Thursday, 2017-05-04

« Wednesday, 2017-05-03 Index Friday, 2017-05-05 »
*** yingjun has joined #openstack-rally00:08
*** winston-d_ has joined #openstack-rally00:46
*** tovin07_ has joined #openstack-rally00:46
*** jkilpatr has quit IRC01:02
*** dave-mccowan has joined #openstack-rally01:03
*** chhavi has joined #openstack-rally01:11
*** yolkfull__ has joined #openstack-rally01:13
winston-d_hi, I have some questions about 'rps' runner.01:18
winston-d_Anybody here is familiar with how 'rps' runner works?01:19
*** yanyanhu has joined #openstack-rally01:27
*** yamamoto has joined #openstack-rally01:31
*** dixiaoli has joined #openstack-rally01:33
*** yamamoto has quit IRC01:35
*** ilbot3 has quit IRC01:46
*** ilbot3 has joined #openstack-rally01:49
*** dave-mccowan has quit IRC02:10
*** chhavi has quit IRC02:13
*** yolkfull has joined #openstack-rally02:13
*** yolkfull_ has joined #openstack-rally02:15
*** yolkfull__ has quit IRC02:16
*** yolkfull__ has joined #openstack-rally02:17
*** yolkfull has quit IRC02:18
*** yolkfull_ has quit IRC02:19
*** dave-mccowan has joined #openstack-rally02:20
*** yolkfull__ has quit IRC02:23
*** yolkfull__ has joined #openstack-rally02:23
*** yamamoto has joined #openstack-rally02:36
*** chhavi has joined #openstack-rally02:46
*** yamamoto has quit IRC02:46
*** dave-mccowan has quit IRC02:51
*** zhubingbing_ has joined #openstack-rally02:55
openstackgerritchenhb proposed openstack/rally master: [WIP]Refactor the related command of task report  https://review.openstack.org/46037403:04
*** shasha_t- has joined #openstack-rally03:06
*** tovin07_ has quit IRC03:16
*** tovin07_ has joined #openstack-rally03:17
*** dixiaoli has quit IRC03:30
*** chhavi has quit IRC04:08
openstackgerritchenhb proposed openstack/rally master: [WIP]Refactor the related command of task report  https://review.openstack.org/46037405:09
*** shausy has joined #openstack-rally05:11
*** yamamoto has joined #openstack-rally05:18
*** yamamoto has quit IRC05:21
*** Serlex has joined #openstack-rally05:33
*** rdas has joined #openstack-rally05:40
*** anshul has joined #openstack-rally05:57
*** yamamoto has joined #openstack-rally06:21
*** yamamoto has quit IRC06:26
*** yamamoto has joined #openstack-rally06:26
*** yamamoto has quit IRC06:28
*** pcaruana has joined #openstack-rally06:45
*** e0ne has joined #openstack-rally06:52
*** yamamoto has joined #openstack-rally06:59
*** yamamoto has quit IRC07:09
*** aarefiev_afk is now known as aarefiev07:24
*** tesseract has joined #openstack-rally07:29
*** aojea has joined #openstack-rally07:48
*** ushkalim_ has quit IRC07:49
openstackgerritStaroverov Anton proposed openstack/rally master: Move 'image-valid-on-flavor' validator to 'openstack/validators.py'  https://review.openstack.org/45726308:00
openstackgerritStaroverov Anton proposed openstack/rally master: Move 'required_neutron_extension' validator to '.../validators.py'  https://review.openstack.org/45926808:15
openstackgerritchenhb proposed openstack/rally master: [WIP]Refactor the related command of task report  https://review.openstack.org/46037408:24
*** e0ne has quit IRC08:24
*** chhavi has joined #openstack-rally08:28
*** ushkalim_ has joined #openstack-rally08:52
openstackgerritStaroverov Anton proposed openstack/rally master: [WIP] Move 'required_contexts' validator to '../validators.py'  https://review.openstack.org/46218608:55
*** tosky has joined #openstack-rally09:11
*** tosky has quit IRC09:11
*** tosky has joined #openstack-rally09:12
*** yamamoto has joined #openstack-rally09:15
*** tovin07 has quit IRC09:18
*** yamamoto has quit IRC09:19
*** yingjun has quit IRC09:31
*** e0ne has joined #openstack-rally09:31
openstackgerritStaroverov Anton proposed openstack/rally master: [WIP] Move 'required_clients' validator to '../validators.py'  https://review.openstack.org/46216709:34
*** rallydev-bot has joined #openstack-rally09:34
rallydev-bot[From Gitter] andreykurilin : @csavignan : hi! I see no problems with. go ahead and update requirements file09:35
