Perl 6 - the future is here, just unevenly distributed

IRC log for #openstack-rally, 2014-10-17

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

All times shown according to UTC.

Time Nick Message
00:23 yingjun joined #openstack-rally
00:30 dmorita joined #openstack-rally
00:46 openstackgerrit Boris Pavlovic proposed a change to stackforge/rally: WIP Improve rally generic cleanup mechanism  https://review.openstack.org/129060
01:10 openstackgerrit Boris Pavlovic proposed a change to stackforge/rally: WIP Improve rally generic cleanup mechanism  https://review.openstack.org/129060
01:25 yaguang joined #openstack-rally
01:39 openstackgerrit Boris Pavlovic proposed a change to stackforge/rally: WIP Improve rally generic cleanup mechanism  https://review.openstack.org/129060
01:42 openstackgerrit Boris Pavlovic proposed a change to stackforge/rally: WIP Improve rally generic cleanup mechanism  https://review.openstack.org/129060
01:45 openstackgerrit Boris Pavlovic proposed a change to stackforge/rally: WIP Improve rally generic cleanup mechanism  https://review.openstack.org/129060
02:02 chandankumar joined #openstack-rally
02:02 openstackgerrit Boris Pavlovic proposed a change to stackforge/rally: WIP Improve rally generic cleanup mechanism  https://review.openstack.org/129060
02:02 yingjun joined #openstack-rally
02:36 jjmb joined #openstack-rally
03:23 yingjun joined #openstack-rally
03:52 rvcehimanshu joined #openstack-rally
03:55 vaidy joined #openstack-rally
04:10 chandankumar joined #openstack-rally
04:15 yingjun joined #openstack-rally
04:30 neeti joined #openstack-rally
04:31 Poornima joined #openstack-rally
04:44 nkhare joined #openstack-rally
04:51 jjmb1 joined #openstack-rally
05:01 oanufriev joined #openstack-rally
05:02 chandan_kumar joined #openstack-rally
05:05 rdas joined #openstack-rally
05:15 oanufriev joined #openstack-rally
05:28 chandankumar joined #openstack-rally
05:41 chandan_kumar joined #openstack-rally
05:55 kevinbenton joined #openstack-rally
05:56 kevinbenton hi, i’m trying to get rally to reference the sample file in this directory in neutron https://github.com/bigswitch/neutron/tree/master/rally-scenarios/plugins
05:56 kevinbenton how do we reference scenarios that live in a different directory like this?
06:02 yingjun joined #openstack-rally
06:09 k4n0 joined #openstack-rally
06:44 chandankumar joined #openstack-rally
06:56 andreykurilin_ joined #openstack-rally
07:15 yingjun joined #openstack-rally
07:18 boris-42 kevinbenton in the same way
07:18 boris-42 kevinbenton just specifying <class_name>.<benchmark_name>
07:19 kevinbenton boris-42: but that plugin directory isn’t in any path is it?
07:19 boris-42 kevinbenton it has to be in specific path
07:19 boris-42 kevinbenton more about that you can read here http://rally.readthedocs.org/en/latest/concepts.html#scenarios-plugins
07:19 boris-42 kevinbenton so basically if it is in /etc/rally/plugins/scenarios or ~/.rally/plugins/scenrios
07:20 boris-42 kevinbenton it will be autoloader and accessable
07:20 kevinbenton boris-42: oh okay. so it won’t work just sitting in the neutron directory like that
07:20 boris-42 kevinbenton in case of neutron "gates" it will
07:20 kevinbenton boris-42: does the gate do that by copying to one of these auto-load directories?
07:21 boris-42 kevinbenton https://github.com/stackforge/rally/blob/master/tests/ci/rally-gate.sh#L24-L27
07:21 boris-42 kevinbenton yep exactly
07:21 kevinbenton boris-42: excellent. thanks for the pointers
07:28 boris-42 kevinbenton so actually one more thing
07:28 boris-42 kevinbenton soon I am expecting update in this stuff
07:28 boris-42 kevinbenton actually in rally plugins are almost everything
07:28 boris-42 kevinbenton SLA_checks, context, scenarios, load generators, ….
07:29 boris-42 kevinbenton so plugins/scenarios -> plugins/*
07:29 kevinbenton boris-42: yeah
07:30 boris-42 kevinbenton but no worries we will update docs =)
07:32 kevinbenton boris-42: one more question. i’m working on a neutron scenario and i need to create a lot of ports for the scenario
07:32 kevinbenton boris-42: but i don’t want to measure the performance of that
07:32 boris-42 kevinbenton so you need something like benchmark.context ?
