On Thu, Dec 12, 2013 at 03:12:34PM +0100, Martin Bähr wrote: > how would this happen practically? i suppose there would be an rmake > instance running the jobs and all developers have access to that to be > able to watch it and report the issues?
mirrorball actually drives rmake. Probably, we'd just run it under cron, and have it mail errors either to this list or to another dedicated list. Then if the errors persist (wasn't just a transient partial mirror fail, for example) we investigate, find the fix, commit it to the controlling Git repositories, and it picks back up where it left off. We'd use mirrorball in a newer mode than was available when we were talking about "boots", one design to import as a rolling release rather than being errata-centric. > my time is also limited, but i am willing to help. Thanks! The goal is to have several people willing to watch errors and participate in determining whether they are transient or require a change. If unsure, documenting what they researched on the mailing list so the next person can jump in and start from what is already known. The bootstrapping process of the initial import is more work; after that initial import, failures should not be a daily occurrence. _______________________________________________ Foresight-devel mailing list [email protected] https://lists.foresightlinux.org/mailman/listinfo/foresight-devel
