Description of problem: Sometimes jobs stay in queued state forever. When rescheduling such job, the newer job is picked and the old one not. Version-Release number of selected component (if applicable): 0.9.4 How reproducible: Occasionally, usually when testing 5.3.LL virt (about 40 jobs submitted in short time). Steps to Reproduce: Unknown Actual results: Jobs in queued state forever (even with high priority). Expected results: Oldest jobs with same priority are picked first and no jobs are queued forever. Additional info: Examples of forever queued jobs https://beaker.engineering.redhat.com/jobs/310815 https://beaker.engineering.redhat.com/jobs/310814 https://beaker.engineering.redhat.com/jobs/310288 Probably different issue than bug 668473.
I'm fairly sure this is the same as bug 701254, which is that the guest recipes are scheduled first into a smaller lab which does not contain any systems which match the host recipe. This will be solved with bug 655009, since we won't be scheduling guest recipes anymore. *** This bug has been marked as a duplicate of bug 701254 ***