| *** stevebaker has quit IRC | 00:00 | |
| *** stevebaker has joined #openstack-qa | 00:06 | |
| *** krtaylor has quit IRC | 00:09 | |
| *** lexinator has quit IRC | 00:28 | |
| openstackgerrit | Matthew Treinish proposed a change to openstack/tempest: Switch to testr serial instead of nose https://review.openstack.org/40723 | 00:37 |
|---|---|---|
| *** lexinator has joined #openstack-qa | 00:38 | |
| *** lexinator1 has joined #openstack-qa | 00:44 | |
| *** lexinator has quit IRC | 00:44 | |
| afazekas | dkranz: cool | 01:09 |
| *** lexinator1 is now known as Lexinator | 01:19 | |
| openstackgerrit | A change was merged to openstack-dev/devstack: misleading source filename in error message https://review.openstack.org/39889 | 01:30 |
| *** wramthun has joined #openstack-qa | 01:58 | |
| *** Lexinator has quit IRC | 02:10 | |
| *** HenryG has joined #openstack-qa | 02:20 | |
| *** LiangZheng has joined #openstack-qa | 02:24 | |
| openstackgerrit | Mike Perez proposed a change to openstack-dev/devstack: Add Cinder V2 API to keystone catalog https://review.openstack.org/22489 | 02:27 |
| *** annegentle has quit IRC | 02:28 | |
| *** wramthun has quit IRC | 03:04 | |
| *** mordred has quit IRC | 03:04 | |
| *** mordred has joined #openstack-qa | 03:07 | |
| kashyap | afazekas, Hey, I'm in Charleston, US, traveling... | 03:18 |
| *** annegentle has joined #openstack-qa | 03:25 | |
| *** anteaya has quit IRC | 03:27 | |
| *** SergeyLukjanov has joined #openstack-qa | 03:46 | |
| *** giulivo has quit IRC | 03:57 | |
| *** SergeyLukjanov has quit IRC | 04:11 | |
| openstackgerrit | Arata Notsu proposed a change to openstack/tempest: Set a preffered region for all services https://review.openstack.org/41026 | 04:33 |
| *** afazekas_ has joined #openstack-qa | 04:47 | |
| *** afazekas_ has quit IRC | 04:48 | |
| *** afazekas_ has joined #openstack-qa | 04:48 | |
| Anju | afazeaks:are u there? | 05:02 |
| Anju | afazekas: i ahve amde the upload function fopr xml also | 05:03 |
| Anju | i have made the upload function fopr xml also | 05:03 |
| Anju | afazekas: should ia dd this | 05:04 |
| Anju | but some two more xml tests cases are missing | 05:04 |
| Anju | for attach and dettach volume | 05:04 |
| Anju | please suggest me | 05:04 |
| *** wramthun has joined #openstack-qa | 05:05 | |
| *** marun has quit IRC | 05:26 | |
| *** marun has joined #openstack-qa | 05:27 | |
| mkoderer | Good morning / Good afternoon india ;) | 05:54 |
| Anju | : ) good morning mkoderer | 05:54 |
| Anju | 11:22 am | 05:54 |
| mkoderer | ohhh.. next time ;) | 05:54 |
| Anju | :) | 05:54 |
| *** sourabh has joined #openstack-qa | 05:59 | |
| sourabh | hi mkoderer | 05:59 |
| sourabh | https://review.openstack.org/#/c/39752/ ?? | 05:59 |
| mkoderer | sourabh: hi, good that you are here | 06:00 |
| mkoderer | sourabh: so we discussed it yesterday.. | 06:00 |
| sourabh | just joined :) | 06:00 |
| mkoderer | the finial plan it to use the existing api / scenario test for stress tests | 06:01 |
| sourabh | ok | 06:01 |
| mkoderer | so with that we don't need to duplicate stuff inside the stress test framework | 06:01 |
| sourabh | yeah good idea!! | 06:02 |
| mkoderer | sourabh: so I am sorry to say, but I think you could abandon you change | 06:02 |
| sourabh | ok, no problem | 06:02 |
| mkoderer | sourabh: but we have a lot of stuff to do... so if you want to help :) | 06:02 |
| sourabh | yes sure... | 06:03 |
| mkoderer | I will try to make a proposal for the new design in the next days | 06:03 |
| mkoderer | but we already have the first step with https://review.openstack.org/#/c/38980/ | 06:04 |
| sourabh | Yes, I am also waiting for it. | 06:05 |
| mkoderer | yes, I will try to push it ;) | 06:05 |
| sourabh | ok good luck ;) | 06:07 |
| *** sourabh has quit IRC | 06:16 | |
| *** yfried has joined #openstack-qa | 06:36 | |
| *** yfried has left #openstack-qa | 06:37 | |
| openstackgerrit | nayna-patel proposed a change to openstack/tempest: Added test case to check floating IP API operations https://review.openstack.org/41038 | 07:22 |
| *** giulivo has joined #openstack-qa | 07:29 | |
| *** fbo_away is now known as fbo | 07:38 | |
| openstackgerrit | Fabien Boucher proposed a change to openstack/tempest: Add tests for swift container listing filters https://review.openstack.org/40416 | 07:42 |
| afazekas_ | mkoderer: I am going to update that patch as we discussed in several hours | 07:43 |
| *** afazekas has quit IRC | 07:44 | |
| *** afazekas_ is now known as afazekas | 07:44 | |
| mkoderer | afazekas: ok great | 07:45 |
| *** SergeyLukjanov has joined #openstack-qa | 07:49 | |
| afazekas | giulivo: for the 'class_setup_per_thread' I need to negate the flags' meaning right ? | 07:51 |
| giulivo | hi afazekas | 07:52 |
| giulivo | morning | 07:52 |
| afazekas | good morning :) | 07:52 |
| giulivo | I'm not sure if that a good naming choice, I just tried to think about a "self explanatory" name | 07:53 |
| afazekas | looks like devstack master is unbreaken on fedora 19 :) | 07:53 |
| afazekas | class_setup_once | 07:54 |
| giulivo | afazekas, also good for me | 07:54 |
| *** jhenner has joined #openstack-qa | 07:55 | |
| afazekas | class_setup_once_per_thread | 07:58 |
| *** thomasbiege has joined #openstack-qa | 07:58 | |
| giulivo | so I'd read _once as it is executed only once and not per thread | 07:59 |
| giulivo | what would _once_per_thread mean? | 07:59 |
| afazekas | in the json file you can specify the number of the worker threads(processes). the setupclass called once per thread or by each run | 08:04 |
| * afazekas note: on linux the tread/process terminology is very tricky, based on the context | 08:07 | |
| openstackgerrit | Yuuichi Fujioka proposed a change to openstack/tempest: Add a test case for instance boot from volume https://review.openstack.org/41048 | 08:29 |
