| *** openstack has joined #openstack-loci | 02:47 | |
| *** ChanServ sets mode: +o openstack | 02:47 | |
| *** spsurya_ has quit IRC | 04:48 | |
| *** spsurya_ has joined #openstack-loci | 04:49 | |
| *** e0ne has joined #openstack-loci | 05:06 | |
| *** spsurya_ is now known as spsurya | 05:07 | |
| *** e0ne has quit IRC | 06:00 | |
| evrardjp | hrw: thanks | 06:40 |
|---|---|---|
| evrardjp | SamYaple: I think bringing new profiles like "healthcheck" or "profiler" are logical (because supported by many project api-paste.ini) but I have the impression if that continues you'll have thousands of profiles. | 06:42 |
| *** ispp has joined #openstack-loci | 07:06 | |
| *** ispp has quit IRC | 07:40 | |
| *** ispp has joined #openstack-loci | 08:05 | |
| *** e0ne has joined #openstack-loci | 08:17 | |
| *** pbourke has quit IRC | 08:39 | |
| *** pbourke has joined #openstack-loci | 08:41 | |
| *** ispp has quit IRC | 09:10 | |
| evrardjp | can someone explain me why is there a pydep and a bindep ? | 09:27 |
| pbourke | evrardjp: I would assume they are optional extras that are available in python format | 09:50 |
| pbourke | im not opposed to various profiles either (Im adding a kolla one :)), though I think we'll hit the limitations of the bindep format as you say | 09:51 |
| evrardjp | still now sure how this doesn't fit profiles | 09:51 |
| evrardjp | yeah. | 09:51 |
| pbourke | maybe eventually we can have various bindep files | 09:51 |
| evrardjp | I am not against that -- I just wanted to understand why in this case pydep.txt exists | 09:52 |
| evrardjp | It would probably be better if upstream services have their own _well managed_ bindep.txt files, which could lead to a simplification of loci's bindep and use loci's bindep for its own purposes | 09:53 |
| evrardjp | that would clean the profiles like cinder keystone | 09:53 |
| pbourke | yeah that would be great | 09:53 |
| pbourke | im finding it a massive trial and error process to figure out what packages each service needs | 09:53 |
| evrardjp | yeah. | 09:54 |
| evrardjp | I am starting to do this for a new distro, so imagine the process :p | 09:54 |
| pbourke | also a service could end up removing the need for a package without us knowing | 09:54 |
| evrardjp | (ofc bindep fails me there on top of it :p ) | 09:54 |
| evrardjp | haha quite a ride :) | 09:54 |
| evrardjp | how are things otherwise pbourke ? | 09:55 |
| evrardjp | so you said you're adding kolla profile -- is that well received here? | 09:55 |
| pbourke | well, SamYaple suggested it, so I assume so :) | 09:55 |
| evrardjp | IMO it should be alright, and we could add some for helm if needed, preventing the code in helm | 09:55 |
| evrardjp | pbourke: for me it makes sense | 09:56 |
| pbourke | Im just a little unclear on what loci wants to gate/publish | 09:56 |
| evrardjp | it's a kind of profile. | 09:56 |
| pbourke | like do we want to gate each profile | 09:56 |
| pbourke | or just the core images | 09:56 |
| evrardjp | pbourke: you can probably gate it in kolla? | 09:56 |
| evrardjp | as this is the same namespace anyway | 09:56 |
| pbourke | i could but then it sort of seems the profile should live in kolla | 09:56 |
| evrardjp | yeah | 09:56 |
| evrardjp | I mean projects using a specific profile could have a job calling those. | 09:57 |
| evrardjp | with the profile | 09:57 |
| evrardjp | you could also add your extra packages if need be | 09:57 |
| evrardjp | well | 09:58 |
| evrardjp | that's my opinion, but who am I right... I just used this project for the first time a few days ago. | 09:58 |
| evrardjp | : p | 09:58 |
| pbourke | evrardjp: whats your process for figuring out what packages are needed for what service? | 09:59 |
| evrardjp | I am lazy, and I am using openstack-ansible patterns | 10:03 |
| evrardjp | but right now I have other things to figure out first -- some bindep quirks | 10:03 |
| evrardjp | pbourke: why do you ask, you have things missing in your images? | 10:04 |
