Perl 6 - the future is here, just unevenly distributed

IRC log for #openstack-rally, 2015-03-12

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

All times shown according to UTC.

Time Nick Message
00:07 dmorita joined #openstack-rally
00:10 openstackgerrit Oleh Anufriiev proposed stackforge/rally: WIP removed task validation from api.Task.start  https://review.openstack.org/158899
00:15 boris-42 joined #openstack-rally
00:17 boris-42 meteorfox: yep better support of benchmarking is work in progress
00:17 boris-42 meteorfox: foreverlongtask
00:18 meteorfox boris-42: cool, yeah I know.
00:19 meteorfox boris-42: hey about the (sec) suggestion? what do you think of this? https://gist.github.com/meteorfox/238136329a9c34ad54e3
00:20 boris-42 meteorfox: looks great
00:21 boris-42 meteorfox: btw maybe we shoul fix these 0.0 to 0.00
00:21 boris-42 or even 0.000
00:21 boris-42 just to unify it with max column
00:21 mwagner_lap joined #openstack-rally
00:22 meteorfox boris-42: ok, I'll fix that, like in the html report.
00:22 meteorfox boris-42: how about this version: https://gist.github.com/meteorfox/238136329a9c34ad54e3#file-gistfile2-txt
00:22 boris-42 ^ first one is more cool=)
00:23 boris-42 meteorfox: ^
00:23 boris-42 meteorfox: apple design lol
00:23 meteorfox boris-42: ok. does cliutils let's me do a header on the table like that?
00:24 meteorfox boris-42: or will I have to write it
00:24 meteorfox ?
00:24 boris-42 meteorfox: I thought you know how to that
00:24 boris-42 meteorfox: lol=)
00:24 meteorfox boris-42: c'mon I know how to do it
00:25 meteorfox boris-42: I was just asking if there was a function to do it already in cliutils =)
00:26 meteorfox boris-42: hey about the avg column, I feel that should separate the other metrics. I feel it should go before the min, or after max
00:30 boris-42 why?)
00:31 boris-42 meteorfox: ^
00:31 meteorfox boris-42: avg is a measurement of central tendency, and not about order
00:31 meteorfox boris-42: min, median, %iles,max talk about order
00:34 boris-42 meteorfox: ok
00:34 boris-42 meteorfox: you operator not me=)
00:34 boris-42 meteorfox: I am just dev=)
00:35 meteorfox boris-42: =)
00:36 boris-42 meteorfox: so you have some amount of time to get your patch in before cut of new release
00:36 meteorfox boris-42: thanks, I'll add the header to the table, and I'll fix the rounding to 3 decimal places
00:36 meteorfox boris-42: when's the new release?
00:37 boris-42 meteorfox: it should be yestarday
00:37 boris-42 meteorfox: but I was too busy and didn't have enough time to make release notes
00:37 meteorfox meteorfox: lol
00:37 boris-42 meteorfox: today I started doing them
00:37 meteorfox boris-42: cool
00:37 boris-42 meteorfox: but didn't finish lol
00:37 boris-42 because to much changes for 39days
00:37 meteorfox boris-42: do you work on Rally full time?
00:38 boris-42 meteorfox: yep (even have team that works on Rally full time)
00:38 meteorfox boris-42: that's cool, with Mirantis right?
00:38 boris-42 meteorfox: yep
00:39 meteorfox boris-42: cool. are you guys using it at Mirantis?
00:39 boris-42 meteorfox: yep for scale/perf testing
00:40 boris-42 meteorfox: that is why I started doing releases
00:40 boris-42 meteorfox: they used for almost half year master=)
00:40 meteorfox boris-42: cool, so I started contributing again to Rally because of this project http://www.rackspace.com/cloud/private/openstack
00:41 meteorfox boris-42: it's pretty much our private cloud offering
00:41 boris-42 meteorfox: so just for testing or certification?
00:41 meteorfox boris-42: my job title really is Software Dev Test - Performance
00:42 meteorfox boris-42: our team does Performance engineering for pretty much all of Rackspace
00:43 meteorfox boris-42: as part of our Kilo release, we want to improve the performance and establish a baseline with Juno
00:43 meteorfox boris-42: I initially was going to use my own tools, or other out there, but based on the things I talked to the architects and management, Rally was a better fit
00:43 boris-42 meteorfox: so things goes worse?
