Tuesday, 2016-01-19

*** salmon_ has joined #openstack-solar08:39
*** dshulyak_ has joined #openstack-solar08:41
openstackgerritJedrzej Nowak proposed openstack/solar: Fixing concurrency problems in history counter  https://review.openstack.org/26923809:04
pigmejhmm, I think we should have possibility to "inherit" from other resources09:06
pigmejimagine I want to alter riak_node.run action, let's say i want to turn some bucket types by default (or whatever), for now I would need to change whole riak_node or write my own.09:06
pigmejwhile for example in Dockerfiles you can "inherit" from other images, and just change/add what you need.09:07
pigmejdshulyak_: salmon_ what do you think ?09:07
openstackgerritJedrzej Nowak proposed openstack/solar: Fixing concurrency problems in history counter  https://review.openstack.org/26923809:11
dshulyak_probably we will need some instruction to execute tasks right after riak_run, and before all other successors of riak run, but im not sure about dockerfile inherit09:17
pigmejdshulyak_: I meant, let's say you have some complicated resource, and you want to alter it a bit09:18
pigmejcurrently you will need to create new resource or copy&paste from original09:18
dshulyak_what kind of modifications? if such things as you said, e.g create/change bucket type, you just need some tasks that will run right after parent and before all dependencies09:22
pigmej dshulyak_ yeah but that was just an example09:25
pigmejmaybe not perfect though because btype requires healhy cluster09:25
dshulyak_we want to release tomorrow of by the end of the week?09:29
salmon_pigmej: you can use Composer file for it, but yes in feature some kind of inheritance may be useful09:33
pigmejdshulyak_: i would bet "never release on friday", so let's make a release on friday ?:D09:34
openstackgerritMerged openstack/solar: Remove ansible.cfg, we use .ssh/config now  https://review.openstack.org/26903510:25
bogdandosalmon_, what is docs-on-rtfd ? https://review.openstack.org/#/c/26755810:35
bogdandocan you  give me RTFM link?)10:35
salmon_bogdando: http://solar.readthedocs.org/en/latest/10:35
openstackgerritJedrzej Nowak proposed openstack/solar: Fixing concurrency problems in history counter  https://review.openstack.org/26923810:36
pigmej^ rebased10:36
openstackgerritMerged openstack/solar: Hardcode ansible version. We are not ready for 2.0  https://review.openstack.org/26923910:53
openstackgerritBogdan Dobrelya proposed openstack/solar: Add tox checkniceness,checksyntax for docs  https://review.openstack.org/26955911:14
bogdandosalmon_, ^^11:15
pigmejbogdando: I added a comment just now11:16
bogdandofor some strange reason we have this https://review.openstack.org/#/c/269559/1/ChangeLog is a bonus as well :-o11:16
pigmejI don't think that it's bonus11:17
bogdandoand this https://review.openstack.org/#/c/269559/1/AUTHORS11:17
bogdandoI think it is cool11:17
bogdandobut I can disable this if you help me how :)11:17
pigmejbogdando: it's full of mess11:18
pigmejsame as automated generated changelog11:18
salmon_bogdando: changelog is generated by pbr, we don't need it in repo11:18
salmon_it should go to .gitignore11:18
bogdandoauthors to keep maybe?..11:20
bogdandoeasy to add ppl to review11:20
pigmejbogdando: ?11:20
bogdandopigmej, what?11:21
bogdandoLet's keep authors to be genereted, thoughts?11:21
pigmejI dislike these "autogenerated authors" usually11:23
openstackgerritBogdan Dobrelya proposed openstack/solar: Add tox checkniceness,checksyntax for docs  https://review.openstack.org/26955911:23
pigmejI also dislike "doc" requirements in test-requirements11:26
openstackgerritBogdan Dobrelya proposed openstack/solar: Add tox checkniceness,checksyntax for docs  https://review.openstack.org/26955911:28
bogdandopigmej, ok, I removed both11:28
salmon_We may add MAINTAINERS later11:28
pigmejbogdando: but I don't see these changes11:29
salmon_afk, ~1h11:30
bogdandopigmej, the patchset 311:30
bogdandoI can see changes11:30
pigmejhttps://review.openstack.org/#/c/269559/3/test-requirements.txt ?11:30
pigmejah you removed files11:30
pigmejhttps://review.openstack.org/#/c/269559/1/test-requirements.txt@19 I added this comment in meantime :)11:31
*** salmon_ has quit IRC11:34
bogdandopigmej, where are those tox env dependencies?11:35
bogdandoI'd like the docs check to be run at CI gates by default11:35
bogdandothere would If I run just tox11:36
openstackgerritBogdan Dobrelya proposed openstack/solar: Add tox checkniceness,checksyntax for docs  https://review.openstack.org/26955911:38
pigmejbogdando: hmm11:39
pigmejI'm not sure if these deps are correctly placed11:40
pigmejalso what's "doc" dependency ?11:40
pigmejah nvm !!11:40
pigmejbogdando: the drawback of that approach that you have, is that we will need to install solar deps 4 times, instead of 211:50
pigmejwhich takes obviously double time11:50
pigmejjust for checking if docs are "cool"11:50
*** salmon_ has joined #openstack-solar12:19
bogdandopigmej, any idea how to fix this?13:09
bogdandopigmej, master branch - tox  40.20s user 2.87s system 301% cpu 14.290 total13:10
bogdandothe docs -  56.03s user 5.36s system 187% cpu 32.777 total13:10
pigmejbogdando: we should probably include these commands in docs step BUT I'm not sure13:11
