| openstackgerrit | Ian Wienand proposed a change to openstack/tempest: Generic skip interface https://review.openstack.org/39417 | 00:24 |
|---|---|---|
| *** adalbas has joined #openstack-qa | 00:33 | |
| *** rhsu has quit IRC | 00:49 | |
| *** dkranz has joined #openstack-qa | 01:00 | |
| *** sthaha has joined #openstack-qa | 01:00 | |
| *** sthaha has joined #openstack-qa | 01:00 | |
| openstackgerrit | Sreeram Yerrapragada proposed a change to openstack-dev/devstack: Add support for vmdk files as glance images https://review.openstack.org/38576 | 01:55 |
| openstackgerrit | minsel proposed a change to openstack/tempest: Add testing of Neutron per tenant quotas API https://review.openstack.org/35724 | 01:59 |
| openstackgerrit | Ivan-Zhu proposed a change to openstack/tempest: prot test_images and test_server_actions part1 https://review.openstack.org/39621 | 02:23 |
| openstackgerrit | Ivan-Zhu proposed a change to openstack/tempest: port test_images and test_server_actions into v3 part2 https://review.openstack.org/39609 | 02:23 |
| *** anteaya has quit IRC | 02:26 | |
| *** HenryG has joined #openstack-qa | 02:34 | |
| *** HenryG has quit IRC | 02:42 | |
| *** SergeyLukjanov has joined #openstack-qa | 02:50 | |
| *** SergeyLukjanov has quit IRC | 03:05 | |
| *** prekarat has joined #openstack-qa | 03:54 | |
| openstackgerrit | Ian Wienand proposed a change to openstack-dev/devstack: Unfubar "unfubar_setuptools" https://review.openstack.org/39628 | 04:25 |
| *** rhsu has joined #openstack-qa | 04:31 | |
| *** afazekas has joined #openstack-qa | 04:34 | |
| Anju | afazekas: good morning | 04:36 |
| afazekas | Anju: good morning | 04:45 |
| openstackgerrit | afazekas proposed a change to openstack/tempest: Remove wait_for_image_resp_code https://review.openstack.org/38995 | 05:00 |
| *** rhsu has quit IRC | 05:38 | |
| *** sthaha has quit IRC | 05:48 | |
| *** sthaha has joined #openstack-qa | 05:51 | |
| *** sthaha has joined #openstack-qa | 05:51 | |
| *** raiesmh08_ has joined #openstack-qa | 06:17 | |
| raiesmh08_ | Hi there, I have add some network api test cases https://review.openstack.org/#/c/39023/ pls have a look at this | 06:19 |
| *** raiesmh08_ has quit IRC | 06:29 | |
| openstackgerrit | Marc Koderer proposed a change to openstack/tempest: Add argument to stop stress test on first error https://review.openstack.org/39524 | 06:40 |
| mkoderer | Good morning | 06:41 |
| Anju | mkoderer:good morning | 06:51 |
| Anju | mkoderer:BTW its afternoon in India | 06:52 |
| mkoderer | Anju: ;) next time it will take this in consideration | 07:00 |
| *** giulivo has joined #openstack-qa | 07:04 | |
| Anju | giulivo : want to ask abut xml/json | 07:11 |
| openstackgerrit | Rohit Karajgi proposed a change to openstack/tempest: Remove duplicate image tests for tenant authZ https://review.openstack.org/39641 | 07:13 |
| giulivo | Anju, so the problem there is that we miss the implementation, for the xml interface of a few methods | 07:14 |
| giulivo | and those will have to be implemented before we can enable the xml interface for the test class | 07:15 |
| Anju | yeah | 07:16 |
| Anju | exactly | 07:16 |
| Anju | so i if i want to make that missing calss ses | 07:16 |
| Anju | how could i | 07:16 |
| Anju | in volume attach/detach ,upload xml api are missing | 07:17 |
| Anju | giulivo : i know the background setting how to call an xm methaods | 07:18 |
| Anju | methods | 07:18 |
| giulivo | Anju, you need to write those | 07:22 |
| Anju | giulivo i wrote : http://pastebin.com/b5eTvx6u | 07:23 |
| Anju | the problem is how to format the response in xml | 07:23 |
| Anju | without knowing what is th output | 07:23 |
| giulivo | ah okay so I suggest you go trough some "manual debug" using curl | 07:24 |
| giulivo | or writing some custom python to dump the returned xml | 07:24 |
| giulivo | after the environment is set up with volumes and instances (you can populate it manually) | 07:25 |
| Anju | a parse_volume function is already created here | 07:25 |
| Anju | giulivo :https://github.com/openstack/tempest/blob/master/tempest/services/volume/xml/volumes_client.py#L44 | 07:26 |
| giulivo | yeah I know there is one, that is used for volumes get and list | 07:27 |
| giulivo | but what you need to implement is upload,attach,detach | 07:28 |
| Anju | giulivo :usually the output comes in a bulk | 07:30 |
| Anju | how could i can know the differences in json and xml response | 07:30 |
| giulivo | in a bulk? | 07:30 |
| Anju | means without any formatting | 07:31 |
| giulivo | I'm sorry I don't see what you mean; the output is formatted by cinder in either xml or json format | 07:31 |
| giulivo | what you're doing in the service class is providing to the test class some "utilities" | 07:32 |
| Anju | giulivo :http://api.openstack.org/api-ref.html#netconn-api | 07:33 |
| Anju | in this document the upload ,dettach attch api request /tesponse is missing | 07:34 |
| Anju | so when i am trying to take the difference of xml and json output is not very clear like here | 07:35 |
| giulivo | yeah okay I get that, that is why I suggested to use curl | 07:36 |
| giulivo | you can issue manually the request and see what the cinder returns | 07:36 |
| Anju | yeah :using curl command | 07:36 |
| giulivo | curl or anything else which is suitable to you | 07:36 |
| giulivo | but the point is to see it printed in the console | 07:37 |
| Anju | i ma not finding the difference in xml/json as it is not very clear | 07:37 |
| giulivo | so from that you learn how that is structured | 07:37 |
| giulivo | difference in how to issue the request? | 07:37 |
| Anju | for formmating | 07:37 |
| Anju | i think we should know | 07:37 |
| Anju | what is the output in xml and json | 07:38 |
| Anju | ok giulivo :can u please suggest me how u have write the | 07:38 |
| Anju | xml api for aggreate tempest/services/compute/xml/aggregates_client.py | 07:39 |
| *** raiesmh08 has joined #openstack-qa | 07:42 | |
| raiesmh08 | hi afazekas | 07:43 |
| afazekas | raiesmh08: hi | 07:43 |
| raiesmh08 | I have added network API test cases https://review.openstack.org/#/c/39023 | 07:43 |
| raiesmh08 | if possible please review this | 07:43 |
| raiesmh08 | afazekas : r u there | 07:45 |
| afazekas | raiesmh08: yes | 07:46 |
| afazekas | LGTM, just a small nit | 07:46 |
| *** jhenner has joined #openstack-qa | 07:47 | |
| raiesmh08 | afazekas: will you please suggest some one who can approve | 07:48 |
| raiesmh08 | afazekas : I have added a bug https://bugs.launchpad.net/openstack-api-site/+bug/1207196 | 07:50 |
| raiesmh08 | please check it out | 07:50 |
| raiesmh08 | andreaf : hi | 07:51 |
| Anju | raiesmh08 :firstly you have to improve the comments given by afazekas | 07:53 |
| Anju | only then it will will be approved by other person | 07:53 |
| afazekas | raiesmh08: https://launchpad.net/~openstack-qa-core/+members | 07:57 |
| raiesmh08 | afazekas : Thanks | 07:57 |
| raiesmh08 | please also check reported bug https://bugs.launchpad.net/openstack-api-site/+bug/1207196 | 07:57 |
| afazekas | FYI: Who frequently does reviews can be a core reviewer .. | 07:58 |
| raiesmh08 | ok | 07:58 |
| *** SergeyLukjanov has joined #openstack-qa | 07:59 | |
| afazekas | I would like to have members into the core group | 08:00 |
| mkoderer | afazekas: cool | 08:01 |
| mkoderer | afazekas: I am already trying to be more visible in the reviews | 08:02 |
| mkoderer | afazekas: but I need more experience with tempest I guess | 08:02 |
| afazekas | mkoderer: cool, devstack changes usually bonus point | 08:03 |