00:44 boris-42 Juno->Kilo
00:44 meteorfox boris-42: we don't know, and that's the thing
00:44 boris-42 I never did such experiments
00:44 boris-42 meteorfox: if you share some results it will be fantastic blogpost=)
00:45 meteorfox boris-42: there hasn't been that much performance testing on previous versions
00:45 echoingumesh joined #openstack-rally
00:45 meteorfox boris-42: so customers ask specific things about the performance, and the truth is that we don't know =)
00:45 boris-42 meteorfox: typical situation=)
00:45 boris-42 meteorfox: just say "it's web scale"
00:46 meteorfox boris-42: so I got buy in from my managers, and leadership, and we are going all in with Rally
00:46 meteorfox boris-42: lol yeah
00:46 meteorfox boris-42: like MongoDb
00:46 boris-42 meteorfox: "MongoDB scale"
00:46 boris-42 =)
00:46 meteorfox boris-42: lol
00:46 meteorfox boris-42: do you remember Hugh, also from Rackspace?
00:47 boris-42 meteorfox: yep hew was Rally core=)
00:47 boris-42 meteorfox: hughsaunders
00:47 boris-42 meteorfox: btw he started VMTask pron
00:47 boris-42 =)
00:47 meteorfox boris-42: yep, so now I work with him
00:47 boris-42 meteorfox: ah nice)
00:47 meteorfox boris-42: he's at UK  though =)
00:48 boris-42 meteorfox: yep I know=)
00:48 meteorfox boris-42: he was pretty excited that I'm going to use Rally
00:48 boris-42 meteorfox: why he is not in IRC?)
00:48 meteorfox boris-42: I don't know, probably he's very busy with this project or something
00:48 meteorfox boris-42: or he has another nickname now
00:49 boris-42 meteorfox: he is in dev chat=)
00:49 boris-42 meteorfox: I think he is afraid to get some interesting task
00:50 boris-42 meteorfox: so he will need to work on it=)
00:50 meteorfox boris-42: lol
00:51 meteorfox boris-42: so my goal for this year is to become a Rally core contributor
00:51 meteorfox boris-42: I hope you'll help with that =)
00:52 boris-42 meteorfox: so yfried becomes recently =)
00:52 boris-42 become*
00:52 meteorfox boris-42: nice
00:52 boris-42 on Sunday=)
00:52 boris-42 meteorfox: just do reviews =)
00:53 meteorfox became*
00:53 meteorfox =)
00:53 boris-42 more good reviews = faster getting
00:53 boris-42 meteorfox: it's tooolate so it's okay=)
00:53 boris-42 meteorfox: 4a.m.=)
00:53 boris-42 not joke=)
00:53 meteorfox boris-42: wow, why are you not sleeping?
00:54 boris-42 meteorfox: working on release
00:54 meteorfox boris-42: at what time do you wake up?
00:54 boris-42 meteorfox: it depends=)
00:54 boris-42 meteorfox: today it was 11 a.m.=)
00:54 meteorfox boris-42: do you work from home?
00:54 boris-42 meteorfox: usually yes=) because it takes too much time to get to office
00:54 boris-42 meteorfox: http://stackalytics.com/report/contribution/rally/30 so this page is review perf tracker
00:55 meteorfox boris-42: ah I see. yeah. I work in the office, but I go there after 9, because of traffic
00:55 boris-42 meteorfox: I just don't want to spend time in transport =)
00:55 boris-42 meteorfox: it takes 1 hour in 1 direction
00:56 meteorfox boris-42: that's bad. that's 2 hours wasted
00:58 meteorfox boris-42: heh, I have 7 reviews. Let's see if I get to 20 before the week ends
01:00 meteorfox boris-42: quick question, once you become core, can you lose being core?
01:01 boris-42 meteorfox: yep
01:01 boris-42 meteorfox: it's enough just to stop reviewing or do small amount of reviews
01:01 boris-42 meteorfox: or do bad reviews (just for stats)
01:02 meteorfox boris-42: ah makes sense.
01:02 boris-42 meteorfox: if you are not reviewing, you can't track changes in code base => you can't make good reviews
01:03 meteorfox boris-42: yep, yep
01:03 meteorfox boris-42: about the HTML report, is there people working on it to improve it?
01:04 boris-42 meteorfox: amaretsky
01:04 boris-42 meteorfox: https://review.openstack.org/#/c/159458/ and https://review.openstack.org/#/c/146814/
01:04 boris-42 meteorfox: you are welcome to review them ^
01:05 meteorfox boris-42: cool I will.
01:05 meteorfox boris-42: I wanted to add like a density plot, or a histogram
01:06 boris-42 meteorfox: they refactor all chart generations to use only additive algorithms
01:06 boris-42 meteorfox: he and me actually*
01:06 meteorfox boris-42: this is d3.js right?
