Perl 6 - the future is here, just unevenly distributed

IRC log for #openstack-rally, 2015-08-10

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

All times shown according to UTC.

Time Nick Message
00:20 kbyrne joined #openstack-rally
01:29 dspano joined #openstack-rally
01:45 psuriset joined #openstack-rally
02:49 sacharya joined #openstack-rally
02:50 psuriset joined #openstack-rally
03:37 PrashantS joined #openstack-rally
04:18 PrashantS joined #openstack-rally
04:52 rdas joined #openstack-rally
04:54 tfreger joined #openstack-rally
05:09 psuriset joined #openstack-rally
05:36 PrashantS joined #openstack-rally
05:55 yfried_ joined #openstack-rally
05:56 yfried__ joined #openstack-rally
06:12 agarciam joined #openstack-rally
06:45 PrashantS joined #openstack-rally
07:03 kiran-r joined #openstack-rally
07:12 psuriset joined #openstack-rally
07:16 abehl joined #openstack-rally
07:17 abehl joined #openstack-rally
07:27 adiantum joined #openstack-rally
07:39 kbaikov joined #openstack-rally
07:39 albertw joined #openstack-rally
07:44 arxcruz joined #openstack-rally
07:45 fhubik joined #openstack-rally
07:48 kbaikov joined #openstack-rally
08:10 e0ne joined #openstack-rally
08:18 zerda joined #openstack-rally
08:32 e0ne joined #openstack-rally
08:44 PrashantS joined #openstack-rally
09:05 PrashantS joined #openstack-rally
09:14 openstackgerrit Oleh Anufriiev proposed openstack/rally: Update rally/contrib to use install_rally.sh  https://review.openstack.org/210137
09:15 klindgren__ joined #openstack-rally
09:18 openstackgerrit Kiran proposed openstack/rally: Add unit test  https://review.openstack.org/210700
09:24 PrashantS joined #openstack-rally
09:28 _kiran_ joined #openstack-rally
09:31 tosky joined #openstack-rally
09:40 openstackgerrit nayna-patel proposed openstack/rally: Add Cinder create_and_update_volume scenario  https://review.openstack.org/208996
09:42 PrashantS joined #openstack-rally
09:44 openstackgerrit Oleh Anufriiev proposed openstack/rally: Update rally/contrib to use install_rally.sh  https://review.openstack.org/210137
09:45 cdent joined #openstack-rally
10:14 PrashantS joined #openstack-rally
10:15 redixin joined #openstack-rally
10:16 rediskin joined #openstack-rally
10:17 redixin joined #openstack-rally
10:19 sunilm joined #openstack-rally
10:26 agarciam1 joined #openstack-rally
10:26 agarciam1 joined #openstack-rally
10:28 PrashantS joined #openstack-rally
10:39 openstackgerrit Illia Khudoshyn proposed openstack/rally: Refactor BenchmarkEngine  https://review.openstack.org/209046
10:49 openstackgerrit Kiran proposed openstack/rally: Add unit test  https://review.openstack.org/210700
10:51 kiran-r joined #openstack-rally
10:53 yfried kiran-r: https://review.openstack.org/#/c/193527 why are vips created using admin client and not regular client?
10:53 jaypipes joined #openstack-rally
10:57 openstackgerrit Alexander Maretskiy proposed openstack/rally: [Reports] Add classes for task HTML report charts  https://review.openstack.org/159458
10:59 kiran-r yfried: In Line #362 https://review.openstack.org/#/c/193527/32/rally/plugins/openstack/scenarios/neutron/utils.py,cm we are trying to use an un-authorized tenant.
11:00 yfried kiran-r: why?
11:01 yfried kiran-r: something here is wrong. these ops should be done on default tenant
11:03 openstackgerrit Alexander Maretskiy proposed openstack/rally: [Reports] Add chunks support to report  https://review.openstack.org/146814
11:05 yfried kiran-r: please go over my comments
11:08 openstackgerrit Alexander Maretskiy proposed openstack/rally: [Reports] Add chunks support to report  https://review.openstack.org/146814
11:08 kiran-r yfried: Sorry I was away
11:09 yfried kiran-r: np
11:09 yfried kiran-r: posted comments to your patch
11:09 kiran-r yfried: I saw. :)
11:10 kiran-r yfried: I will get back to you on the admin_clients thingy soon.
11:12 PrashantS joined #openstack-rally
11:13 PrashantS joined #openstack-rally
11:13 yfried kiran-r: I saw you were asked this already
11:14 kiran-r yfried: using clients will say the ¨pool was not found¨
11:15 kiran-r When iterated around the context created pools
11:16 kiran-r yfried: And I realized that we need to specify tenant_id where the pool was earlier created.
11:17 kiran-r yfried: specifying tenant_id was throwing an error saying trying use unauthenticated tenant. admin_clients solved all the problem.
11:18 yfried kiran-r: then maybe the pools were created badly.
11:22 yfried kiran-r: I'm not sure how, but the lbaas ctx seems to be creating the pools without tenant id
11:22 yfried kiran-r: in that ctx, you should either:
11:23 yfried kiran-r: retrieve the tenant's client from the client manager (don't know how)
11:23 yfried kiran-r: use pool_create with tenant id,
11:23 yfried kiran-r: ^ assuming the client you have in ctx is admin
11:24 openstackgerrit Alexander Maretskiy proposed openstack/rally: [Reports] Add Load Profile chart to report  https://review.openstack.org/169828
11:24 tab_ joined #openstack-rally
11:27 openstackgerrit Illia Khudoshyn proposed openstack/rally: Fix typo in 'new rally input task format' spec  https://review.openstack.org/211107
11:28 kiran-r yfried: You mean I need to create context pools using tenant_id??
11:32 exploreshaifali joined #openstack-rally
11:36 exploreshaifali joined #openstack-rally
11:51 ahirshbeg_ joined #openstack-rally
11:54 pboldin joined #openstack-rally
11:54 pboldin Hi all, please review https://review.openstack.org/183228/
12:04 ahirshbeg_ Hello fellow Rally users. Can some one point or explain the following variables in the runner input task since I wasn't able to find explanation in the code:
12:04 ahirshbeg_ "runner": {         "rps": 1,          "type": "rps",          "timeout": 360,          "times": 200       }
12:04 ahirshbeg_ AFAIK rps = Scenario task per second
12:04 ahirshbeg_ type = ???
12:04 ahirshbeg_ timeout = Total timeout per task (regardless of the fact it is still running)
12:04 ahirshbeg_ times = Total times of task requests to made
12:04 ahirshbeg_ So in theory it will take me 200 seconds to send all task request, am I right here?
12:04 ahirshbeg_ From debug log I see:
12:04 ahirshbeg_ 2015-08-10 14:02:44.256 17810 INFO rally.benchmark.runners.base [-] Task c49e3985-95ac-4e41-a2b5-5ed4ca38ceb7 | ITER: 4 START
12:04 ahirshbeg_ 2015-08-10 14:02:44.257 17811 INFO rally.benchmark.runners.base [-] Task c49e3985-95ac-4e41-a2b5-5ed4ca38ceb7 | ITER: 5 START
12:04 ahirshbeg_ 2015-08-10 14:02:44.257 17811 RALLYDEBUG rally.plugins.common.runners.rps [-] Worker: 2 rps: 0.499915286353 (requested rps: 0.25)
12:04 ahirshbeg_ 2015-08-10 14:02:44.256 17810 RALLYDEBUG rally.plugins.common.runners.rps [-] Worker: 2 rps: 0.499857049339 (requested rps: 0.25)
12:04 ahirshbeg_ 2015-08-10 14:02:44.258 17813 INFO rally.benchmark.runners.base [-] Task c49e3985-95ac-4e41-a2b5-5ed4ca38ceb7 | ITER: 6 START
12:04 ahirshbeg_ 2015-08-10 14:02:44.259 17814 INFO rally.benchmark.runners.base [-] Task c49e3985-95ac-4e41-a2b5-5ed4ca38ceb7 | ITER: 7 START
12:04 ahirshbeg_ 2015-08-10 14:02:44.259 17814 RALLYDEBUG rally.plugins.common.runners.rps [-] Worker: 2 rps: 0.49995562828 (requested rps: 0.25)
12:04 ahirshbeg_ 2015-08-10 14:02:44.258 17813 RALLYDEBUG rally.plugins.common.runners.rps [-] Worker: 2 rps: 0.499929765793 (requested rps: 0.25)
