| *** bmace has joined #kolla | 00:01 | |
| *** achanda has joined #kolla | 00:01 | |
| *** weiyu has quit IRC | 00:01 | |
| *** weiyu has joined #kolla | 00:03 | |
| *** weiyu has quit IRC | 00:06 | |
| *** Haomeng|2 has joined #kolla | 00:07 | |
| *** Haomeng has quit IRC | 00:10 | |
| *** banix has joined #kolla | 00:15 | |
| *** achanda has quit IRC | 00:18 | |
| *** achanda has joined #kolla | 00:25 | |
| *** dolpher has joined #kolla | 00:30 | |
| sdake | jpeeler please review https://review.openstack.org/#/c/212194/ | 00:31 |
|---|---|---|
| *** zhiwei has joined #kolla | 00:39 | |
| *** jasonsb has quit IRC | 00:47 | |
| *** achanda has quit IRC | 00:55 | |
| *** achanda has joined #kolla | 00:58 | |
| openstackgerrit | weiyu proposed stackforge/kolla: dd Dockerfile template for haproxy https://review.openstack.org/212232 | 00:59 |
| *** rhallisey has quit IRC | 01:01 | |
| *** weiyu_ has joined #kolla | 01:01 | |
| *** echoingumesh has quit IRC | 01:09 | |
| *** echoingumesh has joined #kolla | 01:09 | |
| openstackgerrit | weiyu proposed stackforge/kolla: Add Dockerfile template for haproxy https://review.openstack.org/212232 | 01:11 |
| *** echoingumesh has quit IRC | 01:13 | |
| *** echoingumesh has joined #kolla | 01:13 | |
| weiyu_ | sdake: I changed the commit message, please review once again | 01:14 |
| sdake | ack - enjoy | 01:14 |
| weiyu_ | thanks | 01:15 |
| jpeeler | "Job gate-kolla-build-images-centos-rdo not defined" | 01:17 |
| sdake | try looking again? | 01:18 |
| jpeeler | i thought you were going to do a separate commit, not amend it | 01:18 |
| sdake | i want it to havetemplating from the start | 01:18 |
| sdake | rather then rework the entire thing | 01:18 |
| sdake | easier this way | 01:18 |
| jpeeler | anyway, married life doesn't allow me to keep working. will have to look tomorrow if you don't figure it out! | 01:18 |
| sdake | jpeeler it is passing the gate | 01:21 |
| sdake | i think you looked at an old rev | 01:21 |
| sdake | jpeeler does your wife work | 01:22 |
| *** jpeeler has quit IRC | 01:23 | |
| *** weiyu_ has quit IRC | 01:26 | |
| *** weiyu has joined #kolla | 01:28 | |
| *** erkules_ has joined #kolla | 01:31 | |
| *** erkules has quit IRC | 01:34 | |
| *** unicell has quit IRC | 01:35 | |
| *** echoingumesh has quit IRC | 01:40 | |
| *** echoingumesh has joined #kolla | 01:41 | |
| *** echoingumesh has quit IRC | 01:45 | |
| *** wshao has quit IRC | 01:49 | |
| *** achanda has quit IRC | 01:51 | |
| *** erkules_ has quit IRC | 01:52 | |
| *** erkules_ has joined #kolla | 01:54 | |
| *** diogogmt has joined #kolla | 01:54 | |
| openstackgerrit | Swapnil Kulkarni (coolsvap) proposed stackforge/kolla: Add Dockerfile template for Cinder https://review.openstack.org/211919 | 01:55 |
| *** Kennan has joined #kolla | 02:09 | |
| openstackgerrit | Merged stackforge/kolla: Fix missing config in horizon source image https://review.openstack.org/209415 | 02:13 |
| SamYaple | morning | 02:13 |
| Kennan | hi SamYaple: | 02:19 |
| Kennan | does kolla support fedora? | 02:19 |
| Kennan | From doc it said | 02:19 |
| Kennan | NB: Kolla will not run on Fedora 22 or later. Fedora 22 compresses kernel modules with the .xz compressed format. The guestfs system cannot read these images because a dependent package supermin in CentOS needs to be updated to add .xz compressed format support. | 02:19 |
| Kennan | I did not quite understand it. | 02:20 |
| SamYaple | Kennan: centos built images will not work with Fedora 22 | 02:20 |
| sdake | and the fedora images dont build atm | 02:20 |
| SamYaple | Fedora22 built images should work, but Kolla doesnt have fedora22 images | 02:20 |
| SamYaple | additionally fedora is broken atm | 02:20 |
| sdake | use centos images on ubuntu for centos | 02:20 |
| sdake | that is all that works atm kennan | 02:21 |
| sdake | we are sorting it out atm | 02:21 |
| Kennan | ok sdake: so you means kolla container images is based on centos and ubuntu ? | 02:21 |
| Kennan | not have fedora based image | 02:22 |
| Kennan | right ? | 02:22 |
| sdake | kennan our short term plan is to build containers for centos, fedora, oraclelinux, ubuntu, debian, and rhel | 02:22 |
| sdake | in both from-source install and from-rpm/deb install | 02:22 |
| sdake | so the container runtime and binaries match the kernel they are running against | 02:23 |
| sdake | for maximum ioctl, syscall, and netlink compatability | 02:23 |
| sdake | we are in the process of reworking the docker files to ssupport this model | 02:23 |
| sdake | without a million symlinks | 02:23 |
| sdake | when we get done, I doubt we will have any symlinks | 02:23 |
| sdake | yay | 02:23 |
| sdake | the symlink think was an interesting idea but it just is not ideal | 02:24 |
| Kennan | sdake: I checked https://github.com/stackforge/kolla/blob/master/docs/dev-quickstart.md IF I want to start a dev env, what VM should I use to test the kolla compose ? | 02:24 |
| sdake | ya | 02:24 |
| sdake | compose is gone | 02:24 |
| sdake | you have to use ansible to deploy | 02:24 |
| sdake | read ansible-deployment.md | 02:25 |
| sdake | or rst i don't recall which are docs use :) | 02:25 |
| Kennan | OK sdake: the doc in dev-quick not updated if you said use ansible, it is still compose | 02:25 |
| sdake | use either fedora 21, centos 7, or ubuntu 14.04 for your os | 02:25 |
| sdake | ya we have been in a major refactor o the whole codebase in l3 ;) | 02:26 |
| sdake | i am aware dev-quickstart is not up to date | 02:27 |
| sdake | I guss i'll fix that after I finish this review queue | 02:27 |
| sdake | we have like a million patches outstanding | 02:27 |
| Kennan | ok sdake: just this https://github.com/stackforge/kolla/blob/master/docs/ansible-deployment.md I read it, seem not easy for me to understand, is it for dev env? | 02:27 |
| sdake | which part dont you understand | 02:28 |
| openstackgerrit | Sam Yaple proposed stackforge/kolla: Update mariadb confs and ports https://review.openstack.org/210980 | 02:28 |
| openstackgerrit | Swapnil Kulkarni (coolsvap) proposed stackforge/kolla: Add Dockerfile template for Ceilometer https://review.openstack.org/211878 | 02:29 |
| SamYaple | sdake: https://review.openstack.org/#/c/211871/ | 02:29 |
| SamYaple | if youre not busy | 02:29 |
| sdake | just trying to fix the review bcklog ;) | 02:30 |
| sdake | oh i'mgetting to it samyaple | 02:30 |
| Kennan | sdake: I read it, it needs to know ansible , ok that's ok. But If I want to have a setup-by-setup development environment setup | 02:30 |
| Kennan | I did not know how to get that started | 02:30 |
| SamYaple | Kennan: im not sure what you dont understand | 02:30 |
| Kennan | ansible-playbook -i inventory/all-in-one -e @/etc/kolla/defaults.yml -e @/etc/kolla/globals.yml -e @/etc/kolla/passwords.yml site.yml --tags rabbitmq | 02:30 |
| sdake | samyaple I think he wants step by step instructions | 02:31 |
| sdake | kennan i can help you get going | 02:31 |
| SamYaple | You only need to have docker, docker-py, and ansible installed | 02:31 |
| Kennan | just use this ansible command? can we have all services running? | 02:31 |
| sdake | ou have to build the images first | 02:31 |
| sdake | which is undocumented | 02:31 |
| sdake | tools/build.py -t rdo -T 1000 | 02:31 |
| sdake | from master | 02:31 |
| Kennan | The dev-quickstart.md is easy to understand | 02:31 |
| Kennan | in that style and format | 02:31 |
| sdake | yes it is step by step | 02:32 |
| sdake | i get what your saying | 02:32 |
| Kennan | it told you what step and step | 02:32 |
| sdake | i did rework the documentation if your interested in seeing | 02:32 |
| Kennan | yes, like to try kolla, it seems very intersting | 02:32 |
| Kennan | and I think good doc is important for new comers | 02:33 |
| Kennan | if they like to try that | 02:33 |
| sdake | https://review.openstack.org/#/c/210694/1/docs/dev-quickstart.md | 02:33 |
| sdake | have a go with that | 02:33 |
| openstackgerrit | Swapnil Kulkarni (coolsvap) proposed stackforge/kolla: Add Dockerfile template for Cinder https://review.openstack.org/211919 | 02:33 |
| Kennan | sdake: good to know that | 02:33 |
| Kennan | will check that. Thanks | 02:33 |
| sdake | if you get stuck ask, can offer steps and help improve the documentation | 02:34 |
| sdake | after I get through the review queue i'll merge those documents | 02:35 |
| sdake | coolsvap you around | 02:37 |
| openstackgerrit | weiyu proposed stackforge/kolla: Add Dockerfile template for haproxy https://review.openstack.org/212232 | 02:37 |
| openstackgerrit | Merged stackforge/kolla: Cinder-backup is not being configured because config-external spelling error https://review.openstack.org/212160 | 02:37 |
| openstackgerrit | Merged stackforge/kolla: Ansible Cinder needs to create the v2 endpoint https://review.openstack.org/212140 | 02:39 |
| *** achanda has joined #kolla | 02:45 | |
| openstackgerrit | Sam Yaple proposed stackforge/kolla: Implement source fetching for build.py https://review.openstack.org/211954 | 02:45 |
| openstackgerrit | Swapnil Kulkarni (coolsvap) proposed stackforge/kolla: Add Dockerfile template for Magnum https://review.openstack.org/212289 | 02:47 |
| sdake | samyaple i'm not sure the bindmounts are correct on https://review.openstack.org/#/c/211442/ - did you test that change mnaullay? | 02:48 |
| SamYaple | sdake: no i have not tested that change since i did not write it | 02:49 |
| sdake | oh my bad i cant read | 02:49 |
| *** jasonsb has joined #kolla | 02:50 | |
| sdake | ok guys here is the deal on liberty 3 | 02:52 |
| sdake | here is our governance repo entry: https://github.com/openstack/governance/blob/master/reference/projects.yaml#L1062 | 02:53 |
| sdake | that is documented here: | 02:54 |
| sdake | https://github.com/openstack/governance/blob/master/reference/tags/release_independent.rst | 02:54 |
| sdake | the tldr version is we don't have to follow any release schedule or milestone schedule | 02:55 |
| sdake | I would have submitted us for https://github.com/openstack/governance/blob/master/reference/tags/release_cycle-with-milestones.rst | 02:56 |
| sdake | the TLDR of that model is the standard 6 month 3 milestone + rc + freeze cycle | 02:56 |
| sdake | the reason I didn't submit us for this is I wanted flexibility atleast during liberty and possibly during mitaka to push scheduels as necessary since we hve a ton of work to do | 02:57 |