| mkoderer | afazekas: ok, will have a closer look to devstack as well | 08:04 |
| Anju | afazekas: are u there have a look please http://www.fpaste.org/29410/43995137/ | 08:21 |
| afazekas | raiesmh08: does it work with an empty json dict ? '{}' | 08:23 |
| afazekas | Anju: what is the question ? | 08:24 |
| afazekas | giulivo: ^^ | 08:24 |
| Anju | afazekas: i catch the tresponse which is shown to you | 08:24 |
| afazekas | Anju: cool | 08:26 |
| Anju | afazekas: :) but no idea how to proceed | 08:26 |
| giulivo | Anju, I saw your fpaste | 08:27 |
| Anju | yeah | 08:27 |
| giulivo | so that is what you needed to unwrap that from the response | 08:27 |
| giulivo | Anju, you can now use etree.fromstring to parse the response and than return it | 08:32 |
| Anju | giulivo :i did this | 08:33 |
| Anju | giulivo: http://pastebin.com/b5eTvx6u | 08:34 |
| Anju | please look in the upload function | 08:34 |
| Anju | what wrong i am doing | 08:34 |
| giulivo | on line 45 I think you should use xml_to_json rather than self.xml_to_json but that should work | 08:38 |
| giulivo | did you try executing only the volume upload test via xml after that function is added? | 08:38 |
| *** tkammer has joined #openstack-qa | 08:38 | |
| Anju | yeah : as other are not implemnented | 08:40 |
| Anju | giulivo : i ran only test_upload test from volume actions_test | 08:40 |
| giulivo | yeah and did that work? | 08:41 |
| Anju | trying | 08:41 |
| Anju | i will ping u . | 08:41 |
| afazekas | giulivo: is an xml valid without the <?xml> tag ? | 08:45 |
| raiesmh08 | afazekas : I was trying to create with xml formatted body | 08:46 |
| raiesmh08 | I tried <subnet></subnet> | 08:46 |
| raiesmh08 | which was not working | 08:46 |
| afazekas | raiesmh08: I have added neutron to the affected projects | 08:50 |
| giulivo | afazekas, I think Anju can check that | 08:50 |
| giulivo | Anju, the response you dumped | 08:51 |
| giulivo | is that all you get or do you get the starting <xml> tag too? | 08:51 |
| Anju | giulivo :after debugging? | 08:51 |
| raiesmh08 | afazekas : do u think it is a neutron API bug ? | 08:54 |
| raiesmh08 | I think it is a documentation bug | 08:55 |
| raiesmh08 | as without network id parameter subnet can not be created | 08:55 |
| raiesmh08 | and to pass network id body is must to send | 08:55 |
| giulivo | Anju, the xml dump you linked on fpaste | 08:56 |
| giulivo | does not start with the <xml> tag | 08:56 |
| raiesmh08 | neutron handles API request with proper body | 08:56 |
| giulivo | is that the entire output you have pasted, aka do we miss the <xml> completely ? | 08:56 |
| raiesmh08 | so i think there is no problem with neutron | 08:56 |
| raiesmh08 | it is only a documentation bug | 08:56 |
| raiesmh08 | afazekas : how u concluded that it is a neutron bug ? | 08:58 |
| afazekas | raiesmh08: they will invalidate it if it is not, and the document will be fixed | 09:05 |
| raiesmh08 | ok | 09:06 |
| raiesmh08 | :) | 09:06 |
| raiesmh08 | @jaypipes : I have added network API test cases please check https://review.openstack.org/#/c/39023 | 09:07 |
| raiesmh08 | jaypipes : I have added network API test cases please check https://review.openstack.org/#/c/39023 | 09:11 |
| Anju | afazekas,giulivo:http://www.fpaste.org/29424/75348266/ | 09:11 |
| giulivo | thanks Anju , so we get the <xml> starting tag | 09:16 |
| giulivo | how's going with the volume upload test? | 09:16 |
| Anju | yeah | 09:16 |
| Anju | wait for 5 sec | 09:16 |
| raiesmh08 | mtreinish: I have added network API test cases please check https://review.openstack.org/#/c/39023 | 09:19 |
| raiesmh08 | 3 guys reviewed this | 09:20 |
| raiesmh08 | and passed | 09:20 |
| raiesmh08 | please chaeck | 09:20 |
| Anju | sry giulivo :more than 5 sec ...but not running | 09:26 |
| giulivo | and what error is is throwing? | 09:26 |
| Anju | a lot | 09:26 |
| Anju | :) | 09:26 |
| openstackgerrit | Giulio Fidente proposed a change to openstack/tempest: renames the stress test class to include the Volume keyword https://review.openstack.org/39654 | 09:27 |
| Anju | giulio: i am giving you the upper part only http://www.fpaste.org/29427/53495831/ | 09:33 |
| Anju | back in 15 min | 09:33 |
| afazekas | raiesmh08: you can get a +2 vote from me if you fix the whitespaces | 09:41 |
| Anju | giulivo: I am giving you the upper part only http://www.fpaste.org/29427/53495831/ | 09:42 |
| raiesmh08 | afazekas : thanks I will | 09:43 |
| giulivo | Anju, I think you're doing something wrong; VolumesClientXML does have the create_volume method | 09:44 |
| openstackgerrit | Zhi Kun Liu proposed a change to openstack/tempest: add image tests v2 https://review.openstack.org/39212 | 09:47 |
| Anju | giulivo :so u think there is no need to format the response | 09:50 |
| giulivo | Anju, let's do it differently | 09:50 |
| giulivo | how are you running the single test? | 09:51 |
| giulivo | I think the problem you're facing is due to an issue in how you patched the xml client | 09:52 |
| giulivo | you just have to add your upload_volume function in the existing xml client | 09:52 |
| openstackgerrit | raiesmh08 proposed a change to openstack/tempest: Add network api test cases https://review.openstack.org/39023 | 09:52 |
| openstackgerrit | Joe Hakim Rahme proposed a change to openstack/tempest: Adds tests covering Swift's Account Quota middleware https://review.openstack.org/39659 | 09:56 |
| Anju | giulivo :http://www.fpaste.org/29430/75350992/ | 09:56 |
| raiesmh08 | afazekas: I have changed as per your comments and applied patch 2 | 09:56 |
| afazekas | raiesmh08: +2 | 09:58 |
| raiesmh08 | thanks | 10:00 |
| Anju | giulivo : i just add the upload function in https://github.com/openstack/tempest/blob/master/tempest/services/volume/xml/volumes_client.py | 10:03 |
| Anju | this file | 10:03 |
| Anju | giulivo: changes in the base file an d client are already maintained | 10:08 |
| giulivo | Anju looking at the error you pasted it looks like you're posting to the wrong url; this is wrong """ resp, body = self.post('os-volume_upload_image', """ | 10:11 |
| Anju | but this the same in json | 10:12 |
| Anju | giulivo : json and xml uses the same url | 10:12 |
| giulivo | nope, in json it posts to "url" and that is built from previous lines | 10:12 |
| prekarat | afazekas: ping | 10:15 |
| Anju | giulivo :are u checking the json? | 10:16 |
| giulivo | Anju, I wrote the json | 10:16 |
| Anju | :) | 10:16 |
| Anju | giulivo :http://www.fpaste.org/29433/13753521/ | 10:16 |
| Anju | this the json one | 10:16 |
| Anju | and what are u suggesting to me what wrong i am doing? | 10:17 |
| giulivo | yes you see it posts to "url" | 10:17 |
| giulivo | I don't think we can continue the review via IRC | 10:17 |
| prekarat | afazekas: are you aware of any ssh issues in paramiko client? am trying to test some ssh related tests by logging into the instance using RemoteClient, but unable to get any ssh keys generated. | 10:17 |
| Anju | giulivo :so how? | 10:18 |
| giulivo | I suggest you make all your changes to the xml client and also enable the _xml interface for the volume actions class and submit it as a _single_ drafted review | 10:18 |
| prekarat | afazekas: the sever_key returned is null, my problem is here http://paste.openstack.org/show/42848/ | 10:19 |
| giulivo | (drafted means you use git review -D) | 10:19 |
| giulivo | at that point add me or whoever you want as reviewer so we can comment the code inline | 10:19 |
| Anju | okk | 10:19 |
| Anju | i am giving u | 10:19 |
