Earlier  
Later
New Search
Posted Nick Remark
#openstack-sdks - 2018-02-22
16:21:03 SamYaple Tek___: im not sure if this is shade code trying to be clever (finding the bootable volume) or the nova api being nova
16:21:42 elmiko edleafe: lgtm, ship it!
16:22:51 edleafe Done!
17:19:25 samueldmq #link https://github.com/Azure/azure-quickstart-templates
17:19:49 samueldmq thoughts on sdk-quickstart-examples?
17:19:57 samueldmq mordred: ^
17:21:44 samueldmq funny thing is that shade/sdk brings simplicity to the point we wouldn't need a directory for each example, but a single file instead
17:22:00 mordred cmurphy: sorry about that- it should be fixed now
17:22:19 cmurphy mordred: yay
17:22:33 mordred cmurphy: I rechecked that ksa patch
17:22:44 cmurphy ty
17:26:02 mordred Tek___, SamYaple: you are creating a bootable volume from the Debian9 image, then booting from that same image and attaching that volume, the create_server call would say to me "boot from this image and then also attach this volume as an additional volume"
17:28:23 mordred Tek___: if whatyou want to do is boot-from-volume but you want a new bootable volume created from the Debian9 image,...
17:29:31 mordred Tek___: all you need to do is cloud.create_server(name='WS2012-TEST', image=imageTest, flavor=flavor, boot_from_volume=True, wait=True, auto_ip=True, availability_zone='eu-west-0a', network='70b7256f-429e-4bc2-911b-79e0fb26594f', timeout=3600, security_groups='Ouvert',userdata=userdata, key_name='mykeyname')
17:29:54 mordred Tek___: basically, just pass the image name and the flag boot_from_volume and you can skip the create-a-volume step
17:30:33 mordred Tek___: if you want to create the bootable volume as one step and then boot in the second step, leave image= out of the create server call and pass the bootable volume to the boot_volume parameter
17:35:49 mordred Tek___, SamYaple: it occurs to me that this area could use better documentation. :)
19:39:42 openstackgerrit Merged openstack/python-openstacksdk master: Update clustering module's _proxy comment message. https://review.openstack.org/543390
21:09:24 openstackgerrit Monty Taylor proposed openstack/python-openstacksdk master: Honor service type aliases from the catalog https://review.openstack.org/544356
21:09:34 openstackgerrit Monty Taylor proposed openstack/python-openstacksdk master: Mark openstacksdk-functional-devstack-base as abstract https://review.openstack.org/545606
21:12:49 openstackgerrit Monty Taylor proposed openstack/python-openstacksdk master: Run examples tests with functional tests https://review.openstack.org/540374
21:28:44 adriant mordred: what's the default version of identity in the sdk?
21:29:46 adriant because it seems even when built off a very much v3 session, the identity proxy I'm getting is v2
21:30:27 adriant I think I'll need to add some default API version stuff to my constructors
21:31:59 adriant oh, that's interesting. "conn = oi.sdk.get_connection(identity_api_version=3)"
21:32:15 adriant it expects str, so when you give it an int it gets very confused
21:34:09 adriant mordred: http://paste.openstack.org/show/682540/
21:34:42 adriant not sure where that api version param is being parsed, but we should probably stick a caste to string there.
21:34:58 adriant cast*
22:03:27 dhellmann dtroyer, melwitt : what's the story with the osc-placement library and python 3? I see it claims support in the classifiers, but I don't see any test unit jobs configured for it.
22:43:34 mordred adriant: getting the versions touse discovery by default is the next thing on my list
22:43:53 mordred adriant: that said - I just added two tests to try to reproduce your int/string issue and both work for me :(
22:44:15 adriant mordred: odd, I'm running... lemme check what version
22:44:55 openstackgerrit Monty Taylor proposed openstack/python-openstacksdk master: Make sure both string and int versions work https://review.openstack.org/547159
22:45:04 adriant mordred: openstacksdk==0.11.3
22:45:13 mordred adriant: ^^ there's theunit test I tried using to reproduce
22:45:21 adriant maybe that version stuff is in keystoneauth?
22:46:35 mordred it's many places ... but that should be using requests-mock so we should still see the issue - Ill keep poking to see if I can make it fail - what does your get_connection function look like?
22:47:24 adriant mordred: https://github.com/Adrian-Turjak/openstack-interpreter/blob/master/openstack_interpreter/v1/sdk.py
22:48:57 mordred adriant: ok. cool. so we should be able to easily reproduce that issue
22:51:41 adriant I do actually use the sdk outside of my interpreter btw, it's just that the interpreter is the easiest way to test stuff and I use it to figure out how some of the sdk functions actually work and the models returned act/look like :P
22:53:03 mordred adriant: I like it - I might start using it myself :)
22:55:52 adriant mordred: I mostly made it to teach people on our team how to use the various client tools, although that's slowly turning into: "just use the SDK, it's better, and play with it here"
22:56:04 mordred \o/
23:00:49 mordred adriant: WELL - I've discovered at least one other bug - but still can't reproduce your thing
23:00:58 adriant haha
23:01:01 adriant oh well
23:01:07 adriant it's not really that bad, just odd
23:01:22 mordred adriant: I have discovered that 'identity_api_version='red' will get you openstack.identity.v3._proxy.Proxy
23:01:28 adriant I may have to try it out in an entirely clean venv and triple check
23:01:29 adriant hahah
23:01:30 adriant what?1
23:01:31 adriant ?