rallydev-bot[From Gitter] andreykurilin : INFO: rallydev-bot crashed this night, so we lost synchronization with IRC channel. But now everything should be ok.09:36
rallydev-bot[From Gitter] andreykurilin : winston-d_: hi! How can I help you with RPS runner?09:36
openstackgerritTuan Luong-Anh proposed openstack/rally master: Using fixtures instead of deprecated mockpatch module  https://review.openstack.org/46246509:50
*** mvk has quit IRC10:03
openstackgerritCedric Savignan proposed openstack/rally master: Adds Networking-bgpvpn scenarios  https://review.openstack.org/46046510:03
*** tovin07_ has quit IRC10:04
openstackgerritchenhb proposed openstack/rally master: [WIP]Refactor the related command of task report  https://review.openstack.org/46037410:20
*** mvk has joined #openstack-rally10:30
*** yanyanhu has quit IRC10:31
openstackgerritchenhb proposed openstack/rally master: [WIP]Refactor the related command of task report  https://review.openstack.org/46037410:40
*** alexchadin has joined #openstack-rally10:44
*** mrhillsman has quit IRC10:48
*** mrhillsman has joined #openstack-rally10:49
*** jkilpatr has joined #openstack-rally10:53
*** yolkfull__ has quit IRC10:58
winston-d_andreykurilin: hi, I'd like to know how to properly set the 'duration' parameter for RPS runner.10:59
*** chhavi has quit IRC11:00
winston-d_andreykurilin: for example, I'd like to do a loadline test, which RPS starts at 1, ends at 64 and step is 1, what is the relationship between 'times' and 'duration'?11:00
openstackgerritMerged openstack/rally master: Move 'required_neutron_extension' validator to '.../validators.py'  https://review.openstack.org/45926811:01
winston-d_andreykurilin: a sample task YAML with RPS runner: https://paste2.org/FV80YAhW11:03
winston-d_I briefly skim the code, but not sure if I understand the way RPS runner works.11:06
rallydev-bot[From Gitter] andreykurilin : winston-d_: Sorry, you should not read the code, we need to improve https://rally.readthedocs.io/en/latest/plugins/plugin_reference.html to cover all properties with good descriptions.11:07
rallydev-bot[From Gitter] andreykurilin : ok11:07
rallydev-bot[From Gitter] andreykurilin : `times` is a total number of seconds for genarating load.11:07
rallydev-bot[From Gitter] andreykurilin : `duration` is a duration of  the particular step.11:07
winston-d_There are multiple factors may affect how many requests per second RPS runner can issue to a deployment, besides 'start/end/step/duration', there are also 'max_concurrency', # of CPUs, etc.11:07
*** dave-mccowan has joined #openstack-rally11:08
rallydev-bot[From Gitter] andreykurilin : Example: start=X, end=Y, step=Z, duration=M, times=N, max_concurrency=K11:08
winston-d_So the unit for 'start/end/step' is request/sec, for 'duration' and 'times' is seconds?11:09
rallydev-bot[From Gitter] andreykurilin : ok, forgot about max_concurrency) I do not know why we have that parameter11:09
rallydev-bot[From Gitter] andreykurilin : heh11:09
rallydev-bot[From Gitter] andreykurilin : so as for  start=X, end=Y, step=Z, duration=M, times=N, RPS runner will start genarating X request per second during M seconds, after M seconds is reached, RPS runner will increase number of request per second and it will be X+Z. That number of request per seconds will be used again for M seconds.11:11
rallydev-bot[From Gitter] andreykurilin : Number of seconds will be increased each M seconds until N (total time) or the upper limit Z is reached11:12
rallydev-bot[From Gitter] andreykurilin : if Z number of request per second is achieved, but N is not, RPS will generate Z request per second till the end11:13
winston-d_Ah, I see.  But 'constant' runner, the unit for runner is total requests, right?11:13
rallydev-bot[From Gitter] andreykurilin : yes11:13
openstackgerritStaroverov Anton proposed openstack/rally master: [WIP] Move 'required_clients' validator to '../validators.py'  https://review.openstack.org/46216711:13
