Perl 6 - the future is here, just unevenly distributed

IRC log for #openstack-rally, 2014-08-09

| Channels | #openstack-rally index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:17 EmilienM left #openstack-rally
00:59 rediskin left #openstack-rally
02:29 jjmb joined #openstack-rally
03:12 AlexF_ joined #openstack-rally
03:18 oanufriev joined #openstack-rally
03:56 openstackgerrit Joseph Bajin proposed a change to stackforge/rally: Fixes the check for admin role to be case-insenstive.  https://review.openstack.org/112999
03:57 AlexF_ joined #openstack-rally
03:58 RaginBajin joined #openstack-rally
05:10 AlexF_ joined #openstack-rally
07:00 AlexF_ joined #openstack-rally
07:04 andreykurilin_ joined #openstack-rally
07:28 AlexF_ joined #openstack-rally
07:47 openstackgerrit Andrey Kurilin proposed a change to stackforge/rally: Fix the way of obtaining ceilometerclient  https://review.openstack.org/112883
10:21 rediskin joined #openstack-rally
11:33 rook joined #openstack-rally
11:57 rook joined #openstack-rally
12:06 rediskin left #openstack-rally
12:17 openstackgerrit Boris Pavlovic proposed a change to stackforge/rally: Fix rally-gate.sh exit status  https://review.openstack.org/112887
12:19 tzabal joined #openstack-rally
12:25 AlexF_ joined #openstack-rally
12:28 rook joined #openstack-rally
12:56 openstackgerrit Boris Pavlovic proposed a change to stackforge/rally: Fix rally-gate.sh exit status  https://review.openstack.org/112887
12:58 AlexF_ joined #openstack-rally
13:34 openstackgerrit Tzanetos Balitsaris proposed a change to stackforge/rally: Modify config semantic validation of benchmark engine  https://review.openstack.org/112981
13:37 tzabal gates are so empty during weekends :D
13:38 tzabal on*
13:54 boris-42 tzabal yep
13:54 boris-42 tzabal the best  time to work on rally=)
13:54 tzabal boris-42 haha
13:55 boris-42 tzabal we have some evil stuff
13:55 boris-42 tzabal https://review.openstack.org/#/c/112887/
13:56 tzabal boris-42 it will be merged soon
13:56 boris-42 tzabal yep
14:00 openstackgerrit A change was merged to stackforge/rally: Fix rally-gate.sh exit status  https://review.openstack.org/112887
14:12 openstackgerrit Tzanetos Balitsaris proposed a change to stackforge/rally: Modify config semantic validation of benchmark engine  https://review.openstack.org/112981
14:51 tzabal boris-42 whenever you have time please check ^^, I think it ready.
15:37 AlexF_ joined #openstack-rally
15:44 openstackgerrit Joseph Bajin proposed a change to stackforge/rally: Fixes the check for admin role to be case-insenstive  https://review.openstack.org/112999
15:52 boris-42 tzabal great
15:52 boris-42 tzabal I will
15:59 openstackgerrit Lingxian Kong proposed a change to stackforge/rally: Add cinder snapshot scenario tests  https://review.openstack.org/110360
16:13 AlexF_ joined #openstack-rally
16:25 boris-42 tzabal around?
16:43 tzabal boris-42 yes
16:43 boris-42 tzabal one nit
16:43 boris-42 tzabal as you are refactoring that piece of code
16:43 jaypipes joined #openstack-rally
16:57 boris-42 tzabal could you take a look at https://review.openstack.org/#/c/112981/
16:57 boris-42 tzabal it shouldn't take too much time to fix, because it's just renaming
16:59 tzabal boris-42 yes I agree
16:59 boris-42 tzabal I think even you have a troubles with understanding that stuff=)
16:59 boris-42 tzabal at least after couple of months I have=)
16:59 tzabal boris-42 benchmark_config or just plain config you think is better?
