Friday, 2015-08-21

*** juggler has joined #kolla00:01
*** jasonsb has joined #kolla00:06
*** diogogmt_ has joined #kolla00:12
*** dims_ has quit IRC00:12
*** diogogmt has quit IRC00:12
*** diogogmt_ is now known as diogogmt00:12
sdakejpeeler if your around can you ack that last change00:13
*** tfukushima has joined #kolla00:19
*** tfukushima has quit IRC00:23
*** dolpher has joined #kolla00:32
*** banix has joined #kolla00:32
*** tfukushima has joined #kolla00:36
*** diogogmt has quit IRC00:37
*** achanda has quit IRC00:58
*** dims has joined #kolla01:04
*** ccrouch has quit IRC01:05
*** weiyu_ has joined #kolla01:13
*** sdake_ has joined #kolla01:20
*** sdake has quit IRC01:24
*** erkules has joined #kolla01:37
*** erkules_ has quit IRC01:40
*** diogogmt has joined #kolla01:42
*** zhiwei has joined #kolla01:43
*** banix has quit IRC01:44
*** diogogmt has quit IRC01:48
*** diogogmt has joined #kolla01:49
*** erkules has quit IRC01:52
*** daneyon_ has quit IRC01:53
*** erkules has joined #kolla01:55
*** banix has joined #kolla01:56
*** tfukushima has quit IRC02:00
*** tfukushima has joined #kolla02:00
*** maximedev has joined #kolla02:00
maximedevHi guys02:00
*** achanda has joined #kolla02:02
*** tfukushima has quit IRC02:04
*** tummy has joined #kolla02:04
sdake_sup maximedev02:12
*** sdake_ is now known as sdake02:12
*** vern has joined #kolla02:17
*** Haomeng|2 is now known as Haomeng02:41
*** dims has quit IRC02:55
sdakebmace found this gem http://events.linuxfoundation.org/sites/events/files/slides/CloudOpen2015.pdf02:56
*** alisonh has joined #kolla02:57
*** alisonh has quit IRC03:03
*** unicell has quit IRC03:11
*** diogogmt has quit IRC03:13
*** loth has quit IRC03:17
*** alisonh has joined #kolla03:21
*** achanda has quit IRC03:28
*** diga has joined #kolla03:33
SamYaplesdake: you didnt change your vote https://review.openstack.org/#/c/215157/03:38
sdakei wasn't asked to do so03:39
sdakeou asked on a different review which i did03:39
sdakewhere i made an error in the review03:39
SamYapleI responded to your comments in the review and said they were not accurate03:39
sdakeok well i didn't see that03:39
sdakei'll check now hang tight03:39
sdakesamyaple done enjoy03:41
sdakethanks for the info about build-essential03:41
sdakeI don't know alot about ubuntu so its hard for me to evaluate package deletions03:41
SamYapleyea feel free to comment on them, but ths is all tested stuff03:43
sdakei am mainly concerned about breaking centos-binary03:43
sdakebut also trying to help you with the reviews on the ubuntu bits03:44
sdakelike postgresql-devel I doubt that is in essential but i guess itdoesn't matter ;)03:44
SamYapleyea there are no changes to centos-binary03:44
SamYaplesome new if blocks, but same end result for centos03:44
sdakein one of the reviews there was something03:44
sdakei dont recall which03:44
SamYapleno that was about rhel03:44
SamYaplenot centos-binary03:44
SamYaplewe dont do rhel anywhere03:45
SamYaplemostly03:45
SamYaplebut i still fixed that03:45
sdakeya that was just an ask03:45
sdakebecause i can't remmber everything03:45
sdakeold man syndrome :)03:45
*** alisonh has quit IRC03:54
*** alisonh has joined #kolla03:59
sdakesamyaple if your interested in seeing how kolla got kicked off, have a look at this video https://www.youtube.com/watch?v=rZBkCovEIZI04:18
*** dolpher has quit IRC04:20
*** achanda has joined #kolla04:20
sdakesam re the redhat comment i just left04:22
sdakewhat i meant to say is red hat has guidelines about how their trademakr is to be ussd04:22
sdakeand I don't know if "RedHat family" would fit their guideliens04:23
sdakeand RPM is just easier way for me not to have to deal with finding out :)04:23
sdakebut just fyi its Red Hat, not RedHat, they have a shit fit whenever someone says RedHat in legal in any press or code04:23
sdakethe safest way to refer to red hat is "Red Hat, Inc."04:25
Kennanhi sdake: I am trying kolla on ubuntu, it said need kernel 3.19  but seems default can not get such version,/04:26
sdakekennan you need utopic04:26
*** weiyu_ has quit IRC04:26
sdakekennan samyaple is your go to guy for getting rollign on ubuntu04:27
sdakeperosnally i'd recommend running in a centos vm04:27
Kennansdake: utopic seems kernel 3.16 or 3.1404:27
Kennanit not 3.1904:27
sdakekennan shows what I know, SamYaple said Utopic was 3.1904:28
sdakesamyaple with that recent change you made to the config dir, do all config-external files need to change04:29
sdakekennan samyaple is about but he is just getting his workday started so he may take a bit to respond04:30
Kennanok sdake: thanks. for centos04:30
Kennanwhat version do you recommend?04:30
sdakecentos 7 is minimum required version04:31
Kennanok. sdake: let me find a centos 7 if possible04:31
sdakeits a free iso download04:31
sdakekennan diversity results of the big tent: http://ur1.ca/ni02n -> http://paste.fedoraproject.org/257427/3157414404:33
Kennaninteresting sadke:04:39
*** banix has quit IRC04:40
*** weiyu_ has joined #kolla04:52
*** sdake_ has joined #kolla04:52
*** sdake has quit IRC04:56
*** unicell has joined #kolla04:58
*** SourabhP has joined #kolla04:59
*** SourabhP_ has joined #kolla05:00
*** SourabhP has quit IRC05:04
*** SourabhP_ is now known as SourabhP05:04
*** unicell has quit IRC05:05
*** unicell has joined #kolla05:06
sdake_samyaple pbourke harmw wtb reviews on https://review.openstack.org/#/c/213537/05:06
sdake_its blocking the rest of the gating going live05:07
*** weiyu_ has quit IRC05:10
*** absubram has quit IRC05:16
*** sdake_ has quit IRC05:23
SamYapleKennan: hello05:24
SamYapleKennan: did you have a question about the kernel?05:24
SamYapleKennan: i see you have a question about kernel 3.19, that is only if you are using AUFS05:25
Kennanhi SamYaple05:25
KennanI tried to figure out what's requirements for ubuntu05:25
Kennanit said it needs kernel version 3.1905:26
SamYapleWhat version of ubuntu?05:26
Kennanit is too higher05:26
Kennanhttps://wiki.ubuntu.com/UtopicUnicorn/ReleaseNotes05:26
Kennanit is 3.16 for 14.1005:26
SamYapleKennan: I would not recommend 14.10, since it is not an LTS. But in that case your only option is btrfs backing for Docker05:26
Kennanhi SamYaple:05:27
Kennanfor 14.0405:27
SamYaplein 14.04 you can install 3.19 kernel and use AUFS, but btrfs is still faster05:27
Kennanif I use btrfs,05:27
Kennanhow to configure that in kolla ?05:29
KennanI did not know where can configure to use btrfs05:29
SamYapleits docker that you would need t oconfigure05:29
SamYapleyou dont need to configure anything for kolla05:29
SamYapleKolla doesn't care about your Docker backend05:29
Kennanhi SamYaple so you means docker community have docs how to configure with btrfs ?05:30
SamYapleYes Kennan05:30
SamYapleit is refered to as a docker 'storage driver'05:31
SamYapleSo you can look up how to configure the docker storage driver for btrfs05:31
Kennanhi SamYaple what backend do you use for ubuntu ?05:32
SamYapleI use BTRFS and ZFS equally05:34
SamYaplebut with the 3.19 kernel for Btrfs05:34
Kennanhi SamYaple so you compile kernel in ubuntu ?05:43
*** tfukushima has joined #kolla05:45
*** dims has joined #kolla05:47
openstackgerritSteven Dake proposed stackforge/kolla: Convert MD to RST to publish to docs.openstack.org  https://review.openstack.org/21545805:48
*** tfukushima has quit IRC05:50
*** zhiwei has quit IRC05:52
*** dims has quit IRC05:52
*** zhiwei has joined #kolla05:53
*** dims has joined #kolla06:01
*** dims has quit IRC06:05
*** weiyu_ has joined #kolla06:06
openstackgerritSteven Dake proposed stackforge/kolla: Prepare for MD to RST conversion  https://review.openstack.org/21546306:07
openstackgerritSteven Dake proposed stackforge/kolla: Run pandoc to convert the documentation  https://review.openstack.org/21546406:07
openstackgerritSteven Dake proposed stackforge/kolla: Run pandoc to convert the documentation  https://review.openstack.org/21546406:16
*** inc0 has joined #kolla06:20
inc0morning06:20
inc0we don't have single container build yet do we?06:20
*** SourabhP has quit IRC06:26
openstackgerritSteven Dake proposed stackforge/kolla: Run pandoc to convert the documentation  https://review.openstack.org/21546406:45
*** tfukushima has joined #kolla06:46
*** tfukushima has quit IRC06:50
*** diga has quit IRC07:05
*** tfukushima has joined #kolla07:12
openstackgerritSteven Dake proposed stackforge/kolla: Run pandoc to convert the documentation  https://review.openstack.org/21546407:14
*** tfukushima has quit IRC07:16
*** jmccarthy has quit IRC07:24
*** jmccarthy has joined #kolla07:25
*** shardy has joined #kolla07:27
Haomenghi guys07:28
HaomengI run vargant up to setup env, encountered this error - "The box 'puppetlabs/centos-7.0-64-puppet' could not be found.", any comments?07:29
Haomengshould I download this box first?07:34
*** achanda has quit IRC07:35
*** zhiwei has quit IRC07:37
*** zhiwei has joined #kolla07:37
KennanHaoMeng which guide do you refer ?07:38
HaomengKennan: thanks for your helping07:38
HaomengKennan: this guide - https://github.com/stackforge/kolla/blob/master/docs/vagrant.md07:38
HaomengKennan: I think some pre-action should be taken07:38
HaomengKennan: but it is not covered by the guide07:39
Kennando you use a physical machine to setup?07:39
HaomengKennan: I run on VM, will not working with vm?07:40
HaomengKennan: I installed vagrant and virtualbox07:40
KennanI think vagrant  can not work with VM well07:40
Kennanit seems need install on physical machine for vagrant to work well. You may need to check with SamYaple07:41
Kennanvagrant sometimes up action need rely on VT intel or AMD virtuail07:41
Kennanbut VM not have such capability07:41
HaomengKennan: yes, but we can enable vm nested if it supported07:42
Kennandid you check if vagrant support that nested ?07:42
KennanI did not think how can enable such nested VT07:42
HaomengKennan: another question, I just want to play with kolla to deploy openstack, so, vagrant is not nesserary07:42
HaomengKennan: ok07:42
HaomengKennan: so which guide is easy for me to try kolla first07:43
HaomengKennan: https://github.com/stackforge/kolla/tree/master/devenv , is this fine for my fist try?07:44
KennanHaoMeng: I ma also new to kolla07:44
Kennanyes07:44
Kennanhttps://github.com/stackforge/kolla/blob/master/docs/dev-quickstart.md07:44
Kennancould be used , I remember sdake said that before07:44
HaomengKennan: ok, will try, thank you07:44
HaomengKennan: ok07:44
Kennanwcl07:44
HaomengKennan: the reason I use vargant, that because I think it should be more easier:)07:45
*** zhiwei has quit IRC08:03
openstackgerritMichal Jastrzebski (inc0) proposed stackforge/kolla: build.py refactor  https://review.openstack.org/21550408:05
*** diga has joined #kolla08:17
openstackgerritMick Thomspon proposed stackforge/kolla: Added missing configuration to horizon docker template.  https://review.openstack.org/21551108:29
*** inc0 has quit IRC08:35
*** inc0_ has joined #kolla08:35
*** inc0__ has joined #kolla08:37
*** inc0_ has quit IRC08:40
*** inc0_ has joined #kolla08:41
*** inc0__ has quit IRC08:45
*** tfukushima has joined #kolla08:45
*** diga has quit IRC08:49
*** inc0_ has quit IRC08:50
*** inc0_ has joined #kolla08:50
openstackgerritMerged stackforge/kolla: Workaround for existing keepalived.pid file  https://review.openstack.org/21538009:03
