| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-sdks - 2018-06-08 | |||
| 14:00:11 | mriedem | is this where osc talk happens? | |
| 14:00:18 | mriedem | hard to know from https://wiki.openstack.org/wiki/IRC | |
| 14:04:44 | frickler | mriedem: if you want to talk osc, this is the place. not much happening, though | |
| 14:07:15 | openstackgerrit | Sami Makki proposed openstack/python-openstackclient master: Fix the `role implies list` command. https://review.openstack.org/557359 | |
| 14:07:36 | mriedem | frickler: oh i want to talk osc alright | |
| 14:08:21 | mriedem | i'm just going through https://etherpad.openstack.org/p/compute-api-microversion-gap-in-osc so i'm likely to have random questions, and just wanted to make sure this was the right place | |
| 14:08:58 | mriedem | for example, what's the process on changing how options work, i.e. https://review.openstack.org/#/c/460059/ | |
| 14:09:23 | mriedem | openstack server migrate --live <host> requires a host - but we don't actually want people specifying a host during live migration. i suggest we make --live just mean live migration, and add a --host option. | |
| 14:10:12 | mriedem | the problem with live migration with a host is that before microversion 2.30, (which is the default in osc, uses 2.1), it will force the host and bypass the scheduler | |
| 14:10:17 | mriedem | which likely means breaking stuff | |
| 14:11:20 | mriedem | changing those options is clearly a behavior change, but is that ok? with a major version? | |
| 14:11:28 | dtroyer | mriedem: live migrate has a number of problems and needs major cleaning up. there are at least two reviews addressing that and someone who understands how it actually needs to reconcile it all | |
| 14:11:49 | dtroyer | We're planning v4 which will break things like this, I've been sauying that for way too long | |
| 14:12:13 | mriedem | "someone who understands how it actually needs to reconcile it all" yeah that's me | |
| 14:12:15 | dtroyer | IIRC there was at least one proposal to clean up live migrate without breaking currently-working behaviour | |
| 14:12:19 | mriedem | i've commented on chen rui's change above | |
| 14:12:25 | mriedem | that's chenrui's | |
| 14:12:35 | mriedem | it adds a --live-auto option for no host cases | |
| 14:12:42 | mriedem | that's a bit clunky though imo | |
| 14:13:31 | dtroyer | its all clunky. this would help me: write up the way you would handle live migrate for 4.0 (no back-compat) and lets see how much of that can be done without breaking things | |
| 14:14:05 | dtroyer | some of what is there now doesn't work, I'll break that, after all nobody is using it, right? | |
| 14:16:23 | openstackgerrit | Matt Riedemann proposed openstack/python-openstackclient master: Add note about version 2.5 when listing servers using --ip6 https://review.openstack.org/573709 | |
| 14:16:42 | mriedem | dtroyer: yup, already did in the commit message of https://review.openstack.org/#/c/460059/ | |
| 14:16:47 | mriedem | suggest a --live option and a --host option | |
| 14:16:53 | mriedem | and then we can add a --cold option too | |
| 14:17:09 | dtroyer | thanks | |
| 14:52:00 | mriedem | is there an easy-ish way to wrap fields in a show output? http://paste.openstack.org/show/722985/ | |
| 14:52:08 | mriedem | the *_details lists in there wrap hard in a console | |
| 14:53:35 | mriedem | i guess i can use --fit-width | |
| 14:55:01 | mriedem | ooo format_list_of_dicts | |
| 15:00:35 | mriedem | i wish i had abandon auth on these osc compute patches :) lots of really old merge conflict cruft in here | |
| 15:08:19 | mriedem | this looks like a bug https://github.com/openstack/python-openstackclient/blob/master/openstackclient/compute/v2/server.py#L132 - why would show server get the server, then get it again? | |
| 15:09:31 | mriedem | oh i see: https://github.com/openstack/python-openstackclient/blob/4cd702c3610e4e6b40695ff9ea8eecf7c0735c68/openstackclient/compute/v2/server.py#L120 | |
| 15:10:42 | mriedem | sort of | |
| 15:10:54 | mriedem | there is just so much new code to chew on | |
| 15:14:38 | openstackgerrit | Matt Riedemann proposed openstack/python-openstackclient master: Use Server.to_dict() rather than Server._info https://review.openstack.org/573744 | |
| 15:19:11 | slaweq | dtroyer: hi, can You maybe take a look at https://review.openstack.org/#/c/567620/ - I answered to Your comment there and I would like to know what You think about it | |
| 15:49:03 | openstackgerrit | Matt Riedemann proposed openstack/python-openstackclient master: Optimize _prep_server_detail to avoid redundant find_resource https://review.openstack.org/573760 | |