| sdake | the goal is to get to release_cycle_with_milestones | 02:57 |
| sdake | we have been following this exact model | 02:57 |
| sdake | this is why I have been strict witth the rleleases and tagging | 02:57 |
| sdake | liberty 3 concludes aug 31 | 02:58 |
| sdake | it seems clear to me we cant finish teh work we need to finish to meet our objectives for a working system with the basic services we hvae defined | 02:59 |
| sdake | with support for our various distro options as container oses | 02:59 |
| sdake | this is why for example I pushed a bunch of work out | 02:59 |
| sdake | to mitaka | 02:59 |
| sdake | to relieve a bit of the pressure | 02:59 |
| sdake | now, our governance allows us to release on whatever cycle we want | 02:59 |
| *** weiyu has quit IRC | 03:00 | |
| sdake | but for example I dont want us releasing liberty support 3 months after liberty ships :) | 03:00 |
| sdake | lets really focus on the essential/high blueprints for l3 | 03:00 |
| sdake | https://wiki.openstack.org/wiki/Liberty_Release_Schedule | 03:00 |
| sdake | october 15th is the ultimate release date | 03:01 |
| sdake | I'd like to use the month of september to clean up the bugs we are jamming in at a high right, becaue for example people aren't actually testing the code | 03:01 |
| sdake | and implement any medium/low blueprints that we find mandatory to serve our mission | 03:02 |
| sdake | if your working on medium/low blueprints dont interrupt that work | 03:02 |
| sdake | i want to ignore the docker directory for contianers - and focus entirely on "v2" version of docker containers | 03:02 |
| sdake | so we can get multi-distro container implementation done | 03:02 |
| sdake | i want to continue to use ansible to deploy v1 containers | 03:03 |
| sdake | after september 31 (1 mo) I will not be keen on merging any major changes | 03:03 |
| sdake | bug fixes only - that is the 6 mo release cycle RC period | 03:03 |
| SamYaple | whats teh difference between v1 and v2 containers? | 03:04 |
| sdake | we will skip the FFE process for this cycle | 03:04 |
| SamYaple | i dont follow | 03:04 |
| sdake | docker = v1, docker_tempaltes = v2 | 03:04 |
| SamYaple | arent we scraping 'v1' as soon as we template them all? | 03:04 |
| sdake | as soon as we test they work reliably yes | 03:05 |
| SamYaple | right, but thatll be in like a week, right? | 03:05 |
| sdake | also scrapping existing build system as soon as we have confidence in new build system | 03:05 |
| sdake | god I hope so | 03:05 |
| SamYaple | ok good | 03:05 |
| SamYaple | i dont want to have two for any longer than needed | 03:05 |
| sdake | I feel pretty confident at the rate we are going we will be done by next wed | 03:05 |
| sdake | with centos source and binary | 03:05 |
| sdake | we are moving pretty fast | 03:06 |
| sdake | that will give us partiy | 03:06 |
| SamYaple | source is a bit trickier since i only have the tarball stuff done | 03:06 |
| SamYaple | but it works reliably | 03:06 |
| sdake | tarball is good enough to implement with | 03:06 |
| SamYaple | fiar enough | 03:06 |
| sdake | I noticed coolsvap is not implementing from src tho | 03:06 |
| sdake | so we need to make sure people aer implementing from source | 03:06 |
| sdake | its basic cut and paste with a search and replace | 03:06 |
| *** weiyu has joined #kolla | 03:07 | |
| sdake | so lets really shoot for ESSENTIAL/HIGH at aug 31 (3 weeks) and MED/LOW at SEP 31 | 03:08 |
| sdake | after aug31 may start boucning med/low that are not essential | 03:08 |
| SamYaple | sdake: how is he not? | 03:08 |
| SamYaple | he appears to be | 03:08 |
| sdake | i thought i saw some that were not implemented | 03:08 |
| sdake | i coudl be wrong | 03:08 |
| sdake | tired | 03:08 |
| SamYaple | source only installs in the *-base image | 03:08 |
| sdake | right | 03:09 |
| SamYaple | he has ceilometer and cinder | 03:09 |
| sdake | i remember thinking he didn't and then seeing he did in the base | 03:09 |
| SamYaple | hes doing it in both | 03:09 |
| sdake | so I think its fine | 03:09 |
| sdake | my mistake | 03:09 |
| SamYaple | testing isnt happening though | 03:09 |
| sdake | well nobody is deploying containers from v2 that is for certain | 03:10 |
| SamYaple | i did | 03:10 |
| SamYaple | mixed with the v1 that i needed | 03:10 |
| sdake | nice | 03:10 |
| SamYaple | what else would testing mean? | 03:10 |
| sdake | testing that they atlaest build | 03:10 |
| sdake | i am not convinced thats happening either ;) | 03:10 |
| SamYaple | it is not | 03:11 |
| SamYaple | otherwise it would break | 03:11 |
| sdake | to rectify that the gate is going up (review outstanding) to gate the builds | 03:11 |
| SamYaple | the gate should point out non-tested patches | 03:11 |
| SamYaple | yea that | 03:11 |
| sdake | for centos from source and binary | 03:11 |
| sdake | review is outstanding - waiting on infra | 03:12 |
| coolsvap | morning | 03:12 |
| sdake | so goal is to behave like 6 month release cycle | 03:12 |
| sdake | hey coolsvap | 03:12 |
| sdake | but we have flexiblity to release when we want | 03:13 |
| sdake | if we released 3 months after liberty nobody would fault us accordingto our governance | 03:13 |
| coolsvap | sdake, were you talking about trove? | 03:13 |
| sdake | but I think it would be drsaticlly bad for the project | 03:13 |
| sdake | coolsvap your neutron change looks busted - did you do a test deploy | 03:13 |
| sdake | trove was pushed ti mitaka | 03:14 |
| coolsvap | sdake, no i did not, my ansible environment gets screwed every other day | 03:14 |
| sdake | vms ftw | 03:14 |
| sdake | we are not testing our changes - changes to ansible NEED to be tested | 03:15 |
| sdake | changes to container v2 i am less concerned about but it would be handy if people tested those too | 03:15 |
| sdake | any questions? | 03:16 |
| SamYaple | can we stop saying containers v2? | 03:17 |
| SamYaple | first, its 'images' and not containers. second, this is a refactor | 03:17 |
| SamYaple | its bugging me alot | 03:17 |
| sdake | call it whatever | 03:20 |
| SamYaple | templating | 03:20 |
| sdake | samyaple did you finish https://blueprints.launchpad.net/kolla/+spec/containerize-dependencies | 03:20 |
| SamYaple | iamge templating if youre fancy | 03:20 |
| SamYaple | sdake: no | 03:20 |
| sdake | whats left there? | 03:20 |
| *** diogogmt has quit IRC | 03:20 | |
| SamYaple | or rather, i was going to bundle the shade ansible modules into that blueprint since they are deps | 03:21 |
| sdake | it says its in needs code review | 03:21 |
| sdake | is it more like in good progress? | 03:21 |
| SamYaple | yea | 03:21 |
| SamYaple | it probably kicked to code review since there was a prior patch | 03:21 |
| SamYaple | but it was a 'partially' patch | 03:21 |
| sdake | humans kick to code review | 03:22 |
| sdake | not infra | 03:22 |
| SamYaple | probably you then | 03:22 |
| sdake | coolsvap is https://blueprints.launchpad.net/kolla/+spec/ansible-cinder finished or in code review | 03:22 |
| sdake | possible there is no log | 03:22 |
| SamYaple | i manage my blueprints | 03:22 |
| sdake | i htought you were done | 03:22 |
| SamYaple | yea | 03:23 |
| sdake | if it was me | 03:23 |
| coolsvap | sdake, done | 03:23 |
| sdake | merged? | 03:23 |
| SamYaple | im pretty active on launchpad now if you want to trust the status of my blueprints to relieve some burden sdake | 03:23 |
| coolsvap | yes | 03:23 |
| sdake | samyaple cool wfm | 03:23 |
| sdake | tested? | 03:23 |
| coolsvap | rhallisey has copule of bugs on it | 03:23 |
| coolsvap | yes | 03:23 |
| SamYaple | yea its broke right now | 03:23 |
| sdake | cooslvap wfm, can you set to implemented then | 03:24 |
| *** jtriley has joined #kolla | 03:25 | |
| sdake | cooslsvap so your working on ceiloemter atm? | 03:26 |
| sdake | or something else | 03:26 |
| sdake | the onlly thing that concerns me is we are half way through the sprint and atleast half our blueprints are not started :( | 03:27 |
| SamYaple | sdake: do you know the kernel for the gate? | 03:29 |
| sdake | fedora 21, iirc 3.19 | 03:29 |
| SamYaple | ubuntu? | 03:29 |
| SamYaple | or a way to find out | 03:29 |
| coolsvap | sdake yes | 03:29 |
| sdake | our build gates are f2 | 03:29 |
| sdake | 1 | 03:29 |
| SamYaple | im aware | 03:29 |
| SamYaple | we need an ubuntu one though | 03:30 |
| coolsvap | working on ansible-ceilometer | 03:30 |
| sdake | coolsvap nice :) | 03:30 |
| sdake | well keep cracking on that one | 03:30 |
| sdake | i know this template thing has been kind of disruptive | 03:30 |
| *** weiyu has quit IRC | 03:30 | |
| sdake | but i want to get the abi to containers in a stable state | 03:30 |
| sdake | and i really want multi-distro containers | 03:30 |
| sdake | samyaple no idea on ubuntu trusty i think whatever version that is | 03:31 |
| SamYaple | sdake: trusty has several versions | 03:31 |
| sdake | you can ask on #openstack-infra | 03:31 |
| SamYaple | 3.19 being the most recent | 03:31 |
| sdake | if its brand new i highly dobut it has 3.19 kernel | 03:32 |
| sdake | gate oses lag couple releases | 03:32 |
| SamYaple | its not | 03:33 |
| SamYaple | its been in the repo 4 months | 03:33 |
| sdake | still lags | 03:33 |
| SamYaple | but its the hwe kernel | 03:33 |
| SamYaple | ubuntu kernels are different than fedora sdake | 03:33 |
| sdake | ya doubt the gate is running a hardware enablement kernel | 03:33 |
| SamYaple | for the life of 14.04 (trusty) it will remain 3.13 | 03:33 |
| sdake | i dont know for sure | 03:34 |
| sdake | ask on #openstack-infra | 03:34 |
| sdake | you are in need of hwe kernel for overlayfs? | 03:34 |
| *** weiyu_ has joined #kolla | 03:34 | |
| SamYaple | no AUFS | 03:37 |
| SamYaple | its not like the machine has btrfs support | 03:37 |
| SamYaple | well i mean the kernel does but it doesnt have a btrfs disk | 03:37 |
| sdake | can't you use devicemapper? | 03:37 |
| SamYaple | not with ubuntu | 03:37 |
| sdake | really didn't know that | 03:37 |
| SamYaple | old libs | 03:38 |
| sdake | how do people use docker with 14.04 3.13 kernel? | 03:38 |
| SamYaple | you would have to statically build docker for device mapper | 03:38 |
| SamYaple | sdake: i use btrfs | 03:38 |
| SamYaple | can we get a volume attached to the gate? | 03:38 |
| SamYaple | or a an ephemeral disk? | 03:39 |
| sdake | asking wrong dude | 03:39 |
| SamYaple | im askign channel | 03:39 |
| sdake | asking wrong channel :) | 03:40 |
