Perl 6 - the future is here, just unevenly distributed

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

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

All times shown according to UTC.

Time Nick Message
00:01 boris-42 asalkeld actually we can just make a script
00:01 boris-42 asalkeld few bash lines of code with putter proper values =)
00:01 asalkeld sure
00:07 boris-42 asalkeld in any case this is better then nothing=)
00:07 asalkeld yip
00:25 rook joined #openstack-rally
00:51 dmorita joined #openstack-rally
01:26 rook joined #openstack-rally
01:37 Poornima joined #openstack-rally
02:27 rook joined #openstack-rally
03:28 rook joined #openstack-rally
04:13 Poornima joined #openstack-rally
04:29 rook joined #openstack-rally
05:00 nkhare joined #openstack-rally
05:29 rook joined #openstack-rally
06:18 k4n0 joined #openstack-rally
06:30 rook joined #openstack-rally
06:33 openstackgerrit Mikhail Dubov proposed a change to stackforge/rally: Rally info: deploy engines & server providers  https://review.openstack.org/116746
06:47 k4n0 joined #openstack-rally
07:31 rook joined #openstack-rally
07:35 k4n0 joined #openstack-rally
07:58 andreykurilin_ joined #openstack-rally
08:08 openstackgerrit Rohan Kanade proposed a change to stackforge/rally: Collect time duration for Context setup/cleanup  https://review.openstack.org/119606
08:31 rook joined #openstack-rally
08:59 Poornima_ joined #openstack-rally
09:00 tuvenen joined #openstack-rally
09:12 openstackgerrit Boris Pavlovic proposed a change to stackforge/rally: test it  https://review.openstack.org/119716
09:40 andreykurilin left #openstack-rally
09:59 Poornima_ joined #openstack-rally
10:02 openstackgerrit Mikhail Dubov proposed a change to stackforge/rally: Rally info: deploy engines & server providers  https://review.openstack.org/116746
10:16 openstackgerrit Boris Pavlovic proposed a change to stackforge/rally: Improve Rally gate a job  https://review.openstack.org/119584
10:19 ekarlso- boris-42: is there a way to have rally pass cached tokens to the clients ?
10:19 boris-42 ekarlso- nope
10:19 boris-42 ekarlso- now it generates every time
10:19 ekarlso- sadface, it means it's doing roundtrips to keystone each time then :/
10:20 boris-42 ekarlso- one time per iteration
10:20 boris-42 ekarlso- not billions times)
10:20 ekarlso- ah ok
10:21 mmalam joined #openstack-rally
10:21 mmalam boris-42: Hi
10:21 mmalam I was able to run the patches of Ajay, plus run Rally from eclipse. Thanks for the help.
10:22 boris-42 mmalam oh great
10:22 mmalam i will try to list the set of problem.
10:22 mmalam problems*
10:22 mmalam Boris, how are you proceeding in Neutron benchmarking
10:22 mmalam some blueprint document?
10:28 mmalam there?
10:28 mmalam blueprint on neutron benchmarking....
10:29 boris-42 mmalam what kind of blueprints?
10:29 boris-42 mmalam if it is Iperf there is specific blueprint
10:29 mmalam i mean future tasks.
10:29 boris-42 mmalam if it is just neutron api
10:30 boris-42 mmalam then there is another one
10:30 boris-42 mmalam so what one?
10:30 mmalam i remember when I started discussing with you, you said Neutron API level and protocol level benchmarking
10:30 mmalam and you wanted me to start Iperf
10:30 mmalam but as Ajay is already working on
10:30 mmalam what other tasks, I can take it out
10:31 mmalam any thing related to protocol level benchmarking.
10:32 boris-42 mmalam 10 min ok?
10:32 mmalam i am here.
10:32 mmalam please bing me back.
10:46 boris-42 mmalam so what about measuring internet bandwidth using https://pypi.python.org/pypi/speedtest-cli
10:47 mmalam wait a sec
10:47 andreykurilin joined #openstack-rally
10:51 mmalam boris-42: what is the overall scenario?