| openstackgerrit | nayna-patel proposed a change to openstack/tempest: Added test case to check floating IP API operations https://review.openstack.org/41038 | 08:48 |
| openstackgerrit | afazekas proposed a change to openstack/tempest: Unit tests as stress tests https://review.openstack.org/38980 | 08:48 |
| giulivo | afazekas, I knew the functionality | 08:48 |
| giulivo | I was arguing that _once_per_thread isn't easy to read as it doesn't immediately tell if it is executed once OR per_thread | 08:49 |
| giulivo | afazekas, so I'd use "_per_thread" OR "_once" but not both | 08:50 |
| afazekas | giulivo: Please tell mi if something not ok with the above patch, I am not sure I will have time later to update it today | 08:50 |
| giulivo | it's just a rename of the variable! :) | 08:50 |
| giulivo | please! | 08:50 |
| giulivo | _per_thread | 08:50 |
| afazekas | looks like i find bugs what I did not searched, and I need to create some report .. | 08:51 |
| afazekas | s/class_setup_once_per_thread/class_setup_per_thread/ ok ? | 08:52 |
| giulivo | np I've +1 | 08:53 |
| openstackgerrit | Yuuichi Fujioka proposed a change to openstack/tempest: Add a test case for instance boot from volume https://review.openstack.org/41048 | 08:54 |
| *** jhenner has quit IRC | 08:58 | |
| *** dosaboy_ has joined #openstack-qa | 09:01 | |
| *** dosaboy has quit IRC | 09:02 | |
| giulivo | afazekas, if you take a look at this (and re-approve, sorry about) https://review.openstack.org/#/c/38943/ it would be great | 09:04 |
| afazekas | giulivo: both the owner and the other +2 is from RH. I would wait for a non RH +2. If it not get approved or -1 today , I will check how trivial is the rebase | 09:09 |
| afazekas | jaypipes, sdague, cyeoh, mtreinish : can you review this https://review.openstack.org/#/c/38943/ | 09:11 |
| openstackgerrit | Roman Prykhodchenko proposed a change to openstack-dev/devstack: Base support of Ironic https://review.openstack.org/41053 | 09:29 |
| mkoderer | afazekas: could you have a look https://github.com/openstack/tempest/blob/master/tempest/services/volume/json/volumes_client.py#L88 | 09:35 |
| mkoderer | shouldn't this function be called "upload_image"? | 09:36 |
| mkoderer | ahh ok .. now I get it.. it uploaded an volume into glance | 09:36 |
| Anju | mkoderer u can try this using a cli | 09:37 |
| mkoderer | ... I just misread it... | 09:38 |
| Anju | hmm | 09:38 |
| Anju | mkoderer : my question is can u run a cli for attach-volume and dettach-volume | 09:39 |
| Anju | https://github.com/openstack/tempest/blob/master/tempest/services/volume/json/volumes_client.py#L88 | 09:39 |
| Anju | these two are available here | 09:40 |
| afazekas | mkoderer: I will have closer look to that test case soon | 09:42 |
| *** psedlak has joined #openstack-qa | 09:45 | |
| openstackgerrit | Roman Prykhodchenko proposed a change to openstack-dev/devstack: Basic support of Ironic https://review.openstack.org/41053 | 09:45 |
| mkoderer | Anju: did it | 09:46 |
| mkoderer | nova volume-attach 3b1a2175-f717-48bc-afaf-d3c745b505e3 7d86aa2c-d930-4f56-8489-2605e80ff2d4 /dev/bla | 09:46 |
| mkoderer | works | 09:46 |
| Anju | using cinder? | 09:46 |
| mkoderer | no, since volume-attach is not a cinder command ;) | 09:47 |
| Anju | ok checking thanks | 09:47 |
| mkoderer | Anju: http://docs.openstack.org/user-guide/content/boot_from_volume.html | 09:47 |
| Anju | mkoderer: thanks : for nova i was able to run this one :) it good i will take response using this one :) | 09:48 |
| Anju | thanks | 09:48 |
| *** LiangZheng has quit IRC | 10:28 | |
| mkoderer | mtreinish: ping | 10:43 |
| giulivo | afazekas, agreed on the scenario tests ... on the first patch jaypipes reviewed too, let's wait for some more input | 11:08 |
| mkoderer | afazekas: regarding https://review.openstack.org/#/c/38980/ | 11:11 |
| mkoderer | I still have my opinion that this shouldn't be a test action - it should be supported by the driver itself | 11:12 |
| openstackgerrit | Nejc Saje proposed a change to openstack/tempest: Adds client and CRUD tests for Ceilometer alarms https://review.openstack.org/39237 | 11:16 |
| afazekas | mkoderer: AFAIK there are more test families, so it can be plugin | 11:19 |
| afazekas | we might have an arbitrary program (like shell script) plugin as well | 11:19 |
| mkoderer | afazekas: on other thing... which I already forgot - why do we need to run setUp inside of run()? | 11:21 |
| afazekas | mkoderer: why is it problem if you need to specify a handler class instead of selecting the type in another type of variable ? | 11:21 |
| afazekas | mkoderer: If you would like to replicate the setUpclass and tearDown class steps in every run | 11:22 |
| afazekas | for example on server actins you might just wwant to do the actions without a new server creation | 11:23 |
| mkoderer | afazekas: one reason is the logging | 11:23 |
| mkoderer | if you get an exection it will be logged for all classes as "unittest" | 11:23 |
| mkoderer | that's a bit odd.. but maybe a minor issue | 11:24 |
| afazekas | mkoderer: the unittest driver can be improved in the future | 11:24 |
| afazekas | I did not wanted to over complicate it at the first step | 11:25 |
| mkoderer | afazekas: ok got it... so mainly the unit test are creating their resources on setUp... if you fork a new process the will use the same | 11:25 |
| mkoderer | ok I see | 11:26 |
| mkoderer | afazekas: ok that's right .. let's have this simple one at the beginning | 11:27 |
| afazekas | I can't recall which class loading creates a differnt class, and which not, but the class can be loaded to be differnt class even if it's name and package is the same | 11:27 |
| afazekas | the 'address' matters | 11:27 |