| SamYaple | how am i suppose to know | 03:41 |
| SamYaple | oh i mean i was asking infra channel | 03:41 |
| SamYaple | not this channel sdake | 03:41 |
| sdake | oh | 03:41 |
| *** jtriley has quit IRC | 03:49 | |
| *** dolpher has quit IRC | 03:50 | |
| SamYaple | sdake: so it looks liek we can do this | 03:54 |
| SamYaple | how do i get an ubuntu gate? | 03:54 |
| sdake | make it work with tox first | 03:54 |
| sdake | gate second | 03:54 |
| SamYaple | im not sure i follow | 03:55 |
| sdake | atm jpeeler's script has no way of passing arguments to the test script | 03:55 |
| sdake | 2 minutes reviewing | 03:55 |
| sdake | ok tox | 04:00 |
| sdake | https://github.com/stackforge/kolla/blob/master/tox.ini#L45 | 04:00 |
| sdake | you will notie it doesn't pass type or distro | 04:00 |
| SamYaple | i see centos in that name | 04:01 |
| SamYaple | and rdo and src | 04:01 |
| sdake | its the commands that run tox | 04:01 |
| sdake | testr is the run script | 04:01 |
| sdake | testr runs all the test cases in tests directory | 04:02 |
| sdake | i think at the moment it is selected to run only test_build | 04:02 |
| sdake | this is based upon build-all-docker-images; https://github.com/stackforge/kolla/blob/master/tox.ini#L54 | 04:03 |
| sdake | it worked when i tested it | 04:03 |
| sdake | but jeff has recently changed the test script | 04:03 |
| sdake | to run build.py directly | 04:04 |
| sdake | what needs to happen is arguments need to be passed to testr in some way | 04:04 |
| sdake | not sure how | 04:04 |
| sdake | to set distro and type | 04:04 |
| sdake | the testenv: is a inheritence operation | 04:05 |
| sdake | it will run everyting in testenv first | 04:05 |
| sdake | followed by what is in images-centos-rdo | 04:05 |
| sdake | tox -eimages-centos-rdo | 04:05 |
| sdake | to fire it off | 04:05 |
| sdake | this is the gate integration point | 04:05 |
| SamYaple | what do we want this to look like, 3 support 2 network 2 compute | 04:08 |
| sdake | this code runs the job: https://review.openstack.org/#/c/212194/ | 04:08 |
| SamYaple | ? | 04:08 |
| sdake | are you doing docker in docker? | 04:08 |
| SamYaple | yea | 04:08 |
| SamYaple | keeping in mind there is virtually zero overhead for having a different host | 04:08 |
| sdake | let focus on getting the images built first | 04:09 |
| sdake | and expand from there | 04:09 |
| openstackgerrit | Merged stackforge/kolla: Add Dockerfile template for haproxy https://review.openstack.org/212232 | 04:09 |
| SamYaple | well thats what jpeeler is doing is it not? | 04:09 |
| sdake | not sure he is working on ubuntu image building | 04:09 |
| SamYaple | im not either. im wroking on functional testing | 04:09 |
| coolsvap | sdake, can you tell me the the copr for fedora magnum? | 04:09 |
| sdake | coolsvap i linked it in the review | 04:10 |
| coolsvap | oh missed it sorry | 04:10 |
| sdake | you will have to click around to find the repo file | 04:10 |
| coolsvap | got it | 04:11 |
| sdake | we need tox integration for building images before tox integration for docker in docker | 04:11 |
| SamYaple | yea but thats not me, thats jpeeler | 04:12 |
| sdake | so you are dependent on jpeeler for that (or me) | 04:12 |
| SamYaple | im doing the dind | 04:12 |
| *** weiyu_ has quit IRC | 04:12 | |
| sdake | docker in docker will look alot like image building plus running whatever test script or tool ou come up with | 04:13 |
| SamYaple | what is the disconnect here? | 04:14 |
| sdake | image bulding gating isnt' done | 04:14 |
| sdake | its a dependency | 04:14 |
| SamYaple | yea but im not hte one working on that | 04:14 |
| sdake | yes but your blocked because that job isn't finished | 04:14 |
| SamYaple | im not blocked | 04:14 |
| openstackgerrit | Swapnil Kulkarni (coolsvap) proposed stackforge/kolla: Add Dockerfile template for Magnum https://review.openstack.org/212289 | 04:15 |
| SamYaple | how am i blocked | 04:15 |
| SamYaple | I still have to write the scripts | 04:15 |
| sdake | how are you going to build images? | 04:15 |
| sdake | the integration point for dnd shoud be tox as well | 04:15 |
| SamYaple | i dont understand what you think is happening here | 04:16 |
| sdake | preferrably in python | 04:16 |
| SamYaple | it wont be python | 04:16 |
| sdake | we run test_build | 04:16 |
| SamYaple | its bash scripts | 04:16 |
| sdake | thenw erun the dnd | 04:16 |
| SamYaple | dind, not dnd | 04:16 |
| sdake | test build doesn't work for your use case | 04:16 |
| SamYaple | i have no idea what youre talking about sdake. i still have to write the dind images, i still ahve to test the setp | 04:17 |
| SamYaple | what does this have to do with the gate | 04:17 |
| sdake | here is the flow - gate runs tox -eimages-centos-rdo | 04:18 |
| sdake | gate runs tox -e dind | 04:18 |
| sdake | tox runs your script | 04:18 |
| SamYaple | yes | 04:18 |
| SamYaple | i still have to write that script | 04:18 |
| SamYaple | how am i blocked | 04:18 |
| sdake | you aren't blocked on writing the script | 04:19 |
| sdake | you are lbocked on getting it into the gate | 04:19 |
| SamYaple | i know... | 04:19 |
| sdake | we also need a setup_gate_ubutnu.sh | 04:20 |
| sdake | that runs before tox | 04:20 |
| sdake | look at this file: | 04:20 |
| *** dolpher has joined #kolla | 04:20 | |
| sdake | https://review.openstack.org/#/c/212194/6/jenkins/jobs/kolla.yaml | 04:20 |
| sdake | line 39 | 04:21 |
| *** unicell has joined #kolla | 04:21 | |
| SamYaple | we dont need a seperate gate script | 04:21 |
| sdake | when you have dnd, you would add a line 40 which runs tox -e dind-ubuntu-src | 04:21 |
| SamYaple | dind | 04:21 |
| sdake | the current one is fedora specific | 04:21 |
| SamYaple | yea but we dont need a seperate one | 04:22 |
| SamYaple | we can tweak the existing one | 04:22 |
| sdake | i had curl in there for downloading docker - jpeeler -1 the review | 04:22 |
| sdake | now it uses yum which is fedora specific | 04:22 |
| SamYaple | i dont understand whats goign on right now. why are you acting as if there are lots of blockers? | 04:23 |
| SamYaple | that script can have an if state ment, if ubuntu if redhat | 04:23 |
| sdake | agree that is not a blocker | 04:23 |
| SamYaple | this dind script doesnt rely on any of that | 04:23 |
| sdake | just telling you what needs to be done - you did ask | 04:23 |
| SamYaple | no i asked about an ubuntu gate | 04:23 |
| SamYaple | so i could run some commands on it | 04:23 |
| SamYaple | liek ifconfig | 04:23 |
| sdake | changing project-config takes a long time, much faster to use tox integration | 04:24 |
| sdake | that means that structure for that code i inked last is always run | 04:24 |
| SamYaple | i need an ubuntu machine in the gate. if that means changing/adding to project-config we still haveto do that | 04:25 |
| sdake | https://review.openstack.org/#/c/212194/6/jenkins/jobs/projects.yaml | 04:25 |
| sdake | this file runs the job | 04:25 |
| sdake | jobs | 04:25 |
| sdake | I can add another job with ubuntu s;ecific in it | 04:26 |
| SamYaple | ill add it | 04:26 |
| sdake | it would run the 2nd to last pate | 04:26 |
| SamYaple | i believe we need 2 | 04:26 |
| sdake | if you do make it a dep of my patch pls | 04:27 |
| SamYaple | oh thats not merged in yet | 04:27 |
| sdake | precisely | 04:27 |
| sdake | blocker | 04:27 |
| SamYaple | then yea we shoudl just tweak it | 04:27 |
| SamYaple | we need that to be binary, no? | 04:28 |
| SamYaple | 'rdo' isnt going to exist | 04:28 |
| SamYaple | also, why src not source | 04:28 |
| sdake | rdo is the tox job | 04:29 |
| sdake | there is no good reason for it not to be source, but the tox job is src | 04:31 |
| SamYaple | https://review.openstack.org/#/c/212194/6/zuul/layout.yaml | 04:31 |
| SamYaple | but youre not adding it | 04:31 |
| sdake | if you want it changed, change the tox jobs and i'll modify porject-config | 04:31 |
| sdake | i am not adding it as voting | 04:31 |
| sdake | it is nonvoitng | 04:31 |
| sdake | voting means it rejects the patch until it passes | 04:31 |
| SamYaple | i am aware or what it means | 04:32 |
| SamYaple | look you arent adding it as a check | 04:32 |
| sdake | good point i need to edit the patch | 04:33 |
| SamYaple | so can you add the ubuntu stuff now as well? | 04:33 |
| sdake | i think this ptch will possily get a -1 because i've done two gates in one patch | 04:35 |
| sdake | i'm not doing 3 | 04:35 |
| sdake | it would have to be a separate patch | 04:36 |
| SamYaple | i dont understand what that matters? | 04:36 |
| sdake | the infra team suffers from erious reviewer fatigue | 04:36 |
| sdake | the easier the patch is to review the faster it hits the repo | 04:36 |
| SamYaple | i am positive you are just being difficult at the moment | 04:36 |
| sdake | I also dont know how to add ubuntu gates so i will need to consult infra team in morning | 04:36 |
| *** dolpher1 has joined #kolla | 04:37 | |
| sdake | my goal at the moment with this patch is build testing | 04:37 |
| sdake | i really dont know how to add an ubuntu gate | 04:37 |
| sdake | but i can figure it out | 04:37 |
| SamYaple | devstack-ubuntu | 04:37 |
| sdake | but everyone is long asleep | 04:37 |
| sdake | i've been going since 5am and its 10pm | 04:37 |
| sdake | and i'm hungry | 04:37 |
| sdake | had zero food today | 04:38 |
| SamYaple | is that whats causing the difficultness | 04:38 |
| sdake | trying to get wife fed | 04:38 |
| sdake | trying to get wife to feed me atm | 04:38 |
| sdake | and also, reviewer fatigue of the infra team, it needs to be a separate patch | 04:38 |
| sdake | if you want it i will make it happen | 04:38 |
| sdake | i need a tox target to work with | 04:39 |
| *** dolpher has quit IRC | 04:39 | |
| SamYaple | ill submit a patch with a dep on yours, but again I am positive had you wanted it it would be in this patch and you are just being difficult | 04:39 |
| sdake | i hate fucking with project-config - it is really complicated | 04:39 |
| sdake | and it takes on order of 3-5 days to get a merge out of it | 04:40 |
| sdake | so yes i want to keep patches as small as possible to it | 04:40 |
| SamYaple | this changes nothing from what i said | 04:40 |
| sdake | i absolutely do not want it in this patch | 04:41 |
| sdake | i'm good with a dependency patch | 04:41 |
| sdake | i absolutely want functional testing | 04:41 |
| SamYaple | which you are absolutely delaying | 04:41 |