| pbourke | well, I could just copy whats in kolla but im not convinced everything there is needed | 10:12 |
| pbourke | so im taking a real trial and error approach - deploy the container, see what fails, install that package, rinse repeat | 10:12 |
| pbourke | just wondering if there's a better way | 10:12 |
| *** ispp has joined #openstack-loci | 10:24 | |
| evrardjp | pbourke: haha I see you're going the same path as me: https://review.openstack.org/#/c/536355/4 | 10:28 |
| evrardjp | :p | 10:28 |
| evrardjp | pbourke: yeah doing the same for having the minimum packages :p | 10:28 |
| evrardjp | pbourke: don't forget to list them in cleanup ;) | 10:29 |
| pbourke | well, anything I install on top of loci is needed and so cant be cleaned up | 10:29 |
| evrardjp | k | 10:30 |
| evrardjp | pbourke: could you vote on https://review.openstack.org/#/c/584751/ ? | 10:39 |
| evrardjp | thanks | 10:41 |
| *** openstackgerrit has joined #openstack-loci | 10:52 | |
| openstackgerrit | Merged openstack/loci master: Make all the jobs voting by default https://review.openstack.org/584751 | 10:52 |
| *** ispp has quit IRC | 11:48 | |
| *** ispp has joined #openstack-loci | 11:48 | |
| *** openstackgerrit has quit IRC | 12:36 | |
| hrw | pbourke: can you review my patches? https://review.openstack.org/#/c/585408/2 https://review.openstack.org/#/c/583727/ https://review.openstack.org/#/c/583726/ ones | 13:12 |
| *** ispp has quit IRC | 13:21 | |
| pbourke | hrw: looks good | 13:26 |
| pbourke | hrw: just question on the third one | 13:26 |
| hrw | replied | 13:28 |
| pbourke | thanks | 13:29 |
| pbourke | these particular pkgs may not make a big difference | 13:29 |
| hrw | SamYaple: can you review my patches? https://review.openstack.org/#/c/585408/2 https://review.openstack.org/#/c/583727/ https://review.openstack.org/#/c/583726/ ones | 13:29 |
| pbourke | but from my experience with the kolla packages | 13:29 |
| pbourke | there are quite a few differences | 13:29 |
| hrw | pbourke: once we get to the point where it will matter we can handle it | 13:29 |
| pbourke | ok :) | 13:29 |
| hrw | pbourke: many of issues from kolla are handled by making own loci-base image | 13:29 |
| hrw | loci does not have to handle repos, keys, extra packages etc as they land in loci-base image first | 13:30 |
| hrw | pbourke: https://git.linaro.org/ci/job/configs.git/plain/ldcg-loci/Dockerfile | 13:30 |
| pbourke | nice | 13:30 |
| hrw | pbourke: https://git.linaro.org/ci/job/configs.git/plain/ldcg-loci/build.sh is whole build script | 13:31 |
| pbourke | hrw: will the loci cleanup process not purge some of the packages you're installing there? | 13:31 |
| pbourke | mainly git and virtualenv | 13:31 |
| hrw | pbourke: at that moment I do not care is my final image 5MB bigger or smaller | 13:32 |
| SamYaple | heh fair enough | 13:32 |
| hrw | haerwu/loci-nova master-debian-erp cd7003ad189d 46 hours ago 496MB | 13:33 |
| hrw | 14:32 (79718s) linaro@c1n1:scripts$ docker run -it -u root cd7003ad189d bash | 13:33 |
| pbourke | childen in africa could use those 5MB | 13:33 |
| hrw | root@508a95cd2c79:/# dpkg -l git | 13:33 |
| hrw | dpkg-query: no packages found matching git | 13:33 |
| hrw | pbourke: install part in loci-base dockerfile is taken exactly from loci/scripts/install.sh | 13:34 |
| hrw | thanks for reviews | 13:35 |
| *** ispp has joined #openstack-loci | 13:35 | |
| SamYaple | hrw: i think the point was that the git and virtualenv packages are not *available* in the final image | 13:37 |
| SamYaple | even though you are installing them there | 13:37 |
| SamYaple | more of a "just so you know" than a problem | 13:38 |
| hrw | SamYaple: I install them in loci-base just because loci installs them during build phase on each image | 13:38 |
| hrw | SamYaple: kind of 'how to cut few seconds' | 13:38 |
| SamYaple | yep i get that makes sense to me | 13:38 |
| hrw | install.sh is called per image. does "apt update/apt upgrade/apt install few-packages". so why wait ;D | 13:40 |