| mkoderer | afazekas: right you instanciate always a new object | 11:29 |
| *** woodspa has joined #openstack-qa | 11:29 | |
| afazekas | mkoderer: they are in different process and the class loaded after fork (actually clone used without CLONE_VM) | 11:32 |
| mkoderer | afazekas: in case of class_setup_once it will fork afterwards | 11:34 |
| mkoderer | so setUp will be called on the parent and then forked | 11:35 |
| afazekas | mkoderer: thx | 11:38 |
| mkoderer | afazekas: maybe we need to run setUp for every forked process.. otherwise all the processes will run on the same resources | 11:40 |
| afazekas | mkoderer: It should be replaced to inspection based import | 11:42 |
| *** weshay has joined #openstack-qa | 11:44 | |
| mkoderer | afazekas: would this help? At the end it's just important that every process runs it's initialization and generates the resources that is needed for the test | 11:44 |
| *** dkranz has joined #openstack-qa | 12:03 | |
| afazekas | After a reload(sys.modules[klass_module)) you will have a different class | 12:03 |
| *** SergeyLukjanov has quit IRC | 12:08 | |
| * afazekas reevaluate | 12:14 | |
| afazekas | mkoderer: what if we just move the setUpclass call in the run and changing a variable when it ever runed in the process .. ? | 12:17 |
| *** dosaboy_ is now known as dosaboy | 12:18 | |
| * afazekas eat something, after that I add simple variable based thing, what just modified after fork | 12:21 | |
| openstackgerrit | Fabien Boucher proposed a change to openstack/tempest: Add tests for swift container listing filters https://review.openstack.org/40416 | 12:39 |
| *** thomasbiege has quit IRC | 12:41 | |
| *** adalbas has joined #openstack-qa | 12:45 | |
| mkoderer | afazekas: yes I think we need to move it | 12:53 |
| mkoderer | afazekas: how should this variable be exposed? cli or json? | 12:54 |
| afazekas | there will 3 option | 12:54 |
| afazekas | per application, per action, per process | 12:54 |
| mkoderer | ok - upload it when your done.. want to see the code ;) | 12:55 |
| afazekas | The change is done, but I would like to try at least with one option, is it able to start:) | 12:55 |
| mkoderer | ;) right | 12:56 |
| *** jasondotstar has joined #openstack-qa | 13:10 | |
| openstackgerrit | afazekas proposed a change to openstack/tempest: Unit tests as stress tests https://review.openstack.org/38980 | 13:10 |
| *** russellb is now known as rustlebee | 13:14 | |
| *** mriedem has joined #openstack-qa | 13:14 | |
| afazekas | dkranz: https://review.openstack.org/#/c/40897/1/tox.ini | 13:18 |
| afazekas | Do we need an orchestration directory in the scenario ? | 13:19 |
| *** thomasbiege has joined #openstack-qa | 13:19 | |
| afazekas | the others are not structured . | 13:20 |
| dkranz | afazekas: This is for the slow heat tests | 13:20 |
| dkranz | afazekas: There may be slow tests that are not heat tests and we don't want to include them | 13:20 |
| afazekas | so this job wil be osed just only by heat | 13:20 |
| afazekas | and heat will get new folder in the scenario ? | 13:20 |
| dkranz | afazekas: Yes, there will be a separate gate job. | 13:21 |
| dkranz | afazekas: Yes, sbaker is moving the slow tests to scenario | 13:21 |
| afazekas | but the regexp looks like wants a new folder | 13:21 |
| afazekas | \.orchestration does not seams to required | 13:22 |
| dkranz | afazekas: Yes, you are right. | 13:22 |
| dkranz | afazekas: I forgot that scenario does not have these subdirs. | 13:22 |
| dkranz | afazekas: So the regexp part for api needs to be different than scenario | 13:23 |
| dkranz | afazekas: This is ugly. Perhaps there should be orchestration subdir of scenario anyway | 13:23 |
| afazekas | Do we want to limit that job to heat only ? | 13:24 |
| dkranz | afazekas: For now. But some ceilometer might be added. | 13:24 |
| afazekas | if not probably just the flag matching is enough | 13:25 |
| dkranz | afazekas: What do you propose specifically? | 13:25 |
| afazekas | IMHO easier to not limit | 13:25 |
| dkranz | afazekas: I have to step out for about 20min | 13:26 |
| afazekas | IMHO most of the scenario tests touching many thing anyway | 13:26 |
| afazekas | ok | 13:26 |
| openstackgerrit | Fabien Boucher proposed a change to openstack/tempest: Add tests for swift container listing filters https://review.openstack.org/40416 | 13:28 |
| mkoderer | dkranz: should I set my set my review to 0 or is it ok to let it like it is? | 13:33 |
| mkoderer | dkranz: ahh and in requirements.txt nose is also left... | 13:33 |
| mkoderer | dkranz: it's about https://review.openstack.org/#/c/40723/ | 13:34 |
| openstackgerrit | Roman Prykhodchenko proposed a change to openstack-dev/devstack: Basic support of Ironic https://review.openstack.org/41053 | 13:35 |
| openstackgerrit | Mate Lakat proposed a change to openstack-dev/devstack: fix permission issues around dist-packages https://review.openstack.org/41091 | 13:50 |
| dkranz | mkoderer: I am going to mark this WIP until the tests we are looking for are finalized. | 13:50 |
| dkranz | mkoderer, afazekas : Then we will know what the regexp should be. Or maybe we will just list the tests we want explicitly. | 13:51 |
| mkoderer | dkranz: are you talking about https://review.openstack.org/#/c/40897 or https://review.openstack.org/#/c/40723/? | 13:52 |
| dkranz | mkoderer: Sorry, wrong one. Looking at the one you meant how. | 13:53 |
| dkranz | mkoderer: now | 13:53 |
| *** burt has joined #openstack-qa | 13:53 | |
| afazekas | dkranz: just exclude the .orchestration and I will do not have any other question | 13:54 |
| dkranz | mkoderer: I think you could change it to 0 | 13:54 |
| afazekas | This should be merged soon, for the slow PoC jenkins job | 13:54 |
| mkoderer | dkranz: ok | 13:54 |
