| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-sdks - 2018-08-16 | |||
| 14:37:53 | gtema | dtroyer: I have reproduced the problem with python 3.4, 3.5, 3.6 using virtualenv and installing osc with pip | |
| 14:38:12 | mordred | rabel: the library itself should continue to work for forever - but new dev is focused on openstacksdk | |
| 14:39:23 | rabel | mordred: so it will always exist, but in the future it will only be calling openstacksdk ? | |
| 14:39:41 | dtroyer | gtema: pip is the fun part and why capping in cliff isn't enough, it needs to be capped everywhere it is called out in order to guarantee pip does the right thing. | |
| 14:39:59 | mordred | rabel: yes | |
| 14:40:12 | rabel | ok, thanks you two | |
| 14:40:20 | mordred | rabel: the idea is to maintain shade's api contract with its users so that the dev re-org doesn't break anyone | |
| 14:40:21 | dtroyer | making the cap in global-requirements is probably the right thing but that is no longer enforced automatically, we'll have to make that change in osc and osc-lib, etc | |
| 14:41:35 | gtema | dtroyer: change in cliff is not passing checks, so I assume it is required it modify upper_constants in "requirements" | |
| 14:43:03 | dtroyer | gtema: this has shifted recently and I don't have it all in my head anymore, possibly yes. I haven't looked at those cliff jobs yet, if they are the functional test it still may be to requirements in osc-lib or osc | |
| 14:43:59 | gtema | dtroyer: no, it's one the one hand job "requirements-check", which complains on requirements conflict, and then osc-tox-unit-tips | |
| 14:44:22 | gtema | s/osc-tox-unit-tips/cliff-tox-py27-neutronclient-tip/ | |
| 15:32:13 | openstackgerrit | Merged openstack-infra/shade master: Fix format in release notes https://review.openstack.org/592507 | |
| 15:32:24 | mordred | rabel: ^^ thanks! | |
| 15:40:26 | edleafe | cdent: Can you or dtantsur|afk run the API-SIG meeting? I'm pretty distracted by my all-day meeting stuff | |
| 15:42:42 | cdent | aye | |
| 17:07:28 | edleafe | cdent_: Ready for your review: https://etherpad.openstack.org/p/api-sig-newsletter | |
| 17:11:03 | cdent_ | edleafe: add the review in [12] to the pending review section? otherwise ok | |
| 17:11:59 | edleafe | cdent_: ah, good catch. Skipped updating that section. | |
| 17:13:54 | edleafe | Fixed. Good to go? | |
| 17:14:37 | cdent_ | yup, ship it | |
| 17:17:17 | edleafe | done | |
| 18:52:17 | harlowja | anyone seen anything like http://paste.openstack.org/raw/728226/ with shade recently? | |
| 18:53:30 | mordred | harlowja: no- but that doesn't look good | |
| 18:53:43 | harlowja | ya, that popped out of our latest kolla-ansible daily run | |
| 18:54:31 | harlowja | https://imgur.com/a/146tl8f pew pew pew, lol | |
| 18:54:47 | harlowja | `File \"/usr/lib/python2.7/site-packages/shade/********cloud.py` is weird, lol | |
| 18:54:55 | harlowja | secret cloud, lol | |
| 18:55:45 | mordred | SO SECRETS | |
| 18:57:16 | harlowja | ya | |
| 18:57:51 | harlowja | `_normalize.py` seems to have left the building in 1.29 | |
| 19:05:38 | harlowja | ya, hmmm, http://paste.openstack.org/show/728227/ can reproduce with 1.28.0 | |
| 19:06:39 | Shrews | harlowja: i blame Nargles | |
| 19:07:16 | harlowja | http://paste.openstack.org/show/728228/ (from using openstack cli which handles it) | |
| 19:19:41 | openstackgerrit | Joshua Harlow proposed openstack/openstacksdk master: On at least pike clouds metadata can be a json blob https://review.openstack.org/592658 | |
| 19:40:13 | openstackgerrit | Joshua Harlow proposed openstack/openstacksdk master: Glance v2 allows people to overwrite metadata key https://review.openstack.org/592658 | |