01:06 boris-42 meteorfox: it's not so important
01:06 boris-42 meteorfox: the most important code is in python
01:06 boris-42 meteorfox: addtive algorithms for compression of graphs
01:06 boris-42 meteorfox: that works with chunks of data
01:07 meteorfox boris-42: ok, like "paginated" data
01:07 meteorfox ?
01:07 boris-42 meteorfox: you should generate all graphs by 1 iteration over all iterations data (that is chunked)
01:08 boris-42 meteorfox: otherwise you have scale issues=)
01:09 meteorfox boris-42: ok, yeah. it gets slow
01:09 meteorfox https://usercontent.irccloud-cdn.com/file/p2znAmbS/build_times_histograms.png
01:10 meteorfox boris-42: I want something like this
01:10 meteorfox boris-42: I did this with R
01:11 meteorfox https://usercontent.irccloud-cdn.com/file/xhdHGs4R/build_times_density_plots.png
01:11 meteorfox boris-42: and this is the same data, but in a density plot
01:11 meteorfox boris-42: is useful, because you can see how variance changes depending on the load
01:12 boris-42 meteorfox: so first part of work should be those 2 patches
01:12 boris-42 meteorfox: for amaretsky
01:12 meteorfox boris-42: cool
01:12 boris-42 meteorfox: then we can work on any other
01:12 boris-42 meteorfox: but
01:13 boris-42 meteorfox: you have strong limitations on algorithms
01:13 meteorfox https://usercontent.irccloud-cdn.com/file/Ri2AlpOj/build_times_boxplot.png
01:13 meteorfox boris-42: boxplot is also fine
01:13 boris-42 meteorfox: we need to use d3js=)
01:13 boris-42 meteorfox: or nvd3js
01:14 boris-42 meteorfox: because otherwise we will need to do everything from scratch
01:14 meteorfox boris-42: yeah
01:14 boris-42 which is quite long task+)
01:14 meteorfox boris-42: by the way, this data is from rally
01:15 boris-42 meteorfox: hehe=)
01:15 boris-42 meteorfox: so one more thing
01:15 meteorfox boris-42: I collected the "raw" data from the JSON report, and did the aggregation myself in R
01:15 boris-42 meteorfox: we will have parallel scenarios
01:15 meteorfox boris-42: right
01:15 boris-42 meteorfox: so it's one more complexity
01:15 boris-42 meteorfox: I mean Rally should work good when we have 10 iterations
01:16 boris-42 meteorfox: and 10kk iterations
01:16 boris-42 meteorfox: the result should be pretty report with graphs=)
01:16 meteorfox boris-42: yes, hmm.. but I think you'll have to do some sub sampling
01:17 boris-42 meteorfox: sub sampling?
01:17 meteorfox boris-42: sampling from the samples
01:17 boris-42 meteorfox: ?)
01:17 meteorfox boris-42: for example this libray can do it. http://dygraphs.com/
01:18 boris-42 meteorfox: nonon
01:18 boris-42 =)
01:18 meteorfox boris-42:  say you have 1000 samples, instead of showing all 1000, you show sub aggregations of it
01:19 meteorfox boris-42: 100 samples of average of 10
01:19 boris-42 meteorfox: it won't scale enough good
01:19 boris-42 meteorfox: you will get HTML pages with gigabytes of data
01:19 boris-42 meteorfox: what we are doing for now is merging points
01:20 boris-42 meteorfox: so even if you have 100k points they will be merged to 1k
01:20 boris-42 meteorfox: so in HTML report you have info about 1k points not 100k points
01:20 boris-42 meteorfox: which is enough to understand what is happening during the benchmark
01:20 meteorfox boris-42: oh, ok. ah I see
01:20 boris-42 and doesn't kill your browser
01:20 boris-42 =)
01:21 boris-42 meteorfox: so amaretsky is working on refactoring all such merging algorithms to works for one iteration over data
01:21 boris-42 to work*
01:23 boris-42 meteorfox: so developing in such way algorithms will be quite painful=)
01:23 meteorfox boris-42: hmm ok. using d3js, you could use something like this https://github.com/mbostock/d3/wiki/Requests#d3_json
01:23 meteorfox boris-42: or CSV https://github.com/mbostock/d3/wiki/CSV
01:23 boris-42 meteorfox: why?
01:24 boris-42 meteorfox: why do you need that?
01:24 meteorfox boris-42: you don't have to load it all
01:24 boris-42 meteorfox: one of the most important for users feature is to have single file that they can share with managers/customers
01:25 meteorfox boris-42: oh ok.
01:25 boris-42 meteorfox: I mean it's *really* important
01:26 boris-42 meteorfox: I think most important stuff in whole rally(no jokes)
01:26 boris-42 =)
01:26 meteorfox boris-42: so everything is just one big HTML file?
01:26 boris-42 meteorfox: yep
01:26 boris-42 meteorfox: and it should be small and fast working
01:27 boris-42 in any case
01:27 boris-42 you have 1 benchmark with 1kk iterations
01:27 boris-42 or you have 1000 benchmarks with 100 iterations
01:27 boris-42 meteorfox: we have before special command that plots custom graphs from results
01:28 boris-42 meteorfox: but it was complicated and nobody liked/used it
01:28 boris-42 everybody likes 1click solutions
01:28 meteorfox boris-42: yeah that's true
01:29 boris-42 meteorfox: so our goal is to implement everything in autogenerated single page one click without args solution
01:29 boris-42 =)
01:29 boris-42 it's not so simple but it is possible=)
01:29 meteorfox boris-42: for me as power user (perf engineer), I would like to have more detailed graphs. do you think we could something different? perhaps args for  report?
01:30 boris-42 meteorfox: I would like to start from other side
01:30 boris-42 meteorfox: what you would like to have on graphs and why
01:30 boris-42 meteorfox: then we can work on all stuff to make simple interface for that
01:30 boris-42 meteorfox: trying to avoid any choices/arguments
01:31 meteorfox boris-42: yeah. I understand
01:31 boris-42 meteorfox: btw you should keep in mind that we will have 2 more commnds
01:31 boris-42 meteorfox: "rally task compare" and "rally task trend"
01:31 meteorfox boris-42:  as long as I can get the raw data, I'm fine with it
01:31 boris-42 meteorfox: you will be always to do that
01:32 boris-42 meteorfox: be able *
01:33 meteorfox boris-42: yeah, compare and trends are awesome
01:33 boris-42 meteorfox: so now you know everything=)
01:34 boris-42 meteorfox: if you are ok with this let's work on future improvements in HTML reports=)
01:34 meteorfox boris-42: cool, cool. yes definitely
01:34 boris-42 meteorfox: I think that most of part that you need can be implemented even with limitations that we have.
01:35 meteorfox boris-42: awesome
02:41 echoingumesh joined #openstack-rally
02:45 echoingumesh joined #openstack-rally
02:57 echoingumesh joined #openstack-rally
02:58 bkopilov joined #openstack-rally
03:53 rdas joined #openstack-rally
04:27 nkhare joined #openstack-rally
04:34 echoingumesh joined #openstack-rally
04:53 neeti joined #openstack-rally
04:56 bkopilov joined #openstack-rally
05:05 omnibus7 joined #openstack-rally
05:05 kevinbenton_ joined #openstack-rally
05:08 rook-tmp_ joined #openstack-rally
05:09 rmk` joined #openstack-rally
05:09 frickler joined #openstack-rally
05:09 rmk joined #openstack-rally
05:10 aarefiev joined #openstack-rally
05:33 tfreger joined #openstack-rally
06:13 k4n0 joined #openstack-rally
06:35 echoingumesh joined #openstack-rally
06:50 rook-tmp joined #openstack-rally
06:51 yfried|pto joined #openstack-rally
06:56 yfried|pto joined #openstack-rally
07:01 nmagnezi_ joined #openstack-rally
07:08 exploreshaifali joined #openstack-rally
07:20 anshul joined #openstack-rally
07:26 anshul joined #openstack-rally
07:59 yfried boris-42: ping
08:27 tfreger1 joined #openstack-rally
08:37 openstackgerrit Ravikumar Venkatesan proposed stackforge/rally: Adds Neutron floatingip tests  https://review.openstack.org/160312
08:39 arxcruz joined #openstack-rally
08:55 kairat_kushaev joined #openstack-rally
09:04 boris-42 yfried: pong
09:05 neeti joined #openstack-rally
09:05 yfried boris-42: so you know how to create personalize gerrit boards?
09:07 boris-42 yfried: not an expert it this=(
09:07 boris-42 yfried: I am just using standard one
09:07 yfried boris-42: you know who is?
09:07 yfried boris-42: sdague...
09:08 boris-42 yfried: yep he is doing such stuff
09:08 boris-42 yfried: and infra guys
09:15 yfried boris-42: https://review.openstack.org/#/q/project:stackforge/rally+status:open+AND+NOT+owner:self+AND+NOT+label:Verified%253D-1+AND+NOT+label:workflow%253D-1,n,z
09:16 boris-42 yfried: I am reviewing anything
09:16 boris-42 yfried: sometimes people that get from Jenkins -1 are new and they need help with it
09:16 yfried boris-42: I know. but sometimes you are looking for merge candidates
09:22 amaretskiy joined #openstack-rally
09:36 echoingumesh joined #openstack-rally
09:49 boris-42 yfried: I changed settings of gerrit
09:49 boris-42 yfried|afk: it shows 50 patches / page
09:50 tfreger joined #openstack-rally
09:52 rdas joined #openstack-rally
09:55 e0ne joined #openstack-rally
10:10 e0ne joined #openstack-rally
10:18 yfried boris-42: https://review.openstack.org/#/c/159258/4..5/rally/benchmark/scenarios/swift/objects.py,cm
10:18 yfried boris-42: could you please explain your comment about atomic actions in loop?
10:22 boris-42 yfried: did you take a look in our CI gates how it looks?
10:23 boris-42 yfried: in real life
10:23 yfried boris-42: what do you mean?
10:23 boris-42 yfried: our CI (messages from Jenkins)
10:23 yfried boris-42: probably not. could you please explain?
10:24 boris-42 yfried: so Jenkins post messages to every patch
10:24 boris-42 yfried: there are gate-rally-dsvm* jobs they are functional testing
10:25 boris-42 yfried: you can read this https://github.com/stackforge/rally/tree/master/tests/ci for more details
10:25 boris-42 yfried: in any case if you click on gate-rally-dsvm-neutron on that patch you'll see https://github.com/stackforge/rally/tree/master/tests/ci
10:25 boris-42 http://logs.openstack.org/58/159258/4/check/gate-rally-dsvm-neutron-rally/b1a3610/ sry
10:25 boris-42 you can click on HTML report and see all benchmarks in action
10:26 boris-42 yfried: if you take a look at output of swift stuff it will look like: http://logs.openstack.org/58/159258/4/check/gate-rally-dsvm-neutron-rally/b1a3610/rally-plot/results.html.gz#/SwiftObjects.create_and_download_object/overview
10:26 boris-42 yfried: it works but it doesn't look good
10:26 andreykurilin_ joined #openstack-rally
10:27 yfried do you mean that you can't execute the same atomic action twice in the same scenario? (loop or no loop...)
10:28 boris-42 yfried: nope I don't mean that
10:28 boris-42 yfried: as you see you can
10:28 boris-42 yfried: but it looks bad and useless
10:28 boris-42 yfried: you have ScenarioRunner that does loop
10:30 oanufriev joined #openstack-rally
10:35 yfried boris-42: IIUC it loops but each time you execute the same atomic action you get "ACTION (<new num>)" don't you?
10:36 yfried boris-42: should be solved if we display action by exec time and not by name
10:42 boris-42 yfried: I mean it will work
10:42 boris-42 yfried: but it is not how the right way to write benchmarks
10:42 yfried boris-42: yeah, but the data is not too good
10:43 boris-42 yfried: you have runners to repeat actions, you shouldn't do that in iterations
10:43 boris-42 yfried: if you want to create 10 objects, that's find but then it's one action (creating 10 objects)
10:43 boris-42 and you are interested not in getting time of each one in each itteration
10:43 boris-42 but overall how duration of 10 objects is changed over time
10:46 yfried boris-42:  agreed, but what if you want to use the code that's already "atomic" 10 times?
10:46 yfried boris-42: what happens if you wrap the for in a "with atomic" syntax?
10:48 boris-42 yfried: nested atomic actions are not supported
10:49 yfried boris-42: so, you will have to duplicate the code of an already existing atomic, in order to iterate over it?
10:49 yfried boris-42: btw, nested and atomic shouldn't go together...
10:51 boris-42 yfried: hm
10:51 boris-42 yfried: nope you don't need
10:52 boris-42 yfried: just use directly AtomicAction context
10:52 boris-42 yfried: in method and everything will work
11:19 aix joined #openstack-rally
11:41 exploreshaifali joined #openstack-rally
11:47 yfried boris-42: https://review.openstack.org/#/c/160312/4/rally/benchmark/context/cleanup/resources.py,cm
11:48 yfried boris-42: q about cleanup order
11:49 yfried boris-42: since _neutron_order generator is used, what would be the CLEANUP order?
11:49 yfried 1. delete routers
11:49 yfried 2. delete FLIP
11:49 yfried 3. delete subnets
11:49 yfried boris-42: ^?
11:49 yfried or the other way?
11:50 boris-42 yfried: sry was not here=)
11:50 boris-42 yfried: it goes from lower to bigger
11:51 boris-42 floating_ip then subnet
11:51 yfried boris-42: but router first?
11:51 boris-42 yfried: yep
11:51 yfried boris-42: and the cleanup would delete all routers, and THEN all subnets?
11:52 boris-42 router, floatingip, subnets
11:52 boris-42 yep that is what we have on that review
11:53 yfried boris-42: tnx
11:53 openstackgerrit Sergey Skripnick proposed stackforge/rally: WIP: try to launch rally-gate.py  https://review.openstack.org/163785
11:58 openstackgerrit Roman Vasilets proposed stackforge/rally: Add murano benchmarks  https://review.openstack.org/137661
11:58 openstackgerrit Roman Vasilets proposed stackforge/rally: Add base for murano benchmarking  https://review.openstack.org/137650
12:03 cdent joined #openstack-rally
12:07 tfreger joined #openstack-rally
12:09 yfried boris-42: there's a BP for having cleanup search "rally_XXX" prefix, right? has it been implemented yet?
12:10 boris-42 yfried: you are welcome https://review.openstack.org/#/c/139643/
12:12 yfried boris-42: should a patch allow for names that don't match rally_XXXX?
12:13 yfried https://review.openstack.org/#/c/160312/4/rally/benchmark/scenarios/neutron/utils.py,cm L257
12:13 yfried boris-42: ^ if someone would use his own name in "network_dict" we'd have resource leakage
12:17 psd joined #openstack-rally
12:21 k4n0 joined #openstack-rally
12:21 openstackgerrit Roman Vasilets proposed stackforge/rally: Add murano benchmarks  https://review.openstack.org/137661
12:21 openstackgerrit Roman Vasilets proposed stackforge/rally: Add base for murano benchmarking  https://review.openstack.org/137650
12:23 boris-42 yfried: for now we don't have name format
12:25 yfried boris-42: so I shouldn't expect the code to enforce that?
12:25 yfried boris-42: np
12:27 boris-42 yfried: so naming patterns are quite large topic
12:27 boris-42 yfried: I don't see full picture yet
12:34 panbalag joined #openstack-rally
12:34 pboros joined #openstack-rally
12:51 kairat_kushaev joined #openstack-rally
12:52 mwagner_lap joined #openstack-rally
12:58 aix joined #openstack-rally
13:00 yfried_ joined #openstack-rally
13:03 yfried__ joined #openstack-rally
13:03 openstackgerrit Boris Pavlovic proposed stackforge/rally: Release 0.0.2  https://review.openstack.org/163801
13:06 openstackgerrit Roman Vasilets proposed stackforge/rally: Add murano benchmarks  https://review.openstack.org/137661
13:07 boris-42 yfried__: ^^
13:07 boris-42 yfried__: release 0.0.2 stuff
13:08 k4n0_ joined #openstack-rally
13:11 k4n0_ joined #openstack-rally
13:18 openstackgerrit Boris Pavlovic proposed stackforge/rally: Release 0.0.2  https://review.openstack.org/163801
13:29 tfreger joined #openstack-rally
13:44 e0ne joined #openstack-rally
13:52 dpaterson joined #openstack-rally
14:15 kairat_kushaev joined #openstack-rally
14:20 yfried__ joined #openstack-rally
14:27 yfried joined #openstack-rally
14:32 klindgren joined #openstack-rally
14:49 openstackgerrit Boris Pavlovic proposed stackforge/rally: Release 0.0.2  https://review.openstack.org/163801
14:54 obutenko joined #openstack-rally
14:55 panbalag joined #openstack-rally
14:59 yfried joined #openstack-rally
15:00 frobware Can I export rally results and import them on another machine for other ad-hoc processing?
15:01 openstackgerrit David Paterson proposed stackforge/rally: Added 'alias' decorator.  https://review.openstack.org/163861
15:03 panbalag boris-42, Hi
15:04 panbalag boris-42, yfried.. I'm going to start working on the rally verify code to support executing functional tests from projects like swift (Line 17 @ https://docs.google.com/spreadsheets/d/16DXpfbqvlzMFaqaXAcJsBzzpowb_XpymaK2aFY2gA2g/edit?pli=1#gid=0).. Since this is my first time contributing here, I have a couple of questions..
15:05 panbalag boris-42, yfried, What is the usual process? Do I need to update the doc (from discussion tab in the spreadsheet) with the detailed plan on what changes I'm going to make to the code?
15:05 panbalag boris-42, yfried, or Can I start working on the code directly?