| dkranz | mkoderer: This patch was "use testr for serial", not "purge nose from the code base" | 13:54 |
| dkranz | mkoderer: If that is what we really want it can be another patch. | 13:54 |
| mkoderer | dkranz: you're right - thx for the feedback | 13:55 |
| dkranz | mkoderer: NP. Note that I will be out on vacation next week. | 13:55 |
| dkranz | afazekas: slow PoC? | 13:56 |
| afazekas | dkranz: For the POC slow job this is OK, | 13:56 |
| *** anteaya has joined #openstack-qa | 13:56 | |
| afazekas | dkranz: sbaker added two slow tag | 13:56 |
| dkranz | afazekas: Yeah, but I don't see why we need to do this now. We should just wait until the heat tests are ready. | 13:57 |
| dkranz | afazekas: I thought I would get ahead of it but now does not seem worth it especially since I will be out next week. | 13:57 |
| mkoderer | dkranz: happy holidays then... would need some day too ;) | 13:58 |
| afazekas | dkranz: IMHO when the tox entry is there , anybody else can try to setup a job for this | 13:58 |
| afazekas | now those tests are excluded :( | 13:58 |
| dkranz | afazekas: There is more work to do in devstack gate for the heat stuff. Some new env variables to set. | 13:59 |
| dkranz | afazekas: https://review.openstack.org/#/c/40341/ | 13:59 |
| dkranz | afazekas: If you want to take this over while I'm gone that is fine. | 14:00 |
| afazekas | dkranz: I want to do too many thing... I think the tempest part be there is the most critical, the other projects has enough reviewers | 14:01 |
| dkranz | afazekas: The actual real tests are still being blocked by some disk image stuff. | 14:02 |
| dkranz | So this can wait another week I think. | 14:02 |
| *** krtaylor has joined #openstack-qa | 14:06 | |
| * afazekas sorry guys I got life IRQ, I had to leave now.. | 14:08 | |
| mkoderer | have a nice weekend | 14:09 |
| *** afazekas has quit IRC | 14:11 | |
| dkranz | jaypipes: Hi Jay. You gave this a +2 before. Can you +A it please? | 14:14 |
| mtreinish | mkoderer: pong | 14:16 |
| mkoderer | mtreinish: hi | 14:17 |
| mkoderer | mtreinish: I have some questions about testr | 14:17 |
| mkoderer | mtreinish: you are the expert right? | 14:17 |
| mtreinish | mkoderer: sure..... | 14:18 |
| mkoderer | I am trying to add a decorator for the stress test | 14:18 |
| mkoderer | and now I need the test list filtered by this decorator | 14:18 |
| mkoderer | is there a way to do this with a testr function? | 14:19 |
| mkoderer | or am I at the wrong lib? | 14:19 |
| mtreinish | mkoderer: a stress test, test attr decorator | 14:19 |
| mtreinish | kind of like gate, slow, smoke? | 14:19 |
| mtreinish | or something else | 14:19 |
| mkoderer | I need some special parameters | 14:19 |
| mkoderer | so possible a new one | 14:19 |
| mkoderer | like @stresstest(max_threads=5) | 14:20 |
| mkoderer | or do you think a new attr would be better? | 14:20 |
| mkoderer | I just want to reuse the function to select function with a certain decorator | 14:21 |
| mtreinish | mkoderer: well you could just make that stresstest decorator set the attr and do whatever else you need | 14:21 |
| mtreinish | mkoderer: to be able to select the tests you want to run from testr you need to be able to use a regex to filter it from the test id | 14:21 |
| jaypipes | dkranz: which one? | 14:21 |
| *** lexinator has joined #openstack-qa | 14:22 | |
| mtreinish | mkoderer: setting the attr makes that easy because it appends the tag name to the test id | 14:22 |
| dkranz | jaypipes: Sorry :) https://review.openstack.org/#/c/38943/ | 14:23 |
| mkoderer | mtreinish: mhh I have to think about it... my plan is that run_stress uses testr as a library | 14:23 |
| mkoderer | mtreinish: so it just uses to find test cases with the certain decorator | 14:23 |
| jaypipes | dkranz: ah yes, done. | 14:24 |
| mkoderer | mtreinish: do you understand what I mean? | 14:25 |
| dkranz | jaypipes: THx | 14:26 |
| mtreinish | mkoderer: I'm not sure, you want run_stress to call testr to select the tests that have a certain decorator? Does it run them or just dump them somewhere? | 14:26 |
| mkoderer | mtreinish: I just want a list of functions to be called | 14:27 |
| mkoderer | mtreinish: run_stress will use them and create the needed processes | 14:27 |
| mkoderer | mtreinish: maybe testr could be enhanced to be used for it | 14:27 |
| *** thomasbiege has quit IRC | 14:28 | |
| mtreinish | mkoderer: well testr just uses subunit discover to find the tests. Take a look at .testr.conf to see what it runs. | 14:29 |
| mkoderer | mtreinish: aha right.. this one: subunit.run discover -t ./ ./tempest $LISTOPT $IDOPTION | 14:29 |
| mkoderer | but I don't see how the subunit parameter is used there | 14:30 |
| mtreinish | mkoderer: if you run testr run --parallel it'll print out the commands it's using to find the tests and then run the tests. | 14:31 |
| mkoderer | ok let me try | 14:31 |
| *** adalbas has quit IRC | 14:33 | |
| openstackgerrit | Joe Hakim Rahme proposed a change to openstack/tempest: Add more tests for Swift Account Quota https://review.openstack.org/41099 | 14:34 |
| mkoderer | mtreinish: so if I start "testr run --parallel --subunit smoke" | 14:34 |
| mkoderer | I get this output: ${PYTHON:-python} -m subunit.run discover -t ./ ./tempest --list | 14:34 |
| mkoderer | I don't see where my parameter "smoke" is used in subunit | 14:35 |
| *** afazekas has joined #openstack-qa | 14:36 | |
| *** mlavalle has joined #openstack-qa | 14:37 | |
| mlavalle | mtreinish: ping | 14:38 |
| kashyap | afazekas, Just drop me an email if you need me. I'm at Fedora conference. My responses will be sporadic | 14:42 |
| afazekas | kashyap: I just wanted to ask is the latest Fedora image on the main download page, or there is another source ? | 14:43 |