openstackgerritMichal Jastrzebski (inc0) proposed stackforge/kolla: Build.py regex filtering  https://review.openstack.org/21552309:03
openstackgerritMichal Jastrzebski (inc0) proposed stackforge/kolla: Build single image  https://review.openstack.org/21552409:03
inc0_there you go, we all needed that.09:04
*** vbel has quit IRC09:07
*** vbel has joined #kolla09:07
pbourkeinc0_: are your patches alternatives to jpeeler's?09:10
openstackgerritSteven Dake proposed stackforge/kolla: Run pandoc to convert the documentation  https://review.openstack.org/21546409:10
openstackgerritSteven Dake proposed stackforge/kolla: Prepare for MD to RST conversion  https://review.openstack.org/21546309:10
openstackgerritSteven Dake proposed stackforge/kolla: Rework documentaton to actually get people rolling  https://review.openstack.org/21553209:10
inc0_pbourke, I haven't seen Jeffs patches09:11
inc0_I just need this to work and it took me less than an hour09:11
inc0_I'm in "fix all my painpoints with build script" frenzy right now09:11
pbourkesure thing09:11
openstackgerritMerged stackforge/kolla: Add build from source and templating for Zaqar  https://review.openstack.org/20244109:11
pbourkejust fyi he does have a patch09:11
*** sdake has joined #kolla09:11
pbourkeso will have to choose09:11
pbourkemaybe you could review his and see how it compares09:12
pbourkeafter the single image building my next biggest pain point would be the logging09:12
pbourkebut im not sure how that could be solved with the paralell building09:12
pbourkemultiple log files might be nice09:13
pbourkethen you could tail them in separate windows09:13
sdakehey guys09:13
sdakedocs are my biggest pain point :)09:13
*** weiyu_ has quit IRC09:14
sdakewhats going on09:14
inc0_pbourke, is this patch upstream already?09:14
pbourkeinc0_: no its for review09:14
inc0_ah, review09:14
pbourkehttps://review.openstack.org/#/c/215323/09:14
inc0_I must have missed it09:14
pbourkesdake: roger your comment on swift docs09:14
pbourkesdake: was thinking the same, will put something together09:14
sdakepbourke if yu get it into basic shape like a 10-15 liner i can make it int othe docs09:15
sdakehey inc0 jpeeler wrote tht acode already09:15
inc0_logic is a bit different but we'll compare09:15
inc0_hold on, next patch coming right up09:15
inc0_(different one)09:16
openstackgerritMick Thomspon proposed stackforge/kolla: Added missing configuration to horizon docker template.  https://review.openstack.org/21551109:16
Kennansdake: not sleep ?09:16
sdakegod i need to be kennan09:17
sdakekennan try readin g the last commit for docs09:17
sdakeit shoud lget you going09:17
sdakethe one in the queue09:17
Kennansdake: do you know why libvirt is needed in kolla ?09:17
KennanFinally stop libvirt on the host machine. Only one copy of libvirt may be running at a time.09:17
KennanI did not understand that09:18
sdakehttps://review.openstack.org/#/c/215532/09:18
sdakelibvirt runs in a container09:18
sdakeif it were to run outside the container we coudln't do upgrades properly09:18
sdakenor support a feasible deployment model with orrect sharingof data with nova09:18
Kennanok. you make libvirt run in a container09:19
sdakeright09:19
Kennanso it can not run on the host09:19
Kennanright?09:19
sdakeright09:19
Kennanok got it. Thanks09:19
sdakethat is why you have to turn it off on the host09:19
sdakethe new docs are more clear on this point09:19
sdakecoolsvap|away where u at bro :)09:21
sdakeneed reviews!!09:21
sdakekennan id you get aio going yet?09:22
sdakeone thing missing from current master docs is the idea of building imges09:22
sdakeyou need to do that09:22
sdakeits in my docs update09:22
*** stvnoyes has joined #kolla09:23
Kennansdake: ok still fighting for that. just install kolla client now (as some other work in hands, did not do it all the time :))09:23
sdakei think realistically a kolla evaluation should take less then 30 minutes09:23
sdakeour heat and vagrant documentation need love09:24
sdakethe heat documentation is focused around compose still09:24
* sdake groans09:24
openstackgerritMichal Jastrzebski (inc0) proposed stackforge/kolla: Add custom code to base docker templates  https://review.openstack.org/21554109:24
inc0_as I said, solve all the painpoints09:24
pbourkeinc0_: does the regex patch not solve the single case as well?09:25
pbourkedo we need both?09:25
inc0_regex code builds deps09:25
inc0_and I wanted to have option to skip building deps alltogether09:25
Kennansdake: I used quick start guide.09:25
inc0_and just do this single image with no regard of anything09:25
KennanI only have VM in hands now, in your doc it seems need bare-metal09:26
inc0_regex code is useful to "build me latest neutron" use case09:26
pbourkeinc0_: but the cache generally makes it not an issue09:26
pbourkeif deps don't need to be rebuilt they wont be09:26
Kennanis it a must requirements to meet ?09:26
Kennansdake:09:26
sdakekennan vm will work09:26
inc0_pbourke, it shoulding, but if I have custom base image, which I do (proxy and such)09:26
inc0_this keeps biting me09:26
sdakei like hat include header change09:26
pbourkeim not convinced :/09:27
inc0_pbourke, with include header this all will be solvable09:27
pbourkegreat work all the same09:27
inc0_without single09:27
sdakekennan for vm case follow the bare metal model09:27
inc0_feel free to decide what of it is worth merging09:28
inc0_I personally thing regex is useful and convenient tho09:28
sdakeyou will need a libvirt override in /etc/kolla/nova.conf09:28
Kennansdake: is it only one VM is OK for kolla test and deploy? I noticed in some kolla docs. said need more nodes09:28
sdaketo specify that qemu should be used for hypervisor instead of kvm09:28
inc0_as for single...well I need either this one or include one09:28
pbourkeinc0_: love the regex part09:28
sdakekennan you can do an AIO deploy on one vm sure09:28
inc0_I'm perfectly ok with abandoning single if it's not needed09:29
sdakekennan the first step is to get an AIO deployment going09:29
pbourkehow are the single and header related?09:29
inc0_nope09:29
sdakekennan the next step if you like that is to get a multinode deployment going09:29
inc0_just both can solve same problem09:30
sdakebaby steps09:30
inc0_adding proxy support that is09:30
pbourkehow can I solve the proxy support with --single?09:31
inc0_if you build your own manually tinkered base image09:31
inc0_beforehand09:31
inc0_that's how I did till now09:31
inc0_however include would be much better09:31
inc0_https://review.openstack.org/#/c/215504/ and please, review this09:31
inc0_unless this get merged I'll be in rebase hell09:32
sdakenot reviewing at 2:30am aftrer being awake for 20 hurs09:32
pbourkebut i have a manually tinkered base image, i use regex, the cache skips over the manual image and builds the one I specify09:32
pbourkesorry not a big deal just trying to be sure I understand09:32
inc0_cache won't skip it if Dockerfile changes09:32
inc0_and that happends with time09:32
inc0_but, that's why I consider include better option09:33
sdakepbourke any chance you will have those swift comments fixed up in the review queue for today since its end of week?09:33
sdakeI've got a demo wednesday and woudl like to do full monty with swift + heat + horizon09:33
pbourkesdake: lets say yes09:34
sdakecool and some minimal docs09:34
sdakedoesn't even have to be consumable09:34
sdakejustsome way for *me* to  consume them09:34
sdakeguys we are making fantastic progress :)09:35
sdaket - 10 days for liberty 3 :)09:36
openstackgerritVladislav Belogrudov proposed stackforge/kolla: Cinder ansible waits for bootstrap container on all hosts  https://review.openstack.org/21554909:36
pbourkeinc0_: that patch is some slick python09:39
inc0_yeah, but somethigs wrong09:39
pbourkeinc0_: oh?09:39
inc0_there, build tests fail, I'll debug it09:39
sdakehow well does multinode deployment work with master09:39
pbourkeinc0_: was about to ask you in you can convince me it works the same way09:39
pbourkes/in/if09:40
inc0_(strange because I've asserted both logics to be the same)09:40
inc0_it's supposed to, I've compared results from both methods09:40
inc0_but I'll do that again with more options I guess09:40
inc0_we need unit tests for build script09:42
inc0_gonna file a bug for that09:42
pbourke+109:43
inc0_https://bugs.launchpad.net/kolla/+bug/148739009:44
openstackLaunchpad bug 1487390 in kolla "unittests for build script" [Undecided,New]09:44
sdakehey folks we are going to have docso n docsopenstack.org09:45
sdakeeveryone excited about that? :-)09:45
pbourkeif it means more people will use kolla then yes ;)09:46
pbourkelooks like my mail never made it to the ML :(09:46
pbourkewant to get these neutron plugins off my list09:47
sdakemy review-kolla folder has 3k emails in it unread09:51
Kennanhi sdake: for your doc changes09:55
KennanI tried tools/build.py09:55
Kennanit failed with09:55
KennanConnectionError: ('Connection aborted.', error(2, 'No such file or directory'))09:55
sdakeis docker running?09:55
sdakesystemctl status docker-engine09:56
inc0_omg...I've just noticed what's wrong09:57
Kennanby bad, docker not started up itself after curl bash script09:57
KennanI manual start it now09:58
sdakethanks i'll modify the docs to address that in a new itrationkennan09:58
*** dims has joined #kolla10:05
openstackgerritMichal Jastrzebski (inc0) proposed stackforge/kolla: build.py refactor  https://review.openstack.org/21550410:07
inc0_this build script does some crazy memory jumping10:09
inc0_;)10:09
*** akwasnie has joined #kolla10:17
sdakeok guys im off to bed10:22
sdakenght10:22
sdakekennan plese leave feedbak on irc about your aio deploy10:22
sdakeor sk other ppele your qs10:22
sdakewe hae abuthalf our  enginering in europe10:22
sdakeengineering team that is10:22
Kennanok sdake: still building image seems long time10:24
vbelI think most of building is IO-wait :) So parallelizing it does not bring much10:29
vbelunless you are on SSD and 1Gb net :)10:30
vbelnight sdake10:31
vbel"1Gb"->"100Gb"10:35
inc0_pbourke, it seems after fixes refactor works10:40
inc0_now I need some git fun to rebase rest;)10:40
pbourkedoes anyone know how the perms are being set for data containers now10:46
pbourkee.g. /var/lib/rabbitmq seems to be rabbitmq:rabbitmq but cant find the part that does that10:46
inc0_it should be in ansible roles10:46
inc0_rabbitmq/tasks/start.yml10:46
pbourkei see nothing about perms in there10:47
vbelI am checking rabbitmq now10:48
pbourkeits working for rabbit just want to know how to do the same for swift10:49
pbourkeas right now vols are root:root10:49
pbourke*my vols10:49
inc0_look in bootstrap as well10:49
vbelthe problem with that is that this folder is mounted as root and does not have permissions to put anything inside as rabbitmq user10:49
inc0_data containers are started there10:49
inc0_and rabbitmq has single volume /var/lib/rabbitmq according to that10:49
vbel/var/lib/rabbitmq is from data container and belongs to root10:50