12:08 Miouge joined #openstack-rally
12:08 pboldin ahirshbeg_: please use paste.openstack.org
12:09 amaretskiy ahirshbeg_: "type" is runner type - this is one from existing runners
12:10 amaretskiy ahirshbeg_: here are runners that come with rally https://github.com/openstack/rally/tree/master/rally/plugins/common/runners
12:11 amaretskiy ahirshbeg_ `times' is how many iterations to run, e.g. how many times call the scenario (but how exactly to make these  calls  is the runner part)
12:12 amaretskiy ahirshbeg_: `timeout' options is specific to `rps' runner
12:13 amaretskiy ahirshbeg_: https://github.com/openstack/rally/blob/master/rally/plugins/common/runners/rps.py#L40
12:22 openstackgerrit Oleh Anufriiev proposed openstack/rally: Update rally/contrib to use install_rally.sh  https://review.openstack.org/210137
12:23 ahirshbeg_ pboldin: thank for the note, will do next time
12:24 ahirshbeg_ amaretskiy: thanks, can you please explain rps? as we see from the log it's not sending one per second(task 4 & 5 start at the exact same moment)
12:26 ahirshbeg_ http://paste.openstack.org/show/412353/
12:27 amaretskiy ahirshbeg_:  RPS runner starts`rps' iterations per `timeout'
12:29 amaretskiy for example we have times = 200 , rps = 10, timeout = 5, then RPS runner (we definitely have to use type = "rps") will start 10 iteration per each 5 seconds, there will be 20 starts (10 * 20 = 200)
12:30 amaretskiy ahirshbeg_: in opposite to `constant' and `serial' runners, rps runner does not wait until some iterations ends, it just starts interations, so it can generate the most high load
12:35 ahirshbeg_ im a bit confused, please review my input task parms and the output is see in the log:http://paste.openstack.org/show/412356/ the times aren't as you explained
12:45 mwagner_lap joined #openstack-rally
12:47 openstackgerrit Kirill Shileev proposed openstack/rally: Add CobblerProvider provider  https://review.openstack.org/188847
12:57 openstackgerrit nayna-patel proposed openstack/rally: Add Cinder create_and_update_volume_snapshot scenario  https://review.openstack.org/211132
12:59 openstackgerrit Sergey Skripnick proposed openstack/rally: Add new script for rally-gate jobs  https://review.openstack.org/175549
13:02 PrashantS joined #openstack-rally
13:05 Makika_ joined #openstack-rally
13:06 PrashantS joined #openstack-rally
13:09 PrashantS joined #openstack-rally
13:10 openstackgerrit Sergey Skripnick proposed openstack/rally: Add new script for rally-gate jobs  https://review.openstack.org/175549
13:26 ahirshbeg_ amaretskiy: so how can i run one task per second/each second?
13:27 amaretskiy ahirshbeg_: "rps": 1, "timeout": 1
13:32 yfried amaretskiy: mtg now?
13:32 yfried redixin: ^?
13:38 openstackgerrit Illia Khudoshyn proposed openstack/rally: [WIP] Add task format converter  https://review.openstack.org/211160
13:42 ikhudoshyn joined #openstack-rally
13:44 psuriset joined #openstack-rally
13:48 puranamr joined #openstack-rally
13:50 ivasev joined #openstack-rally
13:55 yingjun joined #openstack-rally
13:56 dspano joined #openstack-rally
13:58 ahirshbeg_ amaretskiy: it's seems like it still launch the tasks faster then one per second: http://paste.openstack.org/show/412370/
13:59 mbonell joined #openstack-rally
13:59 amaretskiy ahirshbeg_: add scenario config to paste
14:00 yfried amaretskiy: is there a mtg now?
