Thursday, 2015-08-27

*** jasonsb has quit IRC00:08
*** achanda has joined #kolla00:11
*** weiyu_ has joined #kolla00:15
*** achanda has quit IRC00:19
*** Haomeng has joined #kolla00:19
*** rhallisey has joined #kolla00:25
*** achanda has joined #kolla00:25
*** alisonh has quit IRC00:30
*** zhiwei has joined #kolla00:34
*** sdake has quit IRC00:39
*** alisonh has joined #kolla00:48
*** Haomeng|2 has joined #kolla00:52
*** Haomeng has quit IRC00:55
*** SourabhP has quit IRC01:17
*** unicell has quit IRC01:19
*** jtriley has joined #kolla01:19
*** jasonsb has joined #kolla01:27
*** jasonsb has quit IRC01:28
*** jasonsb has joined #kolla01:29
*** erkules_ has joined #kolla01:30
*** erkules has quit IRC01:32
*** jasonsb has quit IRC01:32
*** zhiwei has quit IRC01:33
*** zhiwei has joined #kolla01:33
*** rhallisey has quit IRC01:37
*** dims has joined #kolla01:43
*** amuller has joined #kolla02:18
amulleranyone alive? I have a few questions about the current state of things, specifically with respect to Neutron and building from source02:19
*** amuller has left #kolla02:30
*** amuller has quit IRC02:30
*** jasonsb has joined #kolla02:35
*** jtriley has quit IRC02:44
*** loth has quit IRC02:51
*** dims has quit IRC02:56
*** alisonh has quit IRC03:21
*** diga has joined #kolla03:23
*** alisonh has joined #kolla03:27
*** weiyu_ has quit IRC04:04
*** achanda has quit IRC04:08
*** achanda has joined #kolla04:44
*** unicell has joined #kolla04:46
*** unicell has quit IRC04:53
*** unicell has joined #kolla04:53
*** SourabhP has joined #kolla05:01
*** SourabhP_ has joined #kolla05:03
*** sdake has joined #kolla05:05
*** SourabhP has quit IRC05:06
*** SourabhP_ is now known as SourabhP05:06
*** sdake_ has joined #kolla05:06
*** blakec has joined #kolla05:07
*** sdake has quit IRC05:10
*** openstackgerrit has quit IRC05:16
*** openstackgerrit has joined #kolla05:16
sdake_evening05:18
*** sdake_ is now known as sdake05:18
coolsvapsdake, hi05:27
sdakehey coolsvap05:28
coolsvapsorry could not attend yesterday's meeting05:28
* coolsvap was busy with family on my bday 05:28
sdakeall good coolsvap05:31
mandrehi05:31
sdakefamily ist ;)05:31
mandrehappy bday coolsvap05:31
coolsvapthanks mandre :)05:31
*** chandankumar has joined #kolla05:34
*** weiyu has joined #kolla05:34
*** dims has joined #kolla05:46
sdakeim off to bed05:50
sdakenight falks05:50
SamYaplenight sdake05:50
*** dims has quit IRC05:52
*** blakec has quit IRC05:56
*** inc0 has joined #kolla05:58
inc0good mornign05:59
SamYaplemorning inc005:59
SamYapletalks are out05:59
SamYapleonly one Kolla :(06:00
inc0yeah, I saw06:00
inc0I hope this won't cause problems with me getting to Tokyo06:01
inc0but it shouldn't...I hope;)06:01
inc0very little technical-meaty sessions landed06:01
SamYapleHey would you mind review https://review.openstack.org/#/c/216691/ inc0 ? I believe you gave it the thumbs up in IRC, but can you give it a +1 so we have a record of agreement?06:01
inc0I feel summits are more and more marketing oriented06:01
SamYaplethey are06:01
SamYaplebut the designones have some nice stuff06:01
inc0it's on you guys to promote kolla then!:P06:02
SamYapleyea i was really hoping for a lab to do just that06:03
inc0I was kinda hoping for "move to kolla howto" session06:03
SamYapleyea that would have been awesome06:04
SamYaplenext summit for sure06:04
sdakei am highlysurprised weddidt get a labb06:07
SamYapleits cause they switched my tracks last second06:07
SamYaplemy lab was in containers and then got moved sdake06:07
SamYapleor thats what im blaming!06:08
sdakethe whitepape s==from the fount is out imy witter steam06:11
SamYapleyea id like to help with that06:11
*** inc0 has quit IRC06:36
*** inc0 has joined #kolla07:05
*** blakec has joined #kolla07:05
*** achanda has quit IRC07:06
*** blakec1 has joined #kolla07:10
*** blakec has quit IRC07:13
*** inc0 has quit IRC07:18
*** inc0 has joined #kolla07:24
inc0SamYaple, heat+ansible is working? (apart of broken heat)07:25
*** sdake has quit IRC07:29
SamYapleinc0: yea07:30
SamYapleand with db upgrade not applitd, it just works07:30
*** shardy has joined #kolla07:31
*** dims has joined #kolla07:35
*** inc0 has quit IRC07:38
*** inc0 has joined #kolla07:38
*** jmccarthy has joined #kolla07:39
*** dims has quit IRC07:40
*** achanda has joined #kolla07:51
*** SourabhP has quit IRC07:54
*** blakec1 has quit IRC07:55
inc0SamYaple, few comments, mostly post-development leftovers07:58
SamYaplethanks inc008:03
SamYaplecoolsvap: https://review.openstack.org/#/c/213397/08:04
SamYaplewould you mind reveiwing now that Kennan has commented?08:04
*** weiyu has quit IRC08:20
*** weiyu has joined #kolla08:23
*** weiyu has quit IRC08:26
*** mandre has quit IRC08:34
*** weiyu_ has joined #kolla08:37
*** akwasnie has joined #kolla08:44
*** weiyu_ has quit IRC08:47
pbourkesad face on summit talks08:49
pbourkereally can't believe only one got picked08:49
SamYapleindeed08:49
SamYaplethere was some last minute track suffling that really hurt us i think08:50
*** achanda has quit IRC08:50
pbourkethe containers on openstack topic seems to get more interest than openstack in containers08:51
*** shardy_ has joined #kolla08:51
SamYaplewhich makes sense when you think about it08:51
pbourkeare all ops that happy these days? :)08:51
SamYaplemost everyone already has a way they deploy08:51
SamYapleno but change is long and hard, they want moer from openstack08:52
pbourkeI think I saw you saying once we have stable working images on dockerhub it will help a lot08:52
pbourkewhich I agree with08:52
SamYaplethere are very few "private cloud" type business out there08:52
coolsvapSamYaple, done, was away for lunch08:52
SamYaplecoolsvap: no worries!08:52
*** shardy has quit IRC08:52
pbourkethe idea was that it should be easier than devstack, right now that's not the case I think08:52
openstackgerritMerged stackforge/kolla: Setup the gate for cross-distro support  https://review.openstack.org/21339708:53
SamYaplewell yea, but then you have to get the individual companies to change the way they deploy08:54
SamYaplevery difficult to do08:54
*** weiyu_ has joined #kolla08:56
*** shardy_ has quit IRC08:56
openstackgerritSam Yaple proposed stackforge/kolla: Remove tiered building in build.py  https://review.openstack.org/21669108:57
*** shardy has joined #kolla08:57
*** weiyu_ has quit IRC08:59
*** weiyu_ has joined #kolla09:03
*** mfalatic has quit IRC09:12
inc0especially that there are less and less ops on summits09:23
inc0which is sad09:23
inc0hopefully Austin will be more welcoming09:23
Kennaninc0: where can find all summits selected this year?09:24
inc0Kennan, you mean talk schedule?09:24
inc0or summit locations?09:24
inc0https://www.openstack.org/summit/tokyo-2015/schedule/ <- schedule09:24
Kennantalk schedule. seems selected talks is finalized09:24
inc0it is09:25
*** mandre has joined #kolla09:30
*** diga has quit IRC09:33
openstackgerritSam Yaple proposed stackforge/kolla: Unify Centos and Ubuntu Source Install - Keystone  https://review.openstack.org/21760109:35
SamYapleplease review the above change, I am about to make alot of those changes and dont want wasted effort09:36
inc0SamYaple, you've said yesterday that you have idea what was wrong with Stevens ansible - ansible not finding interfaces09:38
inc0care to share what's the idea?09:38
SamYaplewhats wrong is he was trying to lookup interfaces on hosts he wasn't addressing in that role09:39
SamYaplea few ways to solve that issue09:39
SamYaplefact caching is one, giving all host to the haproxy role is another09:39
SamYapleill submit a patch for it tonight09:39
SamYapleits easily reproducable09:40
inc0yeah, we've just hit something similar09:40
inc0hence my question09:40
SamYaplebest this i can say for now is add all of your hosts to the network group09:40
SamYaplethats the workaroudn that works with no code change09:40
SamYapleotherwise ill have a patch up tonight09:40
SamYaple(a few hours)09:40
inc0ahh, I see now...yeah that makes sense09:41
inc0but that being said, I don't think its case for us09:41
inc0hold on09:42
SamYapleif you post your error ill look at it09:42
inc0nah, let me have some fun09:42
SamYaplego for it!09:42
SamYaplebut if its an issue post it someone for others too09:42
openstackgerritSam Yaple proposed stackforge/kolla: Unify Centos and Ubuntu Source Install - Glance  https://review.openstack.org/21760509:45
inc0ehh if I'd have +2 keystone would be merged already :(09:47
SamYaplenah i gots to test them all. its fine09:50
*** dims has joined #kolla09:50
coolsvapinc0, we need to utilize/calibrate/test the gates we are setting up09:52
SamYaple^09:52
openstackgerritMerged stackforge/kolla: Remove tiered building in build.py  https://review.openstack.org/21669109:52
inc0yeah, functional testing would be cool actually09:53
inc0I hope we'll get them early M09:53
SamYapleinc0: im going to put them in rc1, sdake doesnt think i have time but i do09:54
inc0somehow I don't doubt that you will Sam09:55
inc0anyway, we all need to discuss how exactly we want them to look like09:56
inc0it's not trivial09:56
inc0and I for one dislike tempest, its faulty09:56
SamYapledoubt me all you want, ive failed nothing!09:57
SamYapleinc0: im talking about different functinoal testing09:57
SamYaplesorry for the confusion09:57
inc0which one then?09:58
SamYapleansible deploy in the gate and boot instance09:58
inc0well, something as simpe as running rally stuff over deployed kolla might be good09:58