inc0_and other rabbit containers have volumes-from10:50
pbourkevbel:10:50
pbourke$ docker exec -it rabbitmq ls -ld /var/lib/rabbitmq10:50
pbourkedrwxr-xr-x 3 rabbitmq rabbitmq 40 Aug 20 10:37 /var/lib/rabbitmq10:50
vbeldepending on your version of rabbit it can su to rabbitmq from the very start and will fail to put its pid into /var/lib/rabbitmq/mnesia10:50
inc0_try changing /var/lib/rabbitmq to /var/lib/rabbitmq:/var/lib/rabbitmq10:51
pbourkeno as I said10:51
pbourkerabbitmq works10:51
pbourkeI just cant see how10:51
vbelyes, it is originally rabbitmq:rabbitmq, but if you mount it it is root:root10:51
vbelso if you run it like  docker run --rm -ti --volumes-from rabbitmq_data centos-binary-rabbitmq:latest /bin/bash10:53
vbelyou will see it becomes root:root10:53
pbourkefound it10:53
vbelI am sure it failes in oracle linux10:53
pbourkeits in config-rabbitmq.sh10:53
vbelyes, true, so it should be not in config-rabbitmq.sh but in run10:54
vbelin start.sh10:54
*** stvnoyes has quit IRC10:55
vbelotherwise only bootstrapping works10:55
vbelanyone to submit the fix?10:56
Haomenganyone can help me, I want to have try to deploy openstack via kolla, what is the guide I should follow? thanks10:57
HaomengI am not sure the steps to play with kolla10:58
pbourkeHaomeng: hi, you can start here: https://github.com/stackforge/kolla/blob/master/docs/dev-quickstart.md10:58
Haomengpbourke: thank you10:58
Haomengpbourke: yes, I try with steps, but some break with fatal error  - http://paste.openstack.org/show/423307/11:00
Haomengpbourke: Error: image kollaglue/centos-binary-kolla-ansible:latest not found11:00
Haomengpbourke: should I buld image first?11:00
pbourkeHaomeng: yes, sorry I think that image is missing11:01
pbourkefrom dockerhub11:01
pbourkeif you build it it should work11:01
Haomengpbourke: ok, np, thanks for your info11:01
Haomengpbourke: run "./tools/build-all-docker-images" right?11:02
pbourkeHaomeng: it would be quicker to run docker/centos/binary/kolla-ansible/build --release11:02
Haomengpbourke: great thanks11:02
pbourkeHaomeng: you will probably need to build base first11:03
Haomengpbourke: ok11:03
pbourkedocker/centos/binary/base/build --release11:03
Haomengpbourke: ok, thank you:)11:03
Haomengpbourke: error - The command '/bin/sh -c yum -d 10 -y install systemd systemd-libs systemd-devel && yum clean all' returned a non-zero code: 111:04
*** inc0_ has quit IRC11:04
pbourkeare you building on ubuntu?11:04
Haomengpbourke: my systems is ubuntu, have to run this on redhat/centos which supports yum?11:04
*** chandankumar has joined #kolla11:04
Haomengpbourke: yes11:04
Haomengpbourke: what is os type required11:05
pbourkeHaomeng: from quickstart:11:05
pbourke"For Ubuntu based systems, do not use AUFS when starting Docker daemon unless you are running the Utopic (3.19) kernel. AUFS requires CONFIG_AUFS_XATTR=y set when building the kernel. On Ubuntu, versions prior to 3.19 did not set that flag. If you are unable to upgrade your kernel, you should use a different storage backend such as btrfs.11:05
pbourke"11:05
Haomengpbourke: ok, got11:06
Haomengpbourke: enable the CONFIG_AUFS_XATTR=y, still get same error - yum-3.4.3-125.el7.centos.noarch has missing requires of yum-plugin-fastestmirror11:08
Haomengpbourke: what is the os requirements11:08
Haomengpbourke: CentOS should be better?11:09
pbourkeHaomeng: yeah im not sure on the specifics of changing the backend. I run ubuntu myself but tend to build in a centos vm11:09
pbourkeHaomeng: hopefully some of this will be smoother after the liberty release11:10
pbourkeHaomeng: we're in a state of flux atm11:10
Haomengpbourke: ok, np:) just want to have a try, thank you11:12
Haomengpbourke: I have an idea, should we have a *kolla docker image*, which is easy to run kolla:)11:13
Haomengpbourke: :)11:13
openstackgerritVladislav Belogrudov proposed stackforge/kolla: Rabbitmq servers fail to start due to wrong permissions  https://review.openstack.org/21558211:13
pbourke:D11:13
vbeljust checked with our ol - now starts properly11:14
pbourkevbel: why is it any different doing it in start.sh vs config-rabbit ?11:15
vbelbootstap container chmod it and exits. new container mounts it as root:root again11:16
pbourkebut config-rabbitmq is sourced from start.sh11:16
vbelin case of KOLLA_BOOTSTRAP11:16
pbourkeah wait, i see it now11:16
pbourkesorry11:16
pbourkewonder why I didnt have this issue though11:17
vbelno problem, it was a bit of mystery :)11:18
vbeldata containers - did we have them for rabbit?11:18
pbourkei was having a problem around this but rebuilding all containers fixed it11:19
pbourkesdake knows the reason11:19
vbelthe reason is data container. it has root:root volume. Before we had rabbitmq as data donor with correct /var/lib/rabbitmq permissions11:21
vbele.g. I do  docker run --rm -ti --volumes-from rabbitmq  centos-binary-rabbitmq:latest /bin/bash11:22
vbeland have correct permissions for rabbit in no time :)11:22
vbelso new 'data' container can be tricky11:23
vbel1 sec11:25
*** diga has joined #kolla11:26
Haomengpbourke: not sure where we can download these kolla base docker image, or have to build on local?11:28
vbelI think I have it :) sometimes it runs sometime does not. we do not wait for bootstrap container and kill it in next task11:30
vbelit runs perfectly with waiting for bootstrap. Bootstrap chmods the volume but sometimes cannot do it because we don't wait for it11:41
*** rhallisey has joined #kolla11:43
openstackgerritVladislav Belogrudov proposed stackforge/kolla: Rabbitmq servers fail to start due to wrong permissions  https://review.openstack.org/21558211:51
vbelanyone could run cinder? Running client gets ERROR: Malformed request url (HTTP 400)12:22
vbelalso cinder_volume says unable to update stats, LVMVolumeDriver not initialized12:23
*** inc0 has joined #kolla12:29
pbourkeHaomeng: its best to build everything locally12:31
Haomengpbourke: ok, will try, thanks12:31
inc0damn, our CI for builds is horrible12:31
inc0no way to check what are errors12:31
Haomengpbourke: now get this error - "failed: [localhost] => {"changes": ["{\"status\":\"Pulling repository docker.io/kollaglue/centos-binary-kolla-ansible\"", looks like there is no such kollaglue/centos-binary-kolla-ansible image in docker.io12:32
pbourkeHaomeng: try tools/build.py12:33
Haomengpbourke: docker search kollaglue/centos-binary-kolla-ansible return nothing12:33
Haomengpbourke: ok12:33
pbourkeHaomeng: it should build everything12:33
Haomengpbourke: ok12:33
Haomengpbourke: run tools/build.py, got "ERROR:__main__:could not find image: no such id: kollaglue/centos-binary-ceilometer-base:latest"12:34
pbourkeinc0: seen this ^12:34
pbourke?12:34
*** Guest94558 has joined #kolla12:35
Haomenghttp://paste.openstack.org/show/423366/12:36
Guest94558Hi Folks, we are looking at containerizing among other control-plane service, openstack too for our deployments12:36
Guest94558and hit upon Kolla projects which aims to just do that12:36
inc0docker images | search ceilometer-base12:36
inc0is there image like that?12:37
Guest94558so, a quick query: is kolla production ready ..?12:37
inc0aslo, are you using ansible?12:37
pbourkeinc0: he's just trying to build all images12:37
Guest94558are there any reference of Kolla in production or such deployments ..?12:37
inc0Guest94558, it is if you want to run master12:37
Haomenginc0: sudo docker images | grep ceilometer-base return nothing12:37
inc0so dependencies are busted somehow12:38
rhalliseyGuest94558, I'm sure if there any reference production deployments12:38
rhalliseyI"m not sure*12:38
Haomengwhich command I should run to build all dependencies, it is "tools/build.py", right?12:38
inc0that should work, right12:38
inc0if it doesn't something is wrong12:38
Haomenginc0: ok, let mey try12:38
inc0try to run ./build.py -d12:38
inc0with debug12:39
inc0aaa12:39
Haomenginc0: ok, thanks12:39
Guest94558inc0: we are definitely interested in Kolla for the obvious benefits of micro-services based deployment12:39
inc0./build.py -d -T 112:39
inc0single thread for better logging12:39
Guest94558however not sure if it is production ready12:39
pbourkeGuest94558: if you're looking to containerise the control plane I would say is the best solution available. As far as "production ready" it's still under quite heavy development12:39
inc0Guest94558, problem (?) of Kolla is that we're master based right12:39
inc0now12:39
Guest94558pbourke: yes, we want in control-plane12:40
inc0so if you're not willing to risk deployment of latest greatest bleeding edge master, Kolla might have problems12:40
inc0however, kolla stable kilo is a thing and it works12:40
pbourkethe liberty release at the end of the month is the first I would recommend for production12:40
SamYaplebut it has no upgrade path inc012:40
inc0but it's very old code and limited comparing to what we have now12:40
Guest94558inc0: so, is there no Kilo/Liberty stable base for Kolla ..?12:40
SamYapleI would not recommend kilo Kolla to anyon12:41
inc0Guest94558, there is no stable Liberty anywhere12:41
inc0Liberty is still in development12:41
SamYaplepbourke: would you consider voting on https://review.openstack.org/#/c/215157/ ?12:41
inc0as for production, I'd suggest to wait till Liberty is released, and run kolla then12:41
SamYapleI believe I have answered your questions12:41
Guest94558inc0: yes, I mean plan for Liberty12:41
Haomenginc0: can you help - http://paste.openstack.org/show/423367/12:42
inc0if you want to deploy liberty - kolla is an answer12:42
Haomenginc0: thank you12:42
Guest94558inc0: in production ..?12:42
inc0yup12:42
vbelcould anyone anytime run cinder service?12:42
inc0It will be as stable as openstack underneath12:42
pbourkeSamYaple: no forgiveness!12:42
Guest94558inc0: ok, sounds good, thanks.12:42
SamYaplepbourke:  :)12:42
Haomenginc0: will run "./build.py -d -T 1" and collect log again12:43
pbourkeSamYaple: i forgot we were using get-pip, so the upgrade is indeed redundant12:43
SamYapleI want to actually move all repos for centos into the base pbourke12:43
SamYapleits safest I think and keeps same package version12:43
pbourkeyeah we should12:43
pbourkerhallisey: I remember you had done some docs on cinder etc, have they disappeared?12:46
rhalliseyuh ya looks like they're gone..12:46
rhalliseyI think they were removed in the latest purge12:47
rhalliseynot sure why12:47
Haomenginc0: the first error is - error: unpacking of archive failed on file /usr/bin/systemd-detect-virt: cpio: cap_set_file12:47
rhalliseywas there a question?12:47
pbourkeHaomeng: this is the same ubuntu issue12:47
SamYaplerhallisey: re ceph12:48
Haomengpbourke: yes, think so12:48
Haomengpbourke: any idea to fix12:48
pbourkerhallisey: not really, just was doing to something short for swift, but then wondered where the other similar docs had gone12:48
rhalliseySamYaple, what's up12:48
SamYaplewhy dont you leave most of that start.sh out since that will need to be handled much different12:48
pbourkeHaomeng: see my comments earlier12:48
Haomengpbourke: ok12:48
rhalliseySamYaple, where would it be handled?12:48