winston-d_Same parameter but different unit, I have to say it's confusing.11:13
rallydev-bot[From Gitter] andreykurilin : winston-d_: we are trying to document as much as possible. constant runner has simple input data, so it was easy to auto-generate docs for it - https://rally.readthedocs.io/en/latest/plugins/plugin_reference.html#constant-scenario-runner (all parameters have clear descriptions there).11:14
rallydev-bot[From Gitter] andreykurilin : As for RPS runner, it has complex logic and our jsonschema->docs convertor did not succed with it11:15
rallydev-bot[From Gitter] andreykurilin : I'll try to fix it soon, so everything will have documentation11:15
winston-d_Thanks for the pointer to doc.  I was looking at https://docs.openstack.org/developer/rally/task/index.html#id2 and it doesn't have as much infor as readthedocs.io one.11:16
openstackgerritMerged openstack/rally master: [image] Extract the common function  https://review.openstack.org/46145711:16
rallydev-bot[From Gitter] andreykurilin : winson-d: docs.openstack.org and readthedocs has the same content11:16
rallydev-bot[From Gitter] andreykurilin : but I like readthedocs theme more:)11:17
rallydev-bot[From Gitter] andreykurilin : I started(https://review.openstack.org/#/c/448027/) work on modifying task page at docs, so it will be more easier to find all info11:17
winston-d_I didn't the description for task runners at https://docs.openstack.org/developer/rally/index.html11:18
winston-d_I didn't find the description for task runners at https://docs.openstack.org/developer/rally/index.html11:19
rallydev-bot[From Gitter] andreykurilin : winston-d_: task page is quite "empty" now. You should follow Plugins->Plugins reference11:19
openstackgerritStaroverov Anton proposed openstack/rally master: Move 'image-valid-on-flavor' validator to 'openstack/validators.py'  https://review.openstack.org/45726311:21
winston-d_hmm, yeah, it's hidden a few levels under the first level section title.11:21
winston-d_Thanks for the help!  Let me try to start a RPS runner.11:21
rallydev-bot[From Gitter] andreykurilin : yes:(11:22
rallydev-bot[From Gitter] andreykurilin : task page will look like "verification" page in near future - https://rally.readthedocs.io/en/latest/verification/index.html11:22
rallydev-bot[From Gitter] andreykurilin : we made a good progress there, I think11:22
rallydev-bot[From Gitter] andreykurilin : so please be patient and I hope in near future we will make our docs great again:)11:23
winston-d_Yes, that'll be much easier to read.11:24
winston-d_Anyway, I started a task with RPS runner and this is how it looks like now: http://paste.openstack.org/show/608847/11:25
winston-d_See the 'start/end/step' is '1/48/1' and 'times' is 300, 'duration' is 5.11:25
winston-d_But look the first 8 requests, they started at the same time.11:26
winston-d_I guess 'max_concurrency' plays a role here.11:26
winston-d_I was expected to see the first 5 seconds, only 5 requests would be sent to deployment.11:28
openstackgerritStaroverov Anton proposed openstack/rally master: [WIP] Move 'required_contexts' validator to '../validators.py'  https://review.openstack.org/46218611:33
winston-d_The 'times' parameter doesn't really seem like to control total duration of time (unit: seconds), but instead it still looks like has the unit of 'requests'11:35
winston-d_andreykurilin: http://paste.openstack.org/show/608848/ I added the result of this run.  As you can see, 'Count' = 300, which matches the value I set to 'times'11:38
openstackgerritStaroverov Anton proposed openstack/rally master: [WIP] Moving of 'required_cinder_services' validator to '../validators.py'  https://review.openstack.org/46204211:43
*** catintheroof has joined #openstack-rally11:47
*** catintheroof has quit IRC11:51
openstackgerritAndrey Kurilin proposed openstack/rally master: Fix an issue with namespace of scenario  https://review.openstack.org/46251611:56
