On Sun, Jul 17, 2016 at 04:32:01PM +0200, Doug Hellmann wrote: > Excerpts from Joshua Harlow's message of 2016-07-15 18:00:37 -0700: > > I assume this is connected into: > > > > https://review.openstack.org/#/c/334014/ > > > > Which was proposed a few weeks ago, I'd be nice to identify the causing > > review and see if we can find the authors to learn about why, > > The KeyError was being raised in error, not as an explicit way to > indicate that a plugin was missing. We classified that as buggy > behavior rather than an API change. There was one other place this was > relying on the behavior and they have fixed their app code (I don't > remember off the top of my head which project that was). > > The best course of action is to change IPA to use the callback mechanism > added as the official API and to require the newer version of stevedore > that supports that callback by raising the minimum in the global > requirements list.
Thanks Doug. g-r change: https://review.openstack.org/#/c/343339/ IPA fix: https://review.openstack.org/#/c/343033/ // jim > > Doug > > > > > If its needed, proposing a requirements block is fine with me (until we > > can get to the root of this). > > > > -Josh > > > > Jay Faulkner wrote: > > > Hi all, > > > > > > I wanted to get a little more light on this bug discovered: > > > https://bugs.launchpad.net/stevedore/+bug/1603542 > > > <https://bugs.launchpad.net/stevedore/+bug/1603542>. Jim posted a > > > summary of it today in #openstack-oslo and didn't get a response, and > > > it's currently breaking IPA's gate. If someone familiar with stevedore > > > could please triage the bug and start moving it towards a resolution, > > > I'd greatly appreciate it, as IPA's gate is broken and stuck until it is > > > fixed. > > > > > > While it's a big hammer I'd like to avoid using, but if there's no > > > moment on this bug by Monday, I'll propose a global-requirements change > > > to blacklist this version of stevedore until a solution is determined > > > since IPA is completely blocked from merging changes. > > > > > > This breakage does not impact the Ironic gate, as Ironic uses prebuilt > > > TinyIPA ramdisk images, and no image has been published with the new > > > breaking stevedore since these problems were caught in CI. > > > > > > Thanks, > > > Jay Faulkner > > > OSIC > > > > > > P.S. Kudos to ironic-inspector folks for writing a great unit test to > > > catch this behavior in our CI rather than having it break users! > > > > > > __________________________________________________________________________ > > > OpenStack Development Mailing List (not for usage questions) > > > Unsubscribe: [email protected]?subject:unsubscribe > > > http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev > > > > __________________________________________________________________________ > OpenStack Development Mailing List (not for usage questions) > Unsubscribe: [email protected]?subject:unsubscribe > http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev __________________________________________________________________________ OpenStack Development Mailing List (not for usage questions) Unsubscribe: [email protected]?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