| Anju | giulivo : just a last one after that http://www.fpaste.org/29433/13753521/ | 10:22 |
| afazekas | prekarat: is it in tempest ? | 10:23 |
| prekarat | afazekas: yes | 10:23 |
| *** adam___ has joined #openstack-qa | 10:23 | |
| prekarat | afazekas: am porting teh test_create_servers.py file to v3 api and tring to enable the instance validation tests by enabling run_ssh true | 10:24 |
| prekarat | afazekas: but my tests fails because authorization fails and i see that the ssh keys are not getting generated | 10:24 |
| prekarat | afazekas: s/authorization/authentication | 10:25 |
| prekarat | afazekas: infact I i enable run_ssh=true in /etc/tempest/tempest.conf and run the v2 tests for test_create_servers.;py it fails there. | 10:27 |
| prekarat | afazekas: am wondering if its specific to my env or there is a bug in paramiko client? | 10:28 |
| afazekas | did you generated an ssh key-pair ? | 10:30 |
| afazekas | As I remember andreaf had pending change to fix it, but it was instable on neutron | 10:30 |
| prekarat | afazekas: no i didn't generate a key-pari | 10:32 |
| afazekas | https://review.openstack.org/#/c/22415/17/tempest/tests/compute/servers/test_server_actions.py | 10:32 |
| afazekas | AFAIK You need to generate a keypair first, (or upload a public key) | 10:33 |
| afazekas | andreaf: ping ^^^ | 10:34 |
| afazekas | https://bugs.launchpad.net/neutron/+bug/1153280 | 10:34 |
| prekarat | afazekas: aah ok , i was under the assumption tha paramiko client is going to do. i think it makes sense now. | 10:35 |
| * prekarat am wondering why folks didn't hit this b4 and how did this code get through our review? | 10:35 | |
| prekarat | afazekas: Thank You. that makes sense. am going to try that now | 10:36 |
| afazekas | I liked that ssh patch, so it would be nice if someone would be able to continue | 10:37 |
| afazekas | another related bug: https://bugs.launchpad.net/tempest/+bug/1132879 | 10:39 |
| afazekas | prekarat: ^^ | 10:40 |
| *** SergeyLukjanov has quit IRC | 10:41 | |
| openstackgerrit | anju Tiwari proposed a change to openstack/tempest: A volume upload test added for xml https://review.openstack.org/39664 | 10:45 |
| openstackgerrit | anju Tiwari proposed a change to openstack/tempest: A volume upload test added for xml https://review.openstack.org/39664 | 10:49 |
| prekarat | afazekasam wondering why htis ssh patch is abandoned. | 10:54 |
| prekarat | afazekas: ^^ | 10:54 |
| afazekas | because of neutron issues | 10:55 |
| afazekas | now the neutron gates are not voting | 10:56 |
| afazekas | IMHO it can be turned off on the neutron gate jobs when it is necessary | 10:57 |
| afazekas | IHMO the instance validations is a very useful thing | 10:57 |
| Anju | giulivo:did u check | 11:06 |
| giulivo | Anju, check what? | 11:09 |
| giulivo | you're posting to the wrong url | 11:09 |
| Anju | giulivo ::http://www.fpaste.org/29433/13753521/ | 11:10 |
| giulivo | anyway, as said before, use a draft submission so we can comment the code inline | 11:10 |
| Anju | i did but couldnt | 11:10 |
| Anju | giulivo : https://review.openstack.org/39664 | 11:11 |
| Anju | this one | 11:11 |
| Anju | sorry i tried like this | 11:11 |
| *** SergeyLukjanov has joined #openstack-qa | 11:11 | |
| Anju | i tried to make it as a draft | 11:11 |
| Anju | but i failede | 11:11 |
| giulivo | no problem, we can keep it public | 11:12 |
| giulivo | so to submit a draft you should have used "git review -D" | 11:12 |
| giulivo | https://wiki.openstack.org/wiki/Gerrit_Workflow | 11:12 |
| Anju | i use the same but git was failing | 11:12 |
| giulivo | so from now on I'll comment on gerrit and we can continue the discussion there | 11:12 |
| Anju | yeah sure | 11:13 |
| Anju | giulivo: please have a look in this https://github.com/openstack/tempest/blob/master/tempest/services/identity/xml/identity_client.py#L97 | 11:16 |
| Anju | giulivo :this 97 line | 11:19 |
| afazekas | sdague: ping | 11:24 |
| sdague | afazekas: pong | 11:24 |
| afazekas | https://review.openstack.org/#/c/39528/3 | 11:25 |
| afazekas | what does the 'unfubar_setuptools' is intended to solve ? | 11:26 |
| sdague | to get a working version of pip and setup tools for pbr we need a pip and a setup tools that's latest, and has been built against each other | 11:28 |
| sdague | that's the pattern to get you there, from mordred | 11:28 |
| afazekas | On F19 it really kills pip and these tools badly | 11:28 |
| sdague | we couldn't get through the gate without it | 11:28 |
| afazekas | I am not sure it is something I should disable on F19 or fix something else | 11:29 |
| raiesmh08 | sdague: I have added network API test cases please have a look on these - https://review.openstack.org/#/c/39023 | 11:30 |
| sdague | I would ping mordred later when he gets on | 11:30 |
| afazekas | sdague: as I remember to goal was to move pip above 1.4 | 11:31 |
| sdague | right | 11:32 |
| sdague | honestly, that section came from monty is us trying to get global requirements applied during devstack install | 11:32 |
| sdague | you know, the thing that we really need to do, is get F19 in the gate | 11:33 |
| afazekas | Ok I'll try to find another way to force 1.4 pip (if it wan't forced originally) | 11:33 |
| sdague | like do the postgresql job on F19 instead of ubuntu | 11:33 |
| afazekas | sounds good to me | 11:33 |
| sdague | so we can get feedback in the gate that this doesn't work | 11:33 |
| sdague | afazekas: someone has to do it though, it's going to be a bunch of work to make it a devstack target | 11:34 |
| sdague | in the gate | 11:34 |
| afazekas | I'll try to clean up my image creation steps and make it compatible with gate scripts | 11:35 |
| afazekas | kashyap: ^^^ | 11:45 |
| kashyap | afazekas, Reading the scroll.. | 11:46 |
| kashyap | afazekas, Cool! I just got Quantum w/ Grizzly bits working, after almost a week of struggle with it. I'm just associating the floating IPs now. Wish me luck :-) | 11:48 |
| afazekas | sdague: could that job use qpid instead of rabbitmq ? | 11:48 |
| sdague | afazekas: yes | 11:51 |
| openstackgerrit | A change was merged to openstack-dev/devstack: Launch ceilometer-alarm-* services. https://review.openstack.org/39300 | 11:51 |
| sdague | I think if we were bringing f19 into the gate, it would make sense to do that on the pg job and use qpid | 11:52 |
| giulivo | sdague++ | 11:52 |
| sdague | it's a lot of work for someone to do that, so a dedicated volunteer would be needed | 11:52 |
| sdague | however I'd support making that gate change | 11:52 |
| *** jasondotstar has quit IRC | 11:54 | |
| afazekas | sdague: I am definitely interested, but I do not know how much work it will be. I'll try to learn more about the infra | 11:57 |
| *** HenryG has joined #openstack-qa | 11:58 | |
| sdague | yeh, get with folks in -infra when they wake up | 11:58 |
| sdague | dprince might have a good idea of what's needed as well | 11:58 |
| sdague | as he did the centos2.6 unit test env | 11:58 |
| *** dkranz has quit IRC | 12:00 | |
| afazekas | I'll ask him, but I guess he has different opinion about the right OS version, but at least we can agree in the OS family | 12:01 |
| sdague | afazekas: well centos was needed for unit tests | 12:02 |
| sdague | because we needed python 2.6 | 12:03 |
| sdague | but for devstack nodes I think it should be fedora | 12:03 |
| sdague | and it couldn't be rhel.... for licensing reasons | 12:03 |
| *** mkollaro has joined #openstack-qa | 12:09 | |
| * afazekas will play kexec + kickstart for create F19 could image with pre installed C/C++ headers and packages | 12:12 | |