14:01 amaretskiy yfried: it should be
14:04 boris-42 joined #openstack-rally
14:04 boris-42 ping kun_huang
14:04 PrashantS joined #openstack-rally
14:04 boris-42 kun_huang: it's meeting time
14:05 boris-42 yfried: ^
14:05 boris-42 redixin: ^
14:06 amaretskiy ahirshbeg_:  please include task config to paste
14:26 openstackgerrit Oleh Anufriiev proposed openstack/rally: Update rally/contrib to use install_rally.sh  https://review.openstack.org/210137
14:32 cas__ joined #openstack-rally
14:44 exploreshaifali joined #openstack-rally
14:47 ushkalim joined #openstack-rally
14:48 yfried boris-42: andreykurilin: amaretskiy: quick q re rps runner
14:48 yfried continuing ahirshbeg's q
14:49 yfried we would like to have an iteration fire over X seconds (without waiting for the previous to finish)
14:50 amaretskiy yfried: RPS runner does not wait until some iterations finished
14:50 yfried but it seems from the logs that they are all fired rapidly regardless of our "timeout=X"
14:50 amaretskiy yfried: give paste with task config included
14:50 yfried amaretskiy: http://paste.openstack.org/show/412356/ config
14:51 openstackgerrit Sergey Skripnick proposed openstack/rally: Add new script for rally-gate jobs  https://review.openstack.org/175549
14:51 yfried amaretskiy: log http://paste.openstack.org/show/412353/
14:52 amaretskiy yfried: I think I made a mistake regarding `timeout', sorry
14:53 yfried amaretskiy: so bug?
14:53 amaretskiy I don't think so
14:53 yfried amaretskiy: or wrong input?
14:53 amaretskiy give me a minute.....
14:58 amaretskiy rps is run per second (one second), and timeout is operation timeout
14:58 amaretskiy https://github.com/openstack/rally/blob/master/rally/plugins/common/runners/rps.py#L41
14:58 amaretskiy I was wrong regarding `timeout'
14:59 amaretskiy so RPS runner starts N iterations per 1 second
14:59 amaretskiy and there is no bug
14:59 yfried amaretskiy: so 1 -  the doc should be improved (we need to apply rally info to runners)
14:59 amaretskiy yfried ahirshbeg_ ^
15:00 ivasev hi guys i am kinda new to rally, i was wondering is there a config where i can specify which versions i want to use like glance v1/v2, keystone and etc
15:00 yfried amaretskiy: 2 - who would I deploy 1 iteration every X seconds?
15:00 yfried amaretskiy: 3 -  what is timeout?
15:02 yfried ivasev: I think that's an openstack issue and not rally - rally doesn't set the openstack backend
15:02 amaretskiy yfried: I think it would be nice to add some descriptive explanations regarding runners to rally doc
15:03 yfried amaretskiy: http://paste.openstack.org/show/412356/ how should the "runner" section be changed to deploy 1 iteration every X seconds?
15:04 agarciam joined #openstack-rally
15:05 amaretskiy I think we need another custom runner for that...
15:05 yfried amaretskiy: what does timeout stand for?
15:05 agarciam joined #openstack-rally
15:05 yfried amaretskiy: according to you, the input I had should have fired 1 every 1sec. it didn't
15:08 PrashantS joined #openstack-rally
15:09 rvasilets yfried, did you tried to run with rps less then 1? https://github.com/openstack/rally/blob/master/rally/plugins/common/runners/rps.py#L41
15:09 yfried rvasilets: no. would it work https://github.com/openstack/rally/blob/master/rally/plugins/common/runners/rps.py#L120-L123 ?