10:51 mmalam to speed test the gateway router?
10:51 boris-42 mmalam you can check that resources are allocated properly
10:52 mmalam am not getting your point.
10:52 mmalam please explain a bit to me.
10:52 boris-42 mmalam you are running this benchmark in VMs of cloud
10:52 mmalam ok
10:52 boris-42 mmalam if you have channel 100mbs
10:53 boris-42 mmalam and you are giving to every vm 10mbs chanell
10:53 boris-42 mmalam are you sure that every of these VM will get 10mbs chanell?
10:53 mmalam then...........isn't very similar to iperf?
10:53 boris-42 mmalam I perf can't check internet
10:54 boris-42 mmalam it can check traffic between 2 VMs no?
10:54 boris-42 mmalam at least in benchmark that is doing a jay
10:54 mmalam so you mean how much bandwidth is available to each VM in order to to be connected to internet?
10:54 boris-42 mmalam yep testing internet connectivity
10:54 boris-42 mmalam and how it's allocated between VMs
10:55 mmalam this will also involve installing speedtest-cli?
10:55 mmalam on each VM
10:57 mmalam boris-42: there?
10:58 boris-42 mmalam we are working on special framework for that
10:58 boris-42 mmalam https://review.openstack.org/#/c/104564/
10:58 boris-42 https://review.openstack.org/#/c/98172/
10:58 boris-42 that will prepare image with installed tool
10:59 boris-42 so we will start our VMs from snapshot
10:59 boris-42 that already has everything installed
11:00 mmalam what should I do then :)
11:03 openstackgerrit Rohan Kanade proposed a change to stackforge/rally: Adds unit tests for rally.cmd.cliutils  https://review.openstack.org/116958
11:05 mmalam there Boris?
11:07 openstackgerrit Rohan Kanade proposed a change to stackforge/rally: Collect time duration for Context setup/cleanup  https://review.openstack.org/119606
11:09 rook joined #openstack-rally
11:25 boris-42 mmalam why not just implementing?
11:25 boris-42 mmalam speedtest-cli
11:25 boris-42 mmalam and we will just refactor it when we finish framework for vms..
11:34 nkhare_ joined #openstack-rally
11:35 mmalam isnt written in python?
11:36 boris-42 mmalam what?
11:36 boris-42 mmalam who?
11:36 boris-42 speedtest-cli is written in python but you'll need to install it in any case
11:39 mmalam not getting boris.
11:40 boris-42 mmalam can you write a code that installs speednet stuff in vm
11:40 boris-42 mmalam and run it
11:40 mmalam yes, that should be possible.
11:40 mmalam that is what iperf was doing IMHO
11:40 boris-42 mmalam like this https://github.com/stackforge/rally/blob/master/doc/samples/tasks/support/instance_dd_test.sh
11:40 mmalam ok i see
11:40 boris-42 mmalam nope Iperf is aboslutelly differnet benchmark
11:40 boris-42 mmalam this is just .sh script
11:40 mmalam yes
11:41 boris-42 mmalam that will be run with VMTask
11:41 boris-42 mmalam so this will have huge benifit
11:41 boris-42 mmalam like small step
11:41 boris-42 mmalam so when we finish framework for preparing images
11:41 boris-42 mmalam it will be really simple to reuse this .sh
11:41 mmalam yes
11:42 boris-42 mmalam so distributed work on the task=)
11:42 mmalam so just sh script and replace it in VMtasks
11:42 boris-42 mmalam nope
11:42 boris-42 mmalam VMTask is one benchmark that can consume these .sh files
11:42 boris-42 mmalam but we are working on other benchmark (that will prepare images with tools) but they will as well need to do similar stuff like in .sh
11:43 boris-42 mmalam so making .sh now is reasonable, so when we finish that framework, we will be able quite fast to make N benchmarks
11:43 boris-42 mmalam based on .sh stuff
11:45 boris-42 mmalam so basically the goal is to have all this stuff http://www.brendangregg.com/Perf/linux_observability_tools.png
11:45 boris-42 mmalam implemented in rally
11:45 boris-42 mmalam as a separated benchmark, that prapres snapshot with this tool
11:45 boris-42 mmalam then runs VMs with it
11:45 boris-42 mmalam benchmarks and unify output
11:46 rook boris-42++
11:46 mmalam but installation code will come in the shell script right?
