Description of problem: linked with bz601170 I met several times problem with submitting job. I do it for tier test purpose for BaseOS, and sometimes happen, that I must add new test, for example tier test have over 200 of scheduled tests. And problematic is, that beaker try to find if all rpms are in beaker repository, and sometimes it fails and show me something like: $ bkr job-submit --convert beaker-xml <Fault 1: "bkr.server.bexceptions.BX:u'Invalid Task: /CoreOS/tzdata/Regression/bz517009-Egypt-DST-2009'"> So I expect some type of fault tolerance option for example --force to do that. Because for me and for tier test is better to schedule it, with missing tasks, than it is not scheduled. Thanks
Placeholder task could be used instead of simple omitting the task as that may put multihost off sync e.g. when a task is unavailable for architecture of one of the machines? Different RPM can be used for these tasks while keep name and role intact. This would also make re-scheduling job after the tasks are uploaded/fixed easier.
(In reply to comment #1) Marian, that a very good point about multihost tests getting out of sync. At the moment though we already have this problem, when a task is excluded because of arch/family: it is just silently removed. So I will do the same thing for this RFE. I will open a separate RFE with your idea of using a placeholder task.
(In reply to comment #1) Bug 674229 is the RFE for inserting placeholder tasks.
http://git.fedorahosted.org/git/?p=beaker.git;a=commitdiff;h=df8b694b709783a5f8884a2df6eeec478e593822
Beaker 0.6.4 has been released.