SamYaplesure, but the fancy testing will probably be in M09:58
inc0will we still be bound by 40mins time limit tho? because well...09:58
SamYaplebarebones, does it deploy? does it boot?09:58
SamYapleinc0: 60 mintutes are the gate09:59
SamYaplewe can get 90 if needed09:59
SamYaplebut we can do it in 6009:59
SamYaplebuild in ~25, deploy in ~5, test == ~30 minutes09:59
inc0more complex test would be to bood volume-backed instance with heat and add floating ip to it10:00
inc0we test pretty much all core projects with this one10:01
inc0and it's single heat stack deploy10:01
SamYapleyea initial test will be cirros - does it boot? does it ping internet? can i ssh to it over floating ip?10:01
SamYapleheat no worky10:01
SamYaplebut im sure sdake will want that :)10:01
inc0I'm kinda afraid we'll end up being CI for openstack10:07
inc0whole openstack10:07
SamYapleafraid?10:07
inc0someone breaks their project, kolla gates are busted10:07
SamYaplethats been my goal from the start10:07
SamYaplenot the way i have it in mind :)10:07
inc0what do you have in mind?10:08
SamYaplewell the kolla gate by its very nature can't be voting. its up to the cores to review the logs of the failed gate10:09
SamYapleadditionally we are going to hit issues with stable and oldstable branches10:09
SamYaplewith versions of things10:09
SamYaplesince we always pull the newest that _will_ break things eventually10:10
SamYapleversion capping will help, but stil itll happen10:10
SamYaplenow for the openstack as a whole ci stuff, each project wont have to build all of openstack, just thier project10:11
SamYapleand there could be daily/hour cached and updated kolla images10:11
SamYaplethat would be far greater for functional testing than anything we currently have10:12
inc0well, if it's non voting10:12
SamYapleit would have to be10:12
SamYapleto many projects, too many changes10:13
SamYapleneutron shouldnt be able to break glance devel10:13
SamYaplebut hey if it works 80% of the time, thats _way_ better than the nothing we have right now10:13
inc0that's essentially my point10:17
inc0we'll have failing gate even if kolla is ok10:17
inc0and it will be increasingly hard for cores to differ between incorrect kolla code and broken service10:17
inc0look at you with heat10:17
inc0I mean it could be doable, we could have repeated CI for kolla master and if something breaks there, that means it's broken in service10:18
SamYaplewait wait10:19
SamYapleyou think kolla code is going to grow?10:19
inc0then all changes to existing containers could be compared to stable kolla errors10:19
inc0actually I do10:19
inc0because openstack changes10:19
SamYaplewell that right there is the problem10:19
pbourkeSamYaple: why do a lot of the services seem to have home directories now?10:19
pbourkejust wondering10:19
SamYaplepbourke: all services always had home directories pbourke10:19
pbourkeummmm10:19
pbourkeI dont think so :)10:19
SamYapleuseradd on centos creates them, but not on ubuntu10:19
pbourkeah10:20
pbourkemaybe we should be doing --system10:20
SamYapleno su for you then10:20
inc0SamYaple, software is finished with death of its last user10:21
inc0kolla will have to be maintained, expanded10:21
SamYaplei disagree on expanded10:21
SamYaplethats feature and scope creep10:21
inc0well right now we have very simplistic ansible10:21
inc0people will want more10:22
SamYaplepeople can write more10:22
SamYaplethis was always a basic ansible template10:22
inc0people will want coreos compatibility10:22
*** weiyu_ has quit IRC10:22
inc0yes, but if they write, why not put it upstream as long as it's correct?10:22
inc0and useful?10:22
inc0I would hate to create another "we don't want it here" project, like nova is right now10:23
inc0I'm not saying accept all, but keep an open mind10:23
pbourkeSamYaple: gotcha10:24
SamYapleopen mind, sure. but you are describing scope creep10:24
SamYapleand feature creep10:24
SamYapleyou started talkign about distrubuted building the other day10:24
SamYaplenever gonna happen10:24
inc0there is a thin line and its up to cores to not cross it10:24
pbourkeits causing failures in OL, Im not sure about centos but id be surprised if not10:24
pbourkeas /home/blah already exists10:24
pbourkewhen it tries to mkdir10:24
SamYapleis that a _failure_?10:25
pbourketis10:25
SamYapleit should just be a warning10:25
pbourkelet me check recent CI builds10:25
SamYaplethe home stuff has been in the code for quite a while10:25
SamYaple(in kolla times)10:25
inc0ok, about distributed build, I throw 400 ideas a minute, 99.9% of them won't ever happen and I'm not saying it should10:25
SamYaplefyi i stopped pushing patches while i test locally10:26
inc0but there is always something that actually should land10:26
SamYaplei dont disagree, but it will slow down was my point10:26
inc0anyway, afk for 40mins, meeting \\o o//10:26
SamYapleand it will10:26
*** Kennan has quit IRC10:38
*** Haomeng|2 has quit IRC10:39
*** zhiwei has quit IRC10:40
*** zhiwei has joined #kolla10:40
SamYaplepbourke: https://review.openstack.org/#/c/217601/10:43
SamYapleit appears to be passing the gate just fine10:43
SamYaplethats centos10:43
SamYapleand it works fine on ubuntu10:43
pbourkeSamYaple: I didnt think it was that review?10:43
pbourkeSamYaple: anyway it might an oraclelinux only thing, im testing atm10:43
SamYapleyea its in there10:44
SamYaplebut from like 2 weeks ago10:44
SamYapleits been merged for a while10:44
pbourkewhy are you -2'ing the above review though10:44
pbourkeits unrelated to the home dir10:44
SamYapleoh right, was just trying to make sure a breaking oraclelinux change didnt slip through10:45
pbourkethe build.py output is bizarre10:45
pbourkereally must look at it10:46
SamYapleits because the logging doesnt spit out each line but saves it all up until the end of the build10:46
SamYaplei want to change that to print everything out as it gets it10:46
SamYaplei can grep for the image name and see the logs10:46
pbourkewell theres that, but there's also -T 110:46
SamYapleyea but thats sllooooo10:46
SamYaplew10:46
pbourkewhich tries to give real time logs, but seems to mix up the output10:46
pbourkeSamYaple: does it ever seem to just hang indefinitely on you10:52
SamYaplepbourke: whats youre docker storage backed?10:52
SamYapleonly time it hangs like that is when docker is applying layer10:53
SamYapleyou should see that if you `ps aux | grep layer`10:53
pbourkewhichever comes default on centos - devicemapper?10:53
SamYaple`docker info`10:53
SamYapleit will tell you10:53
pbourkeyeah its devicemapper10:55
pbourkenothing about layers in ps10:55
*** Kennan has joined #kolla10:55
SamYaplewell thats all i got!10:55
pbourkewe've been experiencing a lot here since moving to build.py10:55
pbourkemaybe yesterdays patch will improve10:55
pbourkeit10:55
SamYapleare you running with -T110:56
SamYaple?10:56
SamYapleto me it sounds like your storage backend cant keep up10:56
SamYaplewhats your iowait times?10:56
vbelgood morning/evening everyone. I am missing generic 'data' container - is there such today? :)10:57
vbelI mean the one with volumes for maria/rabbit/cinder10:57
SamYaplevbel: we dont specify VOLUMES in the dockerfiles anymore10:58
SamYaplewe use ansible to launch a generic data container wit hte needed volumes10:58
jmccarthyTASK: [mariadb | Starting MariaDB data container] *****************************10:58
jmccarthyfailed: [localhost] => {"changes": ["{\"status\":\"Pulling repository docker.io/kollaglue/centos-binary-data\"}\r\n", "{\"errorDetail\":{\"message\":\"Error: image kollaglue/centos-binary-data:latest not found\"},\"error\":\"Error: image kollaglue/centos-binary-data:latest not found\"}\r\n"], "failed": true, "status": ""}10:58
vbelok, I don't see it is build10:58
jmccarthyI suppose I missed one of the arguments to build.py10:59
pbourkejmccarthy: build data10:59
jmccarthyok added data10:59
vbelok, I see, need to build.py  'data'11:00
vbelthanks!11:01
jmccarthyD'oh sorry for the noise11:01
vbelif you don't do full build it can be confusing to find there is no data part ;)11:02
SamYapleif you dont specify any options, it will build all containers11:02
SamYapleincluding data11:02
*** zhiwei has quit IRC11:18
*** rhallisey_ has joined #kolla11:18
pbourkeSamYaple: what backend do you use?11:21
pbourkeSamYaple: also do you build on bare metal?11:21
SamYaplei do build on baremetal11:21
SamYaplei use btrfs or zfs11:21
SamYaplebtrfs is more stable and a bit faster11:22
SamYaplethan zfs*11:22
SamYaplegit branch11:22
SamYaplegit branch11:22
SamYapleoops sorry11:22
openstackgerritSam Yaple proposed stackforge/kolla: Fixes any() in an if statement build-py  https://review.openstack.org/21764611:27
SamYapleguys ^^ there was a syntax issue in the tiering patch11:27
SamYaplei guess i never tested locally replacing len with any pbourke11:27
pbourkemy bad also for not giving correct example11:27
SamYapleits on me, i know how to use nay :(11:28
pbourkearg11:28
pbourke"blah returned a non zero error code"11:28
*** pbourke has quit IRC11:28
*** pbourke has joined #kolla11:29
SamYapleawesome11:42
SamYapleall of the incoming patches have been tested11:43
openstackgerritSam Yaple proposed stackforge/kolla: Unify Centos and Ubuntu Source Install - Ceilometer  https://review.openstack.org/21765511:43
openstackgerritSam Yaple proposed stackforge/kolla: Unify Centos and Ubuntu Source Install - Cinder  https://review.openstack.org/21765611:43