| afazekas | new the ks files just using the stable repo, I will try it again (I had repo sync issue..) | 14:44 |
| mtreinish | mkoderer: so, python -m subunit.run discover -t ./ ./tempest --list will generate the list of tests | 14:44 |
| mtreinish | and then smoke will be a regex filter on the list | 14:44 |
| mtreinish | I can dig up the code for you | 14:44 |
| mtreinish | mlavalle: pong | 14:44 |
| mkoderer | mtreinish: ahhh ok thats easy... | 14:45 |
| mkoderer | mtreinish: thanks for that | 14:45 |
| afazekas | kashyap: I want to switch my image, to something based on the official image | 14:45 |
| mlavalle | mtreinish: thanks for the help with the review. So, what do you suggest now going forward? Wait for your https://review.openstack.org/#/c/40969/ to merge and then continue with mine? | 14:46 |
| mtreinish | mlavalle: I think you're looking for mriedem not me | 14:47 |
| mlavalle | sorry | 14:47 |
| *** adalbas has joined #openstack-qa | 14:48 | |
| mriedem | mlavalle: i'm assuming you're referring to the relation on this? https://review.openstack.org/#/c/39006/ | 14:48 |
| mlavalle | mriedem: yes! | 14:49 |
| mriedem | mlavalle: let's go to openstack-nova | 14:49 |
| mlavalle | see you there | 14:49 |
| *** afazekas has quit IRC | 14:56 | |
| openstackgerrit | gordon chung proposed a change to openstack-dev/devstack: Cinder configuration is not set up for Ceilometer https://review.openstack.org/41108 | 14:56 |
| mtreinish | sdague, dkranz, afazekas: can I get eyes on https://review.openstack.org/#/c/40975/ (dkranz thanks) and https://review.openstack.org/#/c/39099/ | 14:57 |
| mtreinish | they should fix the last big testr race | 14:57 |
| mtreinish | jaypipes: ^^^ you too :) | 14:57 |
| *** burt has quit IRC | 15:02 | |
| openstackgerrit | A change was merged to openstack/tempest: Switch to testr serial instead of nose https://review.openstack.org/40723 | 15:10 |
| openstackgerrit | Matthew Treinish proposed a change to openstack-dev/devstack: Enable debug logging on tempest https://review.openstack.org/41113 | 15:16 |
| jaypipes | mtreinish: \o/ | 15:25 |
| jaypipes | mtreinish: nice work :) | 15:25 |
| openstackgerrit | A change was merged to openstack/tempest: scenario test involving glance, cinder and nova functionalities https://review.openstack.org/38943 | 15:32 |
| *** lexinator has quit IRC | 15:33 | |
| *** meteorfox has joined #openstack-qa | 15:39 | |
| mtreinish | jaypipes: well I doubt it's the last of the races, but it's the last one that is failing most of the runs.... | 15:44 |
| *** lexinator has joined #openstack-qa | 15:56 | |
| *** comstud is now known as bearhands | 16:07 | |
| openstackgerrit | A change was merged to openstack/tempest: Rework class inheritance for scenario tests https://review.openstack.org/40975 | 16:20 |
| *** mriedem has quit IRC | 16:25 | |
| *** krtaylor has quit IRC | 16:29 | |
| *** mriedem has joined #openstack-qa | 16:38 | |
| *** giulivo has quit IRC | 16:42 | |
| *** burt has joined #openstack-qa | 16:43 | |
| *** jog0 is now known as flashgordon | 16:48 | |
| *** lexinator has quit IRC | 16:56 | |
| *** lexinator has joined #openstack-qa | 17:00 | |
| *** jhenner has joined #openstack-qa | 17:07 | |
| *** krtaylor has joined #openstack-qa | 17:11 | |
| *** fbo is now known as fbo_away | 17:26 | |
| *** jhenner has quit IRC | 17:32 | |
| *** hemnafk is now known as hemna | 17:42 | |
| openstackgerrit | Sean Dague proposed a change to openstack-dev/devstack: cleanup potentially installed older oslo.config https://review.openstack.org/41153 | 17:45 |
| sdague | dtroyer: you about? | 17:46 |
| dtroyer | about dead maybe | 17:59 |
| dtroyer | tgif and all that | 17:59 |
| openstackgerrit | Sean Dague proposed a change to openstack-dev/devstack: cleanup potentially installed older oslo.config https://review.openstack.org/41153 | 18:05 |
| sdague | heh | 18:06 |
| sdague | dtroyer: so do you have a better solution than this - https://review.openstack.org/#/c/41091/ ? | 18:19 |
| sdague | because we need some kind of fix | 18:20 |
| dtroyer | not yet, but that's a terrible hack. is this only happening on previously-used devstack machines? | 18:21 |
| sdague | dtroyer: totally clean basic installs hit it | 18:21 |
| sdague | I just started with precise cloud image | 18:22 |
| openstackgerrit | A change was merged to openstack-dev/devstack: Remove unused keystone params from neutron agents' config files https://review.openstack.org/39763 | 18:27 |
| dtroyer | am doing that now too. and I think I've used up all of my old installs, may need to get the de Lorean out to find one to try again | 18:27 |
| openstackgerrit | Matthew Treinish proposed a change to openstack/tempest: Add tenant isolation for scenario tests https://review.openstack.org/39099 | 18:48 |
| openstackgerrit | SlickNik proposed a change to openstack-dev/devstack: Added Trove (Database as a Service). https://review.openstack.org/38169 | 18:55 |
| dtroyer | sdague: so, uh, get this. the permissions issue with prettytable is a *drumrole* permissions issue *IN* prettytable's dist zipfile. | 19:01 |
| dtroyer | it has broken us again | 19:01 |
| sdague | hah! | 19:01 |
| dtroyer | so we can either ban 0.7.2 or try to install it separately and fix it | 19:02 |
| dtroyer | wait…it's been that way since april. wtf? | 19:03 |
| mtreinish | sdague: So, I'm going through the requirements list right now. The first thing I hit on a clean box is that virtualenv and tox aren't installed using the lists. | 19:07 |
| mtreinish | is that something we should include on the requirements list or just assume that people are smart enough to install it | 19:08 |
| sdague | mtreinish: requirements, or devstack package list | 19:11 |
| sdague | it used to work on a clean box | 19:11 |
