Earlier  
Later
New Search
Posted Nick Remark
#openstack-sdks - 2018-07-09
09:35:40 openstackgerrit tianhui proposed openstack/python-openstackclient master: Compute: Add tag support for server add port https://review.openstack.org/580977
11:10:14 frickler mordred: dtroyer: someone on #openstack just reported issues running with python3.7, how do our plans to support that look like?
11:19:29 mordred frickler: I don't think we've made any yet
11:22:21 frickler mordred: ah, I just noticed that it's not even two weeks old, I was thinking it had been a couple of months. but then, yeah, maybe next cycle
11:23:01 mordred frickler: yeah. I think so :)
12:39:02 openstack Launchpad bug 1780587 in devstack "Creds files in accr directory return Identity API errors" [Medium,Confirmed] - Assigned to Dr. Jens Harbott (j-harbott)
12:39:02 frickler mordred: dtroyer: spotz found an interesting issue, please take a look: https://bugs.launchpad.net/devstack/+bug/1780587
12:39:47 mordred frickler: what are acccr files?
12:41:35 frickler mordred: devstack generates these, similar in use to "openrc", but different files per account/project instead of having parameters. http://paste.openstack.org/show/725342/
12:41:41 mordred oh for the love of
12:41:44 frickler didn't know they existed before, either
12:42:00 mordred well, it's still likely a real bug
12:42:16 frickler mordred: bug in OSC/libs?
12:42:19 mordred there's an osc patch I reviewed over the weekend related to identity_api version
12:42:22 mordred yeah
12:42:24 mordred somewhere
12:42:36 mordred there is confusion as to what identity_api_version is supposed to do
12:44:05 mordred there is code in osc that uses it to set some defaults for authentication - which I think we're just going to have to adopt in openstacksdk even though identity_api_versin is _supposed_ to be a setting controlling which version of the CRUD api to use - and auth_type is _supposed_ to control which version of auth one uses
12:44:43 mordred the thing is - literally nobody understands that - so I think the osc attempt at inferring intent from identity_api_version is better than telling people they're wrong
12:44:56 mordred all that said- I have NO IDEA why that would have broken just now
12:45:05 frickler mordred: yes, even setting auth_type=v3password doesn't solve the above issue
12:45:23 mordred we have not cut any releases of anything with the 'make osc-lib use sdk for session' code
12:45:49 frickler mordred: I tried with various older lib versions, none of them helped. so maybe spotz was confused by other clients working, and it was always broken for osc
12:46:09 mordred maybe so
12:46:22 mordred if you set OS_IDENTITY_API_VERSION=3 does it work?
12:46:28 frickler mordred: yes
12:46:38 mordred k. that's where the bug is at least
12:47:07 mordred openstack.config/os-client-config set a default identity_api_version to 2 - I'm betting that is confusing things
12:47:22 mordred we're also getting very close to being able to remove that default value too
12:48:17 mordred anywho - as a bugfix for devstack, adding OS_IDENTITY_API_VERSION=3 to those files will likely fix the specific bug - but we should remember this use case
14:12:31 dhellmann mordred : +2 on https://review.openstack.org/580898
14:12:40 mordred yay!
14:12:44 mordred dhellmann: tyvm
14:12:55 dhellmann did you need another one right away or do we want to wait for that one?
14:13:02 dhellmann and how broken are things? should I single-approve that?
14:13:34 mordred nah - that's the only one - and the openstacksdk gate is broken without it, which is blocking landing the patch to unbreak the osc gate :)
14:13:50 mordred dhellmann: so, if you're comfortable single-coring, that would be great, but also if you're not that's cool
14:15:16 mordred if you're in the reviewing mood though, https://review.openstack.org/#/c/580900 is eventually needed and is one line - but *totally* non-urgent
14:16:43 dhellmann mordred : ok, I'll give dtroyer a little while to come online. If you need me to approve it later today I will
14:16:46 dhellmann and +2 on the other one
14:17:01 dhellmann oh, stand by, if the gate's broken I'm OK with approving it
14:17:03 openstackgerrit Monty Taylor proposed openstack/osc-lib master: Run OSC unit test tips jobs https://review.openstack.org/580875
14:17:31 dhellmann mordred : will we need a release to unbreak the gate?
14:17:42 mordred of sdk - but one is already teed up
14:17:46 dhellmann ok
14:18:03 mordred as soon as we land the osc-lib patch, I can recheck the sdk patch (which is already approved) and then we can cut a release
14:18:11 dhellmann got it
14:18:25 mordred yay!
14:18:26 dhellmann we don't need an osc-lib release, too?
14:18:36 mordred no - it's just the osc-tips jobs that are busted
14:18:42 dhellmann aha, ok
14:19:24 mordred we actually don't need the sdk release for fixing gate - we just were going to do one anyway, so getting in the patch that will be needed before we can release osc or osc-lib seemed like a good idea :)
14:19:50 dhellmann mordred : I'm going to run an errand (because driving while confused is less dangerous than coding). I'll watch for the release request to be updated and ready to go when I get back.