23:01:31 mordred adriant: which I'd argue is ... incorrect behavior
23:01:53 adriant that's amusing
23:05:42 adriant mordred: clean venv, both py2 and py3 cause me the same issue :(
23:07:02 adriant its a minor enough thing though that it probably isn't worth chasing
23:07:26 adriant although I'll admit I'm curious what's the cause
23:08:06 mordred yah. I worry something is wrong at a deeper level - I'm going to try with oi this time
#openstack-sdks - 2018-02-23
00:21:10 mordred adriant: WOOT - found it
00:22:02 adriant mordred: what was it? :P
00:23:19 mordred adriant: the code in oi is passing in an authenticated session, which is a different codepath than I was using in that test. (that's why I didn't get a failure) the underlying issue is that in session codepath the normalize_keys function is not getting called - which is where the stringificatoin of the values happens
00:24:13 adriant ah! that makes sense, and further useful fixes to the session codepath (which I do care about ;))
00:29:31 mordred ++
00:39:58 mordred adriant: https://review.openstack.org/547225 Run normalize_keys on config for session codepath <-- that should fix it
00:42:49 adriant woo!
08:04:38 openstackgerrit Dongcan Ye proposed openstack/python-openstackclient master: Network: Add tag support for floating ip https://review.openstack.org/547316
09:01:36 openstackgerrit Jose Castro Leon proposed openstack/python-openstackclient master: Add support for endpoint group commands https://review.openstack.org/520627
09:33:38 spredzy .close
10:37:45 openstackgerrit Dongcan Ye proposed openstack/python-openstacksdk master: Network: Add tag support for floating ip https://review.openstack.org/547366
10:48:57 openstackgerrit Dongcan Ye proposed openstack/python-openstacksdk master: Network: Add tag support for security group https://review.openstack.org/547369
10:52:35 mardim hello
10:52:46 mardim is this the channel of openstack shade project
10:52:47 mardim ?
10:57:53 openstackgerrit Dongcan Ye proposed openstack/python-openstacksdk master: Network: Add tag support for QoS policy https://review.openstack.org/547372
10:58:50 mardim mordred, Hello I think I am hitting a bug here what do you think : https://github.com/openstack-infra/shade/blob/master/shade/openstackcloud.py#L1317
10:59:19 mardim mordred, the error is TypeError: unsupported operand type(s) for //: 'NoneType' and 'int'
11:39:18 openstackgerrit Jens Harbott (frickler) proposed openstack/python-openstacksdk master: Add support for dns-domain https://review.openstack.org/500660
11:52:43 openstackgerrit Jens Harbott (frickler) proposed openstack/python-openstackclient master: Add support for "--dns-domain" argument https://review.openstack.org/500450
12:18:38 openstackgerrit Jens Harbott (frickler) proposed openstack/python-openstackclient master: Add dns-domain support to Network object https://review.openstack.org/516701
12:22:51 frickler mardim: are you explicitly setting "timeout=None" in your call? otherwise I don't see how this could happen
12:28:16 mardim frickler, hello I think is typo I opened a bug : https://bugs.launchpad.net/python-openstacksdk/+bug/1751251
12:28:16 openstack Launchpad bug 1751251 in OpenStack SDK "Openstack-infra Shade project TypeError" [Undecided,New]
12:28:35 mardim frickler, Maybe this helps you to understand what I mean ^
12:44:55 frickler mardim: can you please add detailed steps how to reproduce the exception?
13:16:29 mardim frickler, To be honest I am not sure how to reproduce the error because I am meeting this bug when I execute some yardstick testcases and that's because OPNFV yardstick
13:16:56 mardim project uses the Shade to create heat Stacks
13:18:05 mardim But I think in general if you find a way yo call the create_stack function the you will hit the error
13:18:33 mardim because when I deleted the '// 60 ' by hand in the code evreything worked
13:18:39 mardim so I think this is the problem
13:26:33 frickler mardim: well, changing the code masks the issue but changes the effect of the timeout parameter in an incompatible way. the "//" is a valid operator on integers and the timeout parameter is explicitly specified as being an int, so the exception you are showing very likely is caused by an illegal caller
13:27:06 frickler mardim: the default value for timeout is 3600, too
13:28:00 mardim frickler, hmm ok let me check something
13:32:49 Shrews well, yes, the error seems to be with the caller, but in other API calls, we let timeout=None mean forever, so i think mardim has found a valid issue we should fix
13:57:34 mardim frickler, Shrews Guys you were right the problem is on the caller
13:57:44 mardim you can delete the bug report if you want
13:58:06 Shrews mardim: i think we should still allow for None values
13:58:19 openstackgerrit David Shrewsbury proposed openstack-infra/shade master: Fix for timeout=None in orchestration API calls https://review.openstack.org/547430
13:58:30 Shrews see ^^^
13:58:38 Shrews just trying to confirm that None means "no timeout"

Earlier   Later