Hello Chris, thanks for the detailed explanations!
Concerning restarting bffe and the build coordinator, I noticed they had not been restarted after a machine upgrade, so I thought it could not disturb to do so. On Sun, Oct 11, 2026 at 09:38:54AM +0100, Christopher Baines wrote: > Clicking back to the derivation in the data service, it shows no builds > for mariadb: > > https://data.qa.guix.gnu.org/gnu/store/lzzmmb1q7pqk1s85wimrqnw60v7hj2w2-mariadb-10.11.14.drv Something strange had happened for mariadb: Probably due to a merge conflict, the top commit of the world-rebuild branch ("Use built-in delete-all-but.") had made an "update" of the package to version 10.11.19, but without changing the hash (and anyway, this would have had nothing to do with this commit). I noticed it on CI, which complained about the hash mismatch. I have silently corrected the version back when rebasing on master. Regardless, mariadb is failing now on CI: https://ci.guix.gnu.org/build/25654695/log >From your explanation, I understand there is a problem in the data service, but do not quite grasp where! The slurm inputs show three failed builds for mariadb here: https://bordeaux.guix.gnu.org/build/97f96879-8f39-4ab9-bc78-e71ff7e71760 But then clicking on "mariadb" leads to the package with only "Unknown" build status: https://data.qa.guix.gnu.org/gnu/store/lzzmmb1q7pqk1s85wimrqnw60v7hj2w2-mariadb-10.11.14.drv Is this not an inconsistency in the database? In the meantime, I will just resubmit mariadb by hand; even if it fails, this should then update the number of blocked rather than unknown builds. Andreas