| openstackgerrit | A change was merged to openstack/tempest: Skip secgroup invalid name/desc tests until bug 1161411 is fixed https://review.openstack.org/40666 | 19:12 |
| mtreinish | sdague: I just spun up a bare ubuntu cloud image guest, installed pip, then ran pip install -r requirements.txt (also test-requires too) | 19:12 |
| sdague | mtreinish: ok | 19:14 |
| sdague | works now? | 19:14 |
| mtreinish | actually I'm not sure it was missing virtualenv looks like it was a permissions issue on httplib2 | 19:14 |
| mtreinish | yeah install_venv_common takes care of installing the virtualenv requirement when you use run tests | 19:18 |
| mtreinish | but easy_install failed on a permissions issue on httplib2 | 19:18 |
| mtreinish | as for tox that still needs to be installed from what I can tell | 19:18 |
| sdague | mtreinish: can we get a good enough tox out of the distro? | 19:21 |
| mtreinish | yeah probably, I'm pretty sure devstack already takes care of that | 19:22 |
| mtreinish | I was testing the requirements list outside of devstack to make sure it was complete | 19:22 |
| mtreinish | sdague: also, looks like I hit the permissions issue with devstack too: http://paste.openstack.org/show/43732/ | 19:24 |
| flashgordon | random question, is it possible to get req-ids in the tempest console output for failed tests? | 19:42 |
| flashgordon | for example http://logs.openstack.org/64/40764/5/check/gate-tempest-devstack-vm-full/ea1d1d1/console.html.gz | 19:42 |
| flashgordon | I have a failed test but have to manaully search the n-api logs to find the matching request | 19:42 |
| mtreinish | flashgordon: does nova include the request id in the response? | 19:43 |
| flashgordon | mtreinish: yes | 19:43 |
| flashgordon | under x-compute-request-id | 19:44 |
| mtreinish | flashgordon: then it should be possible to log it. But, I'm not sure about the fails at the bottom of the run though. That's just a traceback printed by the test runner when a test fails or errors. | 19:45 |
| flashgordon | mtreinish: isn't there a way to log stdout too? | 19:46 |
| flashgordon | so print out every req-id and testr only prints stdout for failed | 19:46 |
| mtreinish | flashgordon: we have testr set to capture stdout and stderror | 19:47 |
| mtreinish | yeah I guess that would work | 19:47 |
| * flashgordon files a tempest bug | 19:47 | |
| flashgordon | mtreinish: https://bugs.launchpad.net/tempest/+bug/1210639 | 19:49 |
| mtreinish | flashgordon: ok cool | 19:55 |
| flashgordon | mtreinish: thanks, my standard debug procedure is find the req-id of failed test and grep for it | 19:56 |
| mtreinish | flashgordon: yeah I do the same thing too. (normally by using the class name to find the isolated creds for the class as a start) | 19:57 |
| mtreinish | it's a good idea I'll probably push something out for it soon | 19:57 |
| mtreinish | it'll help with the parallel debug too... | 19:57 |
| flashgordon | \o/ | 20:00 |
| mtreinish | sdague, dkranz, jaypipes: can you take a look at the new revision for https://review.openstack.org/#/c/39099/ | 20:06 |
| *** psedlak has quit IRC | 20:06 | |
| dkranz | mtreinish: yes | 20:06 |
| mtreinish | dkranz: great thanks | 20:07 |
| *** weshay has quit IRC | 20:13 | |
| openstackgerrit | Monty Taylor proposed a change to openstack-dev/devstack: Protect local dev changes from the update logic https://review.openstack.org/40418 | 20:16 |
| openstackgerrit | Monty Taylor proposed a change to openstack-dev/devstack: Protect local dev changes from the update logic https://review.openstack.org/40418 | 20:17 |
| sdague | mordred: so if you land my other update.py patch, we have a different solution there | 20:18 |
| openstackgerrit | Monty Taylor proposed a change to openstack-infra/devstack-gate: Enable flag to update requirements on run https://review.openstack.org/41192 | 20:19 |
| dkranz | mtreinish: Doesn't this still have the problem in https://review.openstack.org/#/c/39099/13/tempest/scenario/manager.py,unified that Sean pointed out? | 20:19 |
| dkranz | mtreinish: If is_deletion_complete fails the tearDownCLass will not be called? | 20:20 |
| sdague | mtreinish: did you notice that neutron full passed on your run | 20:20 |
| mtreinish | sdague: I fixed neutron? | 20:21 |
| mordred | sdague: oh yeah? | 20:21 |
| mordred | sdague: which one? | 20:21 |
| mtreinish | dkranz: on line 210? | 20:21 |
| sdague | mordred: https://review.openstack.org/#/c/39099/ neutron full passed | 20:21 |
| dkranz | mtreinish: Sorry, yes. | 20:21 |
| sdague | on the update.py thing | 20:21 |
| sdague | mordred: https://review.openstack.org/#/c/40995/ | 20:22 |
| dkranz | mtreinish: If true, not so easy to fix. | 20:22 |
| sdague | so with that you can ./update.py -o global $dir | 20:22 |
| sdague | then get new requirements.txt.global files | 20:22 |
| sdague | which we can use in devstack | 20:22 |
| mtreinish | dkranz: that is in teardownclass | 20:22 |
| dkranz | mtreinish: Yes. | 20:23 |
| mordred | sdague: oh - and then setting REQUIREMENTS_FILES env var? | 20:23 |
| mordred | sdague: col. I like it | 20:23 |
| mtreinish | dkranz: oh I see what you're saying. Yeah this is a problem because we can't do class level addCleanup there isn't really a way to get around it | 20:23 |
| sdague | mordred: I just assumed we'd pip with -r | 20:24 |
| dkranz | mtreinish: OK, I'm cool with that. | 20:24 |
| sdague | on the new files, though if you have another idea, that's cool | 20:24 |
| mordred | sdague: we can do that too | 20:24 |
| dkranz | mtreinish: sdague wanted us to start being anal about this so I was just pointing it out | 20:24 |
| sdague | anyway, take a look at the approach | 20:24 |
| mordred | sdague: except, no, we have to do the env var | 20:24 |
| sdague | mordred: ok, env is fine | 20:24 |
| mordred | sdague: because otherwise the pip install -e is going to pick up the wrong requirements :) | 20:24 |