15:10 rvasilets yfried, I don't know. But its easy to check)
15:11 yfried ushkalim: ^ plz try
15:11 ivasev yfried: not sure about that
15:11 ivasev if it was me i will add a config for the user to chose :)
15:12 ivasev depending on what is available in the catalog, because it seems rally chose instead of me, and i have to change the code
15:12 yfried ivasev: are you talking about using rally to DEPLOY openstack? (currently only with fuel, AFAIK)
15:13 yfried ivasev: I think you are talking about clients and not backends
15:13 ivasev yes clients
15:13 yfried ivasev: "use keystone v3 if endpoint exists" ?
15:15 yfried ushkalim: https://github.com/openstack/rally/blob/master/rally/plugins/common/runners/rps.py#L29-L93
15:15 ivasev yes, or at very least i did not find a way to reconfigure it
15:15 yfried ushkalim: this is the logic rps uses to fire iterations. I have to go. please continue with amaretskiy
15:15 yfried ivasev: not sure. let me take a look
15:18 yfried ivasev: https://github.com/openstack/rally/blob/master/rally/plugins/openstack/scenarios/keystone/utils.py#L41-L43 seems like v3 isn't supported yet
15:18 yfried ivasev: you are welcome to either contribute to it or contact boris-42 with RFE
15:18 yfried ivasev: sorry I can't help you further
15:18 ivasev yfried: i was wondering for glance v2
15:19 ivasev thanks a ton for helping me out!
15:21 yfried ivasev: https://github.com/openstack/rally/blob/master/rally/osclients.py
15:22 yfried ivasev: looks like some of the code allows for init of different client version. not sure how to trigger it from user's end
15:22 boris-42 amaretskiy: yfried about docs
15:22 yfried ivasev: have to leave now. sorry
15:22 yfried boris-42: ?
15:22 ivasev thanks a ton yfried
15:22 boris-42 amaretskiy: yfried I would like to finish swithching to plugin base
15:22 boris-42 amaretskiy: yfried and move info funcationallity to plugin base
15:22 boris-42 amaretskiy: yfried after that we will generate automatically docs about runners and other stuff
15:23 boris-42 amaretskiy: yfried so am I rigth you would like to have runner that runs bunch of items each X seconds
15:23 boris-42 iterations*
15:24 openstackgerrit Sergey Skripnick proposed openstack/rally: Add new script for rally-gate jobs  https://review.openstack.org/175549
15:24 yfried boris-42: yep
15:24 yfried boris-42: seems like rps should be able to do that
15:24 boris-42 yfried: nope
15:24 yfried boris-42:?
15:24 boris-42 yfried: it is another type of runner
15:25 boris-42 yfried: rps just runs iterations each 1/RPS second
15:25 yfried boris-42: doesn't RPS mean N requests per T seconds?
15:25 openstackgerrit Kirill Shileev proposed openstack/rally: Add CobblerProvider provider  https://review.openstack.org/188847
15:25 boris-42 per 1 second
15:25 boris-42 redixin: ^ you shoudl review this
15:25 yfried boris-42: well, if not currently, it should be easily extended
15:26 boris-42 rvasilets: kir patch
15:26 boris-42 yfried: so it is just different type of load
15:26 redixin boris-42, sure
15:26 boris-42 yfried: rps runs jsut each 1/RPS one iteration
15:26 rvasilets rvasilets, you want me to review Kirill patch?
15:27 yfried boris-42: s/"1"/"user input"/
15:27 rvasilets Does RPS should be integer only?
15:27 yfried boris-42: and you're done
15:27 boris-42 rvasilets: I was tyrping redixin
15:27 boris-42 yfried: ?
15:27 boris-42 yfried: we don't need that
15:27 yfried boris-42: I must leave. can we discuss this early tomorrow?
