Le dimanche 17 juin 2007 à 09:27 -0400, Jesse Keating a écrit : > On Saturday 16 June 2007 17:38:34 Nicolas Mailhot wrote: > > Le samedi 16 juin 2007 à 16:39 -0400, Todd Zullinger a écrit : > > > Nicolas Mailhot wrote: > > > > Seems koji was not designed with mock in mind and the koji > > > > repositories are not stable enough for yum (which mock uses). Koji > > > > build repo probably need mirroring or redesign if mock is to be > > > > supported post Fedora 7. > > > > > > I disabled the local repo in the F7 mock config and things worked much > > > better. If this isn't a known problem between mock and koji, which of > > > the two components should a bug be filed under? > > > > It's a problem with koji which static repos have not been designed to be > > remotely accessed through yum > > Actually that's exactly what the static-repos were created for. The > static-repos are publically accessable, perhaps it was a network timeout? > This is the same functionality as the old "local" repo that was in mock, > which pointed at the plague needsign queue. Can you not click on that > repodata file url and download it? Can your system not reach the koji > website? I'm afraid they're not working. They were more or less ok when koji was released and no one was using them (except for a few testers:)) but now koji is in cruise mode and a koji-using mock was released they just fall over all the time. You have to restart yum/mock a dozen times before it manages to download repodata or packages (and I've a proxy so my system never downloads the same rpm twice) (for repodata the fact we have a merged repo and koji includes debuginfo stuff is not helping — koji repodata is huge, but even direct package download is not working 100% of the times, and as yum has been designed to fail at the slightest problem mock/yum koji runs will just fail most of the times) Koji server logs probably trace all the package/repodata interupted downloads. Each one is sufficient to kill a mock/yum run -- Nicolas Mailhot
Attachment:
signature.asc
Description: Ceci est une partie de message =?ISO-8859-1?Q?num=E9riquement?= =?ISO-8859-1?Q?_sign=E9e?=
-- fedora-devel-list mailing list fedora-devel-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/fedora-devel-list