| ispp | SamYaple, are you around? | 13:42 |
| *** ispp has quit IRC | 13:58 | |
| *** ispp has joined #openstack-loci | 13:58 | |
| SamYaple | ispp: im always around | 13:59 |
| hrw | unless you sleep | 13:59 |
| pbourke | he doesn't | 14:01 |
| SamYaple | everytime i blink, im actually sleeping | 14:06 |
| hrw | lucky bastard | 14:06 |
| SamYaple | sometimes i blink for a full 8 hours | 14:06 |
| pbourke | lol | 14:07 |
| hrw | blog post about loci almost done. waiting for zuul now, then CI jobs and then final touch and publish | 14:15 |
| hrw | all merged! YAY! :D | 14:18 |
| SamYaple | +++ | 14:19 |
| hrw | 2:33 was previous build time. now without scipy it should be much faster | 14:19 |
| SamYaple | cassandra is the real beast. it was causing the gate to hit 90m | 14:20 |
| SamYaple | without the parallel build setting that is | 14:20 |
| hrw | https://marcin.juszkiewicz.com.pl/2018/07/26/loci-other-way-of-building-openstack-images/ | 14:23 |
| hrw | what do you think? | 14:25 |
| SamYaple | i think it sums up whats happened quite well | 14:29 |
| SamYaple | no-aarch64 support -> aarch64 support in a few days, thats not bad | 14:29 |
| pbourke | i look forward to the next installment | 14:30 |
| SamYaple | aarch128 | 14:30 |
| pbourke | hrw: if you could blog some figures categorising build time and image size vs kolla images that would be very interesting | 14:30 |
| SamYaple | image size is misleading | 14:31 |
| SamYaple | kolla would be much larger... on paper | 14:31 |
| hrw | pbourke: 200+ images contra 10? | 14:31 |
| SamYaple | but thats not entirely accurate. i would point to the ability to build services *independantly* or rebuild images as long as you have the same requirements base | 14:31 |
| SamYaple | yea its a bit apples/oranges to point to only size. but a keystone->keystone container comparison would be cool | 14:32 |
| hrw | pbourke: I consider loci to be 'really minimal approach to OpenStack in containers' contra 'provide everything to run OpenStack in containers' | 14:32 |
| hrw | looks like loci requires external mariadb (as it is not in bindep) for example | 14:34 |
| pbourke | hrw: loci can do infra containers | 14:34 |
| hrw | no swift, no monitoring | 14:34 |
| pbourke | https://review.openstack.org/#/c/583584 | 14:34 |
| hrw | pbourke: I looked at origin/master | 14:36 |
| SamYaple | hrw: we don't have a complete handle on how we want to do "infra" containers. I, for example, used upstream dockerhub images for the infra services | 14:41 |
| SamYaple | the openstack services are pretty generic, the infra services are far from it. so we arent sure we want that code in our code base when others are doing that work upstream | 14:41 |
| hrw | sure | 14:42 |
| hrw | kolla (on the other hand) provides me with ootb solution. I grab machines, install OS on them and then run kolla-ansible to get cloud running | 14:43 |
| SamYaple | kolla provides a deploy tool. we do not | 14:43 |
| hrw | noticed ; | 14:43 |
| hrw | :D | 14:43 |
| SamYaple | i can't control what a deploy tool will and won't support | 14:43 |
| *** gema has quit IRC | 14:44 | |
| hrw | base+requirements in 24 minutes. when scipy itself was over 40 minutes before... | 14:45 |
| SamYaple | ouch! | 14:45 |
| SamYaple | 24 minutes seems correct | 14:45 |
| SamYaple | 20m requirements builds i would call normal | 14:45 |
| hrw | SamYaple: 96 cpu cores help a lot | 14:46 |
| SamYaple | hahaha | 14:46 |
| hrw | SamYaple: seriously | 14:46 |
| hrw | our CI machine is dual socker thunderx (48 cores cpu) | 14:47 |
| SamYaple | i beleived you | 14:48 |
| *** ispp has quit IRC | 15:36 | |
| hrw | 01:19:20.600 Finished: SUCCESS | 15:38 |
| hrw | base, requirements and set of images | 15:38 |
| SamYaple | nice | 15:40 |
| hrw | 1:19 always beats 2:33 ;D | 15:40 |
| SamYaple | now if you rarely update base, and periodically update requirements then your images should be able to be rebuilt individually | 15:40 |
| hrw | base is same so goes from docker cache in notime | 15:41 |