openstackgerritSam Yaple proposed stackforge/kolla: Unify Centos and Ubuntu Source Install - Designate  https://review.openstack.org/21765711:43
openstackgerritSam Yaple proposed stackforge/kolla: Unify Centos and Ubuntu Source Install - Glance  https://review.openstack.org/21760511:43
openstackgerritSam Yaple proposed stackforge/kolla: Unify Centos and Ubuntu Source Install - Gnocchi  https://review.openstack.org/21765811:43
openstackgerritSam Yaple proposed stackforge/kolla: Unify Centos and Ubuntu Source Install - Heat  https://review.openstack.org/21765911:43
inc0Sam...11:44
openstackgerritSam Yaple proposed stackforge/kolla: Unify Centos and Ubuntu Source Install - Horizon  https://review.openstack.org/21766011:44
inc0really..11:44
openstackgerritSam Yaple proposed stackforge/kolla: Unify Centos and Ubuntu Source Install - Ironic  https://review.openstack.org/21766111:44
openstackgerritSam Yaple proposed stackforge/kolla: Unify Centos and Ubuntu Source Install - Keystone  https://review.openstack.org/21760111:44
inc0stahp11:44
inc0Sam11:44
openstackgerritSam Yaple proposed stackforge/kolla: Unify Centos and Ubuntu Source Install - Magnum  https://review.openstack.org/21766211:44
openstackgerritSam Yaple proposed stackforge/kolla: Unify Centos and Ubuntu Source Install - Murano  https://review.openstack.org/21766311:44
SamYaplenever11:44
openstackgerritSam Yaple proposed stackforge/kolla: Unify Centos and Ubuntu Source Install - Nova  https://review.openstack.org/21766411:44
openstackgerritSam Yaple proposed stackforge/kolla: Unify Centos and Ubuntu Source Install - Swift  https://review.openstack.org/21766511:44
openstackgerritSam Yaple proposed stackforge/kolla: Unify Centos and Ubuntu Source Install - Zaqar  https://review.openstack.org/21766611:44
SamYapleis it over...?11:44
inc0here goes stackalitics review boost11:44
SamYapleactually i hate this many patches11:45
SamYaplebut anytime it gets slighty large people complain that i should have broken it up11:45
inc0it's worse when they touch code that changes often11:45
SamYaplebuild.py latest patch for example11:45
inc0trust me, I know11:45
inc0many patches are good, one big patch is bad11:45
SamYaplei know, but then you end up with this thing that just happened11:46
inc0it's not bad, it's better than single11:46
SamYapleagreed, but you understand11:46
inc0I do11:47
inc0I just noticed that there are keyboard shortcuts in gerrit11:48
SamYapleill never catch up with sdake on reviews :(11:48
inc0brave new world guys, brave new world11:48
SamYaplewhat...11:48
SamYaplewhat do they do11:48
inc0get to code review and press ?11:49
SamYaplecoolsvap mandre if one of you could ack this, it breaks build.py summary https://review.openstack.org/#/c/217646/11:49
SamYapleawesome!11:49
SamYaplewowzers11:50
inc0all I need now is shortcut to +111:51
SamYaplepbourke: have you tested horizon?11:51
inc0-1 I need to do manually, but thats only moment when I need to touch mouse11:51
SamYaplepbourke: it looks like there might still be some icons missing. i thought it was just my browser but idk11:51
SamYaplei would like to start building openvswitch from source11:53
SamYaplewhat are your opinions on this?11:54
SamYapleit wont be the module, that is still on the host system11:54
SamYaplebut the user-utils don't have to match the kernel version (they just should be newer)11:54
SamYaplewell gain benefits from the newer utils even on older kernels accordning to the openvswitch folks11:55
inc0SamYaple, https://review.openstack.org/#/c/217656/1 why do you unlink builds out there?11:57
SamYapleoh thats leftovers from a bad docker -> docker_templates move11:57
SamYaplethose build scripts wont work with the template11:57
SamYaplethats for bash11:57
inc0SamYaple, about ovs...is there concrete benefit of having that?11:58
SamYaplehttp://openvswitch.org/releases/NEWS-2.4.011:58
inc0I know they won't, but we still have non-templates right?11:58
SamYapleyou tell me11:58
SamYapleinc0: yea but thats only in the templates directory inc011:58
SamYaplenot the other directory11:58
inc0a yeah, makes sense11:59
inc0I missed that it's templates directory11:59
SamYaplei just saw them as i was going and removed them11:59
SamYaplehorizon is being super slow in this container11:59
inc0I blame keystone12:00
SamYapleits weird. we are actually getting to the 'i have to debug openstack proper' stage of this project12:00
inc0but that's like automatic reaction for me...somethings broken == neutron fault, somethings slow == keystone fault12:00
SamYaplethats cool12:00
SamYapleim going to do a WIP for openvswitch from source12:01
SamYaplewe already have the build tools installed12:01
SamYapleoh and neutron is now using the ovs python bindings!12:01
SamYaplethey arent wrapping ovs-vsctl anymore!12:01
inc0then do we even need newer ovs?12:01
SamYaplealways12:02
SamYapleyou always need newer ovs12:02
inc0in container I mean12:02
SamYapleyea?12:02
inc0ovs itself will be on host12:02
SamYaplei mean you could do that, yes12:02
SamYapleeven now12:02
SamYaplebut boooo12:02
SamYapleovs works way better in a container12:02
inc0eee?12:03
SamYapleseperating the db from the vswitchd is awesome12:03
inc0that's a new for me...why?12:03
inc0ah, yeah..12:03
SamYaplebecause ovs doesnt really do anything12:03
SamYapleit just talks to the kernel module12:03
SamYapleso we have far greater control over it in a container12:03
SamYaplethe vswitchd can fail, while neutron still updates the db12:03
SamYaplewhen the vswitchd comes back, it will pull in those changes12:04
SamYapleneutron never freaks out12:04
SamYaplethanks rhallisey_12:07
rhallisey_inc0, were using ovs in a container dewd12:07
*** rhallisey_ is now known as rhallisey12:07
inc0I know, but I didn't know why its "better"12:07
SamYapleinc0: ive found it to be more stable12:08
rhalliseyinc0, it actually it very tricky because of the networking stuff we're doing12:08
inc0yeah I know, I mean I know it works12:09
inc0just Sam said it's better and since it's kernel module really, I was wondering what's the difference between container and non-container12:10
inc0but separation of ovsdb makes sense, right12:10
rhalliseydon't know12:10
SamYaplealso that ovsdb is a sqlite database that never gets truncated12:10
SamYaplea new container will truncate it12:10
SamYaplewoo12:10
SamYapleold installs get wieghed down by that in a large environemtn12:11
SamYaple(after years of lots of port changes)12:11
openstackgerritVladislav Belogrudov proposed stackforge/kolla: Add placeholder to unset http proxies  https://review.openstack.org/21767812:11
*** jasonamyers has joined #kolla12:12
*** dims has quit IRC12:14
openstackgerritVladislav Belogrudov proposed stackforge/kolla: Add placeholder to unset http proxies  https://review.openstack.org/21767812:16
SamYapleakwasnie: have you started on https://bugs.launchpad.net/kolla/+bug/1485742 ?12:20
openstackLaunchpad bug 1485742 in kolla "global search and replace /opt/kolla with /var/lib/kolla" [Critical,Confirmed] - Assigned to Alicja Kwasniewska (alicja-kwasniewska)12:20
SamYapleif not, I can take over12:20
SamYapleits due pertty much now for L312:20
openstackgerritMerged stackforge/kolla: Fixes any() in an if statement build-py  https://review.openstack.org/21764612:25
rhalliseySamYaple, how can I create a file on a host in ansible12:25
rhalliseyhave trouble finding an example12:26
SamYapleWhat content do you want?12:26
SamYaplecopy a file?12:26
SamYaplejust touch12:26
SamYaple?12:26
rhalliseySamYaple, this would be to create the json file on a host12:26
SamYaplewhere are you getting the inital content?12:26
SamYaplea string? another file?12:26
rhalliseyI'm thinking a string12:27
openstackgerritVladislav Belogrudov proposed stackforge/kolla: Add placeholder to unset http proxies  https://review.openstack.org/21767812:27
SamYaplecopy module12:27
SamYaplecopy content={{ string }} dest=/file/on/dest/host12:27
rhalliseyok cool12:27
SamYaplecoolsvap: regarding https://bugs.launchpad.net/kolla/+bug/148023112:28
openstackLaunchpad bug 1480231 in kolla "change nova tar url to master" [High,Confirmed] - Assigned to Swapnil Kulkarni (coolsvap)12:28
SamYaplecan you update and/or close that bug?12:28
SamYaplewe do pull nova from master in build.ini12:28
*** dims has joined #kolla12:29
*** openstackgerrit has quit IRC12:31
*** openstackgerrit has joined #kolla12:32
*** inc0 has quit IRC12:38
openstackgerritSam Yaple proposed stackforge/kolla: Fixes matching issue build.py  https://review.openstack.org/21768912:38
pbourkeSamYaple: could you do me a favour?12:41
SamYaplefor you, anything!12:42
pbourke:P12:42
pbourkedoes this work for you12:42
pbourketools/build.py --base centos --type source --template nova-base -T 1 -d12:42
SamYapletesting12:42
SamYapleroot@server02:~/kolla/tools# tools/build.py --base centos --type source --template nova-base -T 1 -d12:42
SamYaple-bash: tools/build.py: No such file or directory12:42
SamYapleim kidding ;)12:42
SamYapleits building now pbourke12:43
pbourkekk12:43
SamYaplewhat am i looking for12:43
pbourkethe mkdir failing the build12:43
SamYapleok12:43
SamYapleim building on master12:44
pbourkeyeah12:44
SamYapleok12:44
pbourkei would just stick mkdir -p in and be done with it, but curious why its building in CI12:44
SamYaplei updated all my patches with a -p12:44
SamYaplebtw12:44
pbourkealso keep an eye out for the logs being jumbled12:45
SamYapleyoull have to defin jumled a bit more12:45
pbourkeI think there's something wrong with the way we're stashing them into image['logs']12:45