rallydev-bot[From Gitter] andreykurilin : winston-d_: I think the problem is in our calculation of the number of request per second which should generate single rally process. By default rally tries to use all processes12:30
rallydev-bot[From Gitter] andreykurilin : try to setup `max_cpu_count` to `12:30
rallydev-bot[From Gitter] andreykurilin : to 112:31
*** chhavi has joined #openstack-rally12:32
openstackgerritStaroverov Anton proposed openstack/rally master: [WIP] Move 'required_services' validator to 'openstack/validators.py'  https://review.openstack.org/45880912:36
*** abalutoiu has joined #openstack-rally12:38
*** abalutoiu has quit IRC12:41
*** alexchadin has quit IRC12:54
*** yamamoto has joined #openstack-rally12:54
*** alexchadin has joined #openstack-rally13:03
winston-d_andreykurilin: how can I change the value of 'max_cpu_count'?13:04
rallydev-bot[From Gitter] andreykurilin : winston-d_: in the same way as times, step and etc13:07
*** yamamoto has quit IRC13:22
*** yamamoto has joined #openstack-rally13:24
openstackgerritMathieu Velten proposed openstack/rally master: Magnum: Move keypair handling at the cluster level  https://review.openstack.org/45013613:27
openstackgerritStaroverov Anton proposed openstack/rally master: [WIP] Move 'required-param-or-context' validator to '../validators.py'  https://review.openstack.org/46255013:27
openstackgerritCedric Savignan proposed openstack/rally master: Adds Networking-bgpvpn scenarios  https://review.openstack.org/46046513:35
openstackgerritCedric Savignan proposed openstack/rally master: Adds Networking-bgpvpn scenarios  https://review.openstack.org/46046513:49
*** rdas has quit IRC14:07
openstackgerritMerged openstack/rally master: Move 'image-valid-on-flavor' validator to 'openstack/validators.py'  https://review.openstack.org/45726314:09
*** alexchadin has quit IRC14:10
*** catintheroof has joined #openstack-rally14:20
*** e0ne has quit IRC14:23
*** catinthe_ has joined #openstack-rally14:23
*** catintheroof has quit IRC14:24
*** e0ne has joined #openstack-rally14:26
*** e0ne has quit IRC14:31
*** e0ne_ has joined #openstack-rally14:31
winston-d_andreykurilin: 'max_cpu_count' doesn't change anything: http://paste.openstack.org/show/608871/14:36
winston-d_still 3 threads began to work at the same time, and total request is 300, is much less than what it supposed to do. (300 seconds, scaling from 1 req/s to 48 req/s).14:38
winston-d_er, I meant to say 8 threads.14:42
*** yamamoto has quit IRC14:44
andreykurilinok. rps runner generate each second new number of requests and it doesn't wait for finishing old launched iterations14:48
andreykurilinthis means that in case of long operations you can have the number of not finished iterations ~~> times14:49
*** yingjun has joined #openstack-rally15:00
openstackgerritStaroverov Anton proposed openstack/rally master: [WIP] Move required-api-versions validator to '../validators.py'  https://review.openstack.org/46259215:09
winston-d_I will try to do a few more rounds with different combination of all parameters.15:11
winston-d_right now, I am doing a round with max_cpu_count and max_concurrency both equals 1.15:11
andreykurilinmax_concurrency should not help in your case15:12
winston-d_But I still observed multiple requests being sent at the same time at the very beginning, with 'start' = 1.15:12
openstackgerritStaroverov Anton proposed openstack/rally master: [WIP] Move 'required-param-or-context' validator to '../validators.py'  https://review.openstack.org/46255015:17
winston-d_andreykurilin: if 'times' = 10, 'start' = 1, 'end' = 10, 'duration' = 1, max_cpu_count/max_concurrency = 1, what's the total number of request rally should send out?15:19
*** oanufrii has quit IRC15:19
winston-d_andreykurilin: it'd be (10+1)*10/2 = 55, right?15:20
openstackgerritStaroverov Anton proposed openstack/rally master: Move 'required-param-or-context' validator to '../validators.py'  https://review.openstack.org/46255015:20
winston-d_andreykurilin: see this full running log: http://paste.openstack.org/show/608883/15:23