| SamYaple | right thats my point | 15:41 |
| hrw | and the whole point of having it as CI job is to not rebuild them individually | 15:42 |
| hrw | just grab a set and use. run a job if you need updates | 15:42 |
| SamYaple | well thats up for debate there | 15:42 |
| hrw | so I can tell testing group 'use loci images build-6 set' | 15:42 |
| SamYaple | yea use cases differ | 15:43 |
| hrw | like I do with kolla: use rocky-54 set for your deployments | 15:43 |
| hrw | you know, I am build engineer not deploy manager | 15:44 |
| SamYaple | ya, im a bit of everything :P smaller team | 15:45 |
| hrw | :D | 15:50 |
| hrw | not that our is huge ;D | 15:51 |
| *** e0ne has quit IRC | 16:02 | |
| *** openstackgerrit has joined #openstack-loci | 16:23 | |
| openstackgerrit | Paul Bourke (pbourke) proposed openstack/loci master: [WIP] Add a profile for kolla infra services https://review.openstack.org/583584 | 16:23 |
| openstackgerrit | Paul Bourke (pbourke) proposed openstack/loci master: [WIP] Add a profile for kolla core services https://review.openstack.org/583255 | 16:27 |
| openstackgerrit | Paul Bourke (pbourke) proposed openstack/loci master: [WIP] Add a profile for kolla infra services https://review.openstack.org/583584 | 16:29 |
| evrardjp | I am very confused, I am trying to build the opensuse image for requirements, so I have this: docker build ./ --no-cache --force-rm --build-arg PROJECT=requirements --build-arg FROM=opensuseleap:15 --build-arg DISTRO=opensuse --tag requirements:opensuse | 16:41 |
| evrardjp | and I have 0 packages listed in my bindep part | 16:42 |
| evrardjp | well | 16:43 |
| evrardjp | it shows git libxml2 | 16:43 |
| evrardjp | very weird | 16:43 |
| SamYaple | moment | 16:43 |
| evrardjp | opensuseleap:15 is my own opensuse image. | 16:43 |
| evrardjp | I will push my code | 16:43 |
| SamYaple | https://github.com/openstack/loci/blob/master/scripts/install.sh#L16 | 16:44 |
| SamYaple | opensuse not there, needs to be added | 16:44 |
| SamYaple | gotta bootstrap a bit to get to using bindep | 16:44 |
| SamYaple | also, we have a super basic distro detect right now for ubuntu/debian/centos https://github.com/openstack/loci/blob/master/scripts/install.sh#L5 | 16:45 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/loci master: [WIP] Add OpenSUSE Leap 15 support for Requirements https://review.openstack.org/586295 | 16:55 |
| evrardjp | SamYaple: wait I haven't been clear | 16:55 |
| evrardjp | that's what I did ^ | 16:55 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/loci master: [WIP] Add OpenSUSE Leap 15 support for Requirements https://review.openstack.org/586295 | 16:56 |
| evrardjp | the problem is not loci I think | 16:56 |
| SamYaple | ahhh cleaner | 16:56 |
| SamYaple | yea makes more sense now | 16:56 |
| evrardjp | I just don't understand why bindep is not showing the full list of packages. | 16:57 |
| evrardjp | is my docker call correct? | 16:57 |
| evrardjp | (sorry I am kinda new) | 16:57 |
| evrardjp | news: it looks like the project REQUIREMENTS is well taken into the CLI. So the bindep call is correct. | 16:58 |
| SamYaple | evrardjp: oh well then i would *first* assume you didnt update bindep.txt https://github.com/openstack/loci/blob/master/bindep.txt#L1 | 16:58 |
| SamYaple | 'platform:zypper' maybe? | 16:58 |
| evrardjp | platform:rpm works | 16:58 |
| evrardjp | should at least | 16:58 |
| SamYaple | if libxml2 is the only thing installed... https://github.com/openstack/loci/blob/master/bindep.txt#L65 | 16:59 |
| evrardjp | yeah | 16:59 |
| SamYaple | that means its failing the other validations | 16:59 |
| SamYaple | i can only assume its platform | 16:59 |
| evrardjp | exactly | 16:59 |
| evrardjp | ok | 16:59 |
| SamYaple | is your bindep patches installed correctly? | 16:59 |
| evrardjp | that's very weird | 16:59 |
| SamYaple | in the container? | 16:59 |
| evrardjp | I will double check | 17:00 |
| evrardjp | that's probably the issue. I will retry | