Earlier  
Later
New Search
Posted Nick Remark
#openstack-sdks - 2019-11-07
17:43:58 gtema_ dtroyer: thanks. wrt pagination: currently if limit is set in osc it return only single page. "paginate" flag in SDK used to fetch everything to the end
#openstack-sdks - 2019-11-08
06:40:39 openstackgerrit Brin Zhang proposed openstack/python-openstackclient master: Microversion 2.79: Add delete_on_termination to volume-attach API https://review.opendev.org/673725
15:54:43 openstackgerrit Matt Riedemann proposed openstack/python-openstackclient master: Provide stderr in exception when check_parser fails https://review.opendev.org/682939
#openstack-sdks - 2019-11-11
03:47:48 openstackgerrit Merged openstack/openstacksdk master: Increase test timeout for 2 tests in TestImageProxy class https://review.opendev.org/692323
07:43:33 danil hello folks. Please help me. Is there a way to get nova client through openstacksdk? I mean that I can not just use novaclient library due to dependencies. Thanks
09:46:23 danil hello folks. Please help me. Is there a way to get nova client through openstacksdk? I mean that I can not just use novaclient library due to dependencies. Thanks
10:08:00 dtantsur danil: something like openstack.connect(...).compute?
10:08:09 dtantsur it won't be a client identical to novaclient, of course
10:14:17 danil dtantsur, thanks . The main method I want to find is something like "interface_attach" in novaclient. This method adds a virtio device to nova instance
10:15:52 dtantsur danil: check https://docs.openstack.org/openstacksdk/latest/user/connection.html maybe it has the call you need?
10:53:53 cdent dtantsur: were you in shanghai? if so, how was it?
11:02:47 dtantsur cdent: I wasn't, in the middle of a move here
11:02:59 cdent ✔
11:03:04 cdent I hope that all goes well
11:03:50 danil dtantsur, yes. looks like it should be there, in openstack.connect(...).compute . I found a method "create_server_interface" in Proxy class
11:04:15 dtantsur cdent: me too :) an infinite stream of small problems
11:04:37 cdent ooph
11:05:39 gtema_ hehe, what you know about small problems. My API monitoring platform got blocked by our internal Firewalls/AntiDDoS/whatsoever and nobody can figure out what and why is blocking.
11:07:27 dtantsur :D
11:07:48 dtantsur not sure if it's as inconvenient as sitting without a kitchen though (and now with a door bell stuck in "mute" position, wtf)
11:09:24 gtema_ he, I was doing this when moved to Germany, so I know perfectly what you mean
11:10:42 gtema_ mordred: still in china? are you allowed to leave or you stay there forever?
11:11:06 mordred gtema_: here til friday. I won't know if they'll let me leave until then :)
11:11:22 gtema_ oki
11:22:24 mordred gtema_: your colleague Nils did a great talk on using sdk btw
11:22:58 gtema_ thanks. He wrote me you talked to each other. Hope we will be able to improve the docs correspondigly
11:23:57 mordred ++
18:26:24 mnaser is there an official page instead of this wiki page? https://wiki.openstack.org/wiki/OpenStackClients
18:26:48 mnaser maybe https://docs.openstack.org/train/language-bindings.html ?
#openstack-sdks - 2019-11-12
00:31:01 mordred mnaser: yeah - I think that's what we've got
06:35:23 openstackgerrit Bin Lu proposed openstack/python-openstackclient master: add optional argument metric to set metric for routes https://review.opendev.org/637658
06:48:48 openstackgerrit Bin Lu proposed openstack/python-openstackclient master: add optional argument metric to set metric for routes https://review.opendev.org/637658
13:13:13 Blinkiz How can I connect to cloud and become role:admin and system_scope:all which is needed if am running with [oslo_policy]/enforce_scope=true?
13:13:17 Blinkiz https://storyboard.openstack.org/#!/story/2006839
13:42:58 frickler Blinkiz: there was some discussion about this here a couple of days ago. IIUC the conclusion was that better support for that was needed in the sdk, but you should be able to set up a dedicated clouds.yaml entry with the required parameters
13:46:11 Blinkiz frickler: hello frickler, thanks for answering. How would that clouds.yaml look like? Am guessing the examples here do not apply because they all defining domain. https://docs.openstack.org/openstacksdk/latest/user/guides/connect_from_config.html
13:46:38 Blinkiz define
13:52:48 frickler Blinkiz: seems you just add "system_scope: all" to the auth section and drop all domain parameters
13:55:14 Blinkiz frickler: Okay. Can you please help me find where this is defined in source code for openstacksdk? Am trying to get deeper knowledge how it works so it would help
13:55:58 gtema it is spread between SDK and Keystoneauth1
13:58:19 Blinkiz oh, I see. Yeah, of course.
