On Sat, Feb 22, 2014 at 07:41:22AM -0500, Michael K. Johnson wrote: > After committing 7 successful builds, rmake fell over. I tried > again and got 12 successful builds before rmake fell over. Our > rmake cluster is currently very small; it was really sized for > people building packages for foresight. It is a bug in rmake, > but also mirrorball was trying to keep 30 jobs at a time running > which is just noise on our cluster that will make it hard to see > with "rmake q" what the current status is. I have reduced that > to 4 slots and it's working much better.
Most of the failures went away after I cleaned up the chroots that were left behind when rmake broke. However, for some reason even when I cleaned out all the roots before restarting the import, the gcc import still failed one of the builds (x86_64) with an error about a chroot that wasn't cleaned up: Cannot create chroot - old root at /srv/rmake/chroots/gcc could not be removed. The kernel package, which is seeing builds as duplicate, has failed in the same manner. There, I think the duplicate flavors are mirrorball configuration problems where the mirrorball configuration for kernel flavors doesn't yet match the set of kernels Fedora ships. Brett, have you seen anything like these before? _______________________________________________ Foresight-devel mailing list [email protected] https://lists.foresightlinux.org/mailman/listinfo/foresight-devel