| 16:22:07 | dtantsur | mordred: back to the topic of retries, ironic and rate-limiting taskmanager. can we implement the taskmanager so that it takes over the retries parameters and implements them instead of passing into Session? | |
| 16:26:07 | mordred | dtantsur: potentially - or I think what I'd suggest is implementing it in the adapter layer so that the adapter is submitting the retries into the taskmanager | |
| 16:26:28 | mordred | which would make sure that a retry doesn't double-hit an api you're rate limiting against | |
| 16:26:56 | mordred | dtantsur: how about I get the complex task manager at least proposed into the sdk today so we've got a sane thing to look at | |
| 16:27:01 | dtantsur | mordred: I guess my question boils down to whether we'll be able to simulate the same interface (session.do_something(status_code_retries=5)) even with rate limiting | |
| 16:27:07 | dtantsur | +++ | |
| 16:27:30 | dtantsur | another random thought: wdyt about wiring things like baremetal_status_code_retries in the configuration? so that users can adjust it? | |
| 16:28:01 | dtantsur | similar to how you wire in default microversions | |
| 16:30:11 | mordred | dtantsur: yes, absolute can do that same interface - that's how it works now at the adapter layer | |
| 16:31:12 | mordred | dtantsur: yes. I think that's a great idea (wiring retries) - I don't think we should make the codes themselves configurable (since those are pretty specific per-service) - but retry count and delay seem like reasonable end-user tunables, yeah? | |
| 16:31:43 | dtantsur | yep (we cannot tweak delays, they're just exponential, but we can tweak retries) | |
| 16:50:43 | openstackgerrit | Dmitry Tantsur proposed openstack/openstacksdk master: Allow configuring status_code_retries and connect_retries via cloud config https://review.openstack.org/573780 | |
| 16:50:45 | dtantsur | mordred: something like this ^^^ won't pass the CI until keystoneauth is released | |
| 18:46:08 | openstackgerrit | Matt Riedemann proposed openstack/python-openstackclient master: Fix server show for microversion 2.47 https://review.openstack.org/573813 | |
| 19:52:48 | openstackgerrit | Matt Riedemann proposed openstack/python-openstackclient master: Fix server show for microversion 2.47 https://review.openstack.org/573813 | |
| 20:40:35 | openstackgerrit | Matt Riedemann proposed openstack/python-openstackclient master: Mention 2.51 and beautify events output for openstack server event show https://review.openstack.org/573830 | |
| 21:11:34 | openstackgerrit | Matt Riedemann proposed openstack/python-openstackclient master: Mention compute API 2.50 in openstack quota show --class https://review.openstack.org/573834 | |
| 21:19:58 | mriedem | on these last 2 changes ^ for some reason when i build the docs, the updated command help isn't showing up in the docs | |
| 21:20:03 | mriedem | via tox -e docs | |
| 21:20:12 | mriedem | maybe how the docs job works for publications is different? | |
| 21:23:47 | mriedem | i guess i still can't add a 2nd task to https://storyboard.openstack.org/#!/story/1751104 - same 500 timeout | |
| #openstack-sdks - 2018-06-09 | |||
| 02:41:43 | openstackgerrit | kaka proposed openstack/microversion-parse master: fix tox python3 overrides https://review.openstack.org/573880 | |
| 04:24:09 | openstackgerrit | Merged openstack/keystoneauth master: fix tox python3 overrides https://review.openstack.org/572906 | |
| 11:17:08 | openstackgerrit | Alessandro Nesta proposed openstack/os-service-types master: fix tox python3 overrides https://review.openstack.org/573950 | |
| 13:16:25 | openstackgerrit | Alessandro Nesta proposed openstack/python-openstackclient master: fix tox python3 overrides https://review.openstack.org/573975 | |
| 13:26:54 | openstackgerrit | Matt Riedemann proposed openstack/python-openstackclient master: Mention 2.51 in help for openstack server event show https://review.openstack.org/573830 | |
| 13:32:18 | openstackgerrit | Matt Riedemann proposed openstack/python-openstackclient master: Mention compute API 2.50 in openstack quota show --class https://review.openstack.org/573834 | |
| 14:05:49 | openstackgerrit | Monty Taylor proposed openstack/openstacksdk master: Honor service-type aliases in config https://review.openstack.org/566929 | |
| 14:15:25 | openstackgerrit | Monty Taylor proposed openstack/openstacksdk master: Pass default_microversion to adapter constructor https://review.openstack.org/557935 | |