| sdake | i offered to do the work on the gate | 04:41 |
| sdake | i need a tox target | 04:41 |
| *** unicell1 has joined #kolla | 04:44 | |
| *** unicell has quit IRC | 04:45 | |
| *** banix has quit IRC | 04:52 | |
| *** weiyu has joined #kolla | 04:58 | |
| *** dolpher has joined #kolla | 05:00 | |
| *** dolpher1 has quit IRC | 05:01 | |
| openstackgerrit | Sam Yaple proposed stackforge/kolla: Remove bootstrap.yml abstraction for haproxy https://review.openstack.org/212328 | 05:08 |
| openstackgerrit | Steven Dake proposed stackforge/kolla: Kolla: add non-voting build-centos-binary job https://review.openstack.org/212330 | 05:09 |
| openstackgerrit | Steven Dake proposed stackforge/kolla: Kolla: add a gate for kolla-centos-source https://review.openstack.org/212331 | 05:09 |
| openstackgerrit | Sam Yaple proposed stackforge/kolla: Fixes typo in nova.conf https://review.openstack.org/212332 | 05:09 |
| openstackgerrit | Sam Yaple proposed stackforge/kolla: Remove bootstrap.yml abstraction for haproxy https://review.openstack.org/212328 | 05:10 |
| * sdake growls | 05:14 | |
| SamYaple | yea thats going to hit them right in the reviewer fatigue sdake | 05:18 |
| sdake | i fixed a bunch of errors | 05:18 |
| sdake | because I have fatigue | 05:18 |
| SamYaple | im going to -2 them just because | 05:19 |
| SamYaple | youre welcome | 05:21 |
| *** weiyu has quit IRC | 05:26 | |
| sdake | samyaple if you build an ubuntu gate use these as deps | 05:37 |
| sdake | https://review.openstack.org/#/c/212346/1 | 05:37 |
| sdake | finally dinner | 05:47 |
| openstackgerrit | Steven Dake proposed stackforge/kolla: Change the tox targtes to match naming https://review.openstack.org/212347 | 05:47 |
| openstackgerrit | Steven Dake proposed stackforge/kolla: Change the tox targtes to match naming https://review.openstack.org/212347 | 05:58 |
| openstackgerrit | Sam Yaple proposed stackforge/kolla: Remove bootstrap.yml abstraction for mariadb https://review.openstack.org/212352 | 06:03 |
| openstackgerrit | Swapnil Kulkarni (coolsvap) proposed stackforge/kolla: Add Dockerfile template for Cinder https://review.openstack.org/211919 | 06:05 |
| *** sdake has quit IRC | 06:07 | |
| *** weiyu has joined #kolla | 06:09 | |
| *** coolsvap_ has joined #kolla | 06:45 | |
| *** coolsvap_ has quit IRC | 06:46 | |
| Kennan | hi SamYaple: | 07:06 |
| Kennan | there? | 07:07 |
| *** weiyu has quit IRC | 07:37 | |
| *** britthou_ has quit IRC | 07:41 | |
| *** openstackgerrit has quit IRC | 07:41 | |
| *** akscram has quit IRC | 07:41 | |
| *** britthou_ has joined #kolla | 07:42 | |
| *** openstackgerrit has joined #kolla | 07:42 | |
| *** akscram has joined #kolla | 07:42 | |
| *** britthouser has joined #kolla | 07:43 | |
| Kennan | hi coolsvap: | 07:43 |
| Kennan | there? | 07:43 |
| coolsvap | Kennan, hi | 07:43 |
| coolsvap | whatsup? | 07:43 |
| Kennan | hi, I checked in https://review.openstack.org/#/c/210694/ | 07:44 |
| Kennan | it seems have some disagree to use ansible | 07:44 |
| Kennan | someone said to use compose | 07:44 |
| Kennan | I am not sure what's the direction for that | 07:44 |
| Kennan | coolsvap: do you know that? | 07:44 |
| *** britthou_ has quit IRC | 07:45 | |
| *** openstackgerrit has quit IRC | 07:45 | |
| *** akscram has quit IRC | 07:45 | |
| coolsvap | i think its still a discussion, but i feel a rework may be expected | 07:46 |
| coolsvap | Kennan, ^^ | 07:46 |
| Kennan | what rework? change to ansible? | 07:46 |
| coolsvap | i am not exactly sure, but as rhallisey said in the last comment, a rework maybe done to make it compatible with config-external | 07:48 |
| *** weiyu has joined #kolla | 07:49 | |
| *** weiyu has quit IRC | 07:54 | |
| *** weiyu has joined #kolla | 07:56 | |
| *** akscram has joined #kolla | 07:58 | |
| *** openstackgerrit has joined #kolla | 08:00 | |
| *** tfukushima has joined #kolla | 08:12 | |
| *** Kennan has quit IRC | 08:14 | |
| *** jmccarthy has joined #kolla | 08:15 | |
| *** kjelly has joined #kolla | 08:15 | |
| *** weiyu has quit IRC | 08:16 | |
| *** akwasnie has joined #kolla | 08:19 | |
| *** Kennan has joined #kolla | 08:21 | |
| *** weiyu_ has joined #kolla | 08:26 | |
| *** weiyu_ has quit IRC | 08:27 | |
| *** gfidente has joined #kolla | 08:29 | |
| *** weiyu_ has joined #kolla | 08:32 | |
| *** achanda has quit IRC | 08:33 | |
| *** weiyu_ has quit IRC | 08:36 | |
| *** weiyu has joined #kolla | 08:49 | |
| *** vbel has quit IRC | 09:09 | |
| *** vbel has joined #kolla | 09:09 | |
| *** tfukushima has quit IRC | 09:13 | |
| *** tfukushima has joined #kolla | 09:14 | |
| *** tfukushima has quit IRC | 09:14 | |
| *** tfukushima has joined #kolla | 09:14 | |
| *** weiyu has quit IRC | 09:17 | |
| *** weiyu has joined #kolla | 09:18 | |
| *** weiyu has quit IRC | 09:23 | |
| *** weiyu has joined #kolla | 09:24 | |
| *** dolpher has quit IRC | 09:30 | |
| openstackgerrit | Alicja Kwasniewska proposed stackforge/kolla: Add Dockerfile template for neutron https://review.openstack.org/212442 | 09:48 |
| SamYaple | Kennan: compose is non-functional at the moment | 10:08 |
| SamYaple | it is unclear if it will become functional again | 10:08 |
| openstackgerrit | Sam Yaple proposed stackforge/kolla: Implement source fetching for build.py https://review.openstack.org/211954 | 10:15 |
| openstackgerrit | Sam Yaple proposed stackforge/kolla: Removed config-internal check.sh for cinder https://review.openstack.org/210873 | 10:18 |
| SamYaple | pbourke: https://review.openstack.org/#/c/212332/ | 10:18 |
| SamYaple | thats currently breaking ansible | 10:18 |
| *** weiyu has quit IRC | 10:19 | |
| *** weiyu has joined #kolla | 10:20 | |
| *** erkules_ is now known as erkules | 10:31 | |
| *** erkules has joined #kolla | 10:31 | |
| openstackgerrit | Sam Yaple proposed stackforge/kolla: Remove bootstrap.yml abstraction for rabbitmq https://review.openstack.org/212457 | 10:35 |
| *** Haomeng|2 has quit IRC | 10:36 | |
| openstackgerrit | Sam Yaple proposed stackforge/kolla: Remove bootstrap.yml abstraction for mariadb https://review.openstack.org/212352 | 10:36 |
| *** Haomeng|2 has joined #kolla | 10:36 | |
| *** zhiwei has quit IRC | 10:37 | |
| *** zhiwei has joined #kolla | 10:37 | |
| *** tfukushima has quit IRC | 10:43 | |
| *** tfukushima has joined #kolla | 10:44 | |
| *** tfukushima has quit IRC | 10:44 | |
| *** tfukushima has joined #kolla | 10:45 | |
| *** Haomeng|2 has quit IRC | 10:51 | |
| *** Haomeng|2 has joined #kolla | 10:52 | |
| openstackgerrit | Alicja Kwasniewska proposed stackforge/kolla: Add Dockerfile template for neutron https://review.openstack.org/212442 | 10:57 |
| *** tfukushima has quit IRC | 11:09 | |
| *** rhallisey has joined #kolla | 11:17 | |
| *** weiyu has quit IRC | 11:21 | |
| *** zhiwei has quit IRC | 11:54 | |
| openstackgerrit | Ryan Hallisey proposed stackforge/kolla: All the Cinder services need to reference the correct config https://review.openstack.org/212157 | 12:05 |
| *** tfukushima has joined #kolla | 12:09 | |
| *** tfukushima has quit IRC | 12:14 | |
| openstackgerrit | Ryan Hallisey proposed stackforge/kolla: Cinder-backup's config-external is missing #!/bin/bash https://review.openstack.org/212503 | 12:17 |
| openstackgerrit | Merged stackforge/kolla: Fixes typo in nova.conf https://review.openstack.org/212332 | 12:20 |
| *** tfukushima has joined #kolla | 12:31 | |
| *** jtriley has joined #kolla | 12:31 | |
| *** Haomeng has joined #kolla | 12:37 | |
| *** Haomeng|2 has quit IRC | 12:39 | |
| *** jtriley has quit IRC | 12:40 | |
| openstackgerrit | Alicja Kwasniewska proposed stackforge/kolla: Add Dockerfile template for neutron https://review.openstack.org/212442 | 12:51 |
| *** sdake has joined #kolla | 12:53 | |
| *** sdake_ has joined #kolla | 12:54 | |
| *** sdake has quit IRC | 12:57 | |
| *** openstackgerrit has quit IRC | 13:01 | |
| *** openstackgerrit has joined #kolla | 13:02 | |
| sdake_ | morning fine folks | 13:07 |
| harmw | SamYaple: you happen to be ansible core dev? | 13:08 |
| harmw | morning sdake_ | 13:08 |
| harmw | or rather, midday or something | 13:08 |
| sdake_ | pbourke ping | 13:16 |
| *** tfukushima has quit IRC | 13:23 | |
| *** athomas has joined #kolla | 13:24 | |
| *** tfukushima has joined #kolla | 13:24 | |
| *** diogogmt has joined #kolla | 13:28 | |
| sdake_ | harmw pretty sure samyaple is not, as he has had to wait on reviews for his submissions prior | 13:32 |
| sdake_ | but he is definately an expert in ansible ;) | 13:32 |
| sdake_ | jpeeler ping | 13:33 |
| pbourke | sdake_: morning | 13:39 |
| sdake_ | hey pbourke | 13:41 |
| sdake_ | so our build from oracle in the templates is busted | 13:41 |
| sdake_ | could you look into repairing it | 13:41 |
| pbourke | sdake_: sure | 13:41 |
| sdake_ | is there an epel like thing for oracle linux? | 13:41 |
| sdake_ | i'd like atleast from source to work for all our distro models | 13:41 |
| *** jtriley has joined #kolla | 13:42 | |
| sdake_ | build.py --template -t source -something oraclelinux | 13:42 |
| sdake_ | is how you build | 13:42 |
| pbourke | we're definitely missing some packages. folks are working internally to get them available from http://public-yum.oracle.com/ | 13:43 |
| pbourke | mostly they should build though, I'll go through and add /bin/false's for the missing ones | 13:43 |
| pbourke | that work for you? | 13:43 |
| harmw | yea sdake_, ansible is lacking a decent amount of cores afaik :p | 13:43 |
| harmw | pbourke: you had a question for me the other day? | 13:44 |
| pbourke | harmw: ah, no, just we were discussing kolla proxy issues | 13:44 |
| sdake_ | from source shouldn't ned a bin false | 13:44 |
| harmw | pbourke: I noticed a bug already | 13:44 |
| sdake_ | the issue with oraclelinux is the bae image | 13:44 |
| pbourke | harmw: and I knew you were one of the ones who is interested in that | 13:44 |
| harmw | in the vagrant setup | 13:44 |
| sdake_ | from rdo is another matter but would be nice if that works as well | 13:44 |
| harmw | hehe yea, well, not anymore :) I'm free from that sucky corporate proxy, thank god | 13:45 |
| sdake_ | that may require some epel-like repo for oraclelinux | 13:45 |
| pbourke | sdake_: hmm ok let see where we're at | 13:45 |
| pbourke | the rate of change is lightning at the moment so hard to stay on top of it all! | 13:46 |
| sdake_ | pbourke the goal is to add gates for all bin/source and distro combos - i'd like to that to happen by 31st | 13:46 |