| *** weshay has joined #openstack-qa | 12:15 | |
| kashyap | afazekas, Please put a note on fedora-cloud list. Matthew Miller is doing some similar work.. Just ensuring there's no duplicate efforts :) | 12:19 |
| prekarat | afazekas: i have created that ssh key path and it works fine now, am going to submit it and requet your review | 12:19 |
| afazekas | kashyap: I found ks script in that image, but it had issues just with the 'Everething' named stable repo, it used different repos as well | 12:22 |
| afazekas | prekarat: cool | 12:23 |
| kashyap | afazekas, Sure, whatever optimized kickstart you end up with, please post it to the list (for historical context, etc...) | 12:24 |
| afazekas | kashyap: I am also thinking about blogging somewhere the kexec install method :) | 12:25 |
| kashyap | afazekas, Wonderful! You haven't added yourself to OpenStack planet, you might want to do so. | 12:26 |
| kashyap | :-) Sure, if you can blog more frequently, that'd be great! | 12:26 |
| afazekas | kashyap: too many thing for today, and I still did not tried reproduce the neutron smoke issues .. | 12:27 |
| kashyap | No worries, didn't mean /today/. | 12:29 |
| *** jasondotstar has joined #openstack-qa | 12:30 | |
| *** woodspa has joined #openstack-qa | 12:35 | |
| *** meteorfox has quit IRC | 12:35 | |
| rahmu | hey, I'm working on writing tempest tests covering swift account quotas. I have one problem with the fact that the admin user should have the ResellerAdmin role. | 12:36 |
| rahmu | Should I add the role in the beginning of the test and then remove it at the end? In that case how to do it? | 12:36 |
| rahmu | Should I create a special user for this case? | 12:36 |
| rahmu | So far I'm doing nothing, and it fails the devstack tests. Should I change devstack so that it grants ResellerAdmin to admin user by default? | 12:37 |
| afazekas | sdague: ^^ | 12:42 |
| sdague | rahmu: I think you should create a special user with the ResellerAdmin role | 12:43 |
| sdague | in the tests themselves | 12:43 |
| rahmu | sdague: how do I do this? | 12:44 |
| sdague | so that it is independent of how the underlying accounts were set up | 12:44 |
| sdague | rahmu: there should be some examples in the code on creating users | 12:44 |
| sdague | anywya, need to drop off to move to the office | 12:45 |
| rahmu | okay, I'm looking into it | 12:46 |
| rahmu | thanks for the help | 12:46 |
| openstackgerrit | Prem Karat proposed a change to openstack/tempest: Support for ssh auth to VMs via public key https://review.openstack.org/39684 | 12:46 |
| *** burt has joined #openstack-qa | 13:04 | |
| burt | sdague: that unfubar_* code is also causing problems for me on Ubuntu Precise 12.04. Not sure why others aren't seeing it yet or why it's not showing up in gate | 13:05 |
| afazekas | mtreinish: FYI looks like the EC2 tests are failed for me because I had http_proxy env variable, (but I had the no_proxy as well, for ignoring localhost) | 13:10 |
| burt | afazekas: the version of boto that comes with devstack probably doesn't support no_proxy | 13:11 |
| openstackgerrit | Fabien Boucher proposed a change to openstack/tempest: Add test for swift ACLs https://review.openstack.org/38823 | 13:13 |
| burt | I had a patch land for that but don't know if it made it in .. oh, it did. Should work in boto v2.9.9 and above | 13:14 |
| afazekas | burt: I guess it has some option to disable the proxy , I'll try to add it, or remove the http_proxy env var before any boto tests | 13:18 |
| burt | no_proxy did not work at all in boto v2.9.8 and below | 13:19 |
| burt | it would just igonre it | 13:19 |
| burt | ignore | 13:19 |
| afazekas | I see, but probably we have another ways to disable fully the proxies, so if you say the next version will fix it , I will not hurry to disable the boto proxy in tempest | 13:23 |
| burt | I don't think there's an option, it used to do "if 'http_proxy' in os.environ" and go from there | 13:27 |
| burt | but I could be wrong | 13:27 |
| *** meteorfox has joined #openstack-qa | 13:28 | |
| *** fbo_away is now known as fbo | 13:29 | |
| *** krtaylor has quit IRC | 13:32 | |
| afazekas | sdague,mordred : is this the solves the issue (at least on F19)http://www.fpaste.org/29471/37536391/ what the https://review.openstack.org/#/c/39528/3 solving on ubuntu ? Are there another things to solve ? | 13:35 |
| sdague | burt: interesting, it works on 12.04 in the gate | 13:45 |
| *** burt has quit IRC | 13:47 | |
| *** dkranz has joined #openstack-qa | 13:47 | |
| *** anteaya has joined #openstack-qa | 13:54 | |
| *** SergeyLukjanov has quit IRC | 14:01 | |
| mtreinish | sdague: about https://review.openstack.org/#/c/39099/ the problem with doing things in a common method or class is that the 2 clients have different input formats and the response format between them is different too. | 14:03 |
| mtreinish | I can do it that way but I thought it looked cleaner to handle it the way I did it | 14:03 |
| mtreinish | but, I'll respin it as a separate class that tries to handle everything uniformly | 14:03 |
| sdague | mtreinish: ok, that would be cool | 14:04 |
| *** Anju_ has joined #openstack-qa | 14:09 | |
| openstackgerrit | afazekas proposed a change to openstack/tempest: task_state must be consider before many action https://review.openstack.org/35165 | 14:14 |
| afazekas | tox -epep8 | 14:14 |
| afazekas | :) | 14:15 |
| openstackgerrit | afazekas proposed a change to openstack/tempest: task_state must be consider before many action https://review.openstack.org/35165 | 14:15 |
| *** krtaylor has joined #openstack-qa | 14:19 | |
| openstackgerrit | David Kranz proposed a change to openstack/tempest: Make testr full runs skip tests with 'slow' tag https://review.openstack.org/39716 | 14:22 |
| afazekas | dkranz: will you add a tox entry just for the slow tag ? | 14:24 |
| dkranz | afazekas: DId you see what I just pushed (WIP)? | 14:24 |
| dkranz | afazekas: oh I see what you mean. | 14:24 |
| dkranz | afazekas: No, I was not anticipating a job that ran jst slow tests | 14:25 |
| dkranz | afazekas: Do you think we need one? | 14:25 |
| afazekas | yes | 14:25 |
| dkranz | afazekas: For what purpose? | 14:26 |
| mtreinish | dkranz: if we plan to have a gate job that runs the slow tag | 14:26 |
| afazekas | dkranz: heat autoscale ? | 14:26 |
| mtreinish | then we will need a tox job that will run the slow tag tests | 14:26 |
| mtreinish | so that devstack-gate will call that tox job | 14:26 |
| dkranz | mtreinish: I thought we were going to have a tox entry that ran the slow heat tests | 14:27 |
| dkranz | mtreinish: I don't see the purpose of "generic slow for anyh project" | 14:27 |
| mtreinish | dkranz: oh, maybe I'm misunderstanding what we're trying to do here | 14:27 |
| dkranz | mtreinish: What's the point of a job that runs the slow tests from every project? | 14:28 |
| dkranz | mtreinish: The nightly should run everything | 14:28 |
| dkranz | mtreinish: And each gate job will run what it wants | 14:28 |
| afazekas | dkranz: usually the slow test are slow because they test things what otherwise not tested | 14:29 |
| mtreinish | dkranz: yeah I see what you're saying now, I was thinking of the tox job for the heat tests | 14:29 |
| dkranz | afazekas: Yes, and we want as many as possible in the gate. | 14:29 |
| dkranz | mtreinish: Yes, I am going to add that as soon as I see the slow skip working | 14:29 |
| dkranz | mtreinish: But we either need to move off of nose in favor of testr-serial or do another skip thing for nose slow. | 14:30 |