SamYapleoh yea im remving that in the patchset incoming12:45
pbourkeit seems you get lots of duplicate output12:45
SamYapleim udating loging12:45
SamYaplebe can discuss it there12:46
SamYaplepeople*12:46
SamYaplejeez why cant i type12:46
pbourkecool12:46
SamYaplevbel: fixed your namespace issue (i think)12:46
SamYaplehttps://review.openstack.org/#/c/217689/12:46
pbourkefor all these --no-cache-dir patches12:47
SamYapleyes12:47
pbourkeare we sure all dependencies are installed in base12:47
pbourkehow can we be sure12:47
SamYaplepip install installs the requirements file12:47
SamYaplethats why currenlty you see a bajillsion "already installed" in the logs12:48
SamYaplebecause we install the requirements, then install the folder (which installs the requirements)12:48
pbourkeso the install -r requirements step was redundant12:48
pbourke?12:48
SamYapleyes12:48
pbourkegotcha12:48
SamYapleINFO:__main__:nova-base:mkdir: cannot create directory '/home/nova': File exists12:57
SamYapleyes pbourke it failes12:57
SamYapleit wont fail with -p though12:57
SamYaplethats the point of that flag kinda12:57
SamYaple      -p, --parents12:58
SamYaple              no error if existing, make parent directories as needed12:58
pbourkeyeah I see the -p's added now, going through those patches12:59
pbourkewhy is upstream building though12:59
SamYaple?12:59
SamYapleoh yea12:59
pbourkesorry, CI12:59
SamYaplei know what youre saying12:59
pbourkealso horizon is busted13:00
SamYapleyea the CI is wierd13:00
pbourkebut not in CI13:00
pbourke:/13:00
SamYaplei dont know whats happening there13:00
SamYaplei notcied it too13:00
SamYaplelike it using old version of kolla or something13:00
pbourkemickt is fixing the horizon bug13:01
pbourkepatch incoming13:01
SamYapleits working for ubuntu fyi13:01
pbourkei dont see how13:01
pbourkeone second13:01
pbourkehttps://github.com/stackforge/kolla/blob/master/docker_templates/horizon/Dockerfile.j2#L4713:02
SamYapleyup13:02
pbourke /etc/openstack-dashboard/local_settings13:02
pbourkedoesn't exist13:02
SamYapleof course it doesnt13:02
SamYapleits copied in with config-external13:02
pbourkebut the build runs first13:03
pbourkewhat am i missing here13:03
SamYaplethat doesnt matter13:03
SamYapleyou can create symlinks pointing to nonexistant things13:03
pbourkeah ok13:03
*** mandre has quit IRC13:03
pbourkethanks13:04
*** vbel has quit IRC13:04
pbourkeneed to get all these -p patches in13:05
*** vbel has joined #kolla13:05
openstackgerritSam Yaple proposed stackforge/kolla: Remove CONFIG_STRATEGY prefix  https://review.openstack.org/21770013:07
akwasnieSamYaple, you asked if I had started on https://bugs.launchpad.net/kolla/+bug/1485742 ?13:09
openstackLaunchpad bug 1485742 in kolla "global search and replace /opt/kolla with /var/lib/kolla" [Critical,Confirmed] - Assigned to Alicja Kwasniewska (alicja-kwasniewska)13:09
SamYapleyes13:09
akwasnieI haven't13:09
SamYapleDo you plan to have it done by L3?13:09
SamYaple(2-3 days?)13:09
akwasniebetter take it over13:10
*** diogogmt has joined #kolla13:10
SamYaplewill do13:10
akwasniegood13:10
openstackgerritMerged stackforge/kolla: Unify Centos and Ubuntu Source Install - Glance  https://review.openstack.org/21760513:20
openstackgerritSam Yaple proposed stackforge/kolla: Improve cleanup script  https://review.openstack.org/21770613:25
openstackgerritSam Yaple proposed stackforge/kolla: Remove CONFIG_STRATEGY prefix  https://review.openstack.org/21770013:26
SamYaplebtw did you guys notice our gate runs 20% faster since the build tiers were removed?13:27
SamYaplethat translates to 5 minutes!13:27
rhalliseycool13:28
rhalliseyanyone else having issuing with haproxy/mariadb lately?13:29
rhalliseyhaproxy sits on 3306 and mariadb never comes up13:29
*** chandankumar has quit IRC13:30
vbelhaproxy should only listen  vip13:30
vbeland mariadb shoudl only listen host interface ip13:31
rhalliseyerr ok so it's something in globals.yml13:31
vbelthere still is a problem with unsetting http proxies - kolla ansible container works for local commands but not for things like creation of endpoints13:32
vbelbecause it does not read kolla-common.sh or anything else13:32
*** jasonamyers has quit IRC13:33
openstackgerritMerged stackforge/kolla: Fixes matching issue build.py  https://review.openstack.org/21768913:35
*** jtriley has joined #kolla13:40
*** diogogmt has quit IRC13:47
rhalliseyvbel, ya I think both a looking at vip13:49
rhalliseyhow can I fix this13:49
rhalliseyI've changed the IP a bunch no difference13:49
vbelmay be mariadb in /etc/my.cnf should only listen to ip, usually folks set bind = 0.0.0.0 or similar, need to check template13:50
SamYaplevbel: yea ansible container wont work if you unset in kolla-common :(13:53
SamYaplewe really need a footer13:53
*** ftcpops has joined #kolla13:53
vbelok, i have idea - just add entrypoint to kolla-ansible which sources kolla-common.sh and exec "$@"13:53
rhalliseythis use to work for me what happened that I have to set that manually?13:57
rhalliseySamYaple, did something change in allinone that would cause haproxy and mariadb to both look at the same ip?13:59
SamYaplevbel: i really dont like that idea, lets see if we can come up with another one and circle back to that one if we cant13:59
*** ccrouch has quit IRC13:59
rhalliseysetting a vip not letting the two run13:59
SamYaplerhallisey: what is your kolla_internal_address?13:59
rhalliseydefault14:00
SamYaplewhat is default?14:00
pbourkeid love to know why its faster14:00
SamYaplei mean it has to be an ip14:00
SamYaplepbourke: the building?14:00
pbourkeyeah14:00
rhallisey10.10.10.25414:00
pbourkejust because its reading from a single queue rather than multiple ones14:00
SamYaplebecause before at the end of the tier only one thread would be running14:00
SamYaplethen the next tier would go14:00
rhalliseyuse to work on 192.168.1.14814:00
SamYaplerhallisey: whats the first ip address of the interface on your box?14:01
rhallisey10.18.57.2014:01
SamYapleim not sure then are you positive its trying to bind the same address?14:01
SamYaplecan you post the galera.cnf?14:01
SamYaplepbourke: now all threads are always active until the last images, before it was bursty14:02
pbourkeah ok14:02
*** Kennan has quit IRC14:02
rhalliseyya just a sec14:02
SamYapleand globals.yml too i suppose14:02
*** Kennan has joined #kolla14:03
rhalliseyglobals http://paste.fedoraproject.org/260101/8422814414:04
rhalliseycontainer logs -> http://fpaste.org/260104/40684294/14:05
rhalliseyfrom mariadb14:05
SamYaplewhat about the galera.cnf?14:05
SamYaple/etc/kolla/galera/galera.cnf14:05
SamYaplei think14:05
SamYaple/etc/kolla/mariadb/galera.cnf14:06
SamYaplethat one14:06
rhalliseyneed to stop haproxy just a second14:06
SamYaplei actually have to go to dmv14:07
SamYapleill be back around just tag me on any pastes14:07
SamYapleill look at them14:07
rhalliseySamYaple, kk14:07
rhalliseyI know what it is, but don't know how this happened14:08
rhalliseywow this is funny..14:09
SamYaplethe vip is the first ip?14:09
SamYaplethats one ive seen14:09
rhalliseyno the file just say the name of the var I"m using for json14:10
rhalliseyI have no idea how14:11
SamYaplelol14:11
rhalliseythat was a surprise lol..14:11
vbelSamYaple: entrypoint does not help because we exec into container directly14:11
SamYaplevbel: i know, but the reason we call the binary directly is to parse the output14:11
rhalliseythanks SamYaple I'm good now14:12
SamYaplepassing it through other things like a start.sh script gave differnet output14:12
SamYapleid like to call ansible directly if possible14:12
SamYaplegotta go14:13
rhalliseySamYaple, good luck14:13
rhalliseydmv sucks14:14
*** jasonamyers has joined #kolla14:14
*** sdake has joined #kolla14:16
*** diogogmt has joined #kolla14:21
*** jtriley has quit IRC14:24
*** diogogmt has quit IRC14:25
*** achanda has joined #kolla14:31
*** jtriley has joined #kolla14:32
*** ccrouch has joined #kolla14:39
*** achanda has quit IRC14:40
*** dims has quit IRC14:51
*** dims has joined #kolla14:51
vbelwhat if we just squash containers to remove whatever build env stuff is included in whatever layers? Pros - clean, 15+% smaller image, Cons - registry will not share common parts15:08
vbel+ security after squashing15:10
vbelI mean it is not for build.py but a guide to clean up stuff from images15:10
*** mfalatic has joined #kolla15:11
*** chandankumar has joined #kolla15:16
pbourkewhat do you mean by squashing?15:24
*** blakec has joined #kolla15:24
pbourkevbel: ^15:25
vbelwe basically remove all layers and ENV, images become 1 layer (sometimes much smaller)15:26
pbourkeis it what's described here? http://jasonwilder.com/blog/2014/08/19/squashing-docker-images/15:27
vbelyes15:27
vbelI do it in bash with couple of commands15:27
vbelID=$(docker run -d oracle/ol-openstack-keystone /bin/bash)15:28
vbeldocker export $ID | docker import - oracle/ol-openstack-keystone:latest15:28
vbeland docker rm $ID15:28
vbelthat's it. NO PROXIES :)))))15:29
vbelno pypi mirrors or magic build flags with internal ips15:29
*** mandre has joined #kolla15:30
vbelthere is disadvantage that it takes more space in registry because sharing of layers is away. The same storage problem would happen with dockers saved as tarballs15:30
pbourkeit seems a bit more complex than just doing the unsets in kolla-common15:33
sdakei'd rather not hack thei mages15:34
sdakejust an fyi ;)15:34
sdakevbel ^^15:34