| 14:15:36 | openstackgerrit | Monty Taylor proposed openstack/openstacksdk master: Rename service_key to service_type https://review.openstack.org/558049 | |
| 14:15:53 | mordred | dtantsur|afk: ^^ I have hopefully addressed your concerns - and thanks! | |
| 14:31:26 | openstackgerrit | Mohammed Naser proposed openstack/python-openstackclient master: Fix incorrect functio name for listing role inferences https://review.openstack.org/573990 | |
| 14:31:38 | openstackgerrit | Mohammed Naser proposed openstack/python-openstackclient master: Fix incorrect function name for listing role inferences https://review.openstack.org/573990 | |
| 14:51:29 | openstackgerrit | Mohammed Naser proposed openstack/python-openstackclient master: Switch to InferenceRuleManager for inferred roles https://review.openstack.org/573997 | |
| 17:11:26 | openstackgerrit | Merged openstack/python-openstackclient master: Add system role functionality https://review.openstack.org/524416 | |
| 17:11:28 | openstackgerrit | Merged openstack/python-openstackclient master: Update role document to include system parameter https://review.openstack.org/573303 | |
| 17:11:29 | openstackgerrit | Merged openstack/python-openstackclient master: Fix urls in README.rst https://review.openstack.org/573140 | |
| 17:11:30 | openstackgerrit | Merged openstack/python-openstackclient master: Add bgp commands to neutron decoder https://review.openstack.org/559742 | |
| 17:38:59 | openstackgerrit | Hongbin Lu proposed openstack/python-openstackclient master: Support filtering port with IP address substring https://review.openstack.org/532324 | |
| 18:16:44 | openstackgerrit | Dean Troyer proposed openstack/python-openstackclient master: Fix RuntimeError when showing project which has extra properties https://review.openstack.org/530207 | |
| 18:43:45 | openstackgerrit | Merged openstack/python-openstackclient master: Remove deprecated ip floating commands https://review.openstack.org/546924 | |
| #openstack-sdks - 2018-06-10 | |||
| 04:35:24 | openstackgerrit | Alessandro Nesta proposed openstack/python-openstackclient master: fix tox python3 overrides https://review.openstack.org/573975 | |
| #openstack-sdks - 2018-06-11 | |||
| 09:10:30 | kimamisa | Hi all. Just doing some advertisement for change, which now includes some functionnal tests ! https://review.openstack.org/#/c/557359/ | |
| 09:10:42 | kimamisa | *my change | |
| 14:55:38 | openstackgerrit | Monty Taylor proposed openstack/openstacksdk master: Import rate limiting TaskManager from nodepool https://review.openstack.org/574285 | |
| 15:02:03 | mordred | dtantsur: ^^ | |
| 15:02:37 | dtantsur | nice! will check after the current meeting | |
| 15:22:18 | openstackgerrit | Monty Taylor proposed openstack/openstacksdk master: Import rate limiting TaskManager from nodepool https://review.openstack.org/574285 | |
| 16:12:27 | arkadefr | hello, I'm facing a problem with CONFIG_SEARCH_PATH here: https://github.com/openstack/openstacksdk/blob/master/openstack/config/loader.py#L48 . I'm running an ansible command with my user X , and the search path are mainly composed by /root/.config/openstack/clouds.yaml instead of my user directory: /home/X/.config/openstack/clouds.yaml . is it a known issue? | |
| 16:14:26 | arkadefr | my library appdirs and openstack etc are installed as system-wide with pip as root. | |
| 16:26:13 | mordred | arkadefr: so - in ansible the context in which you are executing can become confusing ... if you run ansible as you, it's going to connect to the remote system (even if the remote system is localhost) as whatever user is defined in the inventory | |
| 16:26:43 | mordred | arkadefr: the clouds.yaml doesn't get read in the context of the user executing ansible, but as the remote user ansible is using | |
| 16:41:03 | dtantsur | mordred: do you know how often keystoneauth is released? and whom to ping about it? | |
| 16:50:35 | mordred | dtantsur: it's released whenver we need it to be - kmalloc, lbragstad and I are good people to ping | |
| 16:50:49 | dtantsur | PIIIING! :) | |
| 16:51:02 | lbragstad | dtantsur: do you need a release? | |
| 16:51:20 | dtantsur | yeah, with the patch introducing status_code_retries to Session | |
| 16:51:46 | lbragstad | this one ? https://review.openstack.org/#/c/570934/ | |
| 16:51:57 | kmalloc | oh yeah | |
| 16:51:59 | kmalloc | lets release that | |
| 16:52:39 | kmalloc | mordred: i look forward to the email we get to send saying "keystoneclient is deprecated, use SDK instead" :P just FYI | |
| 16:52:57 | kmalloc | mordred: whenever that actually happens, i'm looking forward to it | |