SamYapleceph is going to be tricky since that data can't only exist in a data container since it needs to be sync around to the other nodes12:48
rhalliseyI've already dropped like half of it as it12:49
SamYapleansible has to be able to fetch that data12:49
SamYapleall the generated keys need to be sync'd, the mounting needs to be handled external to the container12:49
rhalliseypbourke, ya not sure why they were removed12:49
rhalliseySamYaple, ok.. this will be tricy12:49
SamYapleindeed12:50
SamYaplei have it working in Yaodu, I can Kolla-fy the ansible code12:50
SamYaplebut it will break some kolla 'best-practices' due to its very natrue12:50
SamYapleno way around that12:50
rhalliseyI was looking at yours + ceph container example12:50
rhalliseythe ceph community has it containerized, but it's not very good practice and we also have ansible12:51
rhalliseyso I'm trying to decipher what goes where when I know nothing about ceph12:51
rhallisey:p12:51
rhalliseyso it's wip12:51
SamYapleyea ive been over ceph-docker but we cant come to agreement on how to do that, i dont agree with them12:51
SamYaplerhallisey: just get the containers up, and we can do the start.sh in the ansible script12:52
rhalliseyya their solution is not very good12:52
SamYapleI keep telling them12:52
SamYapleansible-ceph is decent12:52
rhalliseyit's actually pretty horrible12:52
* SamYaple remembers he has a pull request to finish for ansible-ceph12:52
SamYaplerhallisey: yea im not disagreeing there, but im not running it either12:59
SamYapleinterestingly rhallisey, using ceph in docker containers allows us to avoid several hack jobs that ture-ceph daemons do13:02
SamYaplethe main one is the mount the OSD to read the OSD_ID and then REMOUNT IT  to the correct location once thats found13:02
SamYaplewe dont have too13:02
SamYaplemuch much cleaner13:02
openstackgerritMerged stackforge/kolla: Fix up base for ubuntu  https://review.openstack.org/21515713:06
openstackgerritSam Yaple proposed stackforge/kolla: Fix MariaDB for ubuntu  https://review.openstack.org/21515913:06
openstackgerritMerged stackforge/kolla: Add kolla-ansible for ubuntu  https://review.openstack.org/21511213:07
openstackgerritMerged stackforge/kolla: Fix RabbitMQ for Ubuntu  https://review.openstack.org/21515813:08
inc0sorry guys, was on emergency13:10
openstackgerritSam Yaple proposed stackforge/kolla: Fix MariaDB for ubuntu  https://review.openstack.org/21515913:10
inc0(debug neutron and setup rack...it was productive 15 minutes)13:10
SamYaplecoolsvap|away: sdake test your code!13:14
SamYapleIts literally the only requirement for templating13:15
openstackgerritMichal Jastrzebski (inc0) proposed stackforge/kolla: Build.py regex filtering  https://review.openstack.org/21552313:15
openstackgerritMichal Jastrzebski (inc0) proposed stackforge/kolla: build.py refactor  https://review.openstack.org/21550413:15
openstackgerritMichal Jastrzebski (inc0) proposed stackforge/kolla: Build single image  https://review.openstack.org/21552413:15
openstackgerritSam Yaple proposed stackforge/kolla: Fix broken zaqar template  https://review.openstack.org/21562213:16
SamYaplehey rhallisey pbourke building templates is broke without https://review.openstack.org/21562213:16
*** dims has quit IRC13:16
SamYapleif we could expidate that merge13:16
rhalliseykk13:16
SamYaplethanks guys13:17
rhalliseynp13:17
*** dims has joined #kolla13:17
openstackgerritMichal Jastrzebski (inc0) proposed stackforge/kolla: Add custom code to base docker templates  https://review.openstack.org/21554113:17
*** stvnoyes has joined #kolla13:21
inc0I got lost in git..13:25
openstackgerritMichal Jastrzebski (inc0) proposed stackforge/kolla: Build.py regex filtering  https://review.openstack.org/21552313:25
openstackgerritMichal Jastrzebski (inc0) proposed stackforge/kolla: build.py refactor  https://review.openstack.org/21550413:25
openstackgerritMichal Jastrzebski (inc0) proposed stackforge/kolla: Build single image  https://review.openstack.org/21552413:25
inc0SamYaple, about your -1 on refactor...it is way more clean and idiomatic approach13:26
SamYapleinc0: I disagree13:26
inc0so respectfully, don't -1 refactoring13:26
SamYapleand no lamba ever13:26
inc0lambda is part of language and it's extremaly useful13:26
SamYapleIm -2 on lamba13:27
inc0old code was long, undreadable13:27
inc0what's your argument? I don't accept "people don't understand lambda"13:27
inc0they should13:27
SamYapleTo bad, that is a valid argument13:27
inc0it is not13:27
SamYapleRegardless, im -2 on lamba13:27
inc0that's not magical code, that's idiomatic13:27
*** diga has quit IRC13:29
inc0and in this case, map and filter funcions, its widely used13:30
SamYapleAs the code stands, you dont have to know much about python. Im going to keep it that way. I am -2 on lamba. There will be no swaying that13:31
inc0as coding stands, we should keep high standards13:31
inc0and that's how python code should look like13:32
SamYapleI strongly disagree.13:32
inc0https://www.jeffknupp.com/writing-idiomatic-python-ebook/ good read, I recommend it13:32
SamYaplesure thing. not going to change my mind on lamba13:33
SamYapleIt is not easily understood by most people, I wont have it13:33
pbourkein fairness, inc0's approach is way cleaner13:35
SamYapleAgreed13:35
SamYaplebut redo it without lamba13:35
pbourkethat's ridiculous13:36
pbourkeops are not editing this code13:36
inc0if they want to, better learn python beforehand anyway...13:37
inc0in this use case (and that's pretty much only use case when I use lambda) it's most clean approach I know13:38
*** jtriley has joined #kolla13:38
inc0declaring closure there, and that's what would take to remove lambda would harm readibility13:38
SamYapleIve had the lamba fight. Ive heard all the sides. I am -2 on lamba13:39
inc0I don't accept argument "lambda is too hard". Programming is hard.13:40
inc0also, to be fair SamYaple, things you did with references out there will prove much harder to understand to anyone13:41
SamYapleIm ok with a refactor. Im not ok with lamba13:41
SamYapleIts time for me to sleep again.13:41
inc0go to sleep.13:41
inc0maybe after you get some rest it won't look that bad13:42
inc0you're grumpy person13:42
inc0now13:42
inc0:P13:42
SamYaple:)13:42
rhalliseylol13:42
SamYapleI should clarify my first comment, Im -2 on lamba not the sort_images change13:42
SamYaplewill comment as such13:42
*** Guest94558 has quit IRC13:42
inc0and read this book I linked, really after 7yrs of python experience I've learned to care about that stuff13:43
inc0and yes, lambdas are good if used correctly13:43
SamYapleI completely trust your python inc0. I dont trust everyone else13:43
inc0then maybe that will make them consider this too hard and will ask someone who actually understands python13:44
inc0that will prove beneficial for them13:44
inc0anywayu13:44
SamYaple'lets exlude people from understanding a glorified script' is all i hear13:44
SamYaplethis is a refactor of a bash script, not a true program13:44
inc0nope, I think this should be considered true program right now13:45
inc0including unit tests and such13:45
inc0in fact, this might be one of most critical code in whole project13:45
SamYapleit is certainly not13:45
inc0let's treat it like that, not a bash script13:45
inc0well, if this is broken, gates are broken13:46
SamYapleok like i said its bedtime. I have added another comment clarifying my first13:46
inc0if gates are broken, no new changes comes in13:46
SamYapleI would like to point out you cant merge the build.py code if it breaks the gate....13:46
inc0but it can break the gate afterwards13:47
SamYapleugh im not getting drawn in again. I feel very strongly about no lambda and that isn't going to change13:47
inc0you merge, something is wrong with dir structure, it breaks13:47
SamYapleim ok with a refactor past that your concerns are valid13:47
inc0that's how bugs work13:47
inc0and I think your approach to lambda is zealotus more than rational13:48
inc0:P13:48
inc0lambdas are cool.13:48
inc0if used properly13:48
SamYapleperhaps, but zealotus is unlikely to change in a day if that is the situation13:48
inc0that's why I'll ask for a second opinion:P13:48
inc0when sdake wakes up13:48
*** diogogmt has joined #kolla13:49
inc0let's ask Steven, if he asks me to remove lambda, I'm removing lambda13:49
*** diogogmt has quit IRC13:50
rhalliseyI'm if it's hard to understand we can document the build script.  If people push changes that are incorrect we can block13:51
*** diogogmt has joined #kolla13:52
inc0rhallisey, https://review.openstack.org/#/c/215504/1 is it hard to understand?13:52
SamYaplefyi inc0 jpeeler has his image building code up https://review.openstack.org/#/c/215323/13:53
inc0I've seen that - I've submitted this one before I found out about his13:53
inc0now we have 2 choices, I'm ok with either if it works13:53
rhalliseyI think it's fine13:54
inc0https://review.openstack.org/#/c/215541/ also, unrelated, please review this one - will solve lots of problems13:54
inc0see SamYaple? Everyone thinks lambdas are ok;)_13:54
inc0I'll just submit python code when you're not around13:55
SamYapleyea thats a great thing to say, trying to subvert reviews13:55
inc0your grumpiness killed rest of sense of humor in you13:56
inc0:P13:56
SamYapleno that was the alcohol for the past 10 hours13:56
SamYaplebut that isn't where the lambda thing comes from13:56
SamYaplei need some eggs. im going to sleep guys13:57
SamYapleinc0: I promise to rereview when i awake, but i wont promise a change13:57
inc0I'll ask sdake to solve this issue13:58
inc0let's agree to follow his decision13:58
inc0ok?13:58
SamYapleNo I wont agree to that, but I will have the discussion and I can be swayed.13:58
SamYaplein jpeeler's patch sdake mentions that this may be treated as a library in which case the conversation drastically chagnes for me13:59
SamYapleright now it is a glorified script in my mind, a script that all the ops guys I know can read an modify14:00
openstackgerritPaul Bourke proposed stackforge/kolla: Add Ansible role for memcached  https://review.openstack.org/21459314:03
openstackgerritPaul Bourke proposed stackforge/kolla: Implement Ansible role for Swift  https://review.openstack.org/21506514:03
openstackgerritPaul Bourke proposed stackforge/kolla: Implement Ansible role for Swift  https://review.openstack.org/21506514:05
openstackgerritPaul Bourke proposed stackforge/kolla: Implement Ansible role for Swift  https://review.openstack.org/21506514:08
jpeelerSamYaple: you mean you're opposed to it being treated as a library?14:08
openstackgerritPaul Bourke proposed stackforge/kolla: Implement Ansible role for Swift  https://review.openstack.org/21506514:09
inc0jpeeler, I think that was about my script being unreadable for ops people14:09
SamYaplejpeeler: no thats just a different conversation14:09
inc0and in case this turns into library, ops people won't read it at all14:10
SamYapleinc0: correct14:10
SamYaplethat is my viewpoint14:10
inc0SamYaple, then when we'll have CLI14:10
inc0it will be library14:10
inc0it should end up being library sooner or later14:10
inc0I don't think we should treat this as glorified bash any more14:11