vbelwe cannot workaround ansible containers - they don't read kolla-common.sh and we do exec things into them15:35
SamYaplerhallisey: its worse!15:36
SamYaplemy social security card was stolen a while back15:36
rhalliseyholy15:36
SamYapleso now im at teh social security office15:36
SamYaplepbourke: i figured out a problem15:36
SamYaplewith the build script15:36
vbelso I have found two ways - to run ansible inside thru a script that sources common.sh (SamYaple disagreed here) or stick ENV HTTP* "" at the end of kolla-ansible template15:36
SamYapleyour gonna laugh15:37
*** sdake_ has joined #kolla15:38
SamYapleoh never mind i already fixed that15:39
SamYaplemorning sdake15:39
*** SourabhP has joined #kolla15:40
SamYaplevbel: does your custom docker 1.8 that accepts buildtime env args have api bindings?15:40
*** sdake has quit IRC15:40
SamYapleif so i could extend docker-py for you and have that work15:40
SamYaplethat sounds like the only answer here to be honest15:41
vbelSamYaple: we want abandon that patch :)  Ansible container worked well for ENV HTTP* = ""15:41
vbelso I wil check glance and keystone and if they can work with empty proxies I would prefer include footer15:42
*** jruano has joined #kolla15:43
SamYaplevbel: i like the idea of a footer15:43
SamYaplesuper flexibility15:43
vbelalso it worked well with  "command: docker exec -t kolla_ansible /entrypoint.sh /usr/bin/ansible localhost" . Yes, I like footer idea, will check now15:44
sdake_hey samyaple15:45
openstackgerritMerged stackforge/kolla: Remove CONFIG_STRATEGY prefix  https://review.openstack.org/21770015:46
SamYaplesdake im goldplating what are you gonna do about it?15:47
SamYaple:P15:47
SamYapleno im looking at fixing logging15:47
sdake_goldplating is not an anti-pattern15:49
*** akwasnie has quit IRC15:49
sdake_it is just not an optimal use of resources15:49
openstackgerritMerged stackforge/kolla: Unify Centos and Ubuntu Source Install - Nova  https://review.openstack.org/21766415:49
sdake_anyway that config strategy patch looks like goodgold platingto me :)15:49
openstackgerritMerged stackforge/kolla: Unify Centos and Ubuntu Source Install - Gnocchi  https://review.openstack.org/21765815:49
*** SourabhP has quit IRC15:50
openstackgerritMerged stackforge/kolla: Unify Centos and Ubuntu Source Install - Designate  https://review.openstack.org/21765715:50
sdake_say samyaple15:51
sdake_can you copy the config option for neutron network15:51
sdake_neutron agent\15:51
*** SourabhP has joined #kolla15:51
*** SourabhP_ has joined #kolla15:52
openstackgerritMerged stackforge/kolla: Unify Centos and Ubuntu Source Install - Cinder  https://review.openstack.org/21765615:53
openstackgerritMerged stackforge/kolla: Unify Centos and Ubuntu Source Install - Heat  https://review.openstack.org/21765915:53
SamYaplei dont follow sdake_ ?15:54
SamYapleyou mean openvswitch?15:54
sdake_yes15:54
SamYaplesure thing15:54
SamYaplemoment15:54
openstackgerritSam Yaple proposed stackforge/kolla: Make logging more better  https://review.openstack.org/21777415:54
SamYapleeveryone, try out that logging patch15:54
SamYaplewith grep and searching it should be less confusing15:55
SamYaplebut its the only way i can think of to do proper logging multithreaded15:55
SamYapleits up to you to filter it yourself15:55
SamYaplealso check the gate output once that finishes running15:55
*** SourabhP has quit IRC15:56
*** SourabhP_ is now known as SourabhP15:56
sdake_samyaple the problem with the grep to filter logs is that isnt buidt into the gate15:58
sdake_the bug was about the gate logs not being helpful ;)15:58
openstackgerritSam Yaple proposed stackforge/kolla: Display neutron plugin options in globals.yml  https://review.openstack.org/21777715:58
SamYaplesdake_: but now you can actually use find in a browser15:59
SamYaplewhich you could not before15:59
SamYaplethere is no way these _THREADED_ logs are going to be super readable sdake, filtering what we want is the only option15:59
SamYaplenow we can properly do that filtering15:59
sdake_can't you just filter for the user at the end16:00
*** jasonsb has quit IRC16:00
*** jasonsb has joined #kolla16:01
SamYaplewhat do you mean?16:01
SamYaplethats what everyone is complaining about16:01
sdake_i mean put each log message in a queue for that image16:01
SamYaplethats what we do16:01
sdake_and print out the queues at the end16:01
SamYaplewe collect all logs, printing them all out at the end means no output at all for 10-15 minutes16:02
SamYaplethis is the biggest complaints we get16:02
sdake_i think the fact that th elogs are totally not hepful in a failure should be the biggest complaint ;)16:02
SamYaplenow they will be helpful16:03
SamYapleoh look keystone failed16:03
SamYaplectrl + f keystone16:03
SamYaplehow thats why16:03
SamYapleyou cant do that now16:03
sdake_agree its an improvement if it works propery - hen't tested it16:04
sdake_samyaple i deplyed yesterday during m y demo and ovs locked up16:04
sdake_the part where it says preparing ovs blah blah16:05
sdake_any ideas on that one?16:05
*** jasonsb has quit IRC16:05
SamYaplewould need proper logs for sure16:05
sdake_ok16:06
*** sdake_ is now known as sdake16:06
sdakelet me get my day rolling and i'll see what i can produce16:06
openstackgerritMerged stackforge/kolla: Unify Centos and Ubuntu Source Install - Murano  https://review.openstack.org/21766316:07
*** jpeeler has quit IRC16:08
*** SourabhP has quit IRC16:09
openstackgerritSam Yaple proposed stackforge/kolla: Make logging more better  https://review.openstack.org/21777416:10
pbourkeSamYaple: what was the problem?16:11
SamYaplepbourke: nah i had already fixed that previously16:12
SamYaplei thought the while loop in the thread was running while16:12
pbourkeI know there's been some disucssion around paths16:12
SamYaplebut it blocks on queue.get()16:12
pbourkeI propose we use /etc/kolla everywhere16:12
pbourkelocally, on remote, and in containers16:12
pbourkeI found it extremely confusing when implementing ansible swift, and when explaining to jmccarthy it was not any easier16:13
SamYaplewell pbourke the stuff in containers arent configs16:13
SamYapleand that isn't LSB if we use /etc/kolla16:13
SamYaplewe can _mount_ intinto /etc/kolla for configs16:13
SamYaplebut the scripts have to be in /usr/share/kolla16:13
sdakeyes scripts need to be in /usr/share/kolla16:14
sdakebasically whatever is in /opt should be reworked16:14
SamYaplesdake: ill be submitting that patch tonight16:14
SamYapleitll merge before L316:14
sdakei htink /usr/share/kolla/templates should be in /var/lib/kolla/templates16:14
SamYaplehow so?16:14
sdake /usr/share is supposed to be read only16:14
SamYaplecurious, not fighting16:14
SamYapleok16:15
SamYapleas log as it is out of the way, ill make that change as well!16:15
sdakeya lets get all the path changes done in l316:15
*** jasonamyers has quit IRC16:16
jmccarthySaaam ! Yea agree pbourke - the paths and piecing stuff together broke my tired brain16:18
sdakewell i'd rather stick to lsb standards here rather then jam stuff in oen path to make it easier no developers ;)16:19
sdakeno/on16:19
pbourkewell, also ops, I'm sure will want to understand this stuff16:19
pbourkebut im not suggesting violating LSB16:20
sdakeagree- should be documented16:20
jmccarthyfreudian typo ;P16:20
pbourkehopefully the removal of /opt will simplfy things somewhat16:20
*** unicell has quit IRC16:20
sdakeso just a heads up, i'll let the release go until the 3rd for blueprints for liberty16:24
sdakefor librty-316:24
sdakebut 3rd is cutoff16:24
sdakeso geter done befor then plz16:24
openstackgerritMerged stackforge/kolla: Unify Centos and Ubuntu Source Install - Zaqar  https://review.openstack.org/21766616:29
openstackgerritMerged stackforge/kolla: Unify Centos and Ubuntu Source Install - Magnum  https://review.openstack.org/21766216:30
openstackgerritMerged stackforge/kolla: Unify Centos and Ubuntu Source Install - Swift  https://review.openstack.org/21766516:30
sdakesamyaple this is why the mysql contianer fails to work properly multinode - i'll provide full evidence via docker exec  mariadb_data ls -l /var/lib/mysql and ls -ld /var/lib/mysql16:33
sdakehttps://bugs.launchpad.net/kolla/+bug/148953716:33
openstackLaunchpad bug 1489537 in kolla "mariadb+galera doesn't work multinode" [Critical,Confirmed] - Assigned to Steven Dake (sdake)16:33
sdakei've got a wip patch to sort it out16:34
openstackgerritMerged stackforge/kolla: Unify Centos and Ubuntu Source Install - Horizon  https://review.openstack.org/21766016:34
openstackgerritMerged stackforge/kolla: Unify Centos and Ubuntu Source Install - Ironic  https://review.openstack.org/21766116:36
sdakesamyaple before you begin rework on https://review.openstack.org/#/c/217706/ please read my comments thnaks :)16:37
openstackgerritMerged stackforge/kolla: Unify Centos and Ubuntu Source Install - Keystone  https://review.openstack.org/21760116:39
jmccarthyHmm anyone see 'ProtocolError: ('Connection broken: IncompleteRead(0 bytes read)', IncompleteRead(0 bytes read))' type errors when building containers ?16:44
openstackgerritMerged stackforge/kolla: Unify Centos and Ubuntu Source Install - Ceilometer  https://review.openstack.org/21765516:45
*** jtriley has quit IRC16:45
pbourkethe logging is miles better16:45
pbourkeSamYaple: ^16:45
sdakeimage build down to 22 minutes in the gate = nice work on the gate patch samyaple16:46
sdakehopefully those lvm changes are reliable ;)16:46
pbourkeim a very happy camper with this logging patch16:47
*** jtriley has joined #kolla16:52
sdakesamyaple re the network in all hosts required, you aid you had a fix for that?16:53
*** unicell has joined #kolla16:59
*** sdake_ has joined #kolla16:59
openstackgerritMerged stackforge/kolla: Display neutron plugin options in globals.yml  https://review.openstack.org/21777716:59
*** chandankumar has quit IRC17:00
*** sdake has quit IRC17:02
*** jpeeler has joined #kolla17:03
*** jpeeler has quit IRC17:03
*** jpeeler has joined #kolla17:03
SamYaplesorry, back17:06
SamYaplesocial_security_card.get()17:06
SamYaplesdake_: im almost positive the mysql issue youre seeing will go away when you dont install mysql in the base image17:07
SamYaplewhich is probably why i dont have the issue in ubuntu at app17:07
SamYapleall*17:07
SamYaplepbourke: yea i thought that would be better :)17:07
SamYaplesdake_: i said i know how to fix it, i will fix the networking things17:07
SamYapleit will be in RC117:07
sdake_samyaple undersood thanks17:08
SamYaplei have to think on the best way to do it (there are 3-4 options)17:08
sdake_for centos we install mariadb in the baes?17:08
SamYapleindeed17:08
SamYaplei opened a bug for it17:08
SamYaplethen submitted a patch17:08
*** jasonsb has joined #kolla17:08
SamYaplebut i didnt test centos so i blocked it17:08
SamYaplei probably wont get around to testing centos ever17:08
SamYapleever* being relative to this cycle17:09
sdake_i remember seeing your patch and thinking the reaon we did that was probbly for dependencies17:09
SamYaplesdake_: https://bugs.launchpad.net/kolla/+bug/148637417:09
openstackLaunchpad bug 1486374 in kolla "Remove mysql-server from base" [Medium,Triaged]17:09
SamYaplewell you should not do that :P17:09
sdake_your proposing just mysql in the mariadb container?17:09
SamYaplepull those dependaciesi n better17:09
SamYaplealso move the mariadb repo to the base so all libraries match throughout the environment17:10
SamYaplei install python-mysqldb in my base17:10
SamYaplei think thats all you need17:10
SamYapleno idea what thats calling in centosland17:10
sdake_ok i'll hae a look at your patch and see if that fixes it17:10
SamYapleits old and outdated17:10
SamYaplei just abandonded this morning17:11
sdake_yes i understand17:11
sdake_mysql-server really shouldn't be in the bse in centos17:11
sdake_the fact that our dependencis dont' match may explain why my ssytem blows up on mariadb beyond just the permissions issue17:12
sdake_if you are installing mysql in your own image and not rm -rf /var/lib/mysql this permission problem wont happen17:12
sdake_its the rm -rf /var/lib/mysql that causes the problem because those efiles already hae correct permissions iiuc17:13
sdake_data container copies existing files to new data container17:13
SamYapleyea thats what happened to me in yaodu17:13
SamYapleanyway. I would think that would be highly related. good place to start anyway17:13
sdake_its just a different way to solve the same problem17:14
*** sdake_ is now known as sdake17:14
SamYaplebut the way that doesnt affect ubuntu which works great17:14
sdakei'll look into putting mysql into the proper container17:14
SamYaplei went on a little bug squashing trip earlier sdake. hammered out a few17:19
SamYaplelooks like they have mostly merged now17:19
sdakefwiw your the one that made me delete that directory in the first place ;-)17:19
SamYaplesdake: 19 minute gate time for source!17:19
SamYaplei said remove the contents, not delete17:20
sdakehow do you remove the contents without deleting?17:20
SamYaplerm -rf /folder/*17:20
sdakeyes that is what is in master17:20
SamYapleif you want to be fancy...17:20
SamYapleok thats not removeing the folder17:20
SamYaplejust the content17:21
SamYaplewell non-hidden content17:21
SamYaplebut mysql doesnt have any hidden content17:21
SamYapleif it does we could fix that too17:21
sdakels -a didn't turn anything up17:21
SamYapleyea i checked the first time i told you to use that17:21
SamYapleanyway. i think the issue is related to the base first and foremost17:22
SamYapleand while we will certainyl adjust teh scripts if needed, im hesitant because they work for ubuntu perfectly right now17:22
*** achanda has joined #kolla17:22
sdakewell i'll try your approach first17:23
sdakebut we may need to modify the scripts17:23
SamYaplei highly doubt we will ever _need_ to17:23
sdakethe chown i think needs to be outside the conditional17:23
sdaketime will tell17:23
*** sdake_ has joined #kolla17:26
*** sdake has quit IRC17:29
*** blakec has quit IRC17:30
*** blakec has joined #kolla17:32
*** blakec has left #kolla17:32
openstackgerritSam Yaple proposed stackforge/kolla: Make logging more better  https://review.openstack.org/21777417:37
*** loth has joined #kolla17:37
SamYaplepbourke: ^ that now catches the remaing multi line outputs i believe17:38
SamYapleall lines are prefixed now, even if the message has a \n char17:38
sdake_pbourke vbel did the pull from git work finish?17:38
SamYapleyes sdake17:38
SamYapleits merged17:38
sdake_docker dir is consistenty moving out of sync17:38
sdake_ok time to remove docker dir then ;-)17:39
sdake_what else are we missing17:39
SamYapleagreed17:39
SamYaplei think thats it17:39
SamYaplelets give it a day to confirm though17:39
sdake_pbourke a can you ack that17:39
sdake_what is the percona client equivalent to mariadb package17:40
sdake_mariadb is the client not the server17:40
SamYaplewhat do you mean?17:40
SamYaplewrong...17:40
sdake_maridb-server is the server in centos17:40
SamYaplemariadb is the server17:40
SamYapleit has clients17:40
SamYaplepercona stuff is for galera17:40
SamYapleand there is a repo for centos if you want to add it i believe17:41
sdake_http://ur1.ca/nkrbo -> http://paste.fedoraproject.org/260215/1440697317:42
sdake_samyaple ^^17:42
sdake_two yum commands in there17:42
sdake_one lists out mariadb contents one lists out mariadb-server contents17:42
SamYapleon youre tlaking about names17:42
SamYapleyea package names i dont know17:43
jpeelerSamYaple: did people really not like having non-realtime that much to want jumbled logs?17:43
SamYaplei was saying mariadb _is- a server17:43
SamYaplejpeeler: yea17:43
SamYaplei cant say i completely disagree17:43
sdake_i dont care about nonrealtime17:43
sdake_i care about the fact that the logs don't sync up with the errors coming out of docker!17:43
SamYaplebut that base image takes ~10 minutes to build on alot of systems17:43
SamYaplethats 10minutes of no logs at all17:43
SamYaplefreaks people out17:43
SamYaplesdake_: we prit everything docker gives us17:44
sdake_i think the logs all jammed together is going to confuse people17:44
sdake_yes right as docker gives it rather then in order with the build17:44
jpeeleri think it'll be obvious, just hard to deal with17:44
SamYaplejpeeler: i alwaays redirect to a log file17:44
SamYaplei have for a while because of this17:45
SamYaplethen a simple grep spits out everything i need17:45
SamYaplegrep nova-base and BAM only nova base stuff17:45
SamYapleand all of it too17:45
sdake_i think for the gate that model doesn't work at all samyaple17:45
jpeelerwell i was just thinking about that17:46
jpeeleri mean technically you could wget the logs17:46
SamYaplectrl + f keystone17:46
SamYaplethats all teh keystone logs17:46
jpeelerha17:46
SamYapleeasy to find the error17:46
SamYapleits the last enttry ;)17:46
jpeelerwell, i just thought i would voice it here rather than disagreeing on the review17:46
SamYaplei was think we could pump out the full log in order at teh end of the script _it_ it errors17:47
SamYaplethat should satisfy everyone17:47
sdake_that sounds fantastic17:47
SamYaplewill do17:47
jpeeleryeah i like that17:47
SamYapleeveryone happy?17:47
SamYapleany other thoughts?17:47
sdake_will create double logs17:47
sdake_but thats ok I think ;-)17:47
SamYapleeh17:47
SamYapleits the gate17:47
sdake_we may have to add a flag later17:48
* jpeeler wonders how far SamYaple goes to make me happy17:48
sdake_to select one way or another17:48
SamYaplejpeeler: for a fellow south carolina, i think you know17:48
sdake_but i am not keen on adding flags without alot of requests17:48
jpeelerSamYaple: :)17:48
SamYaplesdake_: ill just make hte extra error output debug info17:48
sdake_that sounds ideal samyaple17:48
sdake_whats going on with that upstream heat bug samyaple17:49
sdake_did any heat cats look at it?17:49
SamYaplei gave up17:49
SamYaplei pdb'd for 15 hours on it17:49
SamYaplei was stepping through sqlalchmey17:50
SamYapleit makes no sense17:50
sdake_code used to be easier to understand17:50
SamYaplei dont tihnk its heat17:50
sdake_second design syndrome17:50
SamYaplefull error log print before or after summary prints?17:51
sdake_before17:51
SamYaplei would think summary should be bottom17:51
SamYapleyea17:51
sdake_that way people can scroll to bottom to see the summary17:52
sdake_make sur the debug output has the data from docker in it ;-)17:52
SamYapleduh17:52
SamYaplethis is why i created the 'logs' key originally17:52
sdake_i haen't looked at or tested your patch ;)17:53
sdake_trying to get multinode in operational state for centos ;-)17:53
SamYapleno the 'logs' key was in the initialy build.py patch17:53
SamYaplei thought ahead!17:53
sdake_since its busted for everyone i appears that have testedit17:53
sdake_jpeeler re the ansibel_eth0 thing17:53
sdake_jpeeler samyaple has a workaround for the moment17:54
sdake_place *ALL* your nodes in the [network] section17:54