| dkranz | And the slow heat test still needs to be moved to scenario I think. | 14:30 |
| afazekas | dkranz: In parallel the slow tests are not too slow :) | 14:31 |
| *** jgallard has joined #openstack-qa | 14:31 | |
| rahmu | hello. I am writing a Tempest test. In the SetUpClass() I would like to create a user and assign it the "ResellerAdmin" role. Here's what I have so far: http://www.hastebin.com/riburevisu.py | 14:32 |
| rahmu | My problem is on line 23. How can I retrieve the role_id? | 14:32 |
| dkranz | afazekas: We shall see. | 14:32 |
| dkranz | afazekas: In theory we could run all tests that there ever are in one parallel gate run but I'm not sure that is the best approach. | 14:33 |
| afazekas | rahmu: try to use the AdminManager | 14:33 |
| mtreinish | rahmu: the way you get a role_id is that you have to list the roles and then pull out the id from the role's name that you're looking for | 14:33 |
| *** adalbas has quit IRC | 14:34 | |
| afazekas | rahmu: list the roles, and filter it | 14:34 |
| mtreinish | rahmu: see: https://github.com/openstack/tempest/blob/master/tempest/test.py#L207 | 14:34 |
| mtreinish | there it's pulling out the admin role id to add the created user to it | 14:34 |
| rahmu | awesome, thanks for both of you. I'll try to do this. (not right now, I've just been called to a meeting. But very soon :) ) | 14:36 |
| openstackgerrit | afazekas proposed a change to openstack/tempest: Disable logging to the stderr https://review.openstack.org/39505 | 14:39 |
| *** jgallard has quit IRC | 14:42 | |
| *** jgallard has joined #openstack-qa | 14:42 | |
| *** psedlak has joined #openstack-qa | 14:43 | |
| *** adalbas has joined #openstack-qa | 14:46 | |
| *** burt has joined #openstack-qa | 14:51 | |
| *** adalbas has quit IRC | 14:51 | |
| openstackgerrit | Monty Taylor proposed a change to openstack-dev/devstack: Unbreak the pip/setuptools install on fedora https://review.openstack.org/39725 | 14:52 |
| openstackgerrit | afazekas proposed a change to openstack/tempest: White space after # in thirdparty https://review.openstack.org/39726 | 14:56 |
| *** prekarat has quit IRC | 14:58 | |
| *** SergeyLukjanov has joined #openstack-qa | 14:59 | |
| openstackgerrit | afazekas proposed a change to openstack/tempest: White space after # in the services https://review.openstack.org/39728 | 14:59 |
| *** krtaylor has quit IRC | 15:03 | |
| openstackgerrit | afazekas proposed a change to openstack/tempest: White space after # in api https://review.openstack.org/39730 | 15:03 |
| *** jhenner has quit IRC | 15:05 | |
| *** adalbas has joined #openstack-qa | 15:08 | |
| *** Anju_ has quit IRC | 15:40 | |
| *** jgallard has quit IRC | 15:41 | |
| *** jgallard has joined #openstack-qa | 15:41 | |
| *** SergeyLukjanov has quit IRC | 15:48 | |
| *** rhsu has joined #openstack-qa | 15:50 | |
| *** tkammer has quit IRC | 15:51 | |
| *** krtaylor has joined #openstack-qa | 15:54 | |
| sdague | mtreinish: http://logs.openstack.org/05/35705/14/check/gate-tempest-devstack-vm-full/f58039f/console.html | 15:56 |
| sdague | thoughts on what we need to change in tempest to get us past that? | 15:57 |
| sdague | that's on the global requirements branch | 15:57 |
| *** sthaha has quit IRC | 15:58 | |
| openstackgerrit | afazekas proposed a change to openstack/tempest: White space after # in the tempest/services https://review.openstack.org/39728 | 16:16 |
| openstackgerrit | afazekas proposed a change to openstack/tempest: White space after # in tempest/api files https://review.openstack.org/39730 | 16:19 |
| *** jgallard has quit IRC | 16:24 | |
| *** SergeyLukjanov has joined #openstack-qa | 16:25 | |
| openstackgerrit | Monty Taylor proposed a change to openstack-dev/devstack: Unbreak the pip/setuptools install on fedora https://review.openstack.org/39725 | 16:25 |
| openstackgerrit | afazekas proposed a change to openstack/tempest: setUpClass/tearDownClass full chain https://review.openstack.org/39346 | 16:31 |
| dkranz | sdague: I just updated tempest and see all the debug spew when I run it. What is the recommended way to shut it up? | 16:39 |
| dkranz | sdague: And shouldn't this be off by default? | 16:39 |
| openstackgerrit | A change was merged to openstack/tempest: Add argument to stop stress test on first error https://review.openstack.org/39524 | 16:39 |
| clarkb | dkranz yes it should be disabled. mtreinish fixes it through devstack and afazekas has a tempest change to fix it more directly | 16:42 |
| openstackgerrit | aguzikova proposed a change to openstack-dev/devstack: Add keystoneclient support for cinder. https://review.openstack.org/39747 | 16:43 |
| dkranz | clarkb: OK, thanks. Think I finally got the regexp https://review.openstack.org/#/c/39716/ :) | 16:43 |
| *** fbo is now known as fbo_away | 16:47 | |
| openstackgerrit | afazekas proposed a change to openstack/tempest: setUpClass/tearDownClass full chain https://review.openstack.org/39346 | 16:51 |
| openstackgerrit | Giulio Fidente proposed a change to openstack/tempest: renames the stress test class to include the Volume keyword https://review.openstack.org/39654 | 16:54 |
| *** mlavalle has joined #openstack-qa | 16:54 | |
| sdague | dkranz: yeh, it's fixed on devstack side, there is a needed tempest config default change as well | 16:56 |
| sdague | if you are running non devstack | 16:56 |
| sdague | mtreinish: you got that proposed somewhere? | 16:56 |
| mtreinish | sdague: the devstack fix or the tempest one? | 16:57 |
| dkranz | sdague: I think one of you just needs to approve https://review.openstack.org/#/c/39505/, no? | 16:57 |
| mtreinish | sdague: the devstack one is merged afazekas just respun the tempest one | 16:57 |
| *** burt is now known as burt-afk | 16:57 | |
| mtreinish | dkranz: yeah that should fix it on the tempest side | 16:57 |
| dkranz | mtreinish: Can you approve now? | 16:58 |
| mtreinish | dkranz: yeah the output was clean right? | 16:58 |
| sdague | done | 16:58 |
| mtreinish | actually that's not a real test of it since the devstack change landed, but it should be fine | 16:59 |
| dkranz | sdague: Thx. | 16:59 |
| dkranz | GUess it is about meeting time. | 16:59 |
| sdague | yep | 16:59 |
| dkranz | mtreinish: I'll confirm after the meeting. | 16:59 |
| mtreinish | dkranz: it's fine those are the 2 options I set in devstack | 17:00 |
| openstackgerrit | saurabh proposed a change to openstack/tempest: Added some more server stress tests https://review.openstack.org/39752 | 17:02 |
| *** fbo_away is now known as fbo | 17:45 | |
| *** fbo is now known as fbo_away | 17:45 | |
| openstackgerrit | Armando Migliaccio proposed a change to openstack-dev/devstack: Remove unused keystone params from neutron agents' config files https://review.openstack.org/39763 | 17:46 |
| *** fbo_away is now known as fbo | 17:54 | |
| giulivo | afazekas, so I'll try to clean up the stress tests naming choice by removing any "duplication", makes sense? | 17:57 |
| sdague | mtreinish: can you take a look at the logs on https://review.openstack.org/#/c/35705/ about what we need to change in tempest requirements.txt to get past it | 17:59 |
| mtreinish | sdague: sure | 18:00 |
| afazekas | mkoderer: ^^^ What do you think ? https://review.openstack.org/#/c/39654/ | 18:03 |
| openstackgerrit | A change was merged to openstack/tempest: Disable logging to the stderr https://review.openstack.org/39505 | 18:05 |