11:46 rook VMTask as it stands is very flexible.
11:46 boris-42 rook imho to flexible=)
11:47 rook lol
11:47 boris-42 rook for end users=)
11:47 boris-42 rook I don't care about all details, i want just results in graphs to show my boss+)
11:47 rook boris-42 it is up to them to prepare the right shell script and json.
11:47 rook haha
11:47 rook I just want a unified tool instead of all these one offs
11:48 boris-42 rook I mean I like unified stuff
11:48 boris-42 rook but we should have doc/samples or hardocded tools as a benchmarks
11:48 yaguang joined #openstack-rally
11:48 boris-42 rook or scripts in doc/samples that does everything
11:49 boris-42 rook I mean it should be all 1 click stuff =)
11:49 boris-42 rook not write 100500 lines of shell/python then do some magic and so on=)
11:49 rook boris-42 i like the idea of samples, but not hard-coded. One click sure, but that is difficult to get a one-click that serves everyones purpose.
11:49 rook haha
11:49 rook I agree.
11:49 rook There needs to be a framework to allow users to implement new benchmarks, that is why I like the sample/doc idea.
11:50 boris-42 rook so new benchmarks without touching code?
11:50 boris-42 rook sounds interesting
11:50 boris-42 rook I need to spend next week on deep diving in that topic=)
11:50 rook roger. /me is about to kick off some rally stuff, bbl =)
11:50 rook i wrote some wrappers for my Rally work
11:51 boris-42 rook hm interesting
11:51 rook to increase load (since that blueprint is still in the works)
11:51 rook also to create some CSVs ;)
11:51 boris-42 rook stress runner?
11:51 rook yup
11:51 rook I will post it to github when i am complete.
11:51 boris-42 rook ah that one is what I am going to think tomorrow
11:51 boris-42 rook so for now I am thinking about chaining output format
11:52 rook it is VERY early code, but it is working to my expectations right now.
11:52 boris-42 rook so rally plot2html in case of stress runner will make sense=)
11:52 mmalam boris-42: so it should install from shell script?
11:52 boris-42 mmalam ya
11:52 boris-42 mmalam install & run for now
11:52 boris-42 mmalam then we will split install & run parts
11:54 mmalam but if we want to find out bugs in the Neutron implementation, what is best for that?
11:54 boris-42 mmalam to write API benchmarks maybe?
11:54 boris-42 mmalam that are missing
11:55 mmalam yes
11:55 mmalam that will be interesting.
11:55 mmalam create_subnet like ?
11:55 mmalam were they termed as API benchmarking?
11:55 boris-42 mmalam yep
11:56 boris-42 mmalam https://github.com/stackforge/rally/tree/master/rally/benchmark/scenarios/neutron
11:56 boris-42 mmalam we have here some amount of them
11:56 boris-42 mmalam and 2 patches
11:56 boris-42 mmalam https://review.openstack.org/#/c/118555/
11:56 mmalam yes, i have got through.
11:56 * boris-42 mmalam https://review.openstack.org/#/c/118555/
11:56 boris-42 mmalam so you can work on missing stuff
11:56 boris-42 mmalam like sec_grroups
11:57 mmalam yes
11:59 nkhare joined #openstack-rally
11:59 boris-42 mmalam so you can work on that if you would like
12:02 mwagner_lap joined #openstack-rally
12:14 RaginBajin joined #openstack-rally
12:27 openstackgerrit Nikita Konovalov proposed a change to stackforge/rally: Sahara Cluster and EDP contexts  https://review.openstack.org/115962
12:27 openstackgerrit Nikita Konovalov proposed a change to stackforge/rally: Sahara Cluster and EDP contexts  https://review.openstack.org/115962
12:28 openstackgerrit Nikita Konovalov proposed a change to stackforge/rally: Sahara Job Execution benchmark  https://review.openstack.org/116881
13:08 Poornima joined #openstack-rally
13:32 zhiyan hi boris-42
13:33 boris-42 zhiyan hi there
13:34 zhiyan boris-42: hi, could you share me the url show osprofiler html report?
