Yes, it does do that. Moving forward I think Zack and I both agree that the roles in ceph-cm-ansible, that are used by teuthology, should work without having to provide a secrets repo. Unfortunately, the testnode role that was the port of all chef functionality was’t initially built with that in mind. It was built to support multiple labs, but the requirement of the secrets repo to do so needs to go away. To be fair, ceph-qa-chef wasn’t very friendly to use in other labs either though. I’ve started today on documenting the testnode role (the primary role in use by teuthology) so that it’s easier for others to use and understand. I also plan to make the testnode role usable without a secrets repo along with that documentation work. You might still need to override some variables though to make it work for your use case. This can be done with the teuthology yaml for the ansible task or by providing your own secrets repo that you’d install on your teuthology node. My apologies for the rough transition, but I hope we can make this better. In the meantime feel free to ping me with any questions or if you need help. I'd appreciate any feedback you can give in your process of setting up a new openstack lab using ansible. On Jul 8, 2015, at 1:05 PM, Loic Dachary <loic@xxxxxxxxxxx> wrote: > I think the main problem here is that the chef / ansible switch introduces a significant and non trivial backward incompatible change that will break all existing teuthology installation if they try to upgrade. > > On 08/07/2015 16:59, Andrew Schoen wrote: >> Yeah, ceph-cm-ansible really *needs* an inventory repo of some kind to work. If not, most of the roles will either noop or fail with undefined variables. >> >> We need to document this better as it’s clearing a confusing aspect of all this. If we can’t make ceph-sepia-secrets public soon to serve as an example then maybe we can fill that gap with more documentation or maybe a ‘skeleton’ inventory repo that somebody could fork and fill in the blanks. >> >> >> On Jul 7, 2015, at 8:11 PM, Dan Mick <dmick@xxxxxxxxxx> wrote: >> >>> The overall issue here is, of course, there's a lot of vars that cephlab >>> relies on that are not generated by the task itself (in the case of no >>> inventory file). I'm not sure what the plan was to cope with all those >>> vars. >>> >>> On 07/07/2015 05:41 PM, Loic Dachary wrote: >>>> Hi Zack & Andrew, >>>> >>>> With Dan's help we hacked passed the sudo problem with: >>>> >>>> cat > /tmp/ansible.yaml <<EOF >>>> overrides: >>>> ansible.cephlab: >>>> branch: wip-fix-defaults >>>> vars: >>>> ansible_sudo: true >>>> interactive-on-error: true >>>> EOF >>>> >>>> teuthology-suite --machine-type openstack --suite-dir $(pwd)/teuthology/test/integration --suite noop /tmp/ansible.yaml >>>> >>>> with full logs at >>>> >>>> http://integration.ceph.dachary.org:8081/ubuntu-2015-07-08_00:31:07-noop-master---basic-openstack/3/ >>>> >>>> Cheers >>>> >>> >>> -- >>> To unsubscribe from this list: send the line "unsubscribe ceph-devel" in >>> the body of a message to majordomo@xxxxxxxxxxxxxxx >>> More majordomo info at http://vger.kernel.org/majordomo-info.html >> > > -- > Loïc Dachary, Artisan Logiciel Libre -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