17:00 boris-42 tzabal I think config
17:00 boris-42 tzabal it's enough explicit
17:04 ajayaa joined #openstack-rally
18:01 rook joined #openstack-rally
18:20 openstackgerrit A change was merged to stackforge/rally: Fixes the check for admin role to be case-insenstive  https://review.openstack.org/112999
18:22 boris-42 tzabal just for my information are you going to finish today patch?
18:24 tzabal boris-42 yeap
18:24 tzabal boris-42 i hope so actually, refactoring unit tests at the moment
18:25 boris-42 tzabal nice
18:25 boris-42 tzabal btw there is another refactoring as well of that piece of code
18:25 boris-42 tzabal coolsvap is working on it
18:26 boris-42 tzabal https://review.openstack.org/#/c/111989/
18:27 tzabal boris-42 nice, so we abandon completely the add method
18:27 boris-42 tzabal just remove*
18:27 tzabal boris-42 good
18:36 boris-42 tzabal continious refacotring
18:38 rmk https://github.com/stackforge/rally/blob/master/rally/benchmark/scenarios/nova/servers.py#L147-L151 and https://github.com/stackforge/rally/blob/master/rally/benchmark/scenarios/nova/utils.py#L100-L108 overlap
18:38 rmk And both are bad assumptions
18:38 boris-42 rmk actually I agree that that looks like shi* especially first one
18:39 boris-42 rmk I wrote it a quite long period of time ago
18:39 boris-42 rmk as a quick fix
18:39 boris-42 rmk and still don't have enough time to refactor it in normal way
18:39 rmk I already remove the second one from my tree but I figured it could be an argument to assign nics or not
18:39 rmk I've had zero success landing patches so far but Ill submit one
18:40 boris-42 rmk =(
18:40 boris-42 rmk I really would like if you could fine to finish your patches
18:40 boris-42 rmk I know it's a long way to do stuff
18:40 rmk yeah I am working on them today, I dont have much time to code during the week
18:40 boris-42 rmk but it's the only way to get code that works for everybody
18:41 rmk I already have merge conflicts with my tree heh
18:41 rmk so I want to get this stuff done
18:41 boris-42 rmk that will be great especially I am waiting paths for entrypoints
18:41 boris-42 patch*
18:41 rmk ok
18:42 boris-42 rmk btw if you have some nit's use cases bugs
18:42 boris-42 rmk we have trello board
18:42 boris-42 rmk https://trello.com/b/DoD8aeZy/rally
18:42 boris-42 rmk you can put in pool of tasks any tasks
18:42 boris-42 rmk I will need just to add you
18:42 boris-42 rmk so somebody from community can start working on stuff that you need
18:43 rmk cool sounds good
18:43 boris-42 rmk you should just give me your nick name
18:43 boris-42 rmk so I'll add you to it
18:44 rook joined #openstack-rally
18:46 openstackgerrit Tzanetos Balitsaris proposed a change to stackforge/rally: Modify config semantic validation of benchmark engine  https://review.openstack.org/112981
18:49 penguinRaider joined #openstack-rally
19:01 rmk boris-42: https://gist.github.com/rmk40/2e5c56cddce1c6ed3bbc
19:01 rmk Is something like that reasonable?
19:02 rmk oops typos but general idea
19:03 boris-42 rmk I like general idea
19:03 boris-42 rmk btw does this work with nova network?
19:03 rmk yeah that's what we're using
19:03 rmk the "default" right now does not work with ours
19:03 boris-42 rmk I mean       auto_assign_nic = kwargs.get('assign_nics', False)
19:03 rmk yeah I just fixed that
19:03 rmk It has to be "True" in my case
19:03 boris-42 rmk okay so I like general idea
19:04 boris-42 rmk but I would like to add this argument
19:04 boris-42 rmk boot_server_from_volme and boot_server
19:04 boris-42 as a explicit
19:04 boris-42 with default value False or True
19:04 boris-42 rmk makes sense?
19:04 rmk OK so not a kwarg, no problem
19:05 boris-42 rmk because it is not std argument
19:05 boris-42 rmk that is passed to nova.servers.create
19:05 rmk makes sense
19:08 boris-42 rmk btw did you see SLA stuff?