07:32 boris-42 kevinbenton to create them before load generation?
07:33 kevinbenton boris-42: oh, maybe that’s what I need
07:33 kevinbenton boris-42: can i still use the built in create_ports helpers?
07:33 boris-42 kevinbenton yep I think so
07:34 kevinbenton boris-42: great, and it won’t delete the ports and recreate them for each benchmark if it’s just context, right?
07:34 boris-42 kevinbenton take a look here
07:34 boris-42 kevinbenton https://github.com/stackforge/rally/blob/master/rally/benchmark/context/images.py
07:34 boris-42 kevinbenton heh it will create them + do one benchmark + delete them
07:34 kevinbenton boris-42: ok, so that’s unavoidable for now then, right?
07:34 boris-42 kevinbenton ?
07:35 boris-42 kevinbenton context are like setUp/cleanUp (that are done before load generation)
07:35 boris-42 so
07:35 kevinbenton boris-42: i was thinking something similar to a fixture
07:35 boris-42 kevinbenton so yep
07:36 boris-42 kevinbenton I think that context are what you are looking for
07:36 boris-42 kevinbenton just every benchmark* has own context
07:36 boris-42 kevinbenton for every benchmark we are creating set of tenants + users + do_some_other_stuff
07:36 kevinbenton boris-42: makes sense
07:36 boris-42 kevinbenton and then we are calling inside this contex multiple times benchmark scenario
07:37 boris-42 kevinbenton after that we are cleanuping
07:37 boris-42 kevinbenton so it's not on every call of benchmark.scenario*
07:37 kevinbenton boris-42: ok, so that’s what this line i copied is doing @base.scenario(context={"cleanup": ["neutron"]})
07:37 kevinbenton boris-42: but right now it’s only doing the cleanup and i’m creating the ports as part of the benchmark
07:38 boris-42 kevinbenton nono=)
07:38 boris-42 kevinbenton that's default context
07:38 boris-42 =)
07:38 kevinbenton boris-42: but it’s what’s doing the cleanup, right?
07:38 boris-42 kevinbenton you can specify default context and can specify via "task"
07:38 boris-42 kevinbenton default context and user specified via task are merged
07:38 boris-42 kevinbenton so basically for ports u don't need to set default one
07:39 kevinbenton boris-42: oh, so is anything cleaning up the ports then right now?
07:39 boris-42 kevinbenton yep this default context "cleanup"
07:40 boris-42 kevinbenton should cleanup ports
07:40 boris-42 kevinbenton actually current approach of cleanup sux
07:40 kevinbenton boris-42: okay, i think i’m getting it
07:40 boris-42 kevinbenton this patch https://review.openstack.org/#/c/129060/
07:40 boris-42 kevinbenton will change cleanup mechanism
07:40 kevinbenton so how do i reference a custom context?
07:40 boris-42 kevinbenton via task
07:41 boris-42 kevinbenton take a look here https://github.com/stackforge/rally/blob/master/rally-scenarios/rally.yaml#L235-L253
07:41 boris-42 kevinbenton we are specifying quotas context
07:41 boris-42 kevinbenton users context is by default specified to create 1 user and 1 teannts
07:41 boris-42 kevinbenton  so actually I have to go, but I'll be back in 1hrs
07:42 boris-42 kevinbenton ping rediskin andreykurilin  msdubov for support=)
07:42 boris-42 kevinbenton until I get back=)
07:42 kevinbenton boris-42: ok, thanks for the pointers
07:50 stannie joined #openstack-rally
07:53 oanufriev joined #openstack-rally
08:22 amaretskiy joined #openstack-rally
08:27 cdent joined #openstack-rally
08:33 rvcehimanshu joined #openstack-rally
08:51 andreykurilin k4n0, here?
08:56 openstackgerrit Timur Nurlygayanov proposed a change to stackforge/rally: Adding Network Context Class  https://review.openstack.org/103306
09:00 openstackgerrit Andrey Kurilin proposed a change to stackforge/rally: Fix wrong argument in rally verify  https://review.openstack.org/129169
09:04 openstackgerrit Andrey Kurilin proposed a change to stackforge/rally: Fix wrong argument in rally verify  https://review.openstack.org/129169
09:04 yaguang joined #openstack-rally
09:45 rdas joined #openstack-rally
09:48 k4n0 andreykurilin, hi
09:48 andreykurilin k4n0, we have critical bug
09:49 andreykurilin k4n0, rally verify is not working
09:49 andreykurilin k4n0, I publish a bug-fix, can you review it?
09:49 andreykurilin *I published
09:52 andreykurilin k4n0, https://review.openstack.org/#/c/129169/
09:55 boris-42 kevinbenton I'm back
09:57 kevinbenton boris-42: hey, still working on some stuff and about to head to bed. i didn’t get around to turning my setup method into a context yet
10:04 openstackgerrit Sergey Murashov proposed a change to stackforge/rally: Add API version parameters to config file  https://review.openstack.org/129195
10:06 boris-42 kevinbenton sure
10:10 openstackgerrit Oleh Anufriiev proposed a change to stackforge/rally: Add detailed description for rally commands  https://review.openstack.org/124910
10:23 openstackgerrit Andrey Kurilin proposed a change to stackforge/rally: Fix wrong argument in rally verify  https://review.openstack.org/129169
11:01 k4n0 andreykurilin, done, the bug came up due my review https://review.openstack.org/#/c/128336/
11:05 Kui joined #openstack-rally
11:06 Kui boris_42: hello, is there a method to debug rally with pdb ?   e.g. insert "import pdb; pdb.set_trace()" in scenario code
11:07 openstackgerrit Mikhail Dubov proposed a change to stackforge/rally: Simplify and improve "rally info"  https://review.openstack.org/126421
11:10 boris-42 Kui yep
11:10 boris-42 Kui you should use "serial" runner
11:11 boris-42 Kui it runs scenario in same procsess/thread
11:16 boris-42 k4n0 please do not merge bug fixes alone
11:16 boris-42 k4n0 it's better to get attention of few cores
11:16 boris-42 k4n0 we need to make sure that there won't be regressions
11:16 k4n0 boris-42, Yes, this one was due to my fix and as andreykurilin mentioned its urgent, hence
11:17 boris-42 k4n0 in any case it's better to get 2 cores
11:17 k4n0 boris-42, Yes
11:17 boris-42 k4n0 if is not small nit / typo in docs and so on
11:17 k4n0 boris-42, got it
11:17 boris-42 k4n0 then it's okay to put +2
11:17 boris-42 k4n0 and workflow +1
11:17 rvcehimanshu joined #openstack-rally
11:18 openstackgerrit James Page proposed a change to stackforge/rally: Add support for creating neutron networks  https://review.openstack.org/129210
11:19 openstackgerrit James Page proposed a change to stackforge/rally: Add support for creating neutron networks  https://review.openstack.org/129210
11:29 openstackgerrit A change was merged to stackforge/rally: Fix wrong argument in rally verify  https://review.openstack.org/129169
11:46 vkmc joined #openstack-rally
11:46 jjmb joined #openstack-rally
12:06 Kui boris-42: I set     "runner": {
12:06 Kui "type": "serial",
12:06 Kui "times": 1
12:06 Kui } in scenario json file, the scenario runs well. But when I insert pdb.set_trace() in "scenarios/cinder/volumes.py", it hang somewhere.
12:06 boris-42 Kui hm
12:06 boris-42 maybe because there is another thread that is fetching results..
12:06 Kui boris-42: 2014-10-17 20:03:40.804 21883 INFO rally.benchmark.context.quotas.quotas [-] Task abea73df-0523-4ce1-8ad1-c16731a441e8 | Starting:  Enter context: `quotas`
12:06 Kui 2014-10-17 20:03:41.164 21883 INFO rally.benchmark.context.quotas.quotas [-] Task abea73df-0523-4ce1-8ad1-c16731a441e8 | Completed: Enter context: `quotas`
12:06 Kui 2014-10-17 20:03:41.165 21883 INFO rally.benchmark.runners.base [-] Task abea73df-0523-4ce1-8ad1-c16731a441e8 | ITER: 0 START
12:07 Kui boris-42: it hangs here
12:10 Kui boris-42: it there a method to make all the threads (including runner and fetching result thread) serialized ?
12:11 boris-42 Kui not sure.. =(
12:11 boris-42 Kui I mean without changes in code No
12:11 openstackgerrit Mikhail Dubov proposed a change to stackforge/rally: Simplify and improve "rally info"  https://review.openstack.org/126421
12:13 Kui boris-42: OK, I will turn to logging / print for debugging.
12:13 boris-42 Kui so I will refactor benchmark.engine and after my patch
12:13 boris-42 Kui we will try to make some single thread mode
12:14 boris-42 Kui so you'll be able to trace with pbr
12:14 boris-42 Kui I don't think that it will be super complicate
12:15 Kui boris-42: Awesome, that will be helpful.  :)
12:37 rvcehimanshu joined #openstack-rally
12:39 cesarm_ joined #openstack-rally
12:45 openstackgerrit Mikhail Dubov proposed a change to stackforge/rally: Simplify and improve "rally info"  https://review.openstack.org/126421
12:46 openstackgerrit Boris Pavlovic proposed a change to stackforge/rally: WIP Improve rally generic cleanup mechanism  https://review.openstack.org/129060
12:46 cdent joined #openstack-rally
13:19 openstackgerrit joined #openstack-rally
13:20 pboros joined #openstack-rally
13:28 openstackgerrit Boris Pavlovic proposed a change to stackforge/rally: WIP Improve rally generic cleanup mechanism  https://review.openstack.org/129060
13:37 chandankumar joined #openstack-rally
13:44 openstackgerrit Boris Pavlovic proposed a change to stackforge/rally: WIP Improve rally generic cleanup mechanism  https://review.openstack.org/129060
13:47 dpaterson joined #openstack-rally
13:53 jamespage joined #openstack-rally
13:54 jamespage boris-42, hey - thanks for pointing me at https://review.openstack.org/#/c/103306
13:54 boris-42 jamespage hi there
13:54 boris-42 jamespage ya it requires more reviewing/testing
13:54 boris-42 jamespage but seems like it's quite ready at least it past unit/functional testing=)
13:54 jamespage boris-42, yeah - about to comment - I had to fixup two things locally
13:54 boris-42 jamespage so left comments there
13:55 boris-42 jamespage that will be really useful
13:55 jamespage boris-42, done!
13:55 boris-42 jamespage nice catches!
13:56 jamespage boris-42, apart from that works well on my test target :-)
13:57 boris-42 jamespage btw what you are using as test target?
13:57 boris-42 jamespage devstack /
13:57 boris-42 ?
13:57 jamespage boris-42, no a juju deployed openstack (running virtually ontop of openstack) using zeromq
13:57 boris-42 jamespage and rally works against that?)
13:57 boris-42 jamespage woot!=)
13:57 jamespage it does
13:58 boris-42 jamespage btw soon we will have third party jobs in rally
13:58 boris-42 jamespage so we can add some periodic tests to check rally against such kind of deployments
13:58 jamespage boris-42, right now I'm just setting up as an existing deployment
13:58 jamespage boris-42, oh nice
13:58 boris-42 jamespage ya we got 2 nodes with 256gb ram=)
13:59 jamespage boris-42, I've also been using rally to grind the current development release of Ubuntu
13:59 boris-42 jamespage and 2 cpu / node=)
13:59 jamespage its good for shaking out things that don't appear during a standard tempest run
13:59 boris-42 jamespage btw do you run tempest by hands or via rally?
14:00 openstackgerrit Andrey Kurilin proposed a change to stackforge/rally: WIP: functional test for rally verify  https://review.openstack.org/129262
14:01 boris-42 ^ andreykurilin wooot
14:01 jamespage boris-42, well not by hand - but not via rally either
14:01 boris-42 jamespage heh we have rally verify command that prepares tempest and runs it
14:01 jamespage boris-42, we have our own CI for deployment/prepare/test right now
14:01 jamespage boris-42, yeah _ I see
14:01 boris-42 jamespage ah oh =)
14:01 andreykurilin boris-42, I want to test ability our gate to launch "rally verify".
14:06 boris-42 jamespage btw soon rally will be much cleaner=)
14:06 boris-42 jamespage https://review.openstack.org/#/c/129060/
14:26 dpaterson boris-42, how's it going?  I've submitted another patch set for the compare feature that incorporates your algorithm.
14:26 dpaterson https://review.openstack.org/#/c/125184/
14:26 boris-42 dpaterson heh I am heavy working on rally cleanups
14:26 boris-42 dpaterson I'll try to take a look at your patch today
14:34 kambiz joined #openstack-rally
14:37 openstackgerrit Andrey Kurilin proposed a change to stackforge/rally: Do not merge: functional test for rally verify --set image  https://review.openstack.org/129282
14:38 dpaterson boris-42, no rush, just letting you know it's been done.
14:39 dpaterson Once merged I will refactor the other verification report to match.