SamYaplethis is temporary17:54
sdake_duh17:54
SamYaplei will have a patch for you guys17:54
jpeeleryeah that makes sense, when i start looking at that again i'll probably just look at focusing on all in one rather than multinode17:54
SamYaplejpeeler: or multinode running all things17:55
SamYaplethats what i do17:55
jpeelerSamYaple: i was trying to get multinode working (with vagrant). this was some time ago though17:56
*** Kennan has quit IRC18:03
*** Kennan has joined #kolla18:03
SamYaplesdake_: youll want to stop using -T100018:07
SamYaplebtw18:07
SamYaplethe threads are in daemon mode now, they stick around for the life of the program18:07
SamYaplethat actually generates 1000 threads now18:07
SamYaple-T50 is more than enough18:08
SamYaplethe defalut is actually good now18:08
SamYapleit wont get much faster i think18:08
openstackgerritSam Yaple proposed stackforge/kolla: Make logging more better  https://review.openstack.org/21777418:08
*** ftcpops has quit IRC18:10
openstackgerritSam Yaple proposed stackforge/kolla: DO NOT MERGE - gate haxoring  https://review.openstack.org/21783118:11
SamYaplehey matt dorn +1'd the logging patch18:12
SamYaplei sit near him18:12
SamYaplehe builds our training courses18:12
sdake_hey sam can you rebase the heat patch with whatever work is missing18:15
sdake_you seemed to imply the rebase went awry18:15
SamYapleim super pissed. i foobar'd that an all my ansible code went away that made heat work right for everyone (not just your machines)18:16
SamYaplei rebased it after a 10 hour pdb session18:16
SamYaplethat was my mistake :(18:16
sdake_bummer18:16
SamYaplebut yea itll get done18:16
sdake_better them rm -rf /usr/lib/python2.7/site-packages18:16
sdake_I've done that several times :)18:16
SamYaplethat would be fine on my ubuntu continer18:17
SamYaplethat is a symlink!18:17
*** SourabhP has joined #kolla18:21
openstackgerritSam Yaple proposed stackforge/kolla: Unify Centos and Ubuntu Source Install - Neutron  https://review.openstack.org/21783618:21
SamYapleoops i missed neutron earlier!18:21
SamYapleall good now18:22
rhalliseyshame18:22
rhalliseyshame18:22
SamYaplenooooooooo18:22
rhallisey(any GoT fans :p)18:22
SamYapleits a reoccuring dream for me18:22
rhalliseylol18:22
SamYapleso guys18:25
SamYapleopenvswitch from source18:25
SamYaplewe should18:25
SamYapleany questinos?18:25
SamYapleit actually might take to long to compile, but if it doesnt we totally should18:26
sdake_lets focus on librty-3 blueprints that are not finished plz :)18:26
SamYaplebuzzkill18:26
SamYaplebtw https://review.openstack.org/#/c/217774/18:27
SamYaplethat 'implements' the build-script blueprint18:27
sdake_ok ok i'll revew it18:27
openstackgerritSam Yaple proposed stackforge/kolla: Unify Centos and Ubuntu Source Install - Neutron  https://review.openstack.org/21783618:29
SamYaplethat one implements install-from-ubuntu blueprint18:30
sdake_looks good sam18:30
sdake_waiting on gate results18:30
SamYapleindeed18:30
SamYapleim watching them18:30
sdake_make sure gate still works18:30
SamYaplelooks good output18:30
sdake_docker cpu = 500% lol18:33
sdake_nice --no-cache-dir didn't know about that18:34
sdake_that should speed up builds alot18:34
SamYapleidk maybe. either way it should prevent double writes18:35
SamYapleits sequential ~50MB write, so a second or two of disk time18:35
SamYaplenot that much18:35
sdake_jpeeler or rhallisey can one of you review https://review.openstack.org/#/c/217836/18:36
sdake_i arrive in tokyo on saturday if anyone wants to hang out ;)18:39
SamYaplegoing a little early arent you18:40
sdake_adjust to tz time18:41
SamYaplestill.. like a few months early18:41
sdake_samyaple after removing mariadb from base i get this18:41
sdake_TASK: [mariadb | Creating haproxy mysql user] *********************************18:41
sdake_fatal: [localhost] => Failed to template {{ (haproxy_user.stdout.split('localhost | SUCCESS => ')[1]|from_json).changed }}: an unexpected type error occurred. Error was expected string or buffer18:41
sdake_any idea where to start debug on that one18:41
SamYaplei think i should start with a good nights sleep18:42
SamYaplei should have been doing that 4 hours ago18:42
SamYaplei have to drink heavily tonight18:42
*** erkules_ is now known as erkules18:42
SamYaplegotta rest up18:42
*** erkules has quit IRC18:42
*** erkules has joined #kolla18:42
sdake_150827 18:39:32 [Warning] Hostname 'bigiron' does not resolve to '192.168.1.100'.18:43
sdake_150827 18:39:32 [Note] Hostname 'bigiron' has the following IP addresses:18:43
sdake_150827 18:39:32 [Note]  - 92.242.140.218:43
sdake_groan18:43
SamYaplelooks like dns problems to me!18:43
sdake_what is haproxy doing when it generates that error18:43
sdake_this is AIO18:43
sdake_which worked previously18:43
SamYapledunno man good thing its a warning18:44
sdake_i am talking about  the task faiure above18:44
SamYaplethats not haproxy failing18:44
SamYaplethat the database not being accessible to create an haproxy user18:44
SamYaplehaproxy has nothing to do with it18:44
sdake_what creates the database?  the ansible_kolla container?18:45
SamYapleim going to bed18:45
SamYaplecant continue at the moment18:45
sdake_enjoy18:45
SamYaplei was only here for the okes18:45
sdake_what is an okes18:46
SamYaplethats the joke18:46
SamYaplei mean i misstypell18:46
SamYaplehey if you guys wanna ack this heat bug after you confirm it that would be great18:47
SamYaplehttps://bugs.launchpad.net/heat/+bug/148877918:47
openstackLaunchpad bug 1488779 in heat "stacktrace on heat stack-delete complaining about isotime()" [High,In progress]18:47
SamYapleget some more dev love on it18:48
sdake_looks like heat is incompatible with our version of mariadb18:53
sdake_is that accurate?18:53
SamYapleno18:54
SamYaplei mentioned below there is something more going on18:54
*** jruano has quit IRC18:54
SamYapleshardy tested against mariadb 10.0.20 no problem (i tested, problem)18:54
SamYapleit looks like the problem is outside the heat code so we need to do a full dep compare18:55
sdake_he tested the galera uilt version?18:55
SamYapleits not the database trust me18:55
sdake_uil/tbuilt18:55
SamYapledatabase has no problem with fractional seconds18:55
SamYaplei tested that by hand18:55
*** jruano has joined #kolla19:00
sdake_samyaple where does this module come from: the python mysqldb module is required19:02
sdake_mysql_user seems to be requiring it19:02
SamYapleon centos i dont know19:02
sdake_there is a mysql_db in the core directory19:02
SamYaplei thinks its MYsql-devel19:02
sdake_does i tmean that or something else?19:02
SamYapleon ubuntu the package name is called python-mysqldb19:03
SamYapleits a distro package sdake19:03
sdake_can you find the upstream of it in the packaging name on ubuntu samyaple?19:03
sdake_using some apt command19:03
SamYaplepython-mysqldb19:04
sdake_i am not finding anything19:04
SamYaplethats the name19:04
sdake_no i mean the upstream pip repo19:04
SamYapleits not in pip19:04
SamYapleits a distro package19:04
sdake_i see19:04
sdake_its not called that on centos19:04
SamYaple19:03 < SamYaple> i thinks its MYsql-devel19:04
sdake_can you paste a list of files it installs on ubuntu19:05
SamYaplei dont know though19:05
sdake_i can find it that wya19:05
sdake_like rpm -qf works19:05
SamYaplenot at the moment19:05
sdake_or mayee its -ql19:05
SamYaplei is in bed19:05
sdake_groan19:05
SamYapleask the internet19:05
SamYapleits a very common package19:05
SamYaplesdake: gate says yes https://review.openstack.org/#/c/217774/19:06
SamYaple(20 minute build time woohoo)19:06
SamYapleyum install MySQL-python19:07
SamYapleliterally the first link in google19:07
openstackgerritMerged stackforge/kolla: Make logging more better  https://review.openstack.org/21777419:09
SamYaplebest commit name ever19:09
*** chandankumar has joined #kolla19:15
*** ftcpops has joined #kolla19:16
sdake_hey ftcops did you get your system rolling19:16
ftcpopsNo, in fact I came home a little bit ago to get around my corporate proxy and it's still failing the same way19:27
ftcpopsI set up a Fedora 21 VM and was getting the containers built over there when I came home.  I'll try to get it working there next19:28
openstackgerritMajor Hayden proposed stackforge/kolla: Fixing CentOS quickstart docs  https://review.openstack.org/21786019:35
openstackgerritMerged stackforge/kolla: Unify Centos and Ubuntu Source Install - Neutron  https://review.openstack.org/21783619:37
sdake_ftcpops could you go into more detail about whatis failing and what command your typing19:40
sdake_ftcpops memory fading :)19:40
ftcpopsdocker_pull_policy: always ==> http://paste.openstack.org/show/428765/ and19:43
ftcpopsdocker_pull_policy: missing ==> http://paste.openstack.org/show/428766/19:43
*** jruano has quit IRC19:57
*** achanda has quit IRC19:58
sdake_your not pulling from a registry19:59
sdake_so y9ouo want to use pull policy missing19:59
sdake_that tells kolla to use the locally registered images in the docker image cache20:00
sdake_you see this erro in the missing20:00
sdake_msg: Docker API Error: Cannot start container 3043d21824f95fad1ecdcc2ea8f343030427ab952383e04d7cddeb2ac41290a0: [8] System error: exec format error20:00
sdake_which version of docker arey ou using?20:00
sdake_i suspect you are out of disk space20:01
sdake_how much disk did you give your vm?20:01
ftcpopshm taht could be20:01
ftcpopsI gave the VM 20Gb20:01
sdake_fedora 21 by default does something wierd like give /home 90% of the disk20:01