19:09 boris-42 rmk https://github.com/stackforge/rally/blob/master/rally-scenarios/rally.yaml#L11-L12
19:09 boris-42 rmk it's convertor to binary results, can be used for regression testing
19:21 openstackgerrit Tzanetos Balitsaris proposed a change to stackforge/rally: Modify config semantic validation of benchmark engine  https://review.openstack.org/112981
19:23 boris-42 tzabal I left in patch set 4 comment
19:23 boris-42 tzabal but it still make sense
19:26 tzabal boris-42 for the first comment, because the validators dont use the task parameter in their body (some of them don't use clients either), the unit tests don't pass this parameter.
19:26 boris-42 tzabal you mean pep check?
19:26 boris-42 tzabal or what?
19:26 tzabal boris-42 for the second comment, you are absolutely correct, let me fix that.
19:26 tzabal boris-42 they fail
19:26 boris-42 tzabal why?
19:26 boris-42 tzabal I do not understand ...
19:27 tzabal boris-42 one moment
19:27 rmk boris-42: yeah main problem for me is not everything is caught a failure yet, like anything that times out or throws an exception
19:28 boris-42 rmk ?
19:28 boris-42 rmk actually failures are catch even in case of bug of atomic_actions
19:28 boris-42 rmk when not all are counted, failure is catch
19:28 rmk https://github.com/metacloud/rally/blob/metacloud/rally/benchmark/scenarios/nova/utils.py#L457
19:29 rmk I cannot get that to be caught as an error
19:29 boris-42 rmk you are getting
19:29 boris-42 rmk iteration is failed
19:29 rmk You mean the whole scenario fails
19:29 rmk right?
19:29 boris-42 rmk iteration of whole scenario*
19:29 rmk Ill check now
19:29 boris-42 rmk check it=)
19:30 boris-42 rmk even more if you run "rally task results"
19:30 boris-42 rmk you'll see full exceptions with traces
19:30 rmk so yeah it might just be the atomic actions not being recording but the scenario itself fails
19:30 tzabal boris-42 for example, with def number_validator(clients, task, **config), these unit tests fail http://paste.openstack.org/show/92569/
19:30 boris-42 rmk yep
19:30 rmk thats good then
19:30 boris-42 rmk we will fix this crap with atomic_actions
19:30 rmk hehe ok
19:31 boris-42 rmk so they will be as well catches
19:31 boris-42 catch*
19:31 boris-42 caught*
19:31 boris-42 tzabal ahh
19:31 boris-42 =)
19:31 boris-42 tzabal that can be fixed easily
19:31 boris-42 tzabal I will left comment in your review
19:32 tzabal boris-42 passing clients and task as None?
19:32 boris-42 tzabal that's wired=)
19:32 boris-42 tzabal one sec
19:32 tzabal boris-42 sure
19:34 boris-42 tzabal replied
19:35 boris-42 rmk so to use SLA checks in gates (openstack ci) https://github.com/stackforge/rally/blob/master/tests_ci/rally-gate.sh#L51
19:35 boris-42 rmk we are calling rally task sla_check at the end
19:37 boris-42 rmk and here is the sample of adding new checks https://github.com/stackforge/rally/blob/master/rally/benchmark/sla/base.py#L77
19:45 tzabal boris-42 why we want to pass to the number_validator method the parameters clients and task when it doesn't use them?
19:46 boris-42 tzabal because we should care about end users
19:46 boris-42 tzabal unification of code makes it simpler to understand how things work
19:47 boris-42 tzabal end user will know okay, validator method always accept client, task & config
19:47 boris-42 tzabal and he can use all of them to make his validation method
19:47 tzabal boris-42 ok
19:47 boris-42 tzabal it's very important to keep things simple as possible
19:48 boris-42 tzabal imho I think that even know Rally in some places is to complicated and should be simplified
19:50 tzabal boris-42 understood. I propose though to have def a_random_validator(config, clients=None, task=None)
19:50 tzabal boris-42 instead of def a_random_validator(clients, task, config)
19:50 tzabal boris-42 actually let me code and leave a comment in patch set better
19:51 boris-42 tzabal ok will do
20:19 openstackgerrit Tzanetos Balitsaris proposed a change to stackforge/rally: Modify config semantic validation of benchmark engine  https://review.openstack.org/112981
20:28 penguinRaider joined #openstack-rally
20:32 rmk boris-42: https://gist.github.com/rmk40/2e5c56cddce1c6ed3bbc
20:32 rmk boris-42: is that what you meant?