SamYaplei think you understand my points inc0, whether you agree or not is a different matter. If I am understood then we can communicate on it further (later)14:11
jpeelerso which patch should continue for the single image building?14:11
SamYaplejpeeler: yours is the non-dependant patch14:12
SamYaplethats the one we require for removing the docker dir14:12
inc0jpeeler, https://review.openstack.org/#/c/215523/ this one adds a bit different functionality14:12
jpeelerit does, yes14:12
inc0one futher on adds "just build that damn image" thing14:13
SamYaplesorry guys im out for real this time. im closing the window14:13
inc0but I think regex building is useful14:13
inc0"build me neutron plz"14:13
inc0its convenient for that stuff14:13
jpeelerright14:13
inc0bbl I'm moving home14:16
inc0cyas14:16
*** jasonamyers has joined #kolla14:20
*** inc0 has quit IRC14:20
openstackgerritMerged stackforge/kolla: Fix broken zaqar template  https://review.openstack.org/21562214:21
*** diogogmt has quit IRC14:26
*** diogogmt has joined #kolla14:31
openstackgerritJeff Peeler proposed stackforge/kolla: Add options for selective build and listing images  https://review.openstack.org/21532314:38
*** banix has joined #kolla14:40
*** pbourke has quit IRC14:58
*** blahRus has joined #kolla14:58
*** pbourke has joined #kolla14:58
sdakemorning14:58
sdakepbourke ping14:59
sdakemorningbtw :) (you dropped off)14:59
pbourkesdake: howdy14:59
pbourkesdake: you mustnt have slept long?14:59
pbourkesdake: i've addressed all but one comment on the swift patch and added docs15:00
pbourkesdake: its good to go imo, it depends how strongly you guys feel on the remaining comment15:00
sdake5 hours?15:00
sdaketypical 5-6 hours enerally15:00
pbourkevery little15:00
sdakeok did you get it in the review queue15:01
sdakei typiccallyget 6 hours of sleep15:01
pbourkeyup its there15:01
sdakeso ya little light15:01
sdakei feel better when I get 8 but the summertime - i wake with the sun :(15:01
sdakein the winter I wke up at 9-10am :)15:02
sdakeany kind of docs available15:02
pbourkeyup15:03
pbourkehttps://review.openstack.org/#/c/215065/9/docs/swift-readme.rst15:03
*** unicell1 has joined #kolla15:03
*** inc0 has joined #kolla15:03
inc0afternoon15:03
sdakepbourke where is the info on what comment is not fixed15:04
sdakejust waking up and i know its your end of day15:04
pbourkesdake: see my most recent comment15:04
sdakeso litttle slow15:04
pbourkeon that review15:04
*** unicell has quit IRC15:04
pbourkenp15:04
inc0sdake, I need your opinion.15:04
inc0https://review.openstack.org/#/c/215504/4 we have small conflict here with Sam15:05
pbourkehonestly the thought of refactoring that config makes me want to go home15:05
pbourkebut its sunny here and its that time of the week15:05
inc0and I would like you to be arbitrator15:05
sdakeoh not rainining for once ;)15:05
sdakedublin never know if its going to rain :)15:06
pbourkeexactly!15:06
inc0I thought you know when it's going to rain15:06
inc0it will.15:06
sdakeok so this is tthe thign where the ring files isn't copied to the target nodes?15:06
sdakesorry if  i seem dense again brain booting15:07
bmaceyou mean the lambda thing inc0?  i think it is a valid python thing to use lambda, especially for sorts and that kind of thing.15:07
inc0yeah, lambda thing15:07
sdakelambda is normal python15:07
sdakeit is perfectly fine to use15:07
pbourkesdake: ive added ansible code to copy the ring files to all nodes. I recommend some coffee and reading the docs ;)15:07
inc0it's not like wizadry...I can do wizadry, this isn't one15:07
sdakeok well whatis missing then? :)15:08
pbourkethe generation of the actual rings15:08
sdakemy inclination at this point is to techdebt bug it15:08
pbourkethat has to be done manually15:08
sdakeoh eneration15:08
sdakeya techdebt bug15:08
pbourkeusing my perfectly documented commands :p15:08
sdakewe can tackle in rc115:08
*** banix has quit IRC15:08
sdakebtw its not coffee that gts me going, i need pepsi ;)15:09
bmacepepsi-max! :)15:09
bmacemaximum caffeine15:10
sdakei would like autogeneration for liberty if possible15:10
sdakebut it doesn't have to be in this patchsset15:10
sdakebmace less cafffeine but more sugar :)15:10
inc0sdake, mind saying that you're ok with lambda under review on which Sam threw -2 bomb?15:10
sdakei probaby wouldn't drink pepsi ifnt' i wasnt addicted to it15:10
bmacebleh.. i prefer the calories in caffeine over the calories in sugar.15:10
sdakeya but i respect other cores -2s15:11
sdakebut i will say lamba is fine to use - got a link15:11
inc0ever tried Yerba Mate?15:11
inc0well, -2 is because there is lambda15:11
sdakewhat is that some drug?15:11
sdakeI dont ned any more drugs15:11
*** SourabhP has joined #kolla15:11
inc0no, it's kind of plant15:11
sdakegimme link to the review15:11
*** banix has joined #kolla15:11
inc0very popular in southern america15:11
inc0they make sort of tea out of it, good stuff15:12
sdakeso is coca leaves15:12
inc0https://review.openstack.org/#/c/215504/415:12
sdakesounds like drug to me :)15:12
inc0no, it's juniper;)15:12
inc0(plant, not the company)15:12
inc0https://en.wikipedia.org/wiki/Yerba_mate15:12
*** SourabhP_ has joined #kolla15:13
inc0they drink it all over souther ameria...and Poland15:13
inc0its suprisingly popular in Poland15:13
inc0that's what I drink when I need caffeine...or I take guarana-extract pills;)15:13
inc0sdake, it doesn't use sorted..it uses filter and map15:16
inc0and that's where lambda is15:16
*** SourabhP has quit IRC15:16
*** SourabhP_ is now known as SourabhP15:16
sdakeinc0 that chemical you mentioned above looks safe enough - only xanthines which are actually known to be neuro-protectants15:18
sdakenot sur eabout polyphenols tho15:19
inc0well, its juniper infusion, and it's being drinked all over that place for several hundred years now15:19
sdakethose may or may not be safe - so I'll stick to my terrile-for-me pepsi :)15:19
sdakepeople been smoking tobacco for hundreds of years too :)15:19
inc0same about pot15:20
inc0I think all should be legal and let evolution solve things15:20
sdakeya well imo pot is fairlly safe minus possible lung damage15:20
sdakeinc0 for once we agree :)15:20
sdakeinc0 you should dig into why samyaple doesn't like lambda15:21
sdakeis it hte particular use, or lambda in general15:21
inc0I did15:21
inc0he considers build.py to be glorified bash script, therefore ops should edit it15:22
inc0and ops are scared of lambda15:22
sdakeya that logic makes sense it is an advanced python  concept15:22
inc0that's just that. Lambda is not an if statement15:22
inc0still, I beg to diffet that it's glorified bash script15:22
sdakeya well i really doubt operators will edit it15:23
inc0I think we should unittest shit out of it and build lib out of it15:23
inc0\\\\\\\\\\\\\\\\\\\\\\\\\\15:23
inc0sorry,...that was my cat15:23
sdakehighly highly doubt15:23
sdakekeyboard cat1115:23
inc0that was actual cat who decided to venture over backslash...and its close to enter on my keyboard15:24
sdakelets be honest, nobody will be able to edit that buid.py script unless they have 1-2 years of python experience15:24
sdakebut lambda may be beyond 1-2 years of python experience15:24
inc0yeah, especially when I actually understood what's happening out there15:24
sdakecan it be done without lambda?15:24
rhalliseylol15:24
sdakei haven't looked at the patch15:24
inc0I can define closure...15:24
inc0it's lambda in map function15:24
inc0so classical use case15:25
sdakewell i would be ok with avoiding conplex language features of python in build.py and in general in kolla15:25
sdakeone of the failures of python is it has too many language features15:25
inc0about this.. https://github.com/stackforge/kolla/blob/master/tools/build.py#L272 this line alters self.images15:26
sdakelambda is one of those complex language features15:26
sdakeinc0 too ealry for me to understand code sorry15:26
inc0I know how references in python works, but thats way beyond 2yrs of python15:26
pbourkeeven inc0's cat can do it15:27
sdakeline 272 is 1-2 years of python experience matieral15:27
inc0I'll remove this for sake of Sam's mental health15:27
*** SourabhP has quit IRC15:28
sdakelambda was the last thing i learned in python if that helps :)15:28
pbourkeIm opposed to removing it15:28
pbourkethe arguments against are not sound15:28
inc0sdake, metaclasses are fun15:28
sdakeif the arguemnt is "ops will edit build.py" that is pure crack talk15:28
*** ccrouch has joined #kolla15:29
inc0that's Sams concern as far as I know, and he confirmed that15:29
pbourkethe code is elegant and is completely standard and valid python15:29
*** ccrouch has quit IRC15:29
*** ccrouch has joined #kolla15:29
sdakeagain I respect other cores -2s15:29
sdakeif his mind can't be changed it can't be changed15:29
inc0I think main difference between me and him is that I think wwe should treat build.by as normal python code which will be extended15:30
inc0and therefore readability counts15:30
pbourkethis is normal python code!15:30
sdakepbourke I think sam's real concern is the use of  advanced langugage features15:30
sdakehe brought this up during the jinja2 teplating thing too15:30
sdakeas if operators will edit jinja2 containers15:31
inc0yeah, lack of includes15:31
inc0I have issues with that too, but not too much15:31
pbourkethis script is beyond the scope of operators15:31
sdaketurns out we dont really need anything besides conditionals15:31
sdakepbourke totally agre15:31
*** ccrouch has quit IRC15:31
inc0sdake, thing is...we never did...its just for convenience and readibility15:31
sdakeoperators are nto going to edit the code base unless they have an engineeirng wizard on staff15:31
*** ccrouch has joined #kolla15:31
inc0we can write every program with nxor, it's turing complete15:32
sdakeinc0 lol15:32
inc0NAND* sorry15:32
sdakewell the openstack process is a -2 kills a review15:32
inc0nand is turing complete15:32
sdakeso pbourke recommend commenting if you think sam is wrong15:33
inc0that's what we should talk with Sam about tbh...its 2nd time he does that to me15:33
inc0-2 I mean15:33
sdakei gave you a -2 i think not sam :)15:33
sdake-2s are perfectly normal part o the process15:33
inc0it was about L115:33
sdakeL1?15:34
inc0but he removed it;) for keepalive container15:34
pbourkeinc0: for you regex patch, should this work? "tools/build.py --template -r *neutron*"15:34
sdakeoh15:34
inc0pbourke, actually I made this so -r neutron is enough15:34
inc0no need for *15:34
sdakepbourke can you comment on the lambda thing if you feel strongly about it in the review15:34
pbourkedoing so now15:34
pbourkeinc0: thanks15:35
inc0its re.search, otherwise it would need to be .*neutron.*15:35
sdakeits also perfectly normal on a -2 for people to comment if they disagree with the core reviewer15:35
inc0but you can do all sorts of regex magic there, it's pretty flexible15:35
sdakebut if he rally doesn't want to remove his -2, then the review is dead as is15:35
sdakethat is just the process15:35
inc0unless I rebase it...15:36
sdakeand we should all respect core reviewers -2s in general15:36
