the problem is not taken care of, a second email is sent outlining the next step. If still not fixed one week later, the packages the user owns will be orphaned and the packages they have approveacls, commit, and watchbugzilla on will remove those acls from the user.

For fas=>bugzilla sync, if after two weeks, the problem is not taken care of, a second email is sent outlining the next step. If still not fixed one week later, the packager will be removed from the packager group (and provenpackager since that depends on it). The script will also have to remove the user from packagedb as outlined above.

This benefits us in admin time (as we don't have to spend time emailing the people, removing account permissions, and orphaning packages manually) but has the potential to be disruptive (as some people don't work on things until '''after''' a deadline has past and by then, they'll have been removed so it'll be too late.)

Alternate proposals welcome.


This sounds fine for cleaning up the current list, but I wonder: could we have a check when someone is sponsored into packager that if they don't have a valid bugzilla email, it just refuses to add them? Then the problem punts back to them and their sponsor...

Hum, I suppose they can change email later and mess it up too. So there would need to be another check at the 'change email' thing.

Also, what should happen in the case (like mine) where the BZ account was an @redhat.com, but is now closed because they left RHT? I currently get an error page with a URL (not even a link) to an internal RT ticket.

kevin took over the nagmail responsibilities so closing this out for now. We can reopen if we want to look again at removing the burden from an actual human's shoulders.

Login to comment on this ticket.

Metadata