20:36 boris-42 rmk yep
20:36 boris-42 rmk -1 random from code nice
20:36 boris-42 =)
20:36 rmk I had to add it to the test though
20:37 rmk https://gist.github.com/rmk40/2e5c56cddce1c6ed3bbc#file-gistfile1-txt-L114-L115
20:37 rmk I don't completely understand that decorator to be honest
20:42 boris-42 rmk you should remove from line 116
20:42 boris-42 rmk mock_choice
20:42 rmk ok
20:42 boris-42 rmk than you can remove that decortaor
20:43 boris-42 rmk mock.patch decorator adds new field in method
20:43 boris-42 rmk that's why it doesn't work if you remove it
20:43 rmk ahh ok
20:43 rmk ill fix that up and submit this
20:44 boris-42 rmk btw it's preferred to use gerrit
20:45 rmk I know :)
20:45 tzabal joined #openstack-rally
20:45 boris-42 rmk so this discussion will read others cores=)
20:45 boris-42 tzabal seems like you have some issue
20:45 tzabal boris-42 yes :/
20:45 rmk I Just wanted an opinion before gerrit
20:45 rmk Im submitting it there now
20:55 openstackgerrit Rafi Khardalian proposed a change to stackforge/rally: Do not assume networks should be assigned  https://review.openstack.org/113096
20:56 rmk ^^ Done.
20:59 openstackgerrit Rafi Khardalian proposed a change to stackforge/rally: Do not assume networks should be assigned  https://review.openstack.org/113096
21:03 boris-42 rmk thanks
21:04 boris-42 tzabal I will try to find why it doesn't work
21:04 boris-42 tzabal be back later
21:05 tzabal boris-42 ok, i am trying to track it too, i think that the problem lies in the validator(fn) function
21:05 boris-42 tzabal it may be
21:06 boris-42 tzabal seems like that all arguments are passed like **kw in some place
22:13 openstackgerrit Tzanetos Balitsaris proposed a change to stackforge/rally: Modify config semantic validation of benchmark engine  https://review.openstack.org/112981
22:44 AlexF_ joined #openstack-rally
22:48 tzabal boris-42 it seems that now is working https://review.openstack.org/112981
22:49 andreykurilin_ joined #openstack-rally
22:54 boris-42 tzabal one more time
22:54 boris-42 tzabal why did you put
22:54 boris-42 tzabal  def image_exists_validator(config, clients=None, task=None):
22:54 boris-42 tzabal clients & task are never None
22:55 boris-42 tzabal or even if they are None they will be passed as a None
22:55 tzabal boris-42 because of the unit tests
22:55 tzabal boris-42 this was the primary reason actually
22:56 boris-42 tzabal heh imho it is better to improve unit tests..