inc0j/k ;)15:36
sdakeif you rebase it without the offending code15:36
sdakeyu mean15:36
sdakei think you mean refactor15:36
inc0well, correct me if I wrong, but simple rebase will throw away -215:36
sdakeno15:36
sdake-2 is locked in15:36
inc0ah ok15:36
sdakethe only way is to remove th change id and make a new patch15:37
sdakethis is highly highy frowned upon15:37
inc0as it should be15:37
sdakeI would -2 to preserve sam's vote in that case ;)15:37
inc0anyway, I'll keep refactoring build.py, I have issues in how wasteful for memory it is15:38
sdakeya a git rebase followed by a git review - the -2 vote stays15:38
sdakeinc0 i would like to use build.py as a library15:38
sdakeso while your refactoring try to keep that in mind15:39
sdakeI'd like to bill it as such as  well15:39
sdakeusing an entrypoint python magic ;)15:39
sdakelong term - doesn't need to happen immediately15:40
sdakeor maybe medium term15:40
inc0if we'd make CLI client, that will come automatically15:40
bmacejust hope nobody -2s magic ;)15:40
sdakefor what its worth, my opinion of python is that it hsa too many language features and "python" code is "hard to read code" :)15:41
sdakealthough pbourke's recommendation to use .get was good - I liked that15:41
inc0sdake, while I agree with lots of features...most of them are syntactic sugar which actually helps to read it15:41
sdakepythonic that is15:41
pbourkesdake: did you add that in the end?15:42
pbourkecan't remember15:42
sdakeyes i did15:42
pbourkenice ;)15:42
sdakealthough jpeeler had a patch which did same thing which doesn't have .get15:42
sdakei just didn't think of it when doing it15:42
sdakebut of course that is th ecorrect way to do it15:42
sdakeit does it in a different spot15:43
sdakebmace I found weikus's slides15:43
sdakelooked solid15:43
bmaceah, from linuxcon? or openstack day?15:43
sdakelinuxcon15:43
bmaceit is mostly accurate.  some minor syntax tweaking is still in the works.15:44
pbourkelink?15:45
*** banix has quit IRC15:46
*** absubram has joined #kolla15:47
jpeelerok so has the decision been made to abandon https://review.openstack.org/#/c/215323/ ?15:49
jpeeleri missed a pep thing, so wondering if it's worth fixing15:49
sdakejpeeler not that i am aware of15:50
sdakei havent even reviewed inc0's work15:50
sdakebut i don't want a new argument and from the discussion it appears a -r argument was added15:50
sdakeby arguemnt i mean to the command line ;)15:51
inc0it's for regex filtering15:51
sdakeit can just be without an argument15:51
inc0I can make if default if you want15:51
sdakebuid.py regex15:51
sdakeyes15:51
inc0ok15:51
sdakei asked same for jeff's work15:51
sdakedoes your patch rebuild parents15:51
inc0it does15:52
sdakeyou should be able to do the following15:52
sdakebuild.py heat ceilometer15:52
sdakeand anything heat and ceilometer elated is rebuilt15:52
sdakethat is what id like to see15:53
sdakeor buid.py heat_engine15:53
sdakennd only heat-engine is build15:53
sdakejpeelers code doesn't build parents15:53
openstackgerritJeff Peeler proposed stackforge/kolla: Add options for selective build and listing images  https://review.openstack.org/21532315:53
pbourkeinc0: dumb question but... could the lambda not be replaced with 'for img in this_tier: images_to_process.remove(img)'15:53
pbourkeinc0: not saying it should15:53
inc0it can ofc15:53
jpeeleri fixed it. feel free to abandon it if inc0's patch gets in15:54
inc0I've used lambda to keep it 1liner15:54
sdakein that case, might as well jus tremove the lambda15:54
pbourke:(15:54
sdakethat is simpler15:54
inc0I will, no worried15:54
sdakebut again the reason is pure crack sauce ;)15:54
pbourkeits not worth arguing about, but I think its a shame15:54
inc0just...we need to have this discussion15:54
*** shardy_ has joined #kolla15:54
sdake(that operators will edit buidl.py)15:55
pbourkeis it going to be a requirement to only use basic code for kolla?15:55
sdakeno15:55
pbourkeif so that may put potential contributors off15:55
sdakepythonic code is good with me15:55
sdakei'll add to agenda for wednesday's meeting so we can have a vote for it15:56
sdakenot that we have alot of python code atm15:56
*** shardy has quit IRC15:56
inc0we will15:56
sdakei said atm15:56
inc0sooner or later15:56
inc0yeah15:56
sdakepython code should be easy to understan15:57
sdakelambda is sometimes *required* to do the job15:57
pbourketbh if an op can understand the tier building mechanism they will definitely understand the lambda15:57
sdakeif its not rquired, my recommendation would be eto avoid it15:57
sdakelambda is an advanced language feature15:58
pbourkewe're probably bikeshedding now15:58
sdakeno we are setting policy15:58
sdakebut we can do it wednesday15:58
pbourkethe regex patch is awesome though15:59
sdakewe have a policy of only using conditionals in .j2 files15:59
sdakethat is a good policy15:59
sdakeoh yu man bikeshedding this patch ;)15:59
*** shardy_ has quit IRC16:00
sdakepbourke bikeshedding which part, the use of -r as an option or lambda?16:00
pbourkethe lambda16:00
pbourkeI just dont like to see someone win because they shout the loudest16:00
pbourkeas I said the arguments against are not good imo16:01
*** shardy has joined #kolla16:01
sdakeagree and we are not setting a policy about lambda today16:01
pbourkethe regex patch is dependant on it thouhg :(16:01
sdakeand agree the arguemnts against are not sound16:01
pbourkeinc0: can you make the regex patch standalone?16:01
pbourkeand we'll postpone the other discussion16:01
sdakelets get it merged16:01
sdakei want stuff going in fast and furious16:01
sdakeneed individual image build like asap :)16:02
pbourke+116:02
pbourkecome back inc0 :P16:03
*** ccrouch has quit IRC16:03
*** ccrouch1 has joined #kolla16:04
*** ccrouch1 has quit IRC16:04
*** ccrouch has joined #kolla16:04
inc0I'm back16:14
inc0:)16:14
inc0I'll remove lambda from refactoring patch16:14
pbourkethought you'd rage quit16:14
pbourke;)16:14
inc0nah, I've played too much of online games to rage quit from something like that16:14
pbourkeinc0: my suggestion. put up the pep8 refactor, regex change, and lambda as 3 independant changes16:15
pbourkewe can merge the first two very easily16:15
pbourkethen vote on the third next wed16:15
openstackgerritVladislav Belogrudov proposed stackforge/kolla: Rabbitmq servers fail to start due to wrong permissions  https://review.openstack.org/21558216:16
sdakewe dont ned to vote on the patch that can be doen in the gerrit reiview16:17
*** chandankumar has quit IRC16:17
sdakewe can vote on whether we accept pythonic code or not in the code base16:17
sdakeincluding advanced language features16:17
sdakeits a policy that might as well be set now since there is disagreemeent among the cores around advanced language features16:18
inc0I'll just remove lambda here16:18
inc0I can do it without lambda easily16:18
inc0and we'll decide on future policy16:19
inc0but...can I get reviews on this?16:20
inc0https://review.openstack.org/#/c/215541/16:20
inc0that would make my life much easyier16:20
sdakeinc0 that is a good course of action16:20
inc0I'm not in texas yet, so I don't need to be redneck;)16:21
inc0so I can go for a compromise16:22
*** unicell1 has quit IRC16:24
sdakeinc0 that patch is reviewed16:25
inc0thanks16:26
*** dims is now known as dimsum__16:26
vbelinc: I request include_footer too :)16:28
vbelto clean up stuff :)16:28
vbelinclude_footer would be at the other end of container hierarchy16:30
inc0sdake, you want to call ./build.py regex?16:30
inc0but that makes regex required16:30
openstackgerritPaul Bourke proposed stackforge/kolla: WIP: Add optional third party Neutron plugins  https://review.openstack.org/20754516:31
*** daneyon has joined #kolla16:31
sdakeinc0 what is your question16:31
sdakei dont understand it16:31
sdakeinc0 if nothign is specified, it should build all16:32
sdakelook at jpeelers patch he made similiar change16:33
sdaketools/builld.py builds all just as happens now16:33
pbourkeon that subject16:33
pbourkeI would prefer no args to print usage16:33
pbourkeprinciple of least surprise16:33
inc0well, then to build all would be ./build.py .16:35
*** vbel has quit IRC16:35
inc0which isn't horrible16:35
*** ashishb has joined #kolla16:35
sdakepbourke agree that would be better but thatwould b ea change in behavior16:36
pbourketrue16:36
sdakei dont mind the changei nbehavior16:36
pbourkenot that we've had any shortage of those in the run up to liberty ;)16:36
sdakelol16:36
sdakeya agree16:36
inc0yeah, after liberty this will be harder16:36
sdakemy consistency engine in my brain isnt working yet, still booting16:36
pbourkesuggest:16:37
pbourke-r regex16:37
pbourke--all16:37
pbourkeno args = print usage16:37
sdakei do not want a -r really16:37
inc0or maybe ./build.py regex as default16:37
inc0./build.py --all for all16:37
sdakeno need for it and w eneed to pass multiple regexes16:37
inc0also, does --all means just centos binary?16:37
sdake--all is good idea16:37
pbourkeinc0: i like that16:37
sdakeinc0 yes just the default16:38
inc0because you know that with regex we can ditch --type binary16:38
pbourkeyes centos binary is default16:38
sdakeinc0 i prefer nto make the building process require regex expereince16:38
inc0ok16:38
sdakeif people want to use some complex regex thats fine16:38
inc0so ./build.py neutorn --type source will work16:38
sdakebut i dont want to make it mandatory theybe a regex wizard16:38
inc0well..this wont, but with "neutron" it would16:39
sdakethe regex shuld be the last thing specified16:43
sdakeand tht shoudl be in the hep documentation16:43
sdakethe lsit of regexes rather16:43
sdake./build.py --type source neutron heat ceilometer16:44
sdakeshould build all neutorn heat ceilometer related junk16:44
sdakepbourke thanks for the swift work :)16:47
sdakei'm really keen to try that out16:47
pbourkesdake: yw hope it works for you16:47
sdakei'll let you know16:47
sdakedoes master work atm?16:47
pbourkeyeah I ran through the AIO from scratch and it worked well16:48
pbourkeI should try multinode too I guess16:48
sdakedid you rm -rf /opt/kolla first?16:48
pbourkeyes16:48
sdakenice16:48
openstackgerritMichal Jastrzebski (inc0) proposed stackforge/kolla: Build.py regex filtering  https://review.openstack.org/21552316:48
openstackgerritMichal Jastrzebski (inc0) proposed stackforge/kolla: build.py refactor  https://review.openstack.org/21550416:48
openstackgerritMichal Jastrzebski (inc0) proposed stackforge/kolla: Build single image  https://review.openstack.org/21552416:48
sdakeok brainstorming session real quick guys16:49
sdakeI found someone that wants to make a big contribution16:49
inc0soo...we want multiple regexes support?16:49
sdakeinc0 yes16:49
inc0k16:49
sdakeso I'mgoing to point them at installable package16:49
sdakeas in setup.py16:49
sdakewhere do we want files installed16:50
openstackgerritMerged stackforge/kolla: Prepare for gating by distro, type, docker type  https://review.openstack.org/21353716:50
inc0uhh let's hold on a moment16:50
pbourkeoff the top of my head most should go in /opt/kolla16:51
pbourkewith confs in /etc/kolla16:51