| mtreinish | dkranz: yeah I looked at doing that once when I first redid the isolated creds. | 20:24 |
| dkranz | mtreinish: There is already plenty of code with similar issues. | 20:24 |
| mtreinish | dkranz: yeah but we don't want to merge more bad code if we can help it :) | 20:25 |
| sdague | mordred: I was going to remove -e :) | 20:25 |
| sdague | and go back to -r on the 2 files | 20:25 |
| sdague | however, env var is fine too | 20:25 |
| mordred | sdague: oh - well, that's fine | 20:25 |
| sdague | either way, it gives us other files | 20:25 |
| sdague | and doesn't polute git | 20:25 |
| mordred | but I want to keep pip -e instead of setup.py develop ... but yes - we can battle over the little bits :) | 20:26 |
| dkranz | mtreinish: In the future we need to be more clear about the cleanup requirements and make it easy for people to not run afoul of them. | 20:26 |
| sdague | yep, that's fine :) | 20:26 |
| dkranz | mtreinish: decorators, more gc-like allocation, etc. | 20:26 |
| mtreinish | sdague: so neutron didn't pass jenkins reporting is broken... | 20:26 |
| mtreinish | it might be related to the move to testr | 20:27 |
| mordred | what? | 20:27 |
| mordred | lifeless: ^^ | 20:27 |
| dkranz | mtreinish: So are you saying I should -1 it to not merge more "bad" code? | 20:27 |
| *** meteorfox has quit IRC | 20:27 | |
| mtreinish | sdague: see: http://logs.openstack.org/99/39099/13/check/gate-tempest-devstack-vm-neutron-full/34856db/testr_results.html.gz | 20:27 |
| sdague | mtreinish: hmmm... it reported success | 20:28 |
| mtreinish | dkranz: hahaha | 20:28 |
| mriedem | mtreinish: test_security_group_create_with_invalid_group_description and test_security_group_create_with_invalid_group_name should be skipped now | 20:28 |
| mriedem | mtreinish: https://review.openstack.org/#/c/40666/ | 20:29 |
| sdague | mtreinish: umm... testr isn't failing right | 20:29 |
| sdague | this is actually very bad | 20:29 |
| dtroyer | mordred, sdague: I'm heading to airport soon but I like what I saw in the scrollback here, especially not messing with the repos. | 20:29 |
| sdague | ok | 20:29 |
| mordred | dtroyer: cool. I just aprv'd sdague's change to requirements | 20:29 |
| dtroyer | BTW, I'm building another script to handle the ugly hacks we need, like the prettytable perms and all of the rhel package nonsense... | 20:29 |
| mordred | once that lands, we can put together a devstack patch to make use of it | 20:29 |
| mtreinish | mriedem: that probably just wasn't on the tree when I pushed out the change | 20:30 |
| sdague | mordred: ok, what did we do wrong on testr? | 20:30 |
| mordred | dtroyer: good. also, jeez | 20:30 |
| dtroyer | should have a proposal soon for just the prettytable stuff | 20:30 |
| sdague | dtroyer: safe travels | 20:30 |
| sdague | dtroyer: cool, I'll check later on it | 20:30 |
| sdague | or tomorrow morning over coffee | 20:30 |
| mtreinish | sdague: it's probably printing testr slowest I bet that's returning a success code | 20:30 |
| mtreinish | in pretty_tox_serial.sh | 20:31 |
| mordred | mtreinish: yup. I 'betcha that's right | 20:31 |
| lifeless | mtreinish: wassup? | 20:31 |
| mordred | lifeless: thought it might be a testr issue - but now we're looking in to be pebkac | 20:32 |
| lifeless | cool. Well, not cool, but ack. | 20:32 |
| lifeless | is it something testr could help you avoid? | 20:32 |
| mtreinish | lifeless: not much we just moved to testr serial in the gate and one of my sloppy wrapper scripts reports every jenkins run as a success. :) | 20:32 |
| lifeless | ah | 20:32 |
| lifeless | whoops :) | 20:32 |
| sdague | mtreinish: so we just dump testr slowest? | 20:33 |
| mordred | sdague: grab $? after the testr | 20:33 |
| mordred | then run slowest | 20:33 |
| mordred | then exit $RET | 20:33 |
| mtreinish | sdague: ^^ yeah that I'm about to push it out | 20:33 |
| sdague | so why aren't we using the integrated --slowest like in nova? | 20:34 |
| mtreinish | that's part of pbr | 20:34 |
| sdague | ok | 20:34 |
| mordred | yeah. that. and also I don't know that there is a way to disable parallel - although I think we've talked about adding a --no-parallel | 20:34 |
| openstackgerrit | Matthew Treinish proposed a change to openstack/tempest: Fix return code for pretty_tox_serial.sh https://review.openstack.org/41195 | 20:35 |
| lifeless | there is a patch up for it | 20:35 |
| mtreinish | mordred: yeah: https://review.openstack.org/#/c/39811/ | 20:35 |
| mtreinish | although i probably should respin it to appease fungi and jgriffith | 20:36 |
| fungi | mtreinish: there is no pleasing me | 20:36 |
| fungi | i *did* +2 that patch | 20:36 |
| fungi | all snark about double negatives aside | 20:37 |
| mtreinish | fungi: all I did was find replace with 'serial' from the previous revision | 20:38 |
| sdague | man, the gate queue is as small as I've ever seen it on Friday | 20:39 |
| mtreinish | sdague: it's this one: https://review.openstack.org/41195 probably want to rush that through | 20:39 |
| sdague | yeh, i just did | 20:40 |
| fungi | mtreinish: well, "serial" was a positive term, whereas "no_parallel" is negative. makes it harder for readers to reason about. that was my only qualm | 20:40 |
| sdague | mtreinish: also, it looks like the full runs are actually passing for reals | 20:40 |
| sdague | so we didn't let anything terribad slip in | 20:40 |
| mtreinish | fungi: no I get that, I was just saying reason and semantics didn't really factor into it | 20:41 |
| mtreinish | just sed | 20:41 |
| fungi | heh, got it ;) | 20:41 |
| lifeless | fungi: --no-foo is a convention for things that have a --foo option. | 20:43 |
| dkranz | mtreinish: OK, I +2 the isolated creds | 20:43 |
