17:00:03 #startmeeting Rally 17:00:03 Meeting started Tue Jun 24 17:00:03 2014 UTC and is due to finish in 60 minutes. The chair is boris-42. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:00:04 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 17:00:07 The meeting name has been set to 'rally' 17:00:25 msdubov hughsaunders marcoemorais ping 17:01:23 boris-42: heya 17:01:29 boris-42: hi 17:01:42 heya 17:01:57 I am here for rally meeting 17:02:04 o/ 17:02:07 =) 17:02:12 * jtaleric lurking 17:02:14 Hi all 17:02:18 hi 17:02:52 hi 17:03:04 let's just wait a couple of minutes 17:03:52 hi 17:04:50 so let's start=) 17:05:16 #topic we should do something with our unit test coverage 17:05:30 Okay recently (today) I added new job in Jenkins 17:05:41 that will create cover report on every patch 17:06:02 job is called rally-coverage 17:06:12 and if you click on it you'll see pretty html report 17:06:26 (red lines means that code is not covered by unit tests) 17:06:28 sorry i late -_- 17:06:32 We should have 100% coverage 17:06:37 Any questions? 17:06:57 boris-42, 100% should be enforced on gates 17:07:03 k4n0 99^ 17:07:36 k4n0 +1 for that to have some kind of 98% as a lower border for coverage 17:07:40 k4n0 otherwise just -1 17:07:51 +1 17:07:57 but first of all we need to cover everything tests=) 17:08:28 okay next topic 17:08:28 We can split up modules and start writing tests , maybe do this on a google doc or something 17:08:41 k4n0 probably we should make some kind of blueprint 17:08:44 k4n0 and workitems 17:08:45 ? 17:08:55 +1 for blueprint 17:09:03 boris-42, blueprint is also good, ill make one after meeting 17:09:10 k4n0 thanks 17:09:25 so we will be able to split work 17:09:30 and work together 17:09:36 okay next topic 17:09:42 ok 17:09:49 #topic osprofiler 17:09:59 #action make blueprint for unit test coverage 17:10:01 Okay guys I made a last changes in API 17:10:14 and cover it 100% by unit tests 17:10:32 Going to launch devstack with it 17:10:39 and if there is no issues I will cut release 17:10:45 Any questions?) 17:11:31 https://github.com/stackforge/osprofiler this is library if somebody doesn't know =) 17:11:43 Okay moving to next topc 17:12:01 #topic Rally SLA 17:12:10 redixin plsss be more verbose 17:12:11 =) 17:12:19 what is Rally SLA ? 17:12:21 redixin and tell us what is it and status 17:13:01 rally sla is just one more section in the task config file 17:13:18 it is possible to specify criterios 17:13:28 like maximum error rate 17:13:34 or time limits 17:13:52 that is great 17:14:01 task is going as usual, but there is special command to check if task was successful or not 17:14:16 according to criterios 17:14:41 here is the patch https://review.openstack.org/#/c/98158/ 17:14:45 redixin btw we will need to add it to HTML 17:15:01 redixin somehow (e.g. does it pass or not criteria) and details 17:15:08 redixin I'll think about it 17:15:20 yes, we can add one more table to html results 17:15:33 and one more table to task detailed outpu 17:15:39 redixin yep 17:16:34 So as well our jobs check-rally-dsvm-rally and check-rally-dsvm-neutron-rally 17:16:43 will be able to check more carefully 17:16:47 benchmarks 17:17:04 e.g. we can setup for most of them that there is no failures 17:17:18 so we won't need to go and take a look every time on all tasks 17:17:27 any questions? 17:20:53 #topic stress runner 17:21:01 olkonami pls could you introduce us? 17:22:19 it is runner which increase concurrency stepwise 17:22:42 https://review.openstack.org/#/c/94806/ 17:22:56 olkonami is it ready? 17:23:56 boris-42: not yet 17:24:06 olkonami ok any questions about it? 17:24:54 olkonami, when does it stop increasing concurrency? 17:25:20 olkonami, i mean if lots of api errors start coming up due to increased load, does it still keep increasing concurrency? 17:26:37 k4n0: we have max_failure_rate parameter for that, if it is exceeded, it will stop increasing 17:26:59 olkonami, ok so that is max failure for atomic action ? 17:27:45 olkonami, I will check the patch in detail, we can move on for now :) 17:28:07 k4n0, sorry, don't understand question 17:28:16 k4n0 nope for atomic action 17:28:19 k4n0 for whole iteration 17:28:26 boris-42, got it 17:28:36 k4n0 so it will run with some constant load 17:28:39 k4n0 for some period 17:28:48 k4n0 before increasing load it will check 17:29:01 k4n0 that success / all is > then some parameter 17:29:21 boris-42, cool , got it 17:29:23 #topic RPS runner (or refactored periodic) 17:29:36 oanufriev hey could you share updates ? 17:29:59 also now parameter for regular stop incrasing is named "growth_count", do someone have better proposes? 17:30:30 start, step, stop 17:30:31 sorry, i can share them in 1.5 haur. Thereis a bug in unit test that causes memory leak 17:31:03 if it's about code 17:31:13 and about works 17:32:35 periodic runner has been redesigned to check rps parameter instead period 17:33:01 so now it should be understanded as N runs per second 17:34:35 also i made extensible pool of multiprocessing.Process instances instead of using fixed lenght list of pools 17:34:58 that was made to decrease memory usage and related failures 17:35:47 questions 17:35:52 ? 17:36:33 oanufriev when can we see any patches related to this? 17:36:54 in 1.5 hour at gerrit ^) 17:36:58 in 1.5 hour at gerrit :) 17:37:00 oanufriev great 17:37:10 any questions from anybody? 17:37:22 marcoemorais hughsaunders rediskin k4n0 olkonami ? 17:37:36 nope 17:37:39 nope 17:38:31 no patchs - no questions :) 17:39:12 lol=) 17:39:16 #topic rally & tempest 17:39:29 k4n0 any news?) 17:39:44 no updates, I am working on fixing review comments on latest patch 17:39:54 I am generating html from json stored in DB 17:40:05 json is coming from tempest subunit logs 17:40:33 Now i need to refactor the code such that we have a command like this "rally verify result -type html -out file.html " 17:41:55 https://review.openstack.org/101917 17:44:31 k4n0 ok great 17:44:40 #topic Open Discussion 17:44:49 Anybody would like to discuss anything? 17:46:05 hey hey hey?) 17:46:21 nothing from my end. I am not feeling well today 17:46:32 I have to get my wisdom tooth removed tomorrow :) 17:46:46 =) 17:46:55 k4n0 tomorrow? 17:47:00 k4n0 or yestardsy?) 17:47:06 tomorrow afternoon 17:48:30 k4n0, get well soon! 17:48:47 thanks 17:48:58 k4n0 oh 17:49:22 k4n0 have the same experience on previous friday=) 17:49:49 I would like to try to avid such stuff in future=) 17:49:58 boris-42, you also removed wisdom tooth? 17:50:05 k4n0 yep 17:50:08 k4n0 4 days ago 17:50:20 boris-42, haha, conincidence 17:50:32 boris-42, anyways, lets talk after meeting :) 17:50:44 k4n0 okay 17:50:51 I think we can end this meeting 17:51:10 thank you all guys=) 17:51:13 boris-42 that's why you so angry =) 17:51:20 oanufriev yep =) 17:51:20 thanks all 17:51:26 oanufriev, haha 17:51:33 #endmeeting