sdakeinstallable package is a dependency for pushing to docs.openstack.org so we are not holding on on this one16:51
inc0if we install kolla packages, that means we eihter break python convention or put dockerifles to dist-packages16:51
sdakedockerfiles can go to /usr/share/kolla/dockerfiles_templates16:51
inc0what that install would do anyway?16:51
*** unicell has joined #kolla16:51
sdakethis is what we are brainstorming inc0 :)16:52
inc0alias wherever/build.py kolla ?16:52
inc0I mean...it will just move files from one place to another16:52
inc0no PATH inclusions16:52
sdakebuild.py would have to searhch for docker files in /usr/share/kolla/dockerfile_templates16:52
sdakewe can do build.py as a separate patch imo16:52
sdakethe idea is to ge tthe structure in place16:52
sdakeso I canget on with my sphinx work16:53
inc0most of sphinx works would be on dockerfiles tho16:53
inc0I mean should16:53
inc0and manuals16:53
inc0I'm not sure how sphinx handles docker16:53
pbourkesdake: i'm about to head off for the weekend. if you find issues I'll look at them first thing monday, Id be surprised if there's anythin major. Would also appreciate a look in on https://review.openstack.org/#/c/207545/16:54
inc0and ansible ofc16:54
sdakesphinx isn't going to build docker16:54
sdakepbourke can yu give me 5 minutes for brainstorm?16:54
sdakeor do you need to roll16:54
pbourkeno that cool16:54
sdakepbourke i wil check out the review16:54
pbourkeis this on where to install stuff?16:55
inc0we could use some ansible magic to make ansible-playbook lots-of-arguments something like kolla install --aio16:55
sdakeok so we have etc that needs installed we have ansible that needs installed we have kolla-ansible that needs installed16:55
inc0this would mean python entrypoints wizadry16:55
sdakewe have build.py that needs to be renamed16:55
sdakeI would be satisified with something basic to begin with16:56
sdakeand iterate16:56
pbourkethere's all the tools16:56
inc0 /var/lib/kolla for dockerfiles and dist-packages for python?16:56
sdake /var/lib is for persistent files16:56
pbourkewell, not all needed but dont forget them16:56
inc0 /opt is ugly for installed stuff I think16:57
sdake thinking /usr/share/kolla/ansible foransible bits16:57
sdake /usr/share/kolla/dockerfile_templates for docker templates16:57
inc0usr/share/kolla/docker for docker then?16:57
sdakeyup16:57
sdake /usr/share is for non variable data :)16:57
inc0make it docker, we're moving to templates as default;)16:57
*** jmccarthy has quit IRC16:57
sdakeok so tools directory goes where?16:58
sdake /usr/bin?16:58
pbourkeyeah16:58
*** achanda has joined #kolla16:58
sdakei dont know if we really want to install all of the tools16:58
inc0usr/local/bin16:58
pbourkethat's what i was saying16:58
sdakenot /usrlocal/bin16:58
pbourkejust some of them16:58
sdakethis is for a poroper install16:58
inc0which ones?16:58
inc0besides build16:58
sdakeya lets make the lsit now16:58
sdakewhat wouold people prpose16:58
sdakekolla-ansible and build.py are what i prpose16:59
sdakeanything else?16:59
inc0how about we turn both of these to python libs and expose single entrypoint?16:59
inc0kolla build, kolla deploy16:59
sdakekolla-ansible is a shell script16:59
sdakeinc0 i am looking for something quick and dirty17:00
inc0simple enough to be turn python in 15mins17:00
sdakenot perfection17:00
sdakejust to unblock my sphinx work17:00
inc0but if we teach people to use cli in liberty, we'll be stuck with it for some time17:00
sdakeinc0 unblock sphinx work is goal here17:00
sdakewe can tackle rest of it in liberty-rc117:01
inc0I'm ok with rc117:01
inc0as long as we'll have more stable cli in liberty itselfd17:01
sdakeso i need basic infrastructure right now though17:01
sdaketo get docs publishedon docs.openstack.org17:01
sdakeok17:01
sdakeso entrypoints are future followon work17:01
sdakewhat to do about etc directory17:02
sdakejust replace what is in /etc/kolla?17:04
inc0yeah, it works nice17:04
sdakeok so whicch toosl do we need installed besides build and kolla-ansible17:07
*** jmccarthy has joined #kolla17:08
*** ccrouch has quit IRC17:09
*** ccrouch1 has joined #kolla17:09
openstackgerritMichal Jastrzebski (inc0) proposed stackforge/kolla: Build.py regex filtering  https://review.openstack.org/21552317:12
openstackgerritMichal Jastrzebski (inc0) proposed stackforge/kolla: Build single image  https://review.openstack.org/21552417:12
inc0I think that's it17:12
*** ccrouch1 has quit IRC17:17
*** ccrouch has joined #kolla17:21
openstackgerritMichal Jastrzebski (inc0) proposed stackforge/kolla: Add custom code to base docker templates  https://review.openstack.org/21554117:22
openstackgerritMichal Jastrzebski (inc0) proposed stackforge/kolla: Add custom code to base docker templates  https://review.openstack.org/21554117:25
inc011 lines of code, 3 bps in partially implements17:25
inc0next one will implement more blueprints than lines of code17:25
inc0ok, I think I've done enough for today17:26
inc0kick SamYaple from me when he wakes up, for that -217:26
inc0good night everyone17:26
*** inc0 has quit IRC17:27
*** ccrouch has quit IRC17:29
*** jmccarthy has quit IRC17:30
*** ccrouch has joined #kolla17:30
*** tfukushima has quit IRC17:32
*** ccrouch has quit IRC17:34
*** ccrouch1 has joined #kolla17:34
*** bradjones has quit IRC17:37
*** ccrouch1 has quit IRC17:37
*** ccrouch has joined #kolla17:38
*** bradjones has joined #kolla17:38
*** bradjones has quit IRC17:38
*** bradjones has joined #kolla17:38
*** akwasnie has quit IRC17:39
*** ccrouch has quit IRC17:41
*** achanda has quit IRC17:44
openstackgerritMerged stackforge/kolla: Added missing configuration to horizon docker template.  https://review.openstack.org/21551117:46
*** jasonamyers has quit IRC17:46
*** loth has joined #kolla17:48
*** apuimedo has quit IRC17:48
*** tummy has quit IRC17:55
*** ashishb has quit IRC18:06
*** achanda has joined #kolla18:07
*** achanda has quit IRC18:08
*** achanda has joined #kolla18:08
*** achanda has quit IRC18:09
openstackgerritMichal Jastrzebski (inc0) proposed stackforge/kolla: Add custom code to base docker templates  https://review.openstack.org/21554118:21
*** ccrouch has joined #kolla18:36
*** absubram has quit IRC18:38
*** absubram has joined #kolla18:41
*** tfukushima has joined #kolla18:42
*** ccrouch has quit IRC18:43
*** sdake has quit IRC18:46
*** ccrouch has joined #kolla18:46
*** ccrouch has quit IRC18:47
*** ccrouch has joined #kolla18:47
*** tummy has joined #kolla19:00
SamYaplealright having read the scrollback I want to point out that whether operators will modify the build.py code or not, they _will_ be reading it19:00
SamYaplereading openstack source is basically part of the job19:00
SamYaplethere is no reason to use lambda in this case it is not well know how to understand it and it makes it less readbale in my opinon19:00
SamYapleall while saving 1 single line19:00
*** banix has joined #kolla19:01
SamYaple1 line....19:01
SamYapleoh and you can do everything in python wihtout a lambda, there is nothing that requires it19:02
SamYapleafter all it is just a function on the fly19:02
*** sdake has joined #kolla19:03
*** tummy has quit IRC19:04
sdakeyo19:04
SamYaplewut wut19:04
sdakesamyaple you left a comment i was missing data containres19:04
sdakedid yu mean i forgot to remove some?19:04
SamYapleyes, was that not clear?19:05
SamYaplei dont remember what i said19:05
sdakeyes just verifying19:05
sdakebefore i pull down the patch and rewor kit19:05
sdakesamyaple i found someone to do setup.py19:06
sdakei need this for sphinx doc generaiton19:06
*** tfukushima has quit IRC19:06
SamYaplewhats his name?19:06
sdakefor docs.openstack.org19:06
sdakesuhail19:06
SamYapleah ok19:06
sdakeinternal dude19:06
SamYaplethought it might have been someone i know19:06
sdakehave any requirements on that work yo uwould like to express19:06
jpeelerwhat exactly is getting packaged?19:07
jpeelerall the testr work done was written to avoid usage of setup.py :/19:07
sdakewe are putting ansible in /usr/locla/kolla/asible19:07
sdakeall etc is going in /etc/kolla19:07
SamYaplejpeeler: the whole kolla project19:07
SamYaplepip install kolla19:07
SamYaplethat would pull down the dockerfiles and everything19:07
sdakebuildl.py and ansible-kolla are going in /usr/bin19:08
SamYapleand install /etc/19:08
sdakedocker is going in /usr/local/koall/docker19:08
sdakedocker_teomplates is going in /usr/local/kolla/docker_templates19:08
SamYaplei have no opinions on this, as long as /etc stays I am good19:08
SamYapledirectory wise19:08
sdakeyou mean existing etc doesn't get overwritten?19:09
sdakeor etc gets installed?19:10
SamYaple/etc/kolla19:10
*** tummy has joined #kolla19:10
sdakethe main goal of his work is to get sphinx in a runnable state19:10
SamYapleyup19:10
sdakeso it may need a bit of rework after19:10
sdakecould you clarify yup on etc, you want it to stay or be copied over?19:11
SamYaplei dont know whats being asked, i defer to you at this time19:11
* SamYaple is a bit screwed up at the moment)19:11
sdakethe plan is to replae whatever is in /etc/kolla with what is in the epo19:11
SamYaplethat will wipe settings and passwords, that is a bad idea19:12
*** tummy has quit IRC19:12
sdakeok woud you propose not copying /etc then?19:12
SamYapleno it just needs to be a smart copy. not overwrite if existing, and (maybe not immediately) but smart upgrading of existing files19:13
SamYapleadding new options/changing names19:13
sdakei dont know if there is a way to conditionally copy /etc19:13
SamYapleoverwritting is really bad19:13
sdakeagreed19:13
sdakelets leave etc out of this patch and file a tech debt bug for that19:14
sdakeagree?19:14
SamYaplesure19:14
*** absubram has quit IRC19:17
sdakeany other requirements for setup.py?19:21
SamYapleI dont think so19:22
SamYaplei dont know19:22
sdakewe can todo any remaining stuff19:23
sdakelike etc19:23
sdakeand the special magic around loading /bin files19:24
sdakesetup.cfg entrypoints is what i'm referring to19:24
sdakeare there other binries we want to install besides kolla-ansible and build.py?19:25
sdakesamyaple comment on https://review.openstack.org/#/c/215504/ for you needs attention if you wouldn't mind19:27
sdake-2 kills a review, i'd lie it unkilled with the proposed change you suggested19:27
SamYaplei -2 because of the talk of getting it merged behind my back :/19:28
sdakethat isn't going to happen19:28
SamYapleWell not _now_19:28
sdakei'll address that in the wednesday meeting if you have logs19:29
sdakecore reviewers should not be merging code behind peoples back19:29
SamYapleIt wasn't cores saying this19:29
SamYapleits fine though, I will update the vote19:29
sdaketx19:29
SamYapleAre you saying -2 blcoks new patches?19:29
SamYaplei wasnt aware of that19:30
sdakenah but it kills the erview19:30
sdakepeople give up on -2 usually :(19:30
SamYapleI did say in the review I would remove it once that was removed19:30
sdakeyou are free to -2 as you like19:30
sdakebut if we find a solution that is agreeable would like otes changed to -1 :)19:30
openstackgerritJeff Peeler proposed stackforge/kolla: Add options for selective build and listing images  https://review.openstack.org/21532319:30