15:07 bkopilov joined #openstack-rally
15:08 boris-42 panbalag: hi there
15:08 boris-42 panbalag: gimme few minutes I will reply
15:08 panbalag boris-42, sure.
15:12 yfried panbalag: boris-42: I think a small spec would be in order, what do you guys think?
15:12 yfried ^ a summary of the doc and a rough plan?
15:12 yfried boris-42: ^ or is it too much?
15:16 boris-42 yfried: so now we have spec mechanism
15:16 boris-42 yfried: So I agree that this discussion should be done in it
15:16 tosky joined #openstack-rally
15:17 boris-42 yfried: as a first step we can move what we have in google doc to spec?
15:17 boris-42 panbalag: ^
15:18 yfried boris-42: agreed
15:19 panbalag yfried,boris-42, sure. What exactly do you mean by moving from google doc to spec?
15:19 openstackgerrit Roman Vasilets proposed stackforge/rally: Add functional test for task samples  https://review.openstack.org/163870
15:19 panbalag boris-42, is there a different place to create the spec?
15:20 panbalag boris-42, or Shall I first update the google docs with a rough plan and we can proceed from there ?
15:20 yfried panbalag: https://review.openstack.org/#/c/159065/
15:21 yfried panbalag: so, googledocs have no official status
15:21 e0ne joined #openstack-rally
15:21 yfried panbalag: they are nice to collaborate between a few peers during planning
15:22 yfried panbalag: specs are the openstack convention for reviewing bp plans
15:22 yfried panbalag: you write them in RST format (many online tools to check RST)
15:22 yfried panbalag: and push them like any other patch.
15:23 boris-42 panbalag: yfried yep it really helps to discuss such stuff
15:23 panbalag yfried, so I need to create a new topic like the one you sent me and then start discussion there?
15:23 boris-42 panbalag: yfried ag I am using them in the moment to disucss new input format
15:23 yfried panbalag: so you can get a proper review and publication of your effort before you dive into code
15:23 boris-42 panbalag: yep you should create new file like that one
15:24 yfried panbalag: there should be a template in spec dir. just follow it
15:25 panbalag boris-42, ok let me create one and then I'll update you. I'll ping if I have any issues.
15:26 openstackgerrit David Paterson proposed stackforge/rally: Added 'alias' decorator  https://review.openstack.org/163861
15:26 boris-42 panbalag: this is template https://github.com/stackforge/rally/blob/master/doc/specs/template.rst
15:27 boris-42 panbalag: copy-paste it and change sections
15:35 meteorfox boris-42: did you even sleep?
15:35 meteorfox boris-42: =)
15:36 andreykurilin meteorfox: boris-42 doesn't know word(and action) sleep. it was deprecated since boris-40 and removed in boris-41
15:36 panbalag boris-42, Thanks. I'll do that.
15:38 meteorfox andreykurilin: lol I mean I was up until almost 5 am, finishing those release notes.
15:38 echoingumesh joined #openstack-rally
15:38 meteorfox boris-42: sorry, I mean him
15:39 meteorfox boris-42: not I, derp
15:39 meteorfox andreykurilin: ^^
15:39 andreykurilin :)
15:40 meteorfox andreykurilin: I can't type, and I haven't had my coffee yet
15:41 boris-42 meteorfox: lol
15:41 boris-42 so actually need to cut release
15:41 boris-42 and unblock work
15:43 meteorfox boris-42: :) but seriously, after this you should sleep or you are going to die young
15:44 boris-42 meteorfox: actually I sleep a bit=)
15:56 openstackgerrit Alexander Maretskiy proposed stackforge/rally: Add classes for benchmark HTML report charts  https://review.openstack.org/159458
16:00 openstackgerrit Roman Vasilets proposed stackforge/rally: Add functional test for task samples  https://review.openstack.org/163870
16:09 krisgan joined #openstack-rally
16:10 andreykurilin joined #openstack-rally
16:17 bkopilov joined #openstack-rally
16:42 openstackgerrit Alexander Maretskiy proposed stackforge/rally: Add chunks support and chart `Parallel iterations' to report  https://review.openstack.org/146814
16:43 openstackgerrit Merged stackforge/rally: Release 0.0.2  https://review.openstack.org/163801
16:44 openstackgerrit Marian Krcmarik proposed stackforge/rally: Add a maximum concurrency option to rps runner  https://review.openstack.org/140150
16:49 Miouge joined #openstack-rally
16:58 openstackgerrit David Paterson proposed stackforge/rally: Added 'alias' decorator  https://review.openstack.org/163861
17:09 boris-42 we made it Rally v.0.0.2 WOOT
17:11 aix boris-42, gj
17:21 vkmc :D
17:21 vkmc congrats guys
17:30 meteorfox alright!