winston-d_rally only issued 10 requests, which proves that 'times' is controlling how many total requests should be sent out during the entire RPS run.15:24
winston-d_it shouldn't have the unit of second.15:24
winston-d_And no matter what value we provided to 'max_cpu_count' and 'max_concurrency', it looks like RPS always use 8 threads.15:25
winston-d_I don't think 8 is a magic number, instead my VM has 8 vcpus, I think that should be the reason I am seeing 8 threads.15:25
winston-d_Still, RPS runner seems like a mystery to me.15:26
winston-d_:/15:26
*** aarefiev is now known as aarefiev_afk15:27
*** yamamoto has joined #openstack-rally15:45
*** e0ne_ has quit IRC15:56
*** yamamoto has quit IRC15:57
andreykurilinwinston-d_: I'll try to digg into RPS runner tomorrow and experiment a bit16:04
*** mvk has quit IRC16:11
*** shausy has quit IRC16:13
*** yamamoto has joined #openstack-rally16:26
*** anshul has quit IRC16:26
*** mrhillsman has quit IRC16:35
*** yamamoto has quit IRC16:37
*** pcaruana has quit IRC16:40
*** ushkalim_ has quit IRC16:42
*** e0ne has joined #openstack-rally16:45
*** mrhillsman has joined #openstack-rally16:46
*** jkilpatr has quit IRC16:48
*** jkilpatr has joined #openstack-rally17:04
rallydev-bot[From Gitter] shihai1991 : @andreykurilin HI, andrey. If some error could be used in cli or server,  we shoud use `4**` or `5**`?17:05
rallydev-bot[From Gitter] andreykurilin : @shihai1991 hi! It depends on exception. If it is NotFound(ResourceNotFound and etc), it  should be 404. So actually, it doesn't metter for Exceptionwhere it is raisen. If it is problem with some input data - it is 4**.17:08
*** e0ne has quit IRC17:09
*** aojea has quit IRC17:10
*** aojea has joined #openstack-rally17:11
*** e0ne has joined #openstack-rally17:12
rallydev-bot[From Gitter] shihai1991 : In my opinion, we should avoid use the code which have been defined in common. And use other status_code to define the error.17:14
*** aojea has quit IRC17:15
*** tesseract has quit IRC17:30
*** yingjun has quit IRC17:48
*** tosky has quit IRC17:54
*** harlowja has quit IRC18:19
rallydev-bot[From Gitter] r-sood : @chenhb-zte  @andreykurilin  @shihai1991 Link with custom scenario and KeyError http://paste.openstack.org/show/608896/18:30
*** zhubingbing_ has quit IRC19:02
*** zhubingbing_ has joined #openstack-rally19:02
*** chhavi has quit IRC19:08
*** aojea has joined #openstack-rally19:27
*** dave-mccowan has quit IRC19:30
rallydev-bot[From Gitter] andreykurilin : @r-sood: I have a quick fix for you19:40
rallydev-bot[From Gitter] andreykurilin : It will work for both new and old rally releases19:40
rallydev-bot[From Gitter] andreykurilin : @r-sood you should put `namespace="openstack"` into configure decorator19:42
*** dave-mccowan has joined #openstack-rally19:43
*** dave-mcc_ has joined #openstack-rally19:45
rallydev-bot[From Gitter] r-sood : @andreykurilin  do you mean something like @context.configure(name="pe", namespace="openstack", order=1000)19:48
*** harlowja has joined #openstack-rally19:48
*** dave-mccowan has quit IRC19:48
*** aojea has quit IRC19:50
*** aojea has joined #openstack-rally19:50
*** aojea_ has joined #openstack-rally19:52
*** aojea has quit IRC19:53
rallydev-bot[From Gitter] andreykurilin : in your case it is `@scenario.configure(namespace="openstack")`20:00
rallydev-bot[From Gitter] andreykurilin : @r-sood ^20:00
rallydev-bot[From Gitter] r-sood : @andreykurilin Thanks! that worked20:02
*** aojea_ has quit IRC20:19
*** dave-mcc_ has quit IRC20:34
*** Serlex has quit IRC20:46
*** e0ne has quit IRC20:59
*** psuriset has joined #openstack-rally21:04
*** psuriset has quit IRC21:26
*** mvk has joined #openstack-rally21:51
*** winston-d_ has quit IRC22:53
*** catinthe_ has quit IRC23:08
*** jkilpatr has quit IRC23:38
*** jkilpatr has joined #openstack-rally23:39
« Wednesday, 2017-05-03 Index Friday, 2017-05-05 »

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!