| lifeless | fungi: testr has --parallel; this is a testr wrapper - thats why I suggested --no-parallel : it's more obvious to folk that have learnt the testr interface (and vice versa, folk that know the wrapper and go to testr have less room for confusion) | 20:43 |
| *** meteorfox has joined #openstack-qa | 20:44 | |
| dkranz | mtreinish: That's a pretty significant change. | 20:44 |
| dkranz | mtreinish: I hope I didn't miss anything. | 20:44 |
| fungi | lifeless: oh i don't disagree there, just the code that went into suddenly had "if not not-thing" | 20:44 |
| fungi | (well, not_thing) | 20:45 |
| mtreinish | dkranz: cool, thanks | 20:47 |
| flashgordon | with tempest how do specify a subset of commands to run? | 20:48 |
| flashgordon | in nova its tox -epy27 test name | 20:48 |
| flashgordon | when I do 'tox -efull test_ec2_instance_run' it runs everything | 20:48 |
| mtreinish | dkranz: I thought about splitting things up but the only logical way to do it was add the full isolated cred blob and then a separate patch for the scenario addition. | 20:49 |
| mtreinish | which wouldn't end up being that different than it is now | 20:49 |
| dkranz | mtreinish: I was not complaining. | 20:49 |
| dkranz | mtreinish: Just saying that I was not as certain everything is ok as I am usually with reviews. | 20:50 |
| mtreinish | flashgordon: the only way to do it now is with manually calling testr with the regex for what you want | 20:50 |
| mtreinish | (or use run_tests.sh with the regex) | 20:50 |
| flashgordon | mtreinish: why no posargs in tox? | 20:50 |
| dkranz | mtreinish: Especially in such a dynamically typed language :) | 20:51 |
| mtreinish | flashgordon: there are but since testr ORs the regex and we already specify one for the full gate job it won't do anything | 20:51 |
| flashgordon | mtreinish: :( | 20:52 |
| mtreinish | flashgordon: actually one sec I'll push out a patch to add posargs to -eall which doesn't specify any regex | 20:53 |
| mtreinish | and then you can use it there | 20:53 |
| mtreinish | but, that will run in parallel. | 20:53 |
| flashgordon | mtreinish: I was doing -efull | 20:54 |
| sdague | mtreinish: I think we'll probably need follow on clean ups, but I'm ok with landing the isolation now | 20:54 |
| mtreinish | sdague: ok cool | 20:54 |
| mtreinish | that's the last of the big race conditions then. Now we can see if there are any hiding behind it | 20:55 |
| *** woodspa has quit IRC | 20:55 | |
| mtreinish | flashgordon: -efull runs it serially | 20:55 |
| lifeless | fungi: ah | 20:55 |
| mtreinish | flashgordon: normally you wouldn't want to use -eall because it runs the broken whitebox tests. But if you add posargs to it'll let you run just a single test using tox. | 20:58 |
| flashgordon | mtreinish: cool | 20:58 |
| *** burt has quit IRC | 21:00 | |
| openstackgerrit | Matthew Treinish proposed a change to openstack/tempest: Fix posargs usage for tox jobs https://review.openstack.org/41198 | 21:03 |
| mtreinish | flashgordon: ^^^ | 21:03 |
| flashgordon | mtreinish: thanks | 21:06 |
| marun | sdague: any pointers on submitting a backport to tempest's stable/grizzly branch? | 21:25 |
| marun | or anyone really | 21:25 |
| marun | dkranz ^ | 21:25 |
| marun | ah, nmind | 21:32 |
| *** mriedem has quit IRC | 21:41 | |
| openstackgerrit | Sean Dague proposed a change to openstack/tempest: Fix return code for pretty_tox_serial.sh https://review.openstack.org/41195 | 21:44 |
| sdague | marun: you need to land it in master, than backport, per same rules of other projects | 21:44 |
| sdague | marun: also, there are a lot of critical neutron bugs in the gate right now, if there are any neutron cycles, would be nice to get someone looking at them | 21:45 |
| marun | sdague: it's been in master a long time, I just didn't know about having to provide the target branch name to git review | 21:47 |
| marun | sdague: I can spare some time next week for neutron bugs. Where can I find them? | 21:48 |
| sdague | marun: the neutron launchpad tracker :) | 21:48 |
| sdague | all the ones marked critical are things that are reseting the gate for other people | 21:48 |
| marun | sdague: ok | 21:49 |
| marun | sdague: *sigh* | 21:49 |
| marun | sdague: I don't get it, how are the changes that break things this bad making it past the gate in the first place? | 21:50 |
| sdague | marun: well one of them we had a slip today | 21:50 |
| sdague | which we're trying to fix | 21:50 |
| sdague | on a lot of them, they only fail 3% of the time | 21:51 |
| sdague | so they can thread the gate | 21:51 |
| sdague | actually a neutron test only needs to pass 50% of the time to let it merge | 21:51 |
| sdague | so the race conditions in neutron just have people recheck to get past them | 21:51 |
| marun | #fail | 21:52 |
| marun | sdague: well, I'll do my best. | 21:53 |
| sdague | thanks | 21:53 |
| *** hemna has quit IRC | 21:54 | |
| sdague | honestly, something you could do that might help other folks, is make the logging a little more clear. So like log requests that come in. The neutron logs are kind of hard to read relative to nova, which is one of the reasons people seem to just recheck them vs. look deeper | 21:54 |
| openstackgerrit | Dean Troyer proposed a change to openstack-dev/devstack: More Python package fixes https://review.openstack.org/41209 | 21:57 |
| *** ianw has quit IRC | 22:22 | |
| *** ianw has joined #openstack-qa | 22:23 | |
| *** jaypipes has quit IRC | 22:47 | |
| *** lexinator has quit IRC | 22:50 | |
| openstackgerrit | A change was merged to openstack/tempest: Fix return code for pretty_tox_serial.sh https://review.openstack.org/41195 | 22:51 |
| *** dkranz has quit IRC | 23:05 | |
| *** jasondotstar has quit IRC | 23:13 | |
| *** mlavalle has quit IRC | 23:50 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!