jpeelershould i just abandon that? ^19:30
SamYapleI just didnt appreciate the 'ill submit when you arent around'19:31
sdakeok fair then samyaple19:31
sdakei'll ddress it in the meeting19:31
sdakeif there was that type of comment i'd keep a -2 on the patch until it was changed too19:31
sdakeuntil the ptl sorted out the conflict19:31
SamYaplejpeeler: it looks like inc0 just pulled in your changes that I wanted (free form args rather than --regex option) so its up to you19:32
SamYapleI am ok with your change, but i see the appeal of inc0 approach19:32
sdakejpeeler i think the dirction is to go with inc0's patch19:32
sdakejpeeler your patch is good too19:32
sdakeits too bad you both did the same job -waste of time19:32
jpeelerwell i read that he was going to wait for me19:32
sdakethis is the problem with global blueprints for features19:32
SamYapleinc0 knew jpeeler was doing this :/19:32
jpeelerso yes, it's a shame that it occurred19:33
SamYaplei was disappointed as well19:33
sdakei dont know what to tell you its not like I control other people's behavior :)19:33
SamYapleif I believe your statement, does that mean you are controling me?19:34
sdakeinfluence and control are two different things19:35
sdakecontrol means "do this" "it gets done"19:35
sdakeinfleunce means "can you do this" "ok I agree"19:35
sdakei do influence peoples behavior :)19:36
sdakejust like you do :)19:36
sdakeyour influencing my behavior around the work suhaiil is doingaround setuppy for example19:36
sdakemake sense?19:37
sdakesamyaple can you approve my doc changes pls or jpeeler19:38
SamYaplelink?19:39
sdakelets iterate on them19:39
sdakesecond19:39
sdakehttps://review.openstack.org/#/c/215532/19:39
sdakeand its children patches19:39
SamYaplesdake: left a review on the final patch https://review.openstack.org/#/c/215532/19:45
SamYaplethe others are good19:45
openstackgerritMerged stackforge/kolla: Prepare for MD to RST conversion  https://review.openstack.org/21546319:47
openstackgerritMerged stackforge/kolla: Run pandoc to convert the documentation  https://review.openstack.org/21546419:47
sdakethanks sam I will adjust19:48
jpeelerlot of spelling errors19:49
jpeeleras far as having multiple dev envs, i'd be happy to see vagrant be the preferred bootstrap tool19:51
*** ccrouch has quit IRC19:51
SamYaplewaht are you takling about jpeeler? he spells the greetest!19:51
jpeelerheh19:51
sdakejpeeler add comments i will fix19:54
jpeeleri did19:54
sdakei dont mind fixing spellings errors i just dont want alot of retructure19:54
sdakei already did the restructure19:54
sdakeif we do more restructure it can be new patches19:54
sdakei'd like vagrant to be the primary development environment butit only works with virutalbox19:55
sdakeand the documentation around heat is not up to date19:55
sdakethat is coming in a followon patch19:55
sdakei wrote those docs at 2am19:55
jpeelersdake: i have a patch to make it work with libvirt19:55
sdakeso was a bit bleary eyed19:55
sdakejpeeler fantastic!!!!19:55
sdakei can't wiat to test it out19:55
jpeelerbut i'm having difficulty verifying multinode19:55
jpeelerso i've been holding off on posting it (as well as what the network should look like)19:56
jpeelerbut yeah, libvirt should become the default in my opinion. great that it works with virtualbox too though19:56
SamYaplejpeeler: a docs patch for libvirt?19:56
sdakei'd like heat to be n option too but it needs updating19:56
sdakesamyaple no its an implementation patch19:57
jpeelerSamYaple: not docs, libvirt requires a modified Vagrantfile and image19:57
sdakevagrant code atm is hard coded to only work with virtual box19:57
SamYapleoh i see19:57
sdakeit should work with both19:57
SamYaplecool19:57
sdakeif we need separate vagrant  files put em in separate dirs19:57
SamYaple+1 on heat19:57
SamYapleI do most of my devel on instances right now19:58
jpeelerhonestly i think it's weird to say, hack on openstack, but you need openstack to do it19:58
sdakeyes we have two development environments plus one bare metal deploy doc19:58
SamYaplethe instances are running on Openstack running Kolla! on my gear here19:58
sdakeyup we have a cisco lab where i can test like 50 nodes with ehat19:58
sdakethat would rock19:58
sdakewith heat that is19:58
sdakebut the heat template needs updating19:59
jpeelerSamYaple: do you know of a way to get rackspace instances for openstack dev?19:59
sdakeits still based upon compose19:59
sdakemaybe the docs just need updating19:59
jpeeleri remember seeing something a while back about $500 credit or something19:59
sdakejpeeler you gotta pay19:59
SamYaplejpeeler: with $$19:59
SamYaplei mean ive got a free account for internal, but i work there ;)19:59
SamYapleI never actually use it19:59
sdakeok19:59
SamYaplejpeeler: have you see the OSIC stuff Intel and RAX are planning?20:01
sdakehey samyaple20:02
SamYaplehi sdake20:02
sdakefeedbck from operators summit, wife said almsot every operator is using linuxbridge not ovs20:02
jpeelerSamYaple: i didn't, but i just read the blog announcement. 1000 node cluster should be useful20:03
SamYaple200020:04
openstackgerritMerged stackforge/kolla: Add custom code to base docker templates  https://review.openstack.org/21554120:05
jpeeler1000*2 yeah. wonder how they'll grant access20:05
SamYaplesdake: who can blame them? very few people use the right versions of OVS, and ubuntu (the most used distro for openstack last i heard) has OVS 2.0.2 (a known broken devel version) in the repo20:05
SamYaplesdake: that being said DVR will never be on OVS and the neutron people only really care about OVS, linuxbridge was always a patch job onto of neutron20:05
SamYapleand it seems to always be a secondary thought20:06
SamYaplejpeeler: as a member of this community (big tent esspecially) youll be able to schedule time to access it20:06
sdakeseems secondary to develdops but primary to operators20:06
SamYaplethats because linuxbridge 'just works'20:06
sdakepointbeing, good thing we have linuxbridge support ;)20:07
SamYaplebut openvswitch has the features and performance needed for the future20:07
SamYapleindeed20:07
sdakeor we would have to scramble to make it happen20:07
SamYapleNah its a piece of cake20:07
SamYapleWait until i setup the configs to do 1 click DVR deploy20:07
SamYaplelets see how many people use OVS then ;)20:08
sdakenice20:08
sdakeanyway i thought that was cool feedback :)20:08
SamYapleit is good. that is suprising though since as late as havana i didnt know anyone using linuxbridge20:08
sdakeare all the docker_templates patches merged?20:08
SamYaplei guess people go burned on early OVS20:09
sdakefor centos + binary20:09
SamYaplelooks like it20:09
sdakeand has anyone tested it?20:09
SamYaplehttps://review.openstack.org/#/q/project:stackforge/kolla+branch:master+topic:bp/dockerfile-template,n,z20:09
sdake(I haven't)20:09
SamYaplei only run templates20:09
SamYaplebeen that way for a week20:09
sdakenice on centos binary?20:09
SamYapleyea, but i havent tested booting for a few days20:09
sdakecool thanks for the info20:10
SamYaplelast time was on mon or tuesday20:10
SamYaplesdake: https://review.openstack.org/#/c/215159/20:10
SamYaplethat change work for you?20:10
sdakewtb project rename already!20:10
sdakelooking20:10
sdakesamyaple i dont have time to verify - willing to +2 / +a but need verification it works with centos-binary20:12
sdakehavey ou tried that part of it out?20:12
SamYapleyea20:12
sdakecool approving then20:12
SamYaplethis doenst change anything for centos20:12
SamYapleyou can actually run the playbooks again without getting a change20:12
sdakeits just all that curdini stuff20:13
sdakethe removals20:13
sdakei was worried about that in galera.cnf.j220:13
SamYapleoh that was cleanup from removing config-interal20:13
sdakeand that definately works post rebuild?20:13
SamYaplei could have done a seperate patch, it just happened20:13
SamYapleyea20:13
sdakeok20:13
SamYaplewe never used those functions for config-external20:13
sdakedidn't know that20:13
SamYaplei only integrated the two that are left with config-external20:14
*** tummy has joined #kolla20:14
*** tummy has quit IRC20:16
openstackgerritMerged stackforge/kolla: Fix MariaDB for ubuntu  https://review.openstack.org/21515920:18
openstackgerritJeff Peeler proposed stackforge/kolla: Add templated dockerfiles for ironic  https://review.openstack.org/21577020:33
sdakejpeeler rockin dude ;)20:36
sdakeprogress ftw20:36
SamYaplejpeeler: is this a new container, or migrations frok docker folder?20:37
jpeeleruntested, but they build20:37
jpeelerSamYaple: new20:37
SamYapleok cool20:38
jpeelerSamYaple: so remove the run line or the entire if endif section?20:40
SamYaplejpeeler: just the entire section20:40
SamYaplethats what we do with the others20:40
openstackgerritJeff Peeler proposed stackforge/kolla: Add templated dockerfiles for ironic  https://review.openstack.org/21577020:45
*** jtriley has quit IRC20:53
sdakeharmw ping20:57
harmwshoot me!20:59
sdake;)20:59
sdakehorizon20:59
harmwyep, tomorrow20:59
sdakein the forecast?20:59
harmwsaturdayevening20:59
sdakenice20:59
sdakestart with the patch up in gerrit20:59
harmwyep20:59
sdakeyou can use the cherry-pick tab to fetch it21:00
sdakeon top of master21:00
sdakeit probably needs a rebae21:00
harmwyes sir21:00
sdakeand rework to remove the start.yml and stuff abstractions21:00
sdakewhatever state you get it into before ou head off to bed21:00
sdakeplease push it21:00
sdakerather git revew it21:00
harmwoh ok21:00
sdakei may pick it up so we can work it 24 hours :)21:00
harmwfair enough21:01
harmw:)21:01
sdakeand then i'll push21:01
sdakeand youcan pick up possibly sunday :)21:01
harmwlets see how far I can get tomorrow21:01
sdakebasically what i'm say8ing is elts work it together21:01
harmwI hear you21:01
sdakei'd like it merged before tuesday if at all possible21:01
harmwill keep that in mind21:02
*** banix has quit IRC21:22
*** weiyu_ has joined #kolla21:24
*** ccrouch has joined #kolla21:33
*** ccrouch has quit IRC21:33
*** ccrouch has joined #kolla21:33
*** dolpher has joined #kolla21:36
*** tummy has joined #kolla21:42
*** bradjones has quit IRC21:42
*** bradjones has joined #kolla21:43
*** bradjones has quit IRC21:43
*** bradjones has joined #kolla21:43
*** jruano has joined #kolla21:56
*** ccrouch has quit IRC21:57
*** dolpher1 has joined #kolla22:01
*** dolpher has quit IRC22:03
*** blahRus has quit IRC22:05
*** weiyu_ has quit IRC22:12
openstackgerritSteven Dake proposed stackforge/kolla: Rework documentaton to actually get people rolling  https://review.openstack.org/21553222:17
sdakejpeeler around?22:25
*** achanda has joined #kolla22:38
*** dolpher has joined #kolla22:42
*** dolpher1 has quit IRC22:44
*** banix has joined #kolla22:47
*** jasonsb has quit IRC22:49
*** tummy has quit IRC22:49
*** banix has quit IRC23:05
*** dolpher1 has joined #kolla23:05
*** dolpher has quit IRC23:07
*** stvnoyes has quit IRC23:25
*** daneyon has quit IRC23:34
*** jruano has quit IRC23:37
*** banix has joined #kolla23:53
*** jruano has joined #kolla23:55

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