| sdake_ | agree we are moving fast but we are all on the same page - so thats a plus :) | 13:46 |
| sdake_ | just to be clear we expect from source to work the same in all containers except base | 13:47 |
| sdake_ | which may need to install the from source build toolchain | 13:47 |
| sdake_ | from binary we expect either to use RDO or distro packaging if its available | 13:47 |
| *** diogogmt has quit IRC | 13:49 | |
| openstackgerrit | Merged stackforge/kolla: Removed config-internal check.sh for cinder https://review.openstack.org/210873 | 13:50 |
| sdake_ | akawasnie small changes neded on https://review.openstack.org/#/c/212442/ - if you get them in today we can get em merged :) | 13:58 |
| pbourke | I really wish we could get live output from the new build script | 14:01 |
| akwasnie | sdake, I will correct them in a while, thanks | 14:06 |
| *** jtriley_ has joined #kolla | 14:06 | |
| sdake_ | cool thanks akawsnie | 14:06 |
| sdake_ | pbourke harmw can you guys ack https://review.openstack.org/#/c/212347/ | 14:07 |
| sdake_ | its needed for the gate | 14:07 |
| sdake_ | I understand if you dont totally understand why - just take my word for it :) | 14:07 |
| *** jtriley has quit IRC | 14:09 | |
| *** devlaps has quit IRC | 14:24 | |
| akwasnie | sdake, I have one question about your second comment here https://review.openstack.org/#/c/212442/3/docker_templates/neutron/neutron-base/Dockerfile.j2 | 14:27 |
| akwasnie | which part from 'source' block won't work for ubuntu/debian? | 14:27 |
| vbel | sdake: build.py --template -t source -something oraclelinux does not work? :) Checking now | 14:28 |
| *** absubram has quit IRC | 14:29 | |
| *** jpeeler has joined #kolla | 14:29 | |
| sdake_ | rhallisey are you good to abaondon https://review.openstack.org/#/c/210417/ | 14:30 |
| sdake_ | akwasnie the part that it wont work in an ubuntu/debian environment. | 14:30 |
| sdake_ | vbel --something is actually a different parameter that i don't recall atm - use --help option to find it | 14:30 |
| *** dwalsh has joined #kolla | 14:31 | |
| sdake_ | rhallisey and this one https://review.openstack.org/#/c/210417/ | 14:33 |
| pbourke | vbel: tools/build.py --template --type source --base oraclelinux | 14:36 |
| vbel | ok thanks! | 14:36 |
| pbourke | vbel: its missing pkgs in base that we dont have in public-yum | 14:36 |
| pbourke | vbel: unsure of best way forward, only looked at it quickly | 14:36 |
| vbel | I see, ok, will look into it further | 14:37 |
| sdake_ | https://blueprints.launchpad.net/kolla/+spec/one-data-container | 14:39 |
| *** unicell has joined #kolla | 14:43 | |
| *** unicell1 has quit IRC | 14:43 | |
| *** tfukushima has quit IRC | 14:46 | |
| openstackgerrit | Alicja Kwasniewska proposed stackforge/kolla: Add Dockerfile template for neutron https://review.openstack.org/212442 | 14:48 |
| *** tfukushima has joined #kolla | 14:48 | |
| rhallisey | sdake_, left a comment there | 14:48 |
| sdake_ | rhallisey rockin thanks | 14:50 |
| sdake_ | i'm asking my docker pro friends | 14:50 |
| sdake_ | can you ask dwalsh the difference between VOLUME and -v /var/lib/mysql:/var/lib/mysql | 14:50 |
| sdake_ | or ask him better yet to comment on that blueprint | 14:50 |
| sdake_ | or some other appropriate expert | 14:51 |
| rhallisey | kk | 14:51 |
| sdake_ | if there is no difference then I think proceeding is solid | 14:51 |
| sdake_ | this works for the case of using compose with tripleo as well | 14:51 |
| sdake_ | since you can use the "name" operation to name the container | 14:51 |
| *** blahRus has joined #kolla | 14:52 | |
| rhallisey | yup agreed | 14:52 |
| * rhallisey walks downstairs | 14:52 | |
| *** achanda has joined #kolla | 14:53 | |
| *** diogogmt has joined #kolla | 14:54 | |
| sdake_ | pbourke you were in that thread on the data containers, please comment on the blueprint https://blueprints.launchpad.net/kolla/+spec/one-data-container | 14:56 |
| *** sdake_ is now known as sdake | 14:56 | |
| vbel | VOLUME and -v /var/lib/mysql:/var/lib/mysql: performance vise or location? | 14:57 |
| vbel | -wise | 14:57 |
| rhallisey | is whether using VOLUME and -v both do the same thing | 15:00 |
| rhallisey | which is bind mount | 15:00 |
| rhallisey | and if they do different things then what is the performance impact | 15:01 |
| sdake | VOLUME clearly is a bit diffeerent in that iit *always* bindmounts | 15:01 |
| sdake | rhallisey what did dwalsh say | 15:03 |
| rhallisey | hes at flock. I forgot he told me on Monday | 15:03 |
| *** achanda has quit IRC | 15:03 | |
| sdake | any other experts you can tap | 15:03 |
| sdake | perhaps you can ask appinfra-strategy list | 15:03 |
| sdake | and point to our blueprint for more information | 15:03 |
| sdake | its a shame that list isn't public | 15:04 |
| sdake | it should be | 15:04 |
| rhallisey | I'll ask | 15:04 |
| sdake | so much good info there | 15:04 |
| sdake | that is the one thing I miss about red hat :) | 15:04 |
| sdake | rhallisey thanks | 15:05 |
| sdake | if you can hunt down docker pros on the storage side too that would help | 15:05 |
| vbel | -v x:y is a folder from host, no need data container at all. We need to backup host folder instead of container what is not very Docker-like | 15:05 |
| sdake | vbel yes but VOLUME doesn't actually change the contents of the volume, it is also a folder on the host imo | 15:06 |
| sdake | rather change the contents of the image | 15:06 |
| sdake | the image is immutable | 15:06 |
| vbel | yes, just layer above | 15:06 |
| sdake | that is like the whole point of docker ;-) | 15:06 |
| sdake | yes but it can't change | 15:06 |
| sdake | the way it changes is by bindmounting a host volume | 15:06 |
| sdake | rhallisey send dwalsh an email he may respond from flocker as well | 15:07 |
| sdake | please and thank you :) | 15:07 |
| vbel | it's a new growing image based on that immutable image | 15:07 |
| sdake | no, images cannot grow in docker | 15:07 |
| sdake | it is a straight up bindmount | 15:07 |
| sdake | atleast that is my understanding | 15:07 |
| vbel | I mean container - we freeze it and save as image :) | 15:07 |
| sdake | so lets talk about the backup case | 15:08 |
| sdake | pbourke you may have interet here too | 15:08 |
| vbel | it's like binary vs process | 15:08 |
| sdake | how do you actually backup a data container | 15:08 |
| sdake | i know its not growing because we can run /bintrue and the data container persists | 15:08 |
| sdake | when used with the volumes_from command | 15:08 |
| sdake | it grows by bindmounting | 15:11 |
| sdake | the only difference I think in VOLUME is it always creates the bindmount | 15:11 |
| sdake | vs -v which is a bindmount create on demand | 15:11 |
| sdake | all the behaviors I have ever seen with -v vs VOLUME show that it is a bind mount | 15:12 |
| sdake | they are both that is | 15:12 |
| sdake | vbel thoughts on that? | 15:12 |
| sdake | our main goal is to be able to backup the data container, perhaps we can test if we can back up a -v specified bindmount data container | 15:12 |
| vbel | yes, agree, -v is just another folder on host, somewhere in /var/lib | 15:13 |
| vbel | sorry volume is just another folder on host, somewhere in /var/lib | 15:13 |
| vbel | so -v and volume are very similar | 15:13 |
| sdake | vbel you seem fast at testing POC ideas, any chance you could test if we can do backup with -v | 15:13 |
| pbourke | the data vol abstraction is useful though | 15:14 |
| sdake | we are keeping the data volume | 15:14 |
| sdake | we are just not keeping VOLUME | 15:14 |
| sdake | does that answer your concern pbourke? | 15:14 |
| sdake | we are making a data volume dynmaically vs statically | 15:14 |
| pbourke | sdake: not sure, can you summarise why? | 15:15 |
| sdake | why do this? | 15:15 |
| sdake | so we can get rid of the data containers that are essentially the same thing | 15:15 |
| sdake | with different VOLUME specifiers | 15:15 |
| sdake | the key is we need to be able to backup and restore the data container with -v | 15:15 |
| sdake | if we can do that, we can use one image | 15:15 |
| sdake | the reason we need multiple images now is we have different VOLUME specifiers in the data containers | 15:16 |
| sdake | btw, I puta link to tihs log in the discussion | 15:16 |
| vbel | backup of data contaners is done with a bit indirection - we do something like docker run --volumes-from dbdata -v $(pwd):/backup centos tar cvf /backup/backup.tar /var/lib/mysql | 15:16 |
| sdake | of the blueprint | 15:16 |
| rhallisey | man no takes on the subject | 15:17 |
| pbourke | -v is bind mount though? | 15:17 |
| sdake | rhallisey hunt the floor at rht, someone has to know :) | 15:17 |
| vbel | -v is telling docker to use backdoor as user specified | 15:17 |
| sdake | -v is a bind mount for sure | 15:17 |
| rhallisey | <agoldste> rhallisey: the dockerfile instruction is the same as running -v /some/path but it's not a bind mount. it just instructs docker not to record diffs to that path when generating layers | 15:17 |
| sdake | fantastic answer | 15:18 |
| sdake | ask him how confident he is in his answer | 15:18 |
| sdake | not to be insulting or anything :) | 15:18 |
| rhallisey | he seems pretty confident | 15:18 |
| rhallisey | lol | 15:18 |
| *** dwalsh has quit IRC | 15:18 | |
| rhallisey | maybe I can phrase this into a question | 15:19 |
| vbel | so if volume is in docker we save data inside and can export, right? | 15:19 |
| vbel | so if volume is in dockerfile we save data inside and can export, right? | 15:19 |
| sdake | vbel if it could be validated that would help :) | 15:20 |
| vbel | I will check | 15:20 |
| sdake | what i'd like to know is if we don't use VOLUME key, and just use -v bindmount | 15:20 |
| sdake | if we can bckupd ata container | 15:20 |
| sdake | if we can, that solves backkup case | 15:21 |
| sdake | since we can use same base image with different names, that provides different -v bindmount points | 15:21 |
| sdake | it would also be good to verify that | 15:21 |
| sdake | docker-dev is AFK ATM :) | 15:21 |
| sdake | agoldstein seems to think VOLUME just tells docker not to store changes in that path | 15:22 |
| sdake | rhallisey that is hte question to phrase to him | 15:22 |
| sdake | "so just to be clear, VOLUME just tells docker not to store changes to that path sepcified in VOLUME?" | 15:22 |