| afazekas | mlavelle: if you have any closer info why to neutron smoke gate failing so frequently let me know. I will try to use some complex debug/tracer solutions if it is necessary to find out the root causes | 18:05 |
| mkoderer | afazekas: I will have a look to it tomorrow | 18:06 |
| mkoderer | afazekas: sounds interesting ;) | 18:07 |
| *** mlavalle has quit IRC | 18:07 | |
| *** burt-afk is now known as burt | 18:08 | |
| mtreinish | sdague: so testtools along with: nose, keyring, testresources, and testrepository aren't in the global requirements.txt | 18:11 |
| mtreinish | so they don't get installed when you run the update requirements | 18:12 |
| mtreinish | I believe that's the problem | 18:12 |
| mtreinish | they are in global test-requirements.txt though | 18:13 |
| mtreinish | but I bet that's the issue since they are in requirements for tempest. | 18:13 |
| mtreinish | mordred: ^^^ do you think that could be the issue? | 18:14 |
| mordred | yes | 18:17 |
| mordred | there was a time when we considered making the global list not be a split list | 18:17 |
| mordred | but be a single list, and let people put those things in to reqs or test-reqs as was appropriate, and make update.py do the rigt thing | 18:18 |
| clarkb | mordred: I still think that is the correct thing to do | 18:18 |
| mordred | clarkb: I agree | 18:18 |
| afazekas | mordred: ping | 18:22 |
| mordred | afazekas: hi! | 18:23 |
| afazekas | https://review.openstack.org/#/c/39725/2/lib/infra | 18:23 |
| mordred | yup | 18:23 |
| mordred | I'm still working on that, btw | 18:23 |
| afazekas | I wonder why is necessary to uninstall the setuptools | 18:23 |
| mordred | because, sadly, we were having problems with system setuptools and pip-installed setuptools conflicting | 18:24 |
| mordred | but I'm actually going investigate the other direction right now, and see if I can prevent them from conflicting | 18:24 |
| afazekas | mordred: does it really causes issues if the distribution does not follows the FHS ? | 18:25 |
| mordred | omg yes | 18:25 |
| mordred | I'm very mad at redhat right now | 18:25 |
| *** mlavalle has joined #openstack-qa | 18:25 | |
| afazekas | https://bugzilla.redhat.com/show_bug.cgi?id=662034 I found this | 18:25 |
| mordred | yup. that's even not the whole thing | 18:25 |
| mordred | they've done sometihng to their python install | 18:26 |
| mordred | which causes 'sudo python setup.py' to install things into /usr/lib/python2.7 | 18:26 |
| mordred | this is categorically evil | 18:26 |
| mordred | because it leads to files overlayed with each other being managed by different systems | 18:26 |
| afazekas | do you think there is an addtional patch in the fedora setuptools to be evil ? | 18:27 |
| mordred | /usr/local/ exists for a reason and has existed for a reason since before redhat was a company | 18:27 |
| mordred | I think it's not fedora setuptools | 18:27 |
| mordred | I think it's fedora python | 18:27 |
| mordred | and it's not the first time that rh have made incompatible changes in their system python | 18:27 |
| mordred | they have patches from the 3.x series backported into 2.7 which cause us to have to patch eventlet in our virtualenv | 18:27 |
| mordred | which is just mean | 18:27 |
| mordred | I'm going to find and fix whatever they did to python and submit it back upstream | 18:28 |
| mordred | and if they won't acept it, I'm going to strongly push that openstack drop support for redhat as a viable platform until it's fixed | 18:28 |
| afazekas | BTW in fedora 19 you do not need to pip install the argparse it comes with the python-libs (after devstack I explicitly unisntall the pip installed one) | 18:28 |
| mordred | because it is categorically unworkable | 18:28 |
| mordred | that's because argparse is a normal part of 2.7 | 18:29 |
| mordred | that's fine :) | 18:29 |
| mordred | argparse is a backport library to help folks on 2.6 | 18:29 |
| mordred | but there's also no harm in installing it on a 2.7 system - it should be mostly a no-op | 18:29 |
| afazekas | mordred: so devstack should not install it from pip on fedora 19 | 18:29 |
| mordred | no. devstack should have no opinion on that | 18:30 |
| afazekas | something contains it in the requirements.txt | 18:30 |
| mordred | yes | 18:30 |
| mordred | argparse is a declared dependency of the project, and the project supports 2.6 | 18:30 |
| mordred | so it's required to be in requirements.txt | 18:30 |
| afazekas | What should ensure on F19 to do not duplicate it ? | 18:30 |
| mordred | the complexity of making it conditional is just not worth it | 18:30 |
| mordred | is it causing a problem? | 18:31 |
| afazekas | I got a warning on several commands when it is duplicated | 18:31 |
| mordred | ok. if you could send those to me, I'd like to sort that out | 18:31 |
| afazekas | Several tempest cli test is failing :) | 18:31 |
| * mordred cries | 18:31 | |
| mordred | they fail on fedora 19 but not on ubuntu? | 18:31 |
| mordred | both with python 2.7? | 18:32 |
| afazekas | python 2.7.5 | 18:33 |
| afazekas | Looks now it does not happens , I'll try it again tomorrow | 18:36 |
| mordred | cool | 18:38 |
| mordred | I was going to be VERY sad if I had yet another python packaging weirdness to investigate | 18:39 |
| afazekas | https://bugzilla.redhat.com/show_bug.cgi?id=662034#c10 | 18:41 |
| afazekas | mordred: do you have link to the back-ported patch ? | 18:46 |
| mordred | for? | 18:47 |
| afazekas | mordred: 'they have patches from the 3.x series backported into 2.7 which cause us to have to patch eventlet in our virtualenv' | 18:48 |
| mordred | ah - no. I gave up on that one a while ago. I've heard that they've fixed the problem not... | 18:48 |
| afazekas | modrder: is there any 'python-setuptools' version, which does not necessary to uninstall ? | 18:50 |
| afazekas | mordred: ^^ | 18:51 |
| afazekas | is the uninstall step skip able, if there is no rpm/deb originated python-setuptools package installed to the system before devstack starts ? | 18:54 |
| mordred | afazekas: not clear - I'm investigating right now | 18:54 |
| mordred | my goal is to remove the uninstall | 18:54 |
| mordred | but I'm going to see if I can get a set of small test cases so that we can prove it works or doesn't | 18:54 |
| openstackgerrit | A change was merged to openstack/tempest: Add testing of Neutron per tenant quotas API https://review.openstack.org/35724 | 18:55 |
| openstackgerrit | Monty Taylor proposed a change to openstack-dev/devstack: Unbreak the pip/setuptools install on fedora https://review.openstack.org/39725 | 18:55 |
| mordred | afazekas: I just put that up as a canary to see what breaks in the gate | 18:56 |
| mordred | afazekas: that's not removing it, and only upgrading setuptools and pip | 18:56 |
| burt | mordred: that just failed for me on my test, but my precise environment wasn't clean | 19:01 |
| mordred | burt: which one? | 19:02 |
| burt | 39725 | 19:02 |
| mordred | ok | 19:02 |
| mordred | I'm trying to dig further in to some root causes here | 19:02 |
| burt | let me try it from the real clean one | 19:02 |
| mordred | burt: becuase putting bandaids on top of bandaids... | 19:02 |
| burt | yeah | 19:02 |
| mordred | burt: the last one I just pushed up doesn't do the remove ... it goes back to the original just-upgrade approach | 19:03 |
| burt | ah | 19:03 |
| mordred | because I want to see the error and see if we can track _that_ down | 19:03 |
| mordred | the workarounds needing workarounds was getting old :) | 19:03 |
| mordred | https://jenkins.openstack.org/job/gate-tempest-devstack-vm-full/38321/console | 19:04 |