13:34 zhiyan your website, iirc
13:35 zhiyan btw, (my) tonight has a trove bp meeting, i'd like to talk about those changes with trove team...
13:35 zhiyan for trove osprofiler integration
13:35 zhiyan would you like join
13:35 zhiyan ?
13:36 boris-42 zhiyan http://boris-42.github.io/ngk.html
13:36 boris-42 zhiyan or even better is to use this blogpost
13:36 boris-42 zhiyan http://ahsalkeld.wordpress.com/2014/09/04/how-to-profile-heat-using-osprofile/
13:38 boris-42 clee hi there
13:38 boris-42 clee sry
13:38 boris-42 coolsvap hi there
13:39 zhiyan boris-42: ah, have heat already integrated with osprofiler now?
13:39 boris-42 zhiyan nope asalkeld is working on that
13:41 zhiyan ok, happy to know that (sorry the delay on it, my colleague promised)
13:41 boris-42 zhiyan so patches are already ready
13:41 boris-42 zhiyan waiting for Kilo as I understand
13:41 zhiyan cool
13:42 boris-42 zhiyan https://review.openstack.org/#/c/118115/
13:42 zhiyan btw, do you have time to take a look on my trove one?
13:42 zhiyan any inputs are welcome
13:43 boris-42 zhiyan I'll try to find
13:44 zhiyan boris-42: https://review.openstack.org/#/c/116653/
13:55 jaypipes joined #openstack-rally
14:12 ilyashakhat joined #openstack-rally
14:41 openstackgerrit Nikita Konovalov proposed a change to stackforge/rally: Adding optional parameters to Sahara cluster  https://review.openstack.org/119778
14:42 openstackgerrit Boris Pavlovic proposed a change to stackforge/rally: Improve Rally gate job  https://review.openstack.org/119584
14:58 openstackgerrit Boris Pavlovic proposed a change to stackforge/rally: Improve Rally gate job  https://review.openstack.org/119584
15:14 openstackgerrit Sergey Skripnick proposed a change to stackforge/rally: Fix CIDR overlapping in _create_network  https://review.openstack.org/119787
15:31 openstackgerrit Olga Kopylova proposed a change to stackforge/rally: Fix rally verify fail (cause of rally KeyError: 'admin')  https://review.openstack.org/119790
15:35 k4n0_ joined #openstack-rally
16:30 RainbowBastion joined #openstack-rally
16:30 boris-42 RainbowBastion hi there
16:30 boris-42 RainbowBastion I left some comment
16:31 RainbowBastion boris-42 I'm looking through it now and am making the appropriate changes.
16:31 boris-42 RainbowBastion great
16:55 openstackgerrit OpenStack Proposal Bot proposed a change to stackforge/rally: Updated from global requirements  https://review.openstack.org/119152
17:00 marcoemorais joined #openstack-rally
17:10 boris-42 k4n0_ around?
17:21 openstackgerrit Kevin Tsang proposed a change to stackforge/rally: Moved context creation to benchmark.engine  https://review.openstack.org/117302
17:22 boris-42 ^ RainbowBastion  you need to rebase your patch
17:23 boris-42 RainbowBastion cause we merged my big one
17:27 RainbowBastion boris-42 Got it
17:36 mwagner_lap joined #openstack-rally
17:36 stannie joined #openstack-rally
17:36 openstackgerrit Kevin Tsang proposed a change to stackforge/rally: Moved context creation to benchmark.engine  https://review.openstack.org/117302
17:39 openstackgerrit Kevin Tsang proposed a change to stackforge/rally: Moved context creation to benchmark.engine  https://review.openstack.org/117302
18:03 RainbowBastion joined #openstack-rally
18:19 boris-42 RainbowBastion you should rebase properly.. you didn't fix merge conflicts..