| sdake | store changes in later layers that is | 15:23 |
| *** dwalsh has joined #kolla | 15:24 | |
| sdake | rhallisey a followup question "if the dockerfile instruction is the same as running -v /some/path:/some/path, how its it not a bindmount? | 15:25 |
| sdake | rhallisey can you ask him to join our channel | 15:27 |
| sdake | so we dont have to play telephone | 15:27 |
| rhallisey | I asked him a bunch of things | 15:27 |
| rhallisey | I think he disappeared | 15:27 |
| sdake | yes but it not logged! | 15:27 |
| sdake | ok | 15:27 |
| sdake | i would disappear if someone griled me too :) | 15:27 |
| rhallisey | give it a moment | 15:27 |
| rhallisey | lol | 15:27 |
| rhallisey | I feel like I'm fishing right now lol | 15:28 |
| rhallisey | see he bites at the question | 15:28 |
| rhallisey | s/he/who | 15:28 |
| sdake | note upstream docs: https://docs.docker.com/userguide/dockervolumes/ | 15:29 |
| sdake | rereading for the 100th time | 15:29 |
| rhallisey | compare that to the link I put into the BP | 15:29 |
| rhallisey | We’re going to look at the two primary ways you can manage data in Docker. | 15:30 |
| rhallisey | Data volumes, and | 15:30 |
| rhallisey | Data volume containers. | 15:30 |
| rhallisey | ^ isn't that the answer? | 15:30 |
| rhallisey | that they differ | 15:31 |
| sdake | still reading | 15:31 |
| rhallisey | I'm going to get some food | 15:32 |
| rhallisey | maybe we will get a response back soon | 15:32 |
| sdake | I think the scenario we are after | 15:32 |
| sdake | is docker run -v /var/lib/mysql mysql_data | 15:32 |
| sdake | docker run volumes_from mysql_data mysql | 15:33 |
| rhallisey | interesting | 15:33 |
| sdake | syntax isn't totally correct but you should get the idea | 15:33 |
| rhallisey | I do | 15:33 |
| pbourke | I dont think that will carry the benefits of a data container | 15:33 |
| pbourke | but you guys have thought about this way more than I have, just my instinct really | 15:34 |
| rhallisey | so what's to stop the argument of: docker run -v /var/lib/mysql mysql_app /bin/sleep infinity | 15:34 |
| rhallisey | pbourke, I'm not yet convinced either, but doing -v outside for the datacontainer is something we arn't doing at the moment | 15:35 |
| sdake | rhallisey the problem with that is then the data container ends up with a bunch of stuff from mysql within it | 15:35 |
| sdake | i'm pretty sure that is what sam has implemented in the ansible data containers | 15:35 |
| sdake | sec let me find a reference for that | 15:35 |
| rhallisey | sdake, I agree with what your saying, but that's what sam's argument is | 15:35 |
| pbourke | bind mounts are poluting the host | 15:35 |
| rhallisey | that it doesn't matter | 15:35 |
| sdake | bind mounts happen with VOLUME operations | 15:36 |
| pbourke | yes but its abstracted under the docker-engine | 15:36 |
| sdake | pbourke had the concern about the extra data | 15:36 |
| sdake | just trying to make the various stakeholders satisified with teh solution :) | 15:36 |
| sdake | https://github.com/stackforge/kolla/blob/master/ansible/roles/glance/tasks/start.yml#L17 | 15:37 |
| sdake | this is the volumes-from operation | 15:37 |
| pbourke | I dont actually see the big deal with maintaining the data images? | 15:38 |
| pbourke | they're very small | 15:38 |
| pbourke | dont change often | 15:38 |
| rhallisey | pbourke, I don't either | 15:38 |
| rhallisey | I don't see them ever changing | 15:38 |
| sdake | https://github.com/stackforge/kolla/blob/master/ansible/roles/glance/tasks/bootstrap.yml#L45 | 15:38 |
| sdake | pbourke i have a general philosophy on engineering to remove anything not absolutely necessary to run the system | 15:39 |
| sdake | the reason I want to tackle this now is because people will start to standaardie on our models | 15:39 |
| sdake | rhallisey sam does just ass I explained above, but uses the various api iamges to create the data containers | 15:40 |
| pbourke | they are necessary though | 15:40 |
| sdake | which results in a bunch of extra crap in the bindmounted directory | 15:40 |
| pbourke | you're just trying to put them in a different form | 15:40 |
| sdake | multipel data containers running is necessary | 15:40 |
| sdake | ithink we can get away with one data container image however | 15:40 |
| rhallisey | I think at the very least we move away from using api images with mounts as out data container | 15:40 |
| pbourke | using the same logic you could say we should just have one dockerfile for everything | 15:41 |
| rhallisey | but, I feel like it's more work to make 1 container than leave it as is | 15:41 |
| rhallisey | since we already have a working set | 15:41 |
| sdake | we could have one dockerfile with the jinja templating but that owuld be overly complex imo :) | 15:41 |
| rhallisey | hehe | 15:41 |
| rhallisey | I noticed that | 15:41 |
| pbourke | yup | 15:41 |
| sdake | if we leave it as is, we will be using tthe api containers | 15:41 |
| rhallisey | sdake, that needs to change imo | 15:42 |
| sdake | this solution is actually simpler then hainga bunch of data containers | 15:42 |
| sdake | yes, I agree, I dont like it | 15:42 |
| pbourke | some of the -v incatations above are similarly complex imo ;) | 15:42 |
| vbel | surprisingly volumes of any kind cannot be exported/saved | 15:42 |
| sdake | that is why the blueprint says to use one data container | 15:42 |
| vbel | https://github.com/docker/docker/issues/7583 | 15:42 |
| sdake | there are instructions on backing up a data container in the link above vbel on the upstream docs | 15:43 |
| sdake | to implement this would require changing this line: | 15:44 |
| sdake | https://github.com/stackforge/kolla/blob/master/ansible/roles/glance/tasks/bootstrap.yml#L43 | 15:44 |
| sdake | to something like image: "{{ data_container_image_full }}" | 15:44 |
| sdake | and making that same change across all ansible bits | 15:45 |
| sdake | how about this, I'll prototype a solution | 15:45 |
| sdake | and we can judge if its too complex :) | 15:45 |
| vbel | agree :) | 15:45 |
| sdake | pbourke rhallisey? | 15:46 |
| pbourke | sounds good | 15:47 |
| rhallisey | sure | 15:47 |
| sdake | rhallisey you noted above what is to stop us from doing docke run /bin/sleep infinity | 15:47 |
| sdake | that is precisely what the ansible implementaiton doe for data containers, just using the api as a base | 15:48 |
| sdake | nice my gate job got a +2 | 15:48 |
| sdake | centos binary gating incoming | 15:49 |
| sdake | got a +2 from andreas, he is a pro reviewer, which means I didn't mess it up:) | 15:49 |
| sdake | so folks for the moment we are going to leave the compose directory in a state of disrepair | 15:51 |
| sdake | there was a patch to remove it but ryan says he needs it ported to config-external for trieplo | 15:52 |
| sdake | brb reconnecting to dev machine | 15:53 |
| *** tfukushima has quit IRC | 15:54 | |
| *** sdake_ has joined #kolla | 15:54 | |
| *** agoldste has joined #kolla | 15:54 | |
| agoldste | sdake: rhallisey: hi | 15:54 |
| *** tfukushima has joined #kolla | 15:55 | |
| sdake_ | hey agoldste | 15:55 |
| sdake_ | so I think we sorted out a course of action | 15:55 |
| sdake_ | you had indicated VOLUME just tells docker to a) do a bindmount to /var/lib/docker for a docker volume and b) not record further change to that directory in the docker image | 15:55 |
| agoldste | sdake: that's not what i said :) | 15:56 |
| agoldste | VOLUME is not a bind mount | 15:56 |
| agoldste | it just indicates don't record changes | 15:56 |
| agoldste | VOLUME /foo == -v /foo | 15:56 |
| agoldste | not -v /foo:/bar | 15:56 |
| agoldste | or whatever | 15:56 |
| sdake_ | ok so isn't -v a bindmount? | 15:56 |
| pbourke | aha | 15:56 |
| pbourke | I thought it was but it's not | 15:57 |
| agoldste | only if it's of the form /host/path:/container/path | 15:57 |
| sdake_ | it usese the kernel bindmount feature to bindmount to /var/lib/docker/whatever-docker-compes-up-with | 15:57 |
| pbourke | subtle | 15:57 |
| sdake_ | that is a host bind mount | 15:57 |
| *** bmace has quit IRC | 15:57 | |
| agoldste | yes, host bind mount | 15:57 |
| sdake_ | the -v /var/lib/mysql is still a bindmount right? | 15:57 |
| sdake_ | jut not a host bind mount | 15:58 |
| *** sdake has quit IRC | 15:58 | |
| sdake_ | it more like a docker bind mount | 15:58 |
| agoldste | i suppose | 15:58 |
| sdake_ | so here is the quetion we really ahve | 15:58 |
| agoldste | the volumes are stored in /var/lib/docker/vfs on the host, so yeah | 15:58 |
| rhallisey | <agoldste> rhallisey: VOLUME /foo is the same as "-v /foo", not "-v /foo:/foo" | 15:58 |
| rhallisey | <agoldste> rhallisey: "-v /foo" is not a bind mount, "-v /foo:/foo" is a bind mount | 15:58 |
| sdake_ | both are bind mounts, one is a host bind mount | 15:59 |
| sdake_ | you can't specify host bind mounts in a dockerfile afaik | 15:59 |
| agoldste | i don't think you can | 16:00 |
| sdake_ | agoldste so essentially what we are after is backup operation on the data container and using one data container image but started with different names and different -v directory operations | 16:00 |
| rhallisey | agoldste, oh I didn't see you here :p | 16:00 |
| vbel | in any case we cannot save/export volume, right? To back it up just hack into it unless we already have backdoor with -v x:y | 16:00 |
| pbourke | sdake_: I think your idea will work quite nicely then | 16:00 |
| rhallisey | just got back from getting food | 16:00 |
| rhallisey | catching up | 16:01 |
| agoldste | sdake_: i think an example would help me understand better, if you can share one | 16:01 |
| sdake_ | we only use host bind mounts for stuff that is mandatory like /var/run to communicate with systemd and /libmodule to load kernel modules | 16:01 |
| sdake_ | agoldste I dont have an example but can explain the problem | 16:01 |
| sdake_ | we have about 10 data containers, all in separate images, all with a separate VOLUME flag | 16:02 |
| sdake_ | we want to turn those into one data container with no VOLUME operation | 16:02 |
| sdake_ | but that canbe used 10 times in different ways so we dont have 10 data container images in our system | 16:02 |
| sdake_ | but one data container image | 16:02 |
| sdake_ | i'm ok with 10 data volumes - that is unavaoidable ;) | 16:03 |
| agoldste | how many actual runnable images do you have? | 16:03 |
| sdake_ | our 10 data containers atm are all separate dockerfiles | 16:03 |
