Closed tfoote closed 7 years ago
Since the problem is within Jenkins I don't see a reason to track this in the issue tracker of this repository. So I am going to close this.
I think the relevant upstream ticket is https://issues.jenkins-ci.org/browse/JENKINS-5150 (which I have already commented on 1.5 yrs ago).
Sounds good. If anyone else runs into this please upvote the upstream jenkins issue and maybe it will get addressed eventually.
I noticed a couple instances of mavros being doubly queued after the latest release.
With the blocking on upstream the first build should not have started while there were still upstream packages building. We have
Block build when upstream project is building
enabled. I believe this is some sort of race condition in the queueing/checking for blocking that possibly should be reported upstream.I've seen this before but never quite as short a trigger queue, nor as repeatable as two arches on the same release.
Details
Build 45 was in the queue, while build 44 was building.
However they were both ultimately triggered by 45539, but it allowed 44 to start despite the fact that upstream jobs were either pending or underway as per the trigger chain for 45.
45 trigger logic
44 trigger logic
Dependency listings:
I saw the same behavior on the armhf from the same release: http://build.ros.org/view/Queue/job/Kbin_uxhf_uXhf__mavros__ubuntu_xenial_armhf__binary/
67 triggers:
66 triggers: