$ curl --silent http://gentoo.mirrors.pair.com/releases/x86/autobuilds/latest-stage3-i686.txt | grep stage3-i686 20130326/hardened/hardened/20130326/stage3-i686-hardened-20130326.tar.bz2 The latest stage3 file points to a hardened stage3. This causes failures if you build a non-hardened system (glibc has issues). A similar issue has happened with the amd64 file pointing to a nomultilib stage3. There must have been some change in how these files are generated, or how builds are made.
This should be fixed now.
(In reply to comment #1) > This should be fixed now. Indeed, it is. There was also the related amd64 bug. Do you believe you've gotten to the root cause on this issue? I'm not sure how it happened in the first place.
(In reply to comment #2) > (In reply to comment #1) > > This should be fixed now. > > Indeed, it is. There was also the related amd64 bug. > > Do you believe you've gotten to the root cause on this issue? I'm not sure > how it happened in the first place. I did find the root cause. I change the location where the stages were stored and that caused confusion for the copy_buildsync.sh script. The amd64 latest stage3 pointing to a nomultilib stage isn't fixed yet. The mirrors are currently ok, but the next time we push a release it's going to break again. I need to fix copy_buildsync.sh or catalyst as the copy_buildsync.sh works if the nomultilib stage is called stage3-amd64+nomultilib*, but not if it's called stage3-amd64-nomultilib*.