18:22 RainbowBastion boris-42 I thought I did, but I'll take a look and see what I did wrong.
18:22 boris-42 RainbowBastion https://review.openstack.org/#/c/117302/14/tests/orchestrator/test_api.py
18:22 boris-42 RainbowBastion you commit merge conflicts..
18:23 boris-42 RainbowBastion in that file ^
18:37 k4n0_ joined #openstack-rally
18:46 openstackgerrit Sergey Skripnick proposed a change to stackforge/rally: Fix CIDR overlapping in _generate_subnet_cidr  https://review.openstack.org/119787
18:54 RainbowBastion joined #openstack-rally
19:07 openstackgerrit joined #openstack-rally
19:10 marcoemorais joined #openstack-rally
19:21 openstackgerrit Kevin Tsang proposed a change to stackforge/rally: Moved context creation to benchmark.engine  https://review.openstack.org/117302
19:21 openstackgerrit A change was merged to stackforge/rally: Improve Rally gate job  https://review.openstack.org/119584
19:22 andreykurilin joined #openstack-rally
19:23 marcoemorais joined #openstack-rally
19:25 openstackgerrit Olga Kopylova proposed a change to stackforge/rally: Fix rally verify fail (cause of rally KeyError: 'admin')  https://review.openstack.org/119790
19:25 olkonami joined #openstack-rally
19:26 penguinRaider joined #openstack-rally
20:00 openstackgerrit Kevin Tsang proposed a change to stackforge/rally: Moved context creation to benchmark.engine  https://review.openstack.org/117302
20:08 rook-out joined #openstack-rally
20:22 openstackgerrit Kevin Tsang proposed a change to stackforge/rally: Moved context creation to benchmark.engine  https://review.openstack.org/117302
20:26 openstackgerrit Boris Pavlovic proposed a change to stackforge/rally: Make rally perf job work in other projects  https://review.openstack.org/119901
20:27 boris-42 RainbowBastion there?
20:28 boris-42 RainbowBastion compare https://review.openstack.org/#/c/117302/11..17/tests/orchestrator/test_api.py
20:30 RainbowBastion boris-42 I wasn't sure which changes were the ones I needed to merge in that file.
20:31 boris-42 RainbowBastion I'll left comment
20:31 RainbowBastion boris-42 thanks
20:32 RainbowBastion boris-42 Reason why I'm pushing up so much for review is because tox still isn't working, both on my Mac and the virtual environment set up last week.
20:32 boris-42 RainbowBastion no worries
20:32 boris-42 RainbowBastion push as much as you want
20:32 RainbowBastion boris-42 thanks
20:40 boris-42 RainbowBastion I left comments
20:46 RainbowBastion boris-42 Remove the entire FakeScenario class?
20:46 boris-42 RainbowBastion it is not used...
20:46 boris-42 RainbowBastion I removed it in patch-set 11
20:46 boris-42 RainbowBastion but you returned it back...
20:46 RainbowBastion boris-42 just checking, will do now
20:47 openstackgerrit Kevin Tsang proposed a change to stackforge/rally: Moved context creation to benchmark.engine  https://review.openstack.org/117302
20:54 openstackgerrit Monty Taylor proposed a change to stackforge/rally: Rename doc environment to docs  https://review.openstack.org/119925
21:15 klindgren_ joined #openstack-rally
21:18 boris-42 RainbowBastion 3 things more
21:22 openstackgerrit Kevin Tsang proposed a change to stackforge/rally: Moved context creation to benchmark.engine  https://review.openstack.org/117302
21:47 openstackgerrit A change was merged to stackforge/rally: Fix rally verify fail (cause of rally KeyError: 'admin')  https://review.openstack.org/119790
21:51 asalkeld joined #openstack-rally
21:55 boris-42 RainbowBastion you still have some errors
21:58 RainbowBastion boris-42 I'll fix the ones related to test_base, but I don't know about test_api
22:02 stannie left #openstack-rally
22:03 marcoemorais joined #openstack-rally
22:11 boris-42 RainbowBastion left comment https://review.openstack.org/#/c/117302/19/tests/orchestrator/test_api.py
22:11 boris-42 RainbowBastion it's quite simple to fix
22:18 asalkeld joined #openstack-rally
22:18 openstackgerrit Kevin Tsang proposed a change to stackforge/rally: Moved context creation to benchmark.engine  https://review.openstack.org/117302
22:20 RainbowBastion test_base wasn't working because the endpoints attribute was removed from base.py, and replaced with the admin attribute.