13:59:19 frickler iiuc this is the code that handles the different scoping options https://opendev.org/openstack/keystoneauth/src/branch/master/keystoneauth1/identity/v3/base.py#L148-L166
#openstack-sdks - 2019-11-13
02:57:22 openstackgerrit Bin Lu proposed openstack/python-openstackclient master: add optional argument metric to set metric for routes https://review.opendev.org/637658
10:58:47 openstackgerrit Riccardo Pittau proposed openstack/openstacksdk master: Add reset_interfaces argument to patch_node https://review.opendev.org/689414
#openstack-sdks - 2019-11-14
16:01:19 elmiko API SIG office hour now starting
17:15:11 elmiko wheeelp, i'm closing up shop. have a nice weekend all o/
18:09:32 openstackgerrit Eric Fried proposed openstack/cliff master: Add autoprogram_cliff_app_dist_name config opt https://review.opendev.org/692464
20:12:14 openstackgerrit Merged openstack/openstacksdk master: Add clustering update_action https://review.opendev.org/682787
#openstack-sdks - 2019-11-15
08:06:54 openstackgerrit Yongli He proposed openstack/python-openstackclient master: Microversion 2.78 - server show topology https://review.opendev.org/680928
09:07:36 openstackgerrit Daniel Bengtsson proposed openstack/python-openstackclient master: Stop configuring install_command in tox. https://review.opendev.org/694484
09:33:55 openstackgerrit Daniel Bengtsson proposed openstack/python-openstackclient master: Stop configuring install_command in tox and stop use pip. https://review.opendev.org/694484
10:43:47 openstackgerrit Daniel Bengtsson proposed openstack/osc-lib master: Stop configuring install_command in tox and stop use pip. https://review.opendev.org/694510
11:48:39 jawad_axd Hi folks, How to get vm console url (as openstack console url show server) from openstackSDK ?
12:45:06 frickler jawad_axd: this seems to be a missing feature. some tried to implement this three years ago, but failed, I haven't found anything happening after that https://review.openstack.org/319828
12:48:28 frickler osc is using novaclient for that action iiuc
12:50:34 jawad_axd frickler: okay..I am looking into link you provided. Thanks for reply.
14:13:27 mriedem dtroyer: when you get a chance, can you weigh in on an option name on this osc-placement patch? https://review.opendev.org/#/c/691422/ - it's updating a set command to add a --no-provider option which means, remove allocations for the given provider (but not all providers),
14:13:40 mriedem the --no-provider name is trying to mirror CLIs like port set --no-tag,
14:13:49 mriedem but at least 2 people (efried and melwitt) were confused by it
14:14:11 mriedem my alternative suggestions are (1) change the option name to be --clear-provider or --remove-provider and keep using the set command,
14:14:21 mriedem or (2) add an unset command with a --provider option
14:15:52 mriedem e.g. openstack resource provider allocation unset --provider prov1 --provider prov2 consumer1 // removes consumer1's allocations from prov1 and prov2
15:20:46 openstackgerrit Nate Johnston proposed openstack/python-openstackclient master: Handle NotFoundException when listing floating IPs https://review.opendev.org/694571
15:28:06 efried mriedem: (2) jumped out as being pretty clean/intuitive. I haven't stared at it yet.
15:31:28 mriedem i left a comment on the review but one fear i have with adding unset is someone expecting it to also have a --no-allocation or some option to also remove specific resource classes from an existing provider allocation combo, which would be really annoying
15:41:10 efried You can specify zero or more
15:41:10 efried You can specify at most one --provider, in which case only that provider is affected.
15:41:10 efried If you don't specify a --provider, all providers are affected.
15:41:10 efried makes enough sense.
15:41:10 efried openstack resource provider allocation unset --consumer C [--provider P] [--resource-class R1 [--resource-class R2 ...]]
15:41:10 efried mriedem: I was about to respond to that. IMO
15:41:21 efried ... --resource-class, in which case only those resource classes are affected.
15:41:55 efried For a first iteration you could leave off the --resource-class arg, but if/when it needs to be added later, does that seem like a reasonable design?
15:42:35 efried If you need to affect a >1 subset of providers, repeat the command. Are we allowed to say that?
15:45:17 mriedem i think consumer and at least one provider should be required, consumer obviously (it's an arg, not an option)
15:45:47 mriedem specifying zero providers means what? clear all allocations?
15:46:25 mriedem and if so, <1.28 means DELETE, >= 1.28 means PUT with empty allocs dict and consumer_generation
15:46:36 mriedem i'm not dealing with that --resource-class garbage
15:47:26 efried Yes, zero providers means remove all allocations, the base use case for 'allocation unset $consumer'.
15:47:54 mriedem and that's only useful over allocation delete b/c of 1.28 and consumer generations?
15:48:45 efried Do you think it's necessary to have a generation'd code path in `allocation delete`?
15:49:06 mriedem not really
15:49:24 mriedem if someone wanted that, you could provide 1.28 and a --safe or --validate option or something like that
15:49:36 efried IMO this kind of reflects the semantics of DELETE vs PUT{}: one you're saying "Blow it away. I don't care what's there." The other you're saying "Remove what I think is there right now. If something changes, hold up, I need to reevaluate"
15:49:38 mriedem openstack --os-placement-api-version 1.28 resource provider allocation delete --safe <consumer_id>
15:49:50 efried But in a CLI like this, the window between the GET and PUT is so small, I'm not sure it adds any value.
15:50:49 efried so no, unset with no provider (and no RC) doesn't really add value. It's just a reducto case of two CLIs providing the same function.
15:51:39 efried If you include (or at least plan for) the RC option, though, it does make sense for --provider to be optional.
15:51:48 efried "Remove all bandwidth allocations"
15:51:55 efried "Remove all GPUs"
15:52:14 efried things that can be on (multiple) nested providers
15:52:52 mriedem congratulations, you've talked me out of wanting to work on this anymore :)
15:55:03 mriedem what i'd probably do is unset with --provider using nargs='+' to start meaning at least one --provider is required,
15:55:11 mriedem if at some point you want to make --provider really optional, you drop the nargs
15:58:13 efried I don't think we should allow more than one provider.
15:58:35 efried mm, hold that thought.
15:59:05 efried I guess you're more likely to want to remove all (nested) allocs from one host, huh.
15:59:25 efried ... unset --root-provider $P ?
15:59:36 efried --tree $P ?
15:59:37 mriedem melwitt specifically wants to be able to remove allocations from more than one provider in a single call

Earlier   Later