15:27 boris-42 yfried: just put in RPS: 0.5 / 0.3 or something else
15:28 rvasilets boris-42, I 've member that there was a patch abouit 0<RPS<1 but not remember what it did
15:28 yfried boris-42: in 15hrs
15:28 boris-42 and you will get 1 iteration per 2 seconds
15:28 boris-42 yfried: ok
15:28 yfried ushkalim: ^^
15:28 boris-42 yfried: see you tomorrow just ping me
15:28 yfried boris-42: if you're online
15:28 yfried boris-42: bye
15:28 rvasilets boris-42, looks like I was right)
15:29 kiran-r joined #openstack-rally
15:29 ushkalim yfried, ;)
15:29 arxcruz joined #openstack-rally
15:47 ushkalim boris-42, amaretskiy  http://paste.openstack.org/show/412388/ debug log    http://paste.openstack.org/show/412389/   config
15:48 ushkalim worker1 is working as expected every 2 sec
15:49 ushkalim but worker 2 and worker 3 are not
15:53 slo_ joined #openstack-rally
16:07 puranamr joined #openstack-rally
16:12 exploreshaifali joined #openstack-rally
16:13 e0ne joined #openstack-rally
16:30 PrashantS joined #openstack-rally
16:38 yfried joined #openstack-rally
16:39 PrashantS joined #openstack-rally
16:47 e0ne joined #openstack-rally
17:05 PrashantS joined #openstack-rally
17:21 puranamr joined #openstack-rally
17:30 PrashantS joined #openstack-rally
17:32 PrashantS joined #openstack-rally
17:39 boris-42 ^ uh seems bad
17:39 boris-42 ^ let me recheck
17:39 dspano joined #openstack-rally
17:44 abehl joined #openstack-rally
17:57 pvaneck joined #openstack-rally
17:58 psuriset joined #openstack-rally
17:59 PrashantS joined #openstack-rally
18:06 e0ne joined #openstack-rally
18:11 flwang joined #openstack-rally
18:12 yfried joined #openstack-rally
18:21 e0ne joined #openstack-rally
18:36 tfreger joined #openstack-rally
18:56 luz_ joined #openstack-rally
19:00 temujin_ joined #openstack-rally
19:11 exploreshaifali joined #openstack-rally
19:19 boris-42 kun_huang: around?
19:20 boris-42 meteorfox: ping
19:20 meteorfox boris-42: hey
19:41 openstackgerrit Oleh Anufriiev proposed openstack/rally: Update rally/contrib to use install_rally.sh  https://review.openstack.org/210137
19:43 echoingumesh joined #openstack-rally
19:44 echoingu_ joined #openstack-rally
19:58 boris-42 meteorfox: could you please help with review of patch
19:58 boris-42 meteorfox: this one https://review.openstack.org/#/c/209046/
19:58 meteorfox boris-42: sure, taking a look now
19:59 echoingumesh joined #openstack-rally
19:59 boris-42 meteorfox: it's just nice refactoring
20:26 meteorfox boris-42: my only 'concern' with that patch would be that it seems the code does 2 things, it consumes the results but it also checks them for SLA. Maybe in the future the SLA check can be refactored out, and perhaps using generators, or something similar, yield the results to another function or object, that runs the checks or any other stuff
20:27 meteorfox boris-42: I know that it was already like that, and he just refactored it to use a ContextManager
20:30 boris-42 meteorfox: seems like harder
20:30 boris-42 meteorfox: than current solution
20:31 meteorfox boris-42: yeah, maybe if one day we have more 'validators' like the SLA checks, then it might be worth pursuing something like that
20:32 boris-42 meteorfox: so I am refactoring validators
20:32 boris-42 meteorfox: but it is to heavy task
20:37 openstackgerrit Merged openstack/rally: Updated from global requirements  https://review.openstack.org/210865
21:00 puranamr joined #openstack-rally
21:21 dspano joined #openstack-rally
21:54 echoingumesh joined #openstack-rally
21:57 mwagner_lap joined #openstack-rally
22:02 echoingumesh joined #openstack-rally
22:16 dspano joined #openstack-rally
22:19 openstackgerrit Boris Pavlovic proposed openstack/rally: Make context plugins decide how to map their data for scenario  https://review.openstack.org/203911
22:22 MaxPC joined #openstack-rally
22:43 openstackgerrit Boris Pavlovic proposed openstack/rally: Make context plugins decide how to map their data for scenario  https://review.openstack.org/203911
23:19 echoingumesh joined #openstack-rally
23:30 baker joined #openstack-rally
23:30 e0ne joined #openstack-rally

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