22:21 boris-42 RainbowBastion yep
22:21 boris-42 RainbowBastion I made that recently
22:21 boris-42 RainbowBastion https://review.openstack.org/#/c/116766/
22:21 boris-42 RainbowBastion so we are quite close =)
22:22 boris-42 RainbowBastion to finish our big big task=)
22:23 boris-42 RainbowBastion so as a next patch you can take a stuff related to removing from scenario runner admin filed
22:23 boris-42 RainbowBastion that is not required now
22:23 RainbowBastion boris-42 So when creating a runner do I change self.fake_endpoints from being passed in to a magic mock object instead?
22:24 boris-42 RainbowBastion hm?
22:24 RainbowBastion boris-42 Since endpoints isn't used anymore
22:24 boris-42 RainbowBastion I don't get about what place you are talking?
22:24 mwagner_lap joined #openstack-rally
22:24 boris-42 RainbowBastion you should remove it from here https://github.com/stackforge/rally/blob/master/rally/benchmark/runners/base.py#L157
22:25 boris-42 RainbowBastion and here https://github.com/stackforge/rally/blob/master/rally/benchmark/runners/base.py#L180
22:25 RainbowBastion http://paste.openstack.org/show/108506/ line 6
22:25 boris-42 RainbowBastion but do this in separated patch
22:25 boris-42 RainbowBastion let's get this in
22:26 boris-42 RainbowBastion yep you'll need to fix related tests as well
22:27 RainbowBastion boris-42 So I remove the admin field?
22:27 boris-42 RainbowBastion not in thi patch
22:27 boris-42 RainbowBastion let merge this frist
22:27 RainbowBastion boris-42 ok
22:27 boris-42 RainbowBastion and then you'll start working on next stuff with removing admin from runners
22:29 RainbowBastion boris-42 my bad, I accidentally took out the endpoint bit from {"endpoint": runner.admin_user} on line 16 of the paste I posted
22:29 RainbowBastion boris-42 I'll add that back in
22:29 openstackgerrit Kevin Tsang proposed a change to stackforge/rally: Moved context creation to benchmark.engine  https://review.openstack.org/117302
22:52 boris-42 RainbowBastion one more nit https://review.openstack.org/#/c/117302/
22:55 marcoemorais joined #openstack-rally
22:56 marcoemorais joined #openstack-rally
22:56 asalkeld joined #openstack-rally
22:58 openstackgerrit Kevin Tsang proposed a change to stackforge/rally: Moved context creation to benchmark.engine  https://review.openstack.org/117302
23:04 marcoemorais1 joined #openstack-rally
23:06 marcoemorais1 joined #openstack-rally
23:17 boris-42 RainbowBastion sry I made a mistake https://review.openstack.org/#/c/117302/
23:18 RainbowBastion boris-42 No problem
23:18 openstackgerrit Kevin Tsang proposed a change to stackforge/rally: Moved context creation to benchmark.engine  https://review.openstack.org/117302
23:20 asalkeld joined #openstack-rally
23:45 yaguang joined #openstack-rally
23:53 openstackgerrit Kevin Tsang proposed a change to stackforge/rally: Moved context creation to benchmark.engine  https://review.openstack.org/117302

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