| *** mkollaro has quit IRC | 19:04 | |
| mordred | in case you want to watch it - but I'm steping out for a sec | 19:04 |
| sdague | mordred: ok, reading through scrollback | 19:09 |
| afazekas | mordred: looks like the argparse workaround is not required now | 19:10 |
| sdague | mordred: so are you working on the update.py updates? | 19:11 |
| sdague | or should I run at it | 19:11 |
| openstackgerrit | Joe Hakim Rahme proposed a change to openstack/tempest: Adds tests covering Swift's Account Quota middleware https://review.openstack.org/39659 | 19:19 |
| openstackgerrit | Dean Troyer proposed a change to openstack-dev/devstack: Fix get_pip_command() https://review.openstack.org/39771 | 19:19 |
| burt | mordred: http://paste.openstack.org/show/42884/ | 19:21 |
| *** mkollaro has joined #openstack-qa | 19:22 | |
| openstackgerrit | David Kranz proposed a change to openstack/tempest: Make testr full runs skip tests with 'slow' tag https://review.openstack.org/39716 | 19:32 |
| rahmu | jeblair: I don't understand the comment you left here: https://review.openstack.org/#/c/39659/. Do you mean that you found errors in the patch set? | 19:36 |
| jeblair | rahmu: no, i left that comment on about 30 changes to re-enqueue them because i restarted zuul | 19:38 |
| rahmu | What's zuul? | 19:39 |
| rahmu | found it: http://zuul.openstack.org/ | 19:41 |
| rahmu | :) | 19:41 |
| adalbas | mtreinish, i linked bug https://bugs.launchpad.net/tempest/+bug/1207498 to the speed up testr bp | 19:50 |
| adalbas | mtreinish, i also see a lot of failures on the job in the gate right now | 19:51 |
| mtreinish | adalbas: ok cool thanks | 19:53 |
| mtreinish | let me take a look at the job's log | 19:53 |
| sdague | rahmu: actually http://status.openstack.org/zuul/ is the nice view | 19:54 |
| afazekas | mtreinish: so the 'tempest.api.compute.admin.test_aggregates:AggregatesAdminTestXML.test_aggregate_add_host_create_server_with_az' is the most problematic testr test right ? | 19:55 |
| mtreinish | afazekas: well it's actually the json version that's failing right now | 19:55 |
| mtreinish | but yeah | 19:55 |
| mtreinish | adalbas: it looks like a good chunk of them are jenkins failures | 19:55 |
| mtreinish | I see a security group one (which I'm working on respinning the fix for right now) and another az fail | 19:56 |
| afazekas | mtreinish: is it tempest or nova side issue , I saw you had nova related change | 19:56 |
| adalbas | mtreinish, right, they show failures on other jobs too | 19:56 |
| mtreinish | afazekas: that was actually for an unrelated race condition that testr uncovered in another aggregates test | 19:57 |
| mtreinish | afazekas: I believe that it's a tempest issue, but I haven't confirmed it yet | 19:57 |
| mtreinish | jog0 and I were starting to look at it last night | 19:57 |
| mtreinish | I tried to add extra debug with https://review.openstack.org/#/c/39596/ but I can't get it to fail there | 19:58 |
| adalbas | mtreinish, want me to also take a look on that aggregates issue? any points where i should start at? | 19:58 |
| afazekas | mtreinish: I will check is the used lock method execve safe | 19:58 |
| mtreinish | adalbas: well the fail is outlined here: http://paste.openstack.org/show/42787/ | 20:00 |
| mtreinish | adalbas: basically when we create a server with an az (which is created right before) nova returns that the az is not available | 20:01 |
| adalbas | mtreinish, ok | 20:02 |
| mtreinish | afazekas: I'm pretty sure the locking is working fine. It logs the proper lock statements at the right times at least. | 20:04 |
| afazekas | mtreinish: so you are seeing lock wait log entries | 20:06 |
| mtreinish | afazekas: I'd have to run it again but that's what I remember. My current theory is that there is another test interfering somewhere. Which is why we only see it some times. | 20:06 |
| mtreinish | afazekas: but I guess it doesn't hurt to double check the locking is working | 20:08 |
| *** mkollaro has quit IRC | 20:09 | |
| afazekas | mtreinish: you can try this for reproducing the issue: https://review.openstack.org/#/c/38980/ # class_setup_in_run=false | 20:11 |
| * afazekas time to go, bye | 20:12 | |
| *** afazekas has quit IRC | 20:13 | |
| *** psedlak has quit IRC | 20:17 | |
| lifeless | dkranz: mtreinish: hi; testr exclusion - what do you need? | 20:18 |
| dkranz | lifeless: Just a way to exclude tests based on a regexp from those already included | 20:18 |
| mtreinish | lifeless: for example this is what we're doing now: https://review.openstack.org/#/c/39716/ | 20:19 |
| dkranz | mtreinish: I have to leave to an appt. Can you take this please? | 20:19 |
| dkranz | mtreinish: SOrry about that. | 20:20 |
| mtreinish | dkranz: sure | 20:20 |
| openstackgerrit | A change was merged to openstack-dev/devstack: Add Midonet plugin support to devstack. https://review.openstack.org/36607 | 20:20 |
| lifeless | mtreinish: you're using a regex to exclude. | 20:20 |
| lifeless | mtreinish: perhaps I don't understand the suggestion? | 20:20 |
| mtreinish | lifeless: have a cli option to specify exclude regex. So we filter with regex like we did before but then exclude a subset of that filtered list with a different regex . | 20:22 |
| mtreinish | so from the example: (?!.*\[.*\bslow\b.*\]) would become something like: --exclude [.*\bslow\b.*\] | 20:23 |
| lifeless | why ? | 20:24 |
| lifeless | is it a convenience thing, or is it a correctness thing? | 20:24 |
| *** dkranz has quit IRC | 20:24 | |
| mtreinish | lifeless: convenience thing mostly. There are people who need to exclude a set of individual tests when running against non devstack environments and the regex gets really unwieldy if they need to figure out one that excludes those tests plus runs the subset we have in the gate | 20:25 |
| lifeless | huh | 20:26 |
| lifeless | how is 'in the gate' defined today ? | 20:28 |
| mtreinish | lifeless: in the gate it's 'tempest.api tempest.scenario tempest.thirdparty tempest.cli' right now we're going to switch it to '^tempest\.(api|scenario|thirdparty|cli)(?!.*\[.*\bslow\b.*\]).*$' to exclude the slow tag | 20:29 |
| *** HenryG has quit IRC | 20:30 | |
| mordred | oh look. it's lifeless! | 20:32 |
| lifeless | mtreinish: that re looks wrong to me | 20:33 |
| lifeless | mtreinish: re.search(r'^tempest\.(api|scenario|thirdparty|cli)(?!.*\[.*\bslow\b.*\]).*$', 'tempest.api.foo(slow)') | 20:33 |
| openstackgerrit | A change was merged to openstack/tempest: Make test_neutron_dhcp_agent_list_hosting_net use net name from conf https://review.openstack.org/39581 | 20:33 |
| lifeless | -> matches | 20:33 |
| mtreinish | lifeless: ok, a good thing that it's still up for review then :) | 20:34 |
| lifeless | mtreinish: I'll have to fiddle around for a bit to test properly - my tempest venv has decided to throw a sad | 20:34 |
| mtreinish | lifeless: according to dkranz the issue is that testr ORs multiple regexp and we need it to AND them to make it more convenient to do multiple excludes. Otherwise the regex just keeps getting longer and easier to make mistakes. | 20:34 |
| mtreinish | lifeless: ok, I think there have been some requirements changes to tempest recently. | 20:35 |
| clarkb | lifeless: that string should match as slow is not surrounded by brackets | 20:35 |
| lifeless | oh, did I pebkac it | 20:36 |
| lifeless | ah I did. cool | 20:36 |
| lifeless | ok so | 20:36 |
| lifeless | mtreinish: right, it does or them. | 20:38 |
| lifeless | so one thing we could do is offer an 'and' combiner - but. | 20:40 |