| sdake_ | about 50 | 16:03 |
| sdake_ | (OpenStack in Docker deployed by Ansible) | 16:03 |
| agoldste | iirc you can start any container from any image you want, have it run something like "true", specify -v for volumes, and then use --volumes-from when starting the real containers | 16:04 |
| sdake_ | right | 16:04 |
| sdake_ | that is the proposal on the table | 16:04 |
| agoldste | i'm not sure you really need a special data container image unless your images don't have simple things like "true" in them that return 0 quickly | 16:04 |
| *** jasonsb has quit IRC | 16:05 | |
| sdake_ | we actually sleep infinity | 16:05 |
| sdake_ | for technical reasons related to ansible | 16:05 |
| sdake_ | I know we don't have to do this | 16:05 |
| agoldste | either works, sure | 16:05 |
| *** jasonsb has joined #kolla | 16:05 | |
| sdake_ | but we do because ansible detects a change if the container exits which causes all kinds of chaos | 16:05 |
| sdake_ | the reason we started with a whole slew of separate data continers is back when I was a noob I thought VOLUME was necessary to create a docker volume | 16:06 |
| sdake_ | didn't know -v was available | 16:07 |
| sdake_ | or we wouldn't even have this problem | 16:07 |
| agoldste | ah, ok | 16:07 |
| sdake_ | cool well it sounds like technically this will work and there are no existing objections | 16:07 |
| sdake_ | any objections? Could I get a vote from the people in this discussion? | 16:09 |
| *** pbourke has quit IRC | 16:09 | |
| rhallisey | ok so looks like a single data container | 16:09 |
| sdake_ | clealry its not set in stone, if this doens't work, we can revert | 16:09 |
| rhallisey | using a bunch of -v | 16:09 |
| *** pbourke has joined #kolla | 16:09 | |
| sdake_ | rhallisey i'll update the compse dir to match the new implementation | 16:09 |
| sdake_ | although compse is not working | 16:10 |
| *** jasonsb has quit IRC | 16:10 | |
| sdake_ | rhallisey just want to confirm you do intend to use compoe for tirpleo | 16:11 |
| sdake_ | because otherwise i'd just straight awa ydelete it ;) | 16:11 |
| rhallisey | sdake_, we are right now, but we changed because kolla was using compose | 16:11 |
| rhallisey | we were origionally using atomic | 16:11 |
| rhallisey | atomic tools | 16:11 |
| sdake_ | compose does serve as a good reference | 16:11 |
| sdake_ | the directory | 16:11 |
| sdake_ | we jsut have no wayt o test it | 16:12 |
| sdake_ | I dont mind shipping it in the repo if config-external rework is done | 16:12 |
| rhallisey | if we change to atomic, I'll make a new dir for atomic | 16:12 |
| rhallisey | and rmeove compose | 16:12 |
| sdake_ | wfm | 16:12 |
| sdake_ | although you probably have to deal with rhel for a long time | 16:12 |
| sdake_ | and atomic is vendor specific | 16:13 |
| sdake_ | as in tripleo wont work in non-atomic environments if we only ship atomic | 16:13 |
| rhallisey | ya I hear ya | 16:13 |
| rhallisey | part of the reason behind changing | 16:13 |
| rhallisey | so I think we will stick with compose but you never know | 16:14 |
| sdake_ | it may amke more sense to maintain compose dir in tripleo's repos somewhere | 16:14 |
| sdake_ | but the leaset we can do is get it into a state where we think its correct | 16:14 |
| sdake_ | ok woring on an implementation now | 16:15 |
| sdake_ | bbifew hours | 16:15 |
| sdake_ | btw for thoe that cae, the gate machines are xeon 8 core cpus | 16:16 |
| sdake_ | but only 8g ig ram | 16:16 |
| sdake_ | rhallisey ifI I could get an easy review of https://review.openstack.org/#/c/212347/ would be appreciated | 16:17 |
| sdake_ | the gate is about to merge | 16:18 |
| sdake_ | and it will not work without this change | 16:18 |
| sdake_ | jpeeler ping | 16:19 |
| sdake_ | jpeeler i need some help on changing around ttest_build.py to take arguments | 16:20 |
| sdake_ | specifically distro and install type | 16:20 |
| rhallisey | sdake_, why should it be maintained elsewhere though? I think it's more relevant here | 16:22 |
| rhallisey | especially if other tools come along and want to use it | 16:22 |
| sdake_ | possibly - we do define the container content | 16:23 |
| sdake_ | ya your right | 16:23 |
| sdake_ | we define the container abi, we need to maintian in our repo | 16:23 |
| *** agoldste has left #kolla | 16:26 | |
| sdake_ | brb need to get into meail | 16:29 |
| *** diogogmt has quit IRC | 16:31 | |
| *** openstackgerrit has quit IRC | 16:31 | |
| *** unicell has quit IRC | 16:31 | |
| *** openstackgerrit has joined #kolla | 16:31 | |
| *** diogogmt has joined #kolla | 16:31 | |
| *** athomas has quit IRC | 16:39 | |
| *** sdake has joined #kolla | 16:42 | |
| *** dwalsh has quit IRC | 16:42 | |
| *** athomas has joined #kolla | 16:43 | |
| *** sdake_ has quit IRC | 16:45 | |
| *** unicell has joined #kolla | 16:59 | |
| *** akwasnie has quit IRC | 17:03 | |
| *** tfukushima has quit IRC | 17:06 | |
| *** gfidente has quit IRC | 17:25 | |
| *** achanda has joined #kolla | 17:29 | |
| *** sdake_ has joined #kolla | 17:29 | |
| *** absubram has joined #kolla | 17:31 | |
| *** sdake has quit IRC | 17:32 | |
| *** dwalsh has joined #kolla | 17:36 | |
| *** gfidente has joined #kolla | 17:37 | |
| *** gfidente has joined #kolla | 17:37 | |
| jpeeler | sdake_: where exactly do you want the arguments to go? i assume you mean with a new tox target | 18:00 |
| sdake_ | ya tox should pass arguments into testr if possible | 18:03 |
| sdake_ | i'm good with hard code of the testr arguments | 18:03 |
| sdake_ | I just dont know how to do it- if you could sort it out we can get the parallel build gate rolling for all distros | 18:03 |
| sdake_ | both our reiews have a +2 | 18:03 |
| *** athomas has quit IRC | 18:05 | |
| sdake_ | samyaple you about | 18:07 |
| jpeeler | passing arguments from tox is difficult because testr was passing stuff internal stuff to the test_build file, which is why i overrode argv | 18:08 |
| *** echoingumesh has joined #kolla | 18:09 | |
| *** echoingumesh has quit IRC | 18:09 | |
| jpeeler | sdake_: where do you want the arguments? tox -e buildimages <arg here> or what? | 18:10 |
| *** echoingumesh has joined #kolla | 18:10 | |
| sdake_ | nah inside the tox command | 18:10 |
| *** dwalsh has quit IRC | 18:10 | |
| sdake_ | so when you run testr hardcode the arguments | 18:10 |
| sdake_ | we can refactor later to what you suggest | 18:11 |
| sdake_ | i don't want to alter the infra review since it has +2's | 18:11 |
| *** unicell has quit IRC | 18:11 | |
| *** unicell has joined #kolla | 18:11 | |
| jpeeler | sdake_: still unsure about where you're talking about. tox calls testr, so what hardcoding are you referring to? | 18:12 |
| sdake_ | moment | 18:12 |
| jpeeler | if you just want to do some informal testing, you can just add arguments to the build_args variable | 18:13 |
| *** SourabhP has joined #kolla | 18:13 | |
| sdake_ | sudo -g docker testr run ^(test_build).* | 18:13 |
| sdake_ | should be sudo -g docker testr run ^(test_build).* centos source | 18:13 |
| *** echoingumesh has quit IRC | 18:14 | |
| jpeeler | i see, well i've never done it before either. and the testr documentation is very light - http://testrepository.readthedocs.org/en/latest/MANUAL.html | 18:17 |
| sdake_ | jpeeler well i could figure it out or you could but i'm busy atm so would appreciate if you could :) | 18:19 |
| jpeeler | i'm looking at it, but i feel like i've tried this before | 18:19 |
| sdake_ | try asking around to some python wizard at rht | 18:20 |
| *** bmace has joined #kolla | 18:25 | |
| *** gfidente has quit IRC | 18:27 | |
| *** echoingumesh has joined #kolla | 18:32 | |
| *** dwalsh has joined #kolla | 18:35 | |
| *** jasonsb has joined #kolla | 18:52 | |
| *** akscram has quit IRC | 18:56 | |
| *** akscram has joined #kolla | 18:58 | |
| *** SourabhP has quit IRC | 19:01 | |
| *** SourabhP has joined #kolla | 19:02 | |
| *** sdake has joined #kolla | 19:09 | |
| openstackgerrit | imain proposed stackforge/kolla: Fix heat configuration. https://review.openstack.org/212706 | 19:12 |
| *** SourabhP has quit IRC | 19:12 | |
| *** sdake_ has quit IRC | 19:12 | |
| openstackgerrit | Ryan Hallisey proposed stackforge/kolla: Fix heat configuration. https://review.openstack.org/212706 | 19:14 |
| *** SourabhP has joined #kolla | 19:18 | |
| *** dwalsh has quit IRC | 19:20 | |
| openstackgerrit | Ryan Hallisey proposed stackforge/kolla: Glance install from source incorrectly places config https://review.openstack.org/212711 | 19:23 |
| *** SourabhP has quit IRC | 19:24 | |
| openstackgerrit | Ryan Hallisey proposed stackforge/kolla: Nova install from source incorrectly places config https://review.openstack.org/212712 | 19:28 |
| *** hoonetorg_ has joined #kolla | 19:28 | |
| *** dwalsh has joined #kolla | 19:30 | |
| *** inc0 has joined #kolla | 19:34 | |
| inc0 | evening | 19:34 |
| *** dwalsh has quit IRC | 19:37 | |
| sdake | hi inc0 | 19:40 |
| *** SourabhP has joined #kolla | 19:40 | |
| openstackgerrit | Ryan Hallisey proposed stackforge/kolla: Cinder install from source incorrectly places config https://review.openstack.org/212718 | 19:42 |
| openstackgerrit | Ryan Hallisey proposed stackforge/kolla: Keystone install from source incorrectly places config https://review.openstack.org/212721 | 19:51 |
| *** achanda has quit IRC | 19:52 | |
| openstackgerrit | Ryan Hallisey proposed stackforge/kolla: Barbican install from source incorrectly places config https://review.openstack.org/212724 | 19:54 |
| openstackgerrit | Merged stackforge/kolla: Change the tox targtes to match naming https://review.openstack.org/212347 | 19:55 |
| openstackgerrit | Merged stackforge/kolla: Cinder-backup's config-external is missing #!/bin/bash https://review.openstack.org/212503 | 19:55 |
| openstackgerrit | Ryan Hallisey proposed stackforge/kolla: Ceilometer install from source incorrectly places config https://review.openstack.org/212727 | 19:57 |
| *** inc0 has quit IRC | 19:59 | |
| openstackgerrit | Ryan Hallisey proposed stackforge/kolla: Designate install from source incorrectly places config https://review.openstack.org/212729 | 20:00 |
| sdake | yay harmw is reviewing ;) | 20:02 |
| openstackgerrit | Ryan Hallisey proposed stackforge/kolla: Gnocchi install from source incorrectly places config https://review.openstack.org/212730 | 20:03 |
| harmw | I honestly don't like the fact that I'm so incredibly non-productive lately | 20:03 |