| 20:38:44 | mnaser | mordred: guilhermesp and i figured out the root cause of that policy issue, turns out sending `enable_snat` is only allowed if you're an admin by neutron policy | |
| 20:39:08 | mnaser | so i think guilhermesp will be opening a github issue soon for this and probably push up a fix for os_router | |
| 20:39:20 | mordred | mnaser: awesome | |
| 20:39:40 | mordred | is it a bug in os_router and not in openstacksdk's 'create_router' method? | |
| 20:39:53 | mnaser | (because the default value in ansible's os_router for enable_snat set to True then that value is always sent) | |
| 20:40:19 | mordred | AH | |
| 20:40:20 | mordred | yeah | |
| 20:40:21 | mnaser | mordred: https://github.com/ansible/ansible/blob/devel/lib/ansible/modules/cloud/openstack/os_router.py#L374 | |
| 20:40:26 | guilhermesp | this is the bad guy mordred https://github.com/ansible/ansible/blob/devel/lib/ansible/modules/cloud/openstack/os_router.py#L310 | |
| 20:40:30 | mnaser | ^ | |
| 20:40:47 | guilhermesp | oh sorry, the bad guy was the link that mnaser sent | |
| 20:40:51 | mnaser | nah its both | |
| 20:41:11 | mnaser | its the default and and the combination that it's always sent | |
| 20:41:11 | mordred | yeah. I grok what you are saying | |
| #openstack-sdks - 2018-08-17 | |||
| 06:35:45 | openstackgerrit | melissaml proposed openstack/openstacksdk master: Remove the duplicated word https://review.openstack.org/592875 | |
| 12:09:43 | openstackgerrit | Merged openstack/openstacksdk master: Remove the duplicated word https://review.openstack.org/592875 | |
| 13:25:01 | openstackgerrit | Chen proposed openstack/python-openstackclient master: WIP: make host command api-version-aware https://review.openstack.org/593049 | |
| 16:00:29 | openstackgerrit | Gage Hugo proposed openstack/keystoneauth master: Add nosec to usage of SHA1 https://review.openstack.org/593094 | |
| 17:39:08 | openstackgerrit | Gage Hugo proposed openstack/keystoneauth master: Change log hashing to SHA256 https://review.openstack.org/593094 | |
| 22:10:35 | openstackgerrit | Merged openstack/keystoneauth master: Change log hashing to SHA256 https://review.openstack.org/593094 | |
| #openstack-sdks - 2018-08-18 | |||
| 07:00:05 | openstackgerrit | Merged openstack/keystoneauth master: add release notes to readme.rst https://review.openstack.org/591943 | |
| #openstack-sdks - 2018-08-20 | |||
| 09:44:37 | dtantsur | morning folks, can I get some reviews on https://review.openstack.org/590770 and https://review.openstack.org/591324 please? | |
| 11:28:56 | openstackgerrit | Deepak Mourya proposed openstack/python-openstackclient master: now we can add description for role creation in OSC https://review.openstack.org/484355 | |
| 14:11:22 | mordred | dtantsur: done | |
| 14:12:06 | mordred | dtroyer: if you get bored this morning ... https://review.openstack.org/#/c/581276 and https://review.openstack.org/#/c/589465 | |
| 14:15:47 | tosky | especially 589465 | |
| 14:15:51 | tosky | I was going to ping about that | |
| 14:15:52 | tosky | :) | |
| 14:42:14 | tosky | mordred: oh, I didn't notice that openstackclient is branchless (while python-openstackclient is branches), so I guess I can use https://review.openstack.org/#/c/588336/ to test it I guess | |
| 14:45:34 | mordred | tosky: yah - should be able to | |
| 14:46:12 | tosky | oki, checking | |
| 15:06:38 | tosky | uhm, something failed | |
| 15:07:03 | tosky | there is a conflict with a version of keystoneclient which seems to come from master (>=3.17.0) | |
| 15:08:13 | tosky | http://logs.openstack.org/36/588336/3/check/openstackclient-check-plugins/7e19a94/ stable/queens | |