| lifeless | that gets kinda complex - is it anding the two adjacent re's, or the entire re so far + the new one? | 20:41 |
| mtreinish | lifeless: that why we figured doing an exclude flag would be easier so by default it works the same. | 20:42 |
| lifeless | mtreinish: the issue with exclude flags is 'do they exclude found so far, or exclude from anything found' | 20:43 |
| mtreinish | but you can use the exclude flag to do remove matches from the list and use AND there | 20:43 |
| lifeless | so you get the same 'how do we build up complex patterns' issue | 20:43 |
| mtreinish | lifeless: I'm not following 'found so far' vs 'anything found' | 20:43 |
| lifeless | should 'testr run foo --exclude bar quux' include or exclude a test called 'bar_quux' ? | 20:45 |
| mtreinish | lifeless: I think it would exclude, you should be explicit saying that exclude is evaluated last. So the test would be picked up because it matches quux, but then excluded because it matches bar. | 20:47 |
| lifeless | so that conflicts with the concept of 'let someone A define a macro that runs testr and let someone B evolve the test selection' | 20:48 |
| lifeless | because if A has an exclude B now cannot escape it. | 20:48 |
| lifeless | which AIUI is the actual challenge here - you don't want devs to have to manipulate an unwieldy regex by hand to tweak the test selection | 20:50 |
| openstackgerrit | Joe Hakim Rahme proposed a change to openstack/tempest: Adds tests covering Swift's Account Quota middleware https://review.openstack.org/39659 | 20:51 |
| *** woodspa has quit IRC | 20:53 | |
| mtreinish | lifeless: yeah for both devs and users. (the whole discussion was started by: https://review.openstack.org/#/c/39417/ which I feel is a terrible idea and the wrong way to do it) | 20:53 |
| *** giulivo has quit IRC | 20:53 | |
| lifeless | mtreinish: oh gosh | 20:54 |
| rahmu | sdague: thanks, bookmarked. | 20:54 |
| lifeless | mtreinish: so it's not terrible | 20:54 |
| lifeless | mtreinish: but it does rather entrench more runner logic into tempest | 20:55 |
| *** meteorfox has quit IRC | 20:55 | |
| mtreinish | lifeless: ok so maybe terrible was a bit harsh. But I feel like we should do it through the test runner and not in tempest. | 20:56 |
| lifeless | me too | 20:56 |
| lifeless | so one thing testr doesn't have today is profiles | 20:57 |
| lifeless | e.g. the config file in total affects every run every time | 20:57 |
| lifeless | and testr doesn't 'understand' that running just slow tests would be a 'full test run' for that partition. | 20:58 |
| *** SergeyLukjanov has quit IRC | 21:00 | |
| mtreinish | lifeless: so if I'm understanding, we could just add profiles that specify the subset of tests that get run? | 21:03 |
| lifeless | its a concept I'm noodling with | 21:03 |
| lifeless | it's not implemented :) | 21:03 |
| clarkb | could the test loader itself make the decision? | 21:04 |
| lifeless | but yeah, you'd have N profiles - say gate and slow and all. When running select a profile, and the profile would define a partition which you could shrink but not grow. | 21:04 |
| lifeless | clarkb: yes, see above. | 21:04 |
| lifeless | mtreinish: suggested a regex style that will be easier to extend | 21:06 |
| mtreinish | lifeless: ok, that's basically what we've been using tox jobs for: https://github.com/openstack/tempest/blob/master/tox.ini Then the filter logic after specifying could be AND instead of OR to make specifiying negative matches easier? | 21:07 |
| mtreinish | (*specifying a profile) | 21:07 |
| openstackgerrit | Matthew Treinish proposed a change to openstack/tempest: Add tenant isolation for scenario tests https://review.openstack.org/39099 | 21:14 |
| *** openstackgerrit has quit IRC | 21:25 | |
| *** openstackgerrit has joined #openstack-qa | 21:25 | |
| *** jasondotstar has quit IRC | 21:38 | |
| *** torandu has quit IRC | 21:40 | |
| *** wramthun has joined #openstack-qa | 21:40 | |
| *** torandu has joined #openstack-qa | 21:41 | |
| *** burt has quit IRC | 21:49 | |
| jog0 | mtreinish: any progress with the aggregates test issue? | 21:58 |
| mtreinish | jog0: not really, the debug print patch hasn't failed testr once, I've done several rechecks on it. | 22:00 |
| clarkb | what purpose do aggregates serve? I was asked that recently by someone and I didn't even know there were aggregates | 22:01 |
| clarkb | are they just logical groupings of nodes. eg rackX or rowY? | 22:01 |
| jog0 | clarkb: mostly | 22:01 |
| jog0 | clarkb: they are a way to assign metadata to nodes. this metadata can be used for scheduling and whatnot | 22:02 |
| jog0 | AZs use aggregates underneath. and you can use aggregates to do thinkgs like say windows VMs go on this group | 22:02 |
| clarkb | jog0: and aggregates aren't specific like an AZ right? | 22:04 |
| jog0 | specific? | 22:04 |
| jog0 | aggregates are ment to be for deployers only and not for users to ever know about | 22:05 |
| clarkb | AZ has a specific meaning right? it is an availability zone which keystone and other things grok | 22:06 |
| clarkb | it seems like with aggregates you have to build a thing that groks what they mean on top. One such thing is AZs | 22:07 |
| openstackgerrit | Matthew Treinish proposed a change to openstack/tempest: Do not merge test for az fail https://review.openstack.org/39596 | 22:07 |
| mtreinish | jog0: ^^^ I added --until-failure to the job so it'll keep on running until it fails (or jenkins times out) so hopefully it'll hit it | 22:08 |
| clarkb | EmilienM: lgtm | 22:09 |
| clarkb | whoops ww | 22:09 |
| *** lcazares has joined #openstack-qa | 22:10 | |
| jog0 | clarkb: yes, correct. This makes things easier for deployers to do custom scheduling etc | 22:11 |
| jog0 | mtreinish: haha I didn't know that was an option | 22:11 |
| *** jasondotstar has joined #openstack-qa | 22:15 | |
| mtreinish | jog0: yeah it's useful for trying to debug these races with parallel tests. I just can never use it on my dev box because I normally have something not working quite right in my environment. | 22:18 |
| openstackgerrit | Matthew Treinish proposed a change to openstack/tempest: Add tenant isolation for scenario tests https://review.openstack.org/39099 | 22:20 |
| openstackgerrit | Kui Shi proposed a change to openstack-dev/devstack: Update tempest test sample in devstack/README.md https://review.openstack.org/39820 | 22:24 |
| lifeless | mtreinish: something like that yes | 22:27 |
| *** plomakin has quit IRC | 22:29 | |
| *** plomakin has joined #openstack-qa | 22:29 | |
| *** krtaylor has quit IRC | 22:33 | |
| *** weshay has quit IRC | 22:36 | |
| openstackgerrit | Chmouel Boudjnah proposed a change to openstack-dev/devstack: Don't add bulk middleware. https://review.openstack.org/39826 | 22:41 |
| openstackgerrit | Dean Troyer proposed a change to openstack-dev/devstack: Add tools/install_pip.sh https://review.openstack.org/39827 | 22:43 |
| *** meteorfox has joined #openstack-qa | 22:44 | |
| openstackgerrit | A change was merged to openstack-dev/devstack: Add support for vmdk files as glance images https://review.openstack.org/38576 | 22:51 |
| openstackgerrit | afazekas proposed a change to openstack/tempest: Generate temepst API doc from source https://review.openstack.org/39291 | 23:14 |
| openstackgerrit | Matthew Treinish proposed a change to openstack/tempest: Add tenant isolation for scenario tests https://review.openstack.org/39099 | 23:20 |
| *** rhsu has quit IRC | 23:54 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!