22:56 tzabal boris-42 one moment
22:56 tzabal boris-42 https://review.openstack.org/#/c/112981/7/tests/benchmark/test_validation.py
22:57 tzabal boris-42 so in the unit tests, we call the validator in lines like: result = validator(config)
22:57 boris-42 tzabal it's wrong
22:57 tzabal boris-42 without keyworded arguments in validators, like the one that you gave above (image_exists_validator)
22:58 boris-42 tzabal I mean we should call it with all arguments
22:58 boris-42 tzabal unit tests should be equal to real life situation
22:58 tzabal boris-42 so: result = validator(config, clients, task)
22:58 boris-42 tzabal or None, None
22:58 boris-42 tzabal config, None, None
22:58 tzabal boris-42 ok with that
22:59 tzabal boris-42 so instead of changing unit tests the call, i changed the parameters in validators with None
22:59 tzabal boris-42 i know that maybe it seems a dump
22:59 boris-42 tzabal nope
22:59 boris-42 tzabal change unit tests
23:00 boris-42 tzabal we have new code, and should have update unit tests
23:00 tzabal boris-42 but while i was making changes, i didn't realize that every time the validator will have clients and task
23:00 boris-42 tzabal not making code that will be used just to pass unit tests
23:00 tzabal boris-42 sure i agree
23:00 boris-42 tzabal at this moment it will always have
23:00 tzabal boris-42 it was just that i didn't realize that every validator takes the clients and tak
23:00 tzabal boris-42 so one moment to change it
23:01 boris-42 tzabal yep that's why refactoring was important
23:01 tzabal boris-42 are you ok with the other parts?
23:01 tzabal boris-42 the bug before was in validator(fn) function
23:03 boris-42 tzabal hmm
23:03 boris-42 tzabal hmm
23:03 tzabal boris-42 :D
23:09 boris-42 tzabal so really not sure=)
23:09 boris-42 tzabal about validator wrapper
23:09 boris-42 tzabal me need to think
23:10 tzabal boris-42 ok, well i think that actually is a bit tricky code
23:10 boris-42 tzabal totally agree=)
23:28 openstackgerrit Tzanetos Balitsaris proposed a change to stackforge/rally: Modify config semantic validation of benchmark engine  https://review.openstack.org/112981
23:29 boris-42 tzabal hm
23:29 boris-42 tzabal could you explain https://review.openstack.org/#/c/112981/8/rally/benchmark/validation.py what is in args and kwargs?
23:30 boris-42 tzabal line 45
23:31 boris-42 tzabal they can be just removed
23:31 tzabal boris-42 args and kwargs is what the wrap_given takes, and args equals with what argument the decorator in the benchmark scenario has (for example, for the@validation.required_services(consts.Service.NOVA, args = ('consts.Service.NOVA'))
23:32 tzabal boris-42 and kwargs is just empty
23:32 tzabal boris-42 it will have something only if we pass from decorator a keyword argument
23:32 boris-42 tzabal ah
23:32 boris-42 tzabal okay
23:32 tzabal boris-42 so i think that args is needed
23:32 boris-42 tzabal both are needed
23:32 tzabal boris-42 but with the currect state, kwargs is not useful
23:32 boris-42 tzabal could you add docstirng
23:33 boris-42 tzabal to wrap_given
23:33 boris-42 tzabal and explain what is it, so others will understand it=)
23:33 boris-42 tzabal faster
23:33 tzabal boris-42 ok
23:34 boris-42 tzabal otherwise looks great
23:34 boris-42 tzabal I like improvements of our framework=)
23:35 tzabal boris-42 yes, refactoring is nice
23:46 boris-42 tzabal so are you going to publish quick new patch with doctoring?)
23:46 boris-42 doc strings*
23:46 tzabal boris-42 yeap, i am writing the doc string
23:47 boris-42 tzabal if so andreykurilin is not sleeping
23:47 boris-42 tzabal so we can get this fast in master=)
23:47 tzabal boris-42 yoohoo :P
23:48 openstackgerrit Tzanetos Balitsaris proposed a change to stackforge/rally: Modify config semantic validation of benchmark engine  https://review.openstack.org/112981
23:49 tzabal boris-42 https://review.openstack.org/#/c/112981/9/rally/benchmark/validation.py any suggestion for first line in docstring of wrap_given?
23:52 boris-42 tzabal just put something like, "scenario validator decorator"
23:53 boris-42 tzabal or scenario validator decorator generator
23:53 boris-42 lol
23:53 boris-42 =)
23:53 tzabal boris-42 hahaha
23:53 tzabal boris-42 better than "Inner function." anyway
23:56 tzabal boris-42 well actually it is the arguments of the decorator of the benchmark scenario

| Channels | #openstack-rally index | Today | | Search | Google Search | Plain-Text | summary