| 15:08:30 | tosky | http://logs.openstack.org/15/588515/3/check/openstackclient-check-plugins/7d684b6/job-output.txt.gz#_2018-08-20_14_52_32_608991 stable/pike | |
| 15:08:33 | tosky | mordred: ^ | |
| 15:13:57 | mordred | tosky: hrm. k. will poke at it in a little bit | |
| 15:18:38 | tosky | thanks | |
| 15:28:50 | openstackgerrit | Merged openstack/openstacksdk master: baremetal: add support for VIF attach/detach API https://review.openstack.org/591324 | |
| 15:29:34 | dtantsur | TheJulia: time to celebrate ^^^ :) | |
| 15:35:47 | openstackgerrit | Merged openstack/openstacksdk master: Correct update operations for baremetal https://review.openstack.org/590770 | |
| 15:35:48 | openstackgerrit | Merged openstack/openstacksdk master: Use the base Resource's JSON patch support in Image https://review.openstack.org/591314 | |
| 16:04:28 | dtantsur | mordred: a chain of baremetal backports when you have a minute: https://review.openstack.org/593685 | |
| 16:05:53 | mordred | dtantsur: all lgtm | |
| 17:16:17 | openstackgerrit | Doug Hellmann proposed openstack/service-types-authority master: import zuul job settings from project-config https://review.openstack.org/593705 | |
| 17:36:48 | edleafe | mordred: there has been some talk of your head exploding (https://etherpad.openstack.org/p/placement-extract-stein, L97). So I'm wondering: is there any concern with "placement" being both the project name *and* the service-type? Is there any reason to create a separate project name? | |
| 17:37:15 | mordred | edleafe: not from me | |
| 17:37:22 | mordred | edleafe: project names are completey meaningless | |
| 17:37:46 | mordred | edleafe: the only thing I care about is that services have unique and consistent service types | |
| 17:38:44 | edleafe | mordred: ok, thanks. I remembered something in the distant past (ceilometer time frame) that we didn't want to ever preclude competition or alternates | |
| 17:39:06 | mordred | well, I have never shared that opinion myself | |
| 17:39:08 | mordred | BUT | |
| 17:39:26 | mordred | Idon't think it matters to much - anything consuming the API should mostly ignore the service_name | |
| 17:40:03 | mordred | unless a cloud has more than one service deployed with the same service_type and thus it's impossible to find the right service _without_ a sevice_name | |
| 17:40:29 | mordred | that said - from a service-types-authority perspective, evenin that case it'snot guaranteed that the project_name will be used as the service_name | |
| 17:40:37 | edleafe | Agreed. The scenario I remembered was a deployment wants to use a different tool that better fit their needs | |
| 17:40:59 | mordred | totes. ceph vs swift for object-store is a good example :) | |
| 17:41:36 | edleafe | yeah. If swift were named 'object-store', it would make it difficult to consider an alternative | |
| 17:41:41 | mordred | edleafe: now - where I *do* think we're making a mistake but it would be too much of a PITA to fix so I've kept my mouth shut ... | |
| 17:42:02 | mordred | is that in nova.conf we have configuration settings for various services using their project name instead of their service-type | |
| 17:42:28 | mordred | (this is also what keystoneauth is producing to inject into oslo.config because it's what nova was already doing when we added the feature) | |
| 17:42:44 | edleafe | oh, yeah - that's terrible. But there are historical $reasons, and we were dumber back then :) | |
| 17:42:48 | mordred | yup | |
| 20:00:30 | openstackgerrit | Merged openstack/openstackclient master: Switch to stestr https://review.openstack.org/581276 | |