14:20:17 mordred dhellmann: awesome. maybe have another coffee as you drive for your errand
14:20:25 dhellmann good plan
14:30:35 dtroyer what a great day to start working late! :)
14:31:35 dtroyer a lot of this stuff is the sort of stuff I was hoping to have just "go away" with OSC 4 and some subtle breakage in how these global options worked… so much for that plan
14:32:18 dtroyer the devstack accrc stuff was added by someone (I don't recall) for a specific purpose, if it's bit-rotted I wouldn't be sad to see it go away
14:33:08 dtroyer this all points out how stupidly fragile OSC is with this handling and why I'm ready for that subtle breakage and cleaning out the cruft
14:47:10 mordred dtroyer: ++
14:48:09 mordred dtroyer: that said - I think I've become convinced that the identity_api_version introspection that osc does that occ/sdk doesn't do is more user friendly and we should add it to occ/sdk
15:27:17 openstackgerrit Merged openstack/osc-lib master: Add release note link in README https://review.openstack.org/578459
16:12:20 openstackgerrit Pavlo Shchelokovskyy proposed openstack/python-openstackclient master: Skip calls to glance and nova when got no servers https://review.openstack.org/568344
16:12:21 openstackgerrit Pavlo Shchelokovskyy proposed openstack/python-openstackclient master: Add --name-lookup option to server list https://review.openstack.org/568345
17:16:26 mordred dtroyer, Shrews: I've gotta be offline for the next few hours - when https://review.openstack.org/#/c/580898/ lands, can one of y'all recheck https://review.openstack.org/#/c/580795/ ?
17:34:31 Shrews mordred: sure
19:25:06 openstackgerrit Merged openstack/osc-lib master: Fix exception in log warning when no cloud config exists https://review.openstack.org/580898
19:36:31 Shrews mordred: rechecked
22:22:37 openstackgerrit Merged openstack/openstacksdk master: Add support for processing insecure https://review.openstack.org/580795
23:01:48 openstackgerrit Feilong Wang proposed openstack/openstacksdk master: Fix Magnum cluster update https://review.openstack.org/581157
#openstack-sdks - 2018-07-10
00:51:11 openstackgerrit Merged openstack/osc-lib master: Pass prompt_for_password to OSC_Config constructor https://review.openstack.org/580900
01:04:51 mordred woot
01:04:57 mordred I'm getting excited about these patches landing
01:05:06 mordred Shrews, dtroyer thanks for the rechecks
01:05:14 mordred remote: https://review.openstack.org/580527 Release 0.15.0 of openstacksdk
02:18:11 openstackgerrit Feilong Wang proposed openstack-infra/shade master: Add Magnum cluster support https://review.openstack.org/577673
02:18:35 openstackgerrit Feilong Wang proposed openstack/openstacksdk master: Fix Magnum cluster update https://review.openstack.org/581157
04:08:07 openstackgerrit Feilong Wang proposed openstack-infra/shade master: Add Magnum cluster support https://review.openstack.org/577673
04:58:52 openstackgerrit Lingxian Kong proposed openstack/openstacksdk master: Support to wait for load balancer to be ACTIVE https://review.openstack.org/581198
05:02:01 openstackgerrit Lingxian Kong proposed openstack/openstacksdk master: Support to wait for load balancer to be ACTIVE https://review.openstack.org/581198
05:09:00 openstackgerrit Merged openstack/osc-lib master: Run OSC unit test tips jobs https://review.openstack.org/580875
07:54:28 openstackgerrit Lingxian Kong proposed openstack/openstacksdk master: Support to wait for load balancer to be ACTIVE https://review.openstack.org/581198
09:03:50 openstackgerrit Vu Cong Tuan proposed openstack/openstackclient master: Switch to stestr https://review.openstack.org/581276
09:23:25 openstackgerrit Vu Cong Tuan proposed openstack/openstackclient master: Switch to stestr https://review.openstack.org/581276
10:13:57 openstackgerrit Lingxian Kong proposed openstack/openstacksdk master: Support to wait for load balancer to be ACTIVE https://review.openstack.org/581198
11:21:37 openstackgerrit Vu Cong Tuan proposed openstack/api-wg master: Switch to stestr https://review.openstack.org/581323
11:22:48 openstackgerrit tianhui proposed openstack/python-openstackclient master: Compute: Add tag support for server add fixed ip https://review.openstack.org/581325
11:37:39 openstackgerrit Vu Cong Tuan proposed openstack/api-wg master: Switch to stestr https://review.openstack.org/581323
13:06:14 Luzi dtroyer, you wanted to know whether the osc-included image signing would work, with an image provided via stdin, right?
13:06:31 dtroyer Luzi: yes
13:07:54 Luzi well we face another problem with that: we are currently prompting for a password for the private key (if it is encrypted) via stdin.
13:10:11 Luzi we use the osc_lib/utils get_password therefore, which collides with the image, when that is uploaded via stdin
13:13:41 Luzi is there another way to handle the prompting for passwords? because we need the password to access the key which we use to encrypt the image (in that order)
13:14:11 Luzi so when the image is provided via stdin, i see no way to also provide the password
13:21:12 mordred I don't see a good way to prompt for password and also accept key on stdin
13:21:50 mordred so you might want the answer there to be that if you want to pass the image on stdin, you need to have password somewhere like clouds.yaml - and if you want to be prompted for password, you should maybe have the image in a file
13:22:32 mordred Luzi: (also, sorry I haven't gotten back with you on the image signing code in sdk - a few things broke last week and have been putting the pieces back together)
13:24:37 Luzi mordred, ah that's reasonable, also a lot of people have vacation, so i didn't worry :)
13:24:54 mordred Luzi: also - your patch is failing CI tests - looks like your cryptography addition "cryptography!=1.3.0,>=1.0 # BSD/Apache-2.0" does not match the entry in global-requirements "cryptography!=2.0" - and something with pep8
13:25:07 mordred Luzi: yes - vacation season! :)

Earlier   Later