| harmw | so yes, reviewing some quick ones | 20:04 |
| sdake | rhallisey | 20:08 |
| sdake | I think all those patches are wrong | 20:08 |
| sdake | what should happen is the cp to /etc should be removed | 20:08 |
| sdake | because config-external will create the correct config files | 20:08 |
| rhallisey | sdake, let me check into that | 20:08 |
| rhallisey | using config external with the existing doesn't work you get /etc/heat/heat/heat.conf | 20:09 |
| rhallisey | I'm open to fixing it a diff way | 20:09 |
| sdake | you may git that directory | 20:09 |
| sdake | but that config file isn't used | 20:09 |
| sdake | the daemon uses /etc/heat/heat-containersubname.conf | 20:09 |
| sdake | I was going to make a patch set that removed /etc copies for source | 20:10 |
| rhallisey | ok | 20:10 |
| sdake | part of blueprint remove-external-config | 20:10 |
| rhallisey | I can just change these | 20:10 |
| rhallisey | to do that | 20:10 |
| sdake | but since you already have a patch set | 20:10 |
| sdake | perhaps youcan | 20:10 |
| rhallisey | ya I'll just alter them | 20:10 |
| sdake | and close all those bugs as invalid and tke them out of liberty/liberty-3 | 20:11 |
| sdake | I agree your solution is correct | 20:11 |
| sdake | but we dont need to copy the config files | 20:11 |
| rhallisey | ya that makes sense | 20:11 |
| rhallisey | fixing | 20:11 |
| sdake | well we may need to if there is policy.json actually | 20:11 |
| rhallisey | should I reference the BP | 20:12 |
| rhallisey | instead of those biug | 20:12 |
| sdake | let me think a moment | 20:12 |
| sdake | i dont have a deployment handy, mine ni borked because i am doing this data container thing | 20:12 |
| rhallisey | kk | 20:12 |
| sdake | can you look at some dirs in nova and etc | 20:12 |
| rhallisey | I'll just leave as is | 20:12 |
| rhallisey | I gota leave in like 3min | 20:12 |
| sdake | and see if there are extra files | 20:13 |
| rhallisey | kk | 20:13 |
| rhallisey | I'll do that real quick | 20:13 |
| sdake | and if so ill ack em | 20:13 |
| Slower | sdake: so are just going to be bind mounting eg /etc/heat then? | 20:13 |
| sdake | we bindmount /opt/kolla/configs/heat/heat.conf:ro | 20:14 |
| sdake | and copy it in in config-external.sh | 20:14 |
| rhallisey | ya /etc/nova has more than just nova.conf | 20:14 |
| rhallisey | I think we need to cp all | 20:14 |
| sdake | ok, but the bug infomratino is wrong ;-) | 20:14 |
| sdake | you should update each one with a link to this irc log | 20:15 |
| sdake | click on right side where the timestamp is | 20:15 |
| *** britthou_ has joined #kolla | 20:15 | |
| sdake | an cut and paste the discussion | 20:15 |
| sdake | the link | 20:15 |
| sdake | of the discussion | 20:15 |
| sdake | so you don't end up with -1s in your reviews | 20:15 |
| rhallisey | sure | 20:16 |
| rhallisey | I can just change the commit messages to be more clear | 20:16 |
| openstackgerrit | Ryan Hallisey proposed stackforge/kolla: Magnum install from source incorrectly places config https://review.openstack.org/212740 | 20:17 |
| *** britthouser has quit IRC | 20:18 | |
| sdake | if your looking for more easy work the docker-templates file also has these problems | 20:18 |
| openstackgerrit | Merged stackforge/kolla: Add Dockerfile template for Cinder https://review.openstack.org/211919 | 20:19 |
| openstackgerrit | Merged stackforge/kolla: Remove bootstrap.yml abstraction for haproxy https://review.openstack.org/212328 | 20:19 |
| openstackgerrit | Merged stackforge/kolla: Remove bootstrap.yml abstraction for mariadb https://review.openstack.org/212352 | 20:19 |
| openstackgerrit | Merged stackforge/kolla: Implement Murano source images https://review.openstack.org/209905 | 20:19 |
| Slower | sdake: ah ok I misunderstood | 20:19 |
| openstackgerrit | Merged stackforge/kolla: Update mariadb confs and ports https://review.openstack.org/210980 | 20:19 |
| *** britthou_ has quit IRC | 20:19 | |
| sdake | do docker_templates as one patch plz | 20:20 |
| sdake | this clicking sucks | 20:20 |
| openstackgerrit | Ryan Hallisey proposed stackforge/kolla: Neutron install from source incorrectly places config https://review.openstack.org/212750 | 20:20 |
| sdake | needs rebase: https://review.openstack.org/#/c/212157/ | 20:21 |
| openstackgerrit | Ryan Hallisey proposed stackforge/kolla: Murano install from source incorrectly places config https://review.openstack.org/212771 | 20:23 |
| openstackgerrit | Ryan Hallisey proposed stackforge/kolla: Swift install from source incorrectly places config https://review.openstack.org/212775 | 20:24 |
| sdake | harmw a whole lot of rubber stamping awaits | 20:25 |
| harmw | for real.. | 20:26 |
| openstackgerrit | Ryan Hallisey proposed stackforge/kolla: All the Cinder services need to reference the correct config https://review.openstack.org/212157 | 20:26 |
| rhallisey | sdake, k I gota go | 20:27 |
| rhallisey | I'll sort out the bug tagging tmr | 20:27 |
| *** rhallisey has quit IRC | 20:27 | |
| harmw | why are those bugs pointing to different bugs, and not just some BP | 20:27 |
| harmw | I take it that was discussed? | 20:28 |
| harmw | or are they just that, bugs :) | 20:28 |
| sdake | it really could have just been 1 commit imo :) | 20:29 |
| harmw | yep, that to | 20:29 |
| *** jasonsb has quit IRC | 20:29 | |
| *** jasonsb has joined #kolla | 20:29 | |
| *** sdake_ has joined #kolla | 20:32 | |
| *** achanda has joined #kolla | 20:33 | |
| *** jasonsb has quit IRC | 20:34 | |
| jpeeler | sdake: i see no way to do what you're wanting. since testr is designed to run multiple tests in parallel, i suspect passing arguments to each test isn't supported | 20:35 |
| *** sdake has quit IRC | 20:36 | |
| jpeeler | i bet he didn't see that. | 20:36 |
| sdake_ | i did see it | 20:38 |
| jpeeler | oh good | 20:38 |
| sdake_ | another optoin is to make a superclass of your current test and make a subclass for each distro option | 20:38 |
| jpeeler | yes, though are you talking about using the same tox target? | 20:39 |
| sdake_ | have the subclass feed the distro/type hardcoded int othe super | 20:39 |
| sdake_ | no instead of build_images it would be build_images_centos_binary | 20:39 |
| sdake_ | that would be your new python test name | 20:39 |
| sdake_ | each distro+os type would be a 5 line python function which supered out to the superclass | 20:40 |
| sdake_ | not same tox target | 20:41 |
| sdake_ | i dont care if there are 10 tox targets in the tox.ini file | 20:41 |
| sdake_ | there are already 3 for budling | 20:41 |
| sdake_ | tox -e build_images_centos_binary | 20:42 |
| sdake_ | would run testr (build_images_centos_binary) | 20:42 |
| sdake_ | which is a test in your build testing script | 20:42 |
| jpeeler | so do you need me to do that work? | 20:46 |
| jpeeler | almost done, just need to test | 21:09 |
| *** jtriley_ has quit IRC | 21:16 | |
| sdake_ | cool | 21:25 |
| openstackgerrit | Steven Dake proposed stackforge/kolla: WIP - Use one data container distro-type-data for all data containers https://review.openstack.org/212804 | 21:42 |
| openstackgerrit | Steven Dake proposed stackforge/kolla: Use one data container distro-type-data for all data containers https://review.openstack.org/212804 | 21:45 |
| openstackgerrit | Steven Dake proposed stackforge/kolla: Use one data container distro-type-data for all data containers https://review.openstack.org/212804 | 21:47 |
| openstackgerrit | Steven Dake proposed stackforge/kolla: Use one data container distro-type-data for all data containers https://review.openstack.org/212804 | 21:58 |
| sdake_ | anyone seen this before? | 21:59 |
| 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 buffer | 21:59 |
| sdake_ | FATAL: all hosts have already failed -- aborting | 21:59 |
| sdake_ | samyaple when you come alive if you could look ^^ | 21:59 |
| sdake_ | ccrouch you around | 22:01 |
| *** sdake has joined #kolla | 22:12 | |
| jpeeler | sdake_: is there a blueprint for multiple distro support for image building? | 22:12 |
| sdake | there is a blueprint for ubuntu | 22:13 |
| sdake | and one for rhel | 22:13 |
| sdake | beyond that no | 22:13 |
| jpeeler | can i just put this under the build-script blueprint? | 22:13 |
| sdake | wfm | 22:13 |
| sdake | or jut leave it unmarked | 22:13 |
| sdake | its gating | 22:13 |
| sdake | there is a gating blueprint actually | 22:13 |
| sdake | use that | 22:13 |
| *** sdake_ has quit IRC | 22:16 | |
| *** dolpher has joined #kolla | 22:16 | |
| openstackgerrit | Jeff Peeler proposed stackforge/kolla: Use inheritance for overriding of builder args https://review.openstack.org/212814 | 22:19 |
| jpeeler | sdake: hopefully that's exactly what you were looking for | 22:19 |
| sdake | nice work jpeeler! | 22:21 |
| *** dolpher has quit IRC | 22:21 | |
| sdake | can you add a tox.ini as a dependent patch tht builds both centos for srcbinary | 22:23 |
| sdake | shuld only take a couple mins | 22:23 |
| sdake | two tox targets | 22:23 |
| sdake | build-all-docker-imagesgoing byebye | 22:24 |
| sdake | yay ;) | 22:24 |
| jpeeler | very close! | 22:24 |
| sdake | our gate patches will likely merge tonight | 22:24 |
| sdake | well i don't know if ansible is busted or my data container patch is busted | 22:28 |
| sdake | my data container patch is very straightforward | 22:28 |
| openstackgerrit | Jeff Peeler proposed stackforge/kolla: Change tox targets to use new build.py https://review.openstack.org/212821 | 22:29 |
| jpeeler | ok that should be good to go now. | 22:29 |
| jpeeler | see everybody tomorrow | 22:30 |
| sdake | thanks jeff! | 22:30 |
| sdake | now samyaple can do dind gating it sounds like | 22:30 |
| *** sdake_ has joined #kolla | 22:42 | |
| *** absubram has quit IRC | 22:43 | |
| *** sdake has quit IRC | 22:46 | |
| openstackgerrit | Steven Dake proposed stackforge/kolla: Use one data container distro-type-data for all data containers https://review.openstack.org/212804 | 23:03 |
| openstackgerrit | Steven Dake proposed stackforge/kolla: Remove data containers and adjust compose files https://review.openstack.org/212830 | 23:03 |
| *** blahRus has quit IRC | 23:23 | |
| *** SourabhP has quit IRC | 23:44 | |
| *** dwalsh has joined #kolla | 23:51 | |
| *** diogogmt has quit IRC | 23:51 | |
| *** jtriley has joined #kolla | 23:53 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!