| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-sdks - 2017-10-11 | |||
| 14:05:01 | kuzko | well that's nice, I used pip to fech the libray, should I update using pip or git clone? | |
| 14:05:15 | kuzko | *library... | |
| 14:10:01 | mordred | kuzko: I'm overdue for a release - so I'll try to get one cut tomorrow | |
| 14:24:02 | umbSublime | mordred, if you can add that feature for custom api call like we discussed yesterday in the next release (if it's ready that is) it would be awesome !! | |
| 14:25:46 | mordred | umbSublime: yes - I fully intend to get that in to the next release | |
| 14:26:18 | umbSublime | awesome!! | |
| 14:26:21 | mordred | umbSublime: well - next sdk release - not next shade release | |
| 14:26:45 | umbSublime | ohh right you guys were talking about shade, my bad :) | |
| 14:27:15 | mordred | umbSublime: well - hopefully that confusion will not persist for TOO much longer :) | |
| 14:27:21 | umbSublime | hehe | |
| 14:27:25 | mordred | but that certaily won't get done by tomorrow :) | |
| 14:27:41 | umbSublime | no worries, I have my work-around working :) | |
| 14:37:34 | kuzko | is shade supposed to become the SDK at a later date? ^^ | |
| 14:38:34 | mordred | kuzko: the current proposed plan is that we merge the shade code into the sdk tree, then make the shade library a thin backwards-compat shim that wraps and uses the equivilent code in the sdk | |
| 14:38:59 | mordred | kuzko: so it's still totally safe to use shade today - and I do not intend to break it, well, ever | |
| 14:39:25 | kuzko | nice, because I think we might use it for our operations going forward | |
| 14:40:04 | kuzko | simplicity allow for quick newbs onboarding and teaching | |
| 14:46:21 | mordred | ++ | |
| 14:47:38 | mordred | kuzko: the overall intent is to have all three types of access co-exist using the same Connection object - the object-oriented API that sdk currently has - the more procedural business-logic 'easy' api of shade - and fallback/passthrough layer for just making direct REST calls if you need something that doesnt' have a higher-level wrapper | |
| 14:48:42 | mordred | so you can do conn.list_servers() ... conn.compute.server.servers() and conn.compute.get('/servers') depending on your need | |
| 14:49:04 | kuzko | oh right, that would be sweet | |
| 14:49:57 | umbSublime | indeed ^^ | |
| 14:50:04 | mordred | that way for times like today if you're like 'hi, I need to set bootable flag' I can say "crap, well, for now just do conn.block_storage.post('/volumes/{id}/action'.format(volume['id']), {'os-set_bootable': True}) | |
| 14:50:43 | mordred | I've got most of that working locally but not passing tests yet | |
| 14:53:29 | mordred | (which is mostly just due to mocking issues in the tests - not actual issues) | |
| 14:56:37 | mordred | kuzko: so your workaroud is just to create the volume using cloud.create_volume(image=imageMA, name='ATEL-Modele-20170926',size=50) and that did properly wind up with a bootable volume yeah? | |
| 14:59:53 | mordred | kuzko: did you specifically want to do that in two steps? asking because if your goal is 'create a volume from this image and then boot a server from it' - you can totally do that all just in the create_server call | |
| 15:00:53 | kuzko | it's in two step because I want to reuse the volume to upload it at a later date, it's part of an image update script... and no, it's not doing a proper bootable volume... | |
| 15:01:17 | mordred | gotcha. | |
| 15:01:18 | kuzko | *at a later date ... in the script... | |
| 15:01:21 | mordred | yah | |
| 15:03:31 | mordred | hrm. creating the volume from an image is supposed to set bootable ... lemme go ask someone something | |
| 15:29:35 | mordred | kuzko: SOOO ... I've got some patches queued up to deal with this | |
| 15:29:57 | mordred | kuzko: but - I chatted with the cinder folks and also just did a test real quick against vexxhost doing v=c.create_volume(image='dib-ubuntu-trusty', wait=True, size=50) ... and I get v['is_bootable'] == True | |
| 15:30:44 | mordred | kuzko: can you verify that newvolume['is_bootable'] in your script is False? or, if it is showing True, can you paste me the error you're getting when you get it? | |
| 15:31:01 | mordred | (trying to figure out if there is a deeper bug somewhere) | |
| 15:32:48 | kuzko | 'bootable': False | |
| 15:34:32 | mordred | AWESOME | |
| 15:35:19 | kuzko | you seem awfuly entertained by a bug | |
| 15:35:34 | mordred | well - I'm sick in the head | |
| 16:29:36 | mordred | kuzko: AHA! I think I know what the issue is | |
| 16:30:36 | kuzko | mordred : so how bad is it? | |
| 16:30:41 | mordred | kuzko: your create volume call is not waiting for the volume to be ready - and cinder does not set the bootable attribute on the image until later in processing | |
| 16:30:55 | kuzko | oh... | |
| 16:31:09 | kuzko | I just have to set a wait? | |
| 16:31:45 | mordred | kuzko: yah. try changing cloud.create_volume(image=imageMA, name='ATEL-Modele-20170926',size=50) to cloud.create_volume(image=imageMA, name='ATEL-Modele-20170926',size=50, wait=True) | |