14:39 openstackgerrit Andrey Kurilin proposed a change to stackforge/rally: Do not merge: functional test for rally verify --set image  https://review.openstack.org/129282
14:39 boris-42 dpaterson great
14:41 isviridov joined #openstack-rally
15:11 klindgren__ joined #openstack-rally
15:15 vkmc boris-42, hi, how are you?
15:16 vkmc boris-42, is it possible to use the context on initialization?
15:36 vkmc nevermind :) find a workaround I think
15:50 openstackgerrit Sergey Skripnick proposed a change to stackforge/rally: Cli arguments refactoring(WIP)  https://review.openstack.org/129306
15:56 andreykurilin_ joined #openstack-rally
16:02 chandankumar joined #openstack-rally
16:17 marcoemorais joined #openstack-rally
16:17 marcoemorais joined #openstack-rally
16:19 boris-42 vkmc hi there
16:19 boris-42 vkmc so what do you mean context on initialization?)
16:21 vkmc boris-42, I meant getting the context before instantiating the client
16:21 boris-42 vkmc still not sure that I get=)
16:21 vkmc but I found and alternative way to pass the tenant_id to the client
16:21 vkmc using this https://github.com/stackforge/rally/blob/master/rally/benchmark/scenarios/quotas/quotas.py#L44
16:22 vkmc in here https://github.com/stackforge/rally/blob/master/rally/osclients.py#L266
16:22 boris-42 vkmc you can do the similar thing in your scenario
16:22 vkmc :D
16:22 boris-42 vkmc it will be just self.context()["user"]["tenant_id"]
16:22 vkmc of course, the thing was that I didn't know how to access the auth_conf from there
16:23 vkmc I found out that and now I'm debugging something else
16:25 boris-42 so you can get endpoint
16:26 boris-42 self.context()["user"]["endpoint"]
16:26 boris-42 vkmc this is what you have as users https://github.com/stackforge/rally/blob/master/rally/benchmark/context/users.py#L134-L136
16:26 boris-42 vkmc this is endpoint https://github.com/stackforge/rally/blob/master/rally/objects/endpoint.py#L21-L37
16:27 vkmc cool, thanks boris-42!
16:27 boris-42 vkmc and this is very important plac
16:27 boris-42 vkmc https://github.com/stackforge/rally/blob/master/rally/benchmark/runners/base.py#L56-L89
16:28 boris-42 vkmc this is how one iteration is executed
16:28 * vkmc clicks
16:48 openstackgerrit Alexander Maretskiy proposed a change to stackforge/rally: (WIP) Improve HTML report  https://review.openstack.org/128991
16:48 jamespage boris-42, is there a nice way to disable cleanup? I specifically don't want to delete all the many 1000's of instances I've created during benchmarking
16:54 chandankumar joined #openstack-rally
16:56 rook jamespage not currently, but you can modify the code to do it ;)
17:14 rvcehimanshu joined #openstack-rally
17:37 boris-42 jamespage not yet
17:37 boris-42 jamespage it will be
17:38 boris-42 jamespage in future
17:38 boris-42 jamespage btw you can request this feature
17:38 boris-42 jamespage could you please add a request here https://github.com/stackforge/rally/tree/master/doc/feature_request
17:59 rvcehimanshu joined #openstack-rally
18:00 rvcehimanshu left #openstack-rally
18:01 marcoemorais joined #openstack-rally
18:06 mwagner_lap joined #openstack-rally
18:26 marcoemorais joined #openstack-rally
18:28 marcoemorais joined #openstack-rally
19:21 openstackgerrit Victoria Martínez de la Cruz proposed a change to stackforge/rally: Adds Zaqar Basic scenarios  https://review.openstack.org/129038
19:23 openstackgerrit Victoria Martínez de la Cruz proposed a change to stackforge/rally: Adds Zaqar Basic scenarios  https://review.openstack.org/129038
19:30 openstackgerrit Victoria Martínez de la Cruz proposed a change to stackforge/rally: Adds Zaqar Basic scenarios  https://review.openstack.org/129038
19:34 openstackgerrit Victoria Martínez de la Cruz proposed a change to stackforge/rally: Adds Zaqar Basic scenarios  https://review.openstack.org/129038
19:40 oanufriev joined #openstack-rally
20:02 marcoemorais joined #openstack-rally
20:04 Kui joined #openstack-rally
20:30 oanufriev joined #openstack-rally
20:48 klindgren_ joined #openstack-rally
21:51 andreykurilin_ joined #openstack-rally
22:04 openstackgerrit joined #openstack-rally
22:51 jjmb joined #openstack-rally

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