17:30 meteorfox boris-42: great job!
17:33 andreykurilin_ joined #openstack-rally
17:42 yfried joined #openstack-rally
17:46 echoingumesh joined #openstack-rally
17:51 amalinow joined #openstack-rally
17:55 exploreshaifali joined #openstack-rally
18:00 amalinow hi all, I'm preparing benchmark in rally to measure live migration speed of multiple servers. Right now my test shows quite long timings. I supose it's because calling server.live_migrate(...) in loop is waiting for results for every server in this loop, and in my scenario it should call live migration asynchronously one after another. I'm stuck here, because I don't know how to make it async. Any advice? Could multiproce
18:06 mwagner_lap joined #openstack-rally
18:12 amalinow sorry, my bad, I realized that looking for valid host is taking long time, after prepering this task before actually calling live migration it started to work quite fast
18:21 bkopilov joined #openstack-rally
18:51 boris-42 amalinow: hi there
18:51 boris-42 meteorfox: yep yep we need to cut them every 2 weeks
18:59 boris-42 vkmc: thank you=)
18:59 boris-42 vkmc: how are you?)
18:59 vkmc :)
18:59 vkmc boris-42, fine and you?
18:59 boris-42 amalinow: so how can I help you?
19:00 boris-42 amalinow: I mean if you put in runner section propre values you will get live migration of multple VMs in parallel
19:00 boris-42 vkmc: great=)
19:00 boris-42 vkmc: are you going to attend next summit?)
19:00 vkmc boris-42, happy to hear
19:00 vkmc boris-42, yes sir
19:01 boris-42 vkmc: are you going to make some sessions related to trove?)
19:01 boris-42 zaqar?)
19:01 vkmc boris-42, this time with a map
19:01 vkmc I have a presentation for Trove... for the how to contribute track, looking forward for the results :)
19:02 vkmc boris-42, and you? will you be presenting about the features in Rally v0.0.2?
19:03 boris-42 vkmc: so actually I am not sure about extactly that version=)
19:03 boris-42 vkmc: because next release should be in 2-3 weeks=)
19:03 boris-42 vkmc: it was quite hard to write release notes for 45 days https://rally.readthedocs.org/en/latest/release_notes/v0.0.2.html =)
19:04 vkmc boris-42, I saw your blogpost :)
19:05 boris-42 vkmc: ya party hard
19:05 boris-42 vkmc: I need release manager lol+)
19:05 vkmc haha
19:12 openstackgerrit Carlos L. Torres proposed stackforge/rally: Adds median metric to summary report of task  https://review.openstack.org/161037
19:25 meteorfox boris-42: let's see if my median patch goes in before 0.0.3 =)
19:25 meteorfox boris-42: will versioning have semantics?
19:32 bkopilov joined #openstack-rally
19:36 boris-42 meteorfox: that is the goal
19:36 boris-42 meteorfox: http://semver.org/ to use these rules
19:36 boris-42 meteorfox: after we cut release 1.0.0 we will fully support it
19:38 meteorfox boris-42: cool
19:39 boris-42 meteorfox: we are just preparing everything
19:39 e0ne joined #openstack-rally
19:39 boris-42 meteorfox: deprecation mechanisms
19:40 boris-42 meteorfox: upgrades for DB
19:40 boris-42 meteorfox: and other sutff
19:40 boris-42 meteorfox: CI that tests all such stuff in all combinatiosn
19:40 openstackgerrit joined #openstack-rally
19:41 boris-42 meteorfox: so baby steps=)
20:05 exploreshaifali joined #openstack-rally
20:06 arxcruz joined #openstack-rally
20:21 boris-42 arxcruz: ping
20:34 e0ne joined #openstack-rally
20:58 echoingumesh joined #openstack-rally
21:04 harlowja joined #openstack-rally
21:10 meteorfox boris-42: I made the changes we talked about when printing the tables, what do you think? http://logs.openstack.org/37/161037/5/check/gate-rally-dsvm-cli/656b7b5/rally-cli-output-files/TaskTestCase/test_report_bunch_files.txt
21:22 echoingumesh joined #openstack-rally
21:23 echoingumesh joined #openstack-rally
21:26 exploreshaifali joined #openstack-rally
22:04 harlowja joined #openstack-rally
22:32 andreykurilin_ joined #openstack-rally
22:38 krisgan joined #openstack-rally
23:39 echoingu_ joined #openstack-rally
23:44 boris-42 meteorfox: looks nice=)

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