| 16:32:04 | kuzko | mordred : hang on, that makes no sense, the default value of wait is true | |
| 16:32:09 | mordred | oh. hrm | |
| 16:32:16 | mordred | kuzko: ok. nevermind. ignore me :) | |
| 16:32:23 | kuzko | def create_volume( | |
| 16:32:23 | kuzko | self, size, | |
| 16:32:23 | kuzko | wait=True, timeout=None, image=None, **kwargs): | |
| 16:32:28 | mordred | yup. you're very right | |
| 16:34:37 | mordred | kuzko: the cinder team are curious as to what version of cinder you're running - if you happen to know | |
| 16:35:28 | kuzko | :~$ cinder --version | |
| 16:35:29 | kuzko | 3.2.0 | |
| 16:38:28 | kuzko | mordred : should I use an other? | |
| 16:39:20 | mordred | kuzko: nah - if you have time, would you mind running with http logging set like: http://paste.openstack.org/show/623372/ and pasting the (likely annoying long) results? | |
| 16:39:50 | kuzko | mordred : paste not running | |
| 16:42:54 | mordred | kuzko: https://etherpad.openstack.org/p/ihJXPoJon5 | |
| 16:42:57 | kuzko | mordred : alright there you go https://pastebin.com/ZK990uxi | |
| 16:43:12 | mordred | kuzko: sweet - thankyou very much - I appreciate your help here | |
| 16:47:21 | kuzko | tell me if there is something more i can do about it | |
| 16:48:27 | mordred | ooh- look at line 59 - it's not passing imageRef - I think the reason it's not bootable is that it's not even creating it from the image | |
| 16:48:54 | kuzko | hu... what? | |
| 16:53:02 | mordred | yah. so - that'll be the next rabbit hole to dive down | |
| 16:54:25 | mordred | kuzko: I think I see it ... your script does a get_image for 'ATEL-Modele-Latest' ... but I do not see that image in your image list | |
| 16:54:49 | mordred | kuzko: so I thinkn imageMA is coming back None, which you are then passing to create_volume | |
| 16:55:47 | kuzko | indeed, I am also a dumbass for not checking my code | |
| 16:55:57 | kuzko | and my tenant | |
| 16:56:29 | mordred | well - also shade is failing at figuring out that you requested something it can't do | |
| 16:57:30 | mordred | I have an idea for solving that ... | |
| 16:57:36 | mordred | at least for this use case | |
| 17:01:37 | kuzko | print("dude, you're doing it wrong, that image doesn't exist") | |
| 17:01:51 | kuzko | mordred : maybe? | |
| 17:03:05 | mordred | :) | |
| 17:03:07 | mordred | kuzko: amusingly enough - if you had just passed image='ATEL-Modele-Latest' to create_volume, you would get an exception | |
| 17:03:57 | kuzko | ah... yeah, that makes sense | |
| 17:07:31 | kuzko | god, yeah, i'm stupid, so much time for a freaking varaible misshap | |
| 17:07:36 | kuzko | *variable | |
| 17:07:50 | kuzko | anyhow, thank you mordred | |
| 17:18:01 | mordred | kuzko: well - it certainly shows a place where there is potential for very confusing user mishap - and we've also gotten 2 shiny new patches pushed up | |
| 17:18:05 | mordred | kuzko: so thanks! | |
| 17:23:06 | kuzko | anyway, gotta go thanks for the help ! | |
| 17:52:24 | hyang | hi there, I'm working on openstackclient and wondering how we deal with password cache in current version? I know we used to have keyring support long time ago but that was removed. | |
| 17:53:55 | hyang | If we don't want store password in env and not using the token flow, is there any other option to avoid typing password every time? | |
| 19:12:07 | openstackgerrit | Eric Fried proposed openstack/keystoneauth master: Add EndpointData.__str__ for debugging https://review.openstack.org/489413 | |
| 19:13:58 | openstackgerrit | Eric Fried proposed openstack/keystoneauth master: WIP: Return the endpoint_override from EndpointData https://review.openstack.org/491947 | |
| 19:56:29 | dtroyer | hyang: yes, you can put it in ~/.config/openstack/clouds.yaml | |
| 20:14:07 | enis_ | I'm trying to retrieve details about a specific project using it's name but `conn.identity.projects(name='pr-name')` returns all the projects even though a project by the specified name exists. Is there a different parameter I should be using? | |
| 20:15:48 | enis_ | Using conn.keystone.projects.list(name='pr-name') works but I'd like to use single sdk. | |
| #openstack-sdks - 2017-10-12 | |||
| 12:31:09 | elmiko | cdent: were you referring to the openstack-sigs ml? | |
| 12:31:15 | cdent | ya | |
| 12:31:20 | elmiko | ack | |
| 12:33:18 | cdent | elmiko: is not your bad, nor anybody’s bad | |
| 12:33:26 | cdent | just occured to me that we’ve got that list, may as well use it | |
| 12:37:23 | elmiko | you are correct though, we should be talking about it on that list | |
| 12:37:40 | elmiko | i certainly don't want to contribute to anything that would exclude apac folks | |
| 12:37:58 | elmiko | i can see that having the discussion at our meeting doesn't really help them =( | |
| 12:38:24 | elmiko | part of this too is that i have some baggage about the last apac meetings we held, i need to drop that | |