pigmejin all other cases we will make 4 tox envs, with a lot of dependencies.13:12
pigmejalso for docs you probably don't need all our dependencies, isn't it ?13:12
pigmejyou probably just need few of them ? Like in our 'docs' env, where we only require13:13
openstackgerritDmitry Shulyak proposed openstack/solar: Zerorpc worker for orchestration modules  https://review.openstack.org/26916613:13
pigmejalso I'm kinda sure that adding these to env isn't enough if you want to check them automaticaly on openstack-infra13:13
pigmejbecause afair commands are predefined and they are `tox -epy27` tox -edocs`13:14
pigmejbut salmon_ can say more about that13:14
salmon_docs are build by jenkins with something like this: tox -e testenv python setup.py build_sphinx13:20
salmon_I don't see a way to run this checks automaticly13:23
dshulyak_please review https://review.openstack.org/269166, to test - you can run solar-worker in foreground13:30
pigmejdshulyak_: with pleasure but I need to fix mess that we did in transports :P13:31
dshulyak_i will do concurrency patches in 30 mins13:32
dshulyak_what mess?13:32
pigmejnot only13:32
pigmejright now each transport returns different stuff13:32
pigmejyou removed fabric partially, I had partially working wrapper for fabric,13:32
pigmejso let's say that we have 3 different return values from transports :D13:32
pigmejwith 3 different API ;D13:33
pigmejso I'm improving that situation, after that errros :)13:34
pigmejwe broke it over time, or improve because it depends from which side will we look ;D13:38
openstackgerritJedrzej Nowak proposed openstack/solar: Unify returned transport results  https://review.openstack.org/26965313:59
pigmejit's WIP still13:59
pigmejhmm dshulyak_ so you started that "errors" in fact some time ago (when you removed fabric)14:26
dshulyak_what do you mean? some time ago i added unified handling of erros for puppet/ansible/bash14:27
dshulyak_when i partially removed fabric14:27
pigmejah ok14:28
pigmejI wondered when this happened ;D14:28
pigmejok, then I know wtf ;)14:29
openstackgerritLukasz Oles proposed openstack/solar: Update Wordpress tutorial  https://review.openstack.org/26967314:29
pigmejdshulyak_: hmm, how did you get that 'exists' error?14:39
dshulyak_i used - next(solar_models.NegativeCounter.get_or_create(str(i)))14:41
pigmejit seems that I did something wrong during rebase :D14:43
dshulyak_i cant find exists property on Counter - https://github.com/basho/riak-python-client/blob/master/riak/datatypes/counter.py14:43
pigmejbecause it's not14:44
dshulyak_it shouldnt be Counter?14:44
pigmejtrying to see wtf14:44
pigmejit shouldn't be get_or_create that's for sure14:45
dshulyak_get_or_create was used in LogItem14:47
pigmejyaeh yeah, I see, I wonder what's going on currently :)14:48
pigmejMaybe I patched riak client for a while and I forgot about it...14:48
pigmejhmm, something is definitely wrong with this counter patch, I will fix it dshulyak_ now14:50
dshulyak_when this is possible if siblings[0]._get_encoded_data() == siblings[1]._get_encoded_data():14:50
pigmejit's in conflict resolution right?14:51
pigmejyeah it's there, so there are at least one place where we can have this situation, one is counter on that branch14:53
pigmejalso I had some problems with system log updates afair, so I added this there, "just in case", because I thought that 2 the same => ok14:53
pigmejand when I was experimenting with some reducing load in CRDT values, there was a chance that you had 2 the same childs, then I craeted separate resolver for it completely14:54
pigmejbrb dinner arriwed14:59
dshulyak_i will need to spend more time on crdtish locking, it is not very clear why it became so complex :)15:01
dshulyak_afk for 1-1.3h15:01
*** dshulyak_ has quit IRC15:02
*** dshulyak_ has joined #openstack-solar15:06
pigmejdshulyak_: we can switch later to riak native locking15:10
pigmejriak native crdt but we will need it for sqlite too15:10
pigmejin fact it's simple, it's just of operations :)15:11
openstackgerritJedrzej Nowak proposed openstack/solar: Fixing concurrency problems in history counter  https://review.openstack.org/26923815:19
pigmejdshulyak_: adjusted :)15:19
pigmejsorry :(15:19
pigmej"something went wrong" :P15:19
openstackgerritJedrzej Nowak proposed openstack/solar: Unify returned transport results  https://review.openstack.org/26965315:56
openstackgerritJedrzej Nowak proposed openstack/solar: Unify returned transport results  https://review.openstack.org/26965316:15
pigmejok, unify seems to be ok, it's required step for https://bugs.launchpad.net/solar/+bug/153287216:16
openstackLaunchpad bug 1532872 in Solar "Handle errors from transports" [Undecided,In progress] - Assigned to Jedrzej Nowak (jnowak)16:16
pigmejsalmon_: hmm, so we have some bot there...16:16
openstackgerritJedrzej Nowak proposed openstack/solar: Fixing concurrency problems in history counter  https://review.openstack.org/26923817:16
pigmejsalmon_: I hate git rebase....17:16
salmon_Yeah, me too :)17:17
pigmejI don't remember that I had conflict on solar.yaml... but it show up after rebase....17:17
pigmejweird :)17:18
*** dshulyak_ has quit IRC19:46
*** salmon_ has quit IRC22:42

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