sdake_you have to mnaully specify your disk ormat20:01
sdake_you need ~100gb of disk space20:01
sdake_maybe more20:01
ftcpopsthen 20Gb won't be enough ;-)20:01
ftcpopshehe20:01
ftcpopsokay, let me do some resizing and I'll try again20:02
sdake_yup that is what the exec format is about20:02
*** chandankumar has quit IRC20:02
sdake_make sure /var filesystem has enough diskspace20:02
sdake_again fedora21 will put most storage into /home20:02
ftcpopsok, will do20:02
*** jasonsb has quit IRC20:06
*** jasonsb has joined #kolla20:06
jpeelerdocker info should give insight into disk space usage20:07
openstackgerritRyan Hallisey proposed stackforge/kolla: WIP:Replace config-external with a JSON file  https://review.openstack.org/21687420:07
jpeeleri definitely didn't think exec format had anything to do with disk space - if so, that error is horrible20:07
rhalliseyeh that's not completely right..20:08
jpeelerdoes anybody know of a service that modifies the config file of another service?20:08
jpeeler(in kolla)20:08
sdake_jpeeler i dont think that is possible20:09
jpeeleruh oh20:09
jpeeleri mean i'm sure it's possible somehow...20:09
sdake_tell me what your tyring to do20:09
sdake_is this related to ironic?20:09
jpeelerironic requires nova configuration changes20:09
sdake_ok that is possible20:10
sdake_in the nova/defaults dir under ansible20:10
sdake_ /home/sdake/kolla/ansible/roles/nova/templates20:10
sdake_there is a file claled nova.conf.j220:11
sdake_use a jinja2 conditional if ironic - configure for ironic else configure for not ironic20:11
jpeelerah ok20:11
openstackgerritRyan Hallisey proposed stackforge/kolla: WIP:Replace config-external with a JSON file  https://review.openstack.org/21687420:13
*** akwasnie has joined #kolla20:15
sdake_jpeeler i thhink you are goin to have to end up with two nova containers20:15
sdake_one for ironic and one for not ironic20:15
sdake_unless one nova server can handle vm and bare metal20:16
jpeelerhonestly don't know how that would work, but the config conditional would be the same either way at least20:16
openstackgerritRyan Hallisey proposed stackforge/kolla: WIP:Replace config-external with a JSON file for Galera  https://review.openstack.org/21788020:21
*** achanda has joined #kolla20:22
*** achanda has quit IRC20:22
*** achanda has joined #kolla20:23
openstackgerritRyan Hallisey proposed stackforge/kolla: WIP:Replace config-external with a JSON file  https://review.openstack.org/21687420:27
openstackgerritMerged stackforge/kolla: Fixing CentOS quickstart docs  https://review.openstack.org/21786020:28
sdake_rhallisey more comments for you looking better ;)20:34
rhalliseysdake_, thanks20:35
rhalliseyI fixed a bunch of those20:35
rhalliseyI wrote some c++ in there20:35
rhalliseylost my mind20:35
rhalliseysdake_, you can test it out with galera20:37
rhalliseyworks20:37
sdake_focused on liberty-3 atm no time to test ;)20:37
*** jruano has joined #kolla20:37
rhalliseythis could make it20:37
sdake_rhallisey <ryansb>sdake_: from pwd import getpwnam20:38
sdake_[13:37:53]  <ryansb>getpwnam('ryansb').pw_uid20:38
sdake_[13:38:12]  <sdake_>ryansb thanks!20:38
sdake_thats the proper way to do it20:38
rhalliseycool thx20:38
sdake_if we are to merge this i want json lists if possible20:40
rhalliseythat's fine, it will come in following iterations20:40
sdake_rhallisey where is the start.sh change made?20:41
sdake_lets get itin the bse patch20:41
sdake_rther then reworking evrything later20:41
rhalliseywhat change are you referring to?20:42
sdake_the stuff thatruns set_configs.py20:42
rhalliseyhttps://review.openstack.org/#/c/217880/1/docker/common/galera/start.sh20:42
*** jasonsb has quit IRC20:43
rhalliseythe 2nd patch I put up20:43
sdake_i see - your dependencies in gerrit are broken20:45
sdake_i believe instead of template there is a copy command20:45
rhalliseykeystone has template20:46
*** jasonsb has joined #kolla20:46
rhalliseybut it does it differently20:46
rhalliseyhttps://github.com/stackforge/kolla/blob/master/ansible/roles/keystone/tasks/config.yml#L1920:46
rhalliseysdake_, ^ that format ok?20:46
sdake_i am reviewing your patch moment20:47
rhalliseysdake_, k let me know when you're done got some more additions20:48
sdake_ok done20:52
sdake_get your patches with proper dependencies in gerrit pls :)20:52
sdake_so they are not outdated20:53
sdake_i am pretty sure there is a straight up copy operation20:53
sdake_you are right this coudl make liberty-320:54
sdake_if you hussle ;)20:54
rhalliseysdake_, once I have a template for this ya20:54
rhalliseysdake_, whats the deadline like tmr?20:54
sdake_aug31 is soft deadline20:54
sdake_sept 3rd is hard deadline20:54
rhalliseykk20:54
sdake_depends where we are in that range20:55
sdake_i am ok with not autogenerating the json for the moment20:55
sdake_we can tackle that in rc320:55
rhalliseysdake_, so I looked into that.  I didn't find ansible very friendly for that kind of operation20:56
sdake_http://docs.ansible.com/ansible/copy_module.html20:57
sdake_for which operation, genrating the json on the fly?20:57
rhalliseyit would look like:  context: { { "base_file":  { "source": "/opt/kolla/config" } }}20:57
rhalliseyso for neutron it would be 10 bigger20:57
rhallisey10x20:57
rhalliseyya on the fly20:58
sdake_ok well lets tackle generation on the fly later20:58
sdake_lets tackle list format now tho20:58
rhalliseyya sounds good20:58
sdake_we dont need a dictionary20:58
rhalliseytesting w/ copy20:58
sdake_what you want is a list of dictionaries21:01
sdake_I think [ {"source": "/optkolla/config", "dest": "destdir", "owner"="rootin", "perms" = "perms"} , { .. }, { .. } ]21:02
rhalliseyya looks good I have it on another branch will pull in after this round of testing21:03
sdake_ok well get a patch up so i can beat on the format of the code21:03
sdake_one with the pwname21:03
rhalliseyhow can I get gid?21:03
*** jtriley has quit IRC21:03
sdake_try pw_gid21:04
sdake_instead of pw_uid21:05
*** sdake has joined #kolla21:23
*** sdake_ has quit IRC21:25
*** akwasnie has quit IRC21:26
*** sdake_ has joined #kolla21:26
sdake_rhallisey close to done with another patch21:26
sdake_?21:26
rhalliseysdake_, I was but I added a few things21:27
rhalliseytest takes like 5 min21:27
rhalliseycause the build21:27
rhalliseysometimes21:27
rhalliseyif I change anything in the base image (set_configs.py) takes a few minutes21:28
*** sdake has quit IRC21:30
sdake_you put set_configs.py in the bae image?21:32
rhalliseysdake_, ya21:41
sdake_soundsgood21:42
openstackgerritRyan Hallisey proposed stackforge/kolla: Replace config-external with a JSON file for Galera  https://review.openstack.org/21788021:50
sdake_why do you need to specify SERVICE?21:56
sdake_rhallisey ^^21:56
rhalliseysdake_, oh my bad I didn't commit my comments21:57
rhalliseysdake_, k they're there now21:57
*** jruano has quit IRC22:08
sdake_rhallisey you didn't relaly address many of my comments dude22:12
sdake_like the getpwname thing for example22:13
sdake_all the trailing periods22:13
sdake_etc22:13
sdake_i've marked them again please address all comments so we can get down to the nitty gritty22:13
sdake_I realize its painful but this will be used in every container I want to make sure the code is flawless22:13
rhalliseysdake_, I'm aware a set behind you22:14
rhalliseyon comments22:14
sdake_i see22:14
sdake_well let try to get er done today22:14
rhalliseytesting base changes sucks22:14
sdake_so pepole can have a non wip to review before friday22:15
rhallisey5min a pop22:15
sdake_do this make all the changes22:15
sdake_and when i'm done with the review22:15
sdake_then test ;)22:15
rhalliseyoh I didn't push a new patch for the first one22:15
sdake_like when I think the code is gtg22:15
sdake_that might explain it22:15
rhalliseysdake_, that's why there are no changes addressed22:15
*** SourabhP has quit IRC22:24
*** SourabhP has joined #kolla22:26
*** ftcpops has quit IRC22:28
*** ccrouch has quit IRC22:36
*** unicell1 has joined #kolla22:42
*** unicell has quit IRC22:42
*** akwasnie has joined #kolla22:43
rhalliseysdake_, I can give you an untested patch if you want to look at it while I head out for a few minutes22:44
sdake_yes push it up i'll review22:45
sdake_push up both your patches please22:45
sdake_and i'll review and make comments22:45
sdake_lets get it iterated on22:45
sdake_then you can test22:45
sdake_make sure to label them Wip22:46
openstackgerritRyan Hallisey proposed stackforge/kolla: WIP:Replace config-external with a JSON file  https://review.openstack.org/21687422:49
rhalliseyok next one..22:49
sdake_why not do git review on both at same tiem ? :)22:49
rhalliseyI didn't fix the other22:50
rhalliseyyet22:50
rhalliseyonly will take a sec22:50
openstackgerritRyan Hallisey proposed stackforge/kolla: WIP: Replace config-external with a JSON file for Galera  https://review.openstack.org/21788022:56
rhalliseysdake_, ok I'll be back in an hour22:56
rhalliseysdake_, I'll come back and test it22:57
rhalliseythanks for the reviews22:57
rhalliseybrb22:57
*** rhallisey has quit IRC22:57
sdake_more comments22:59
*** ccrouch has joined #kolla22:59
*** jruano has joined #kolla23:09
*** dims_ has joined #kolla23:18
*** akwasnie has quit IRC23:20
*** rhallisey has joined #kolla23:20
*** dims has quit IRC23:22
*** ccrouch has quit IRC23:35
*** dims_ has quit IRC23:38
*** SourabhP has quit IRC23:43
*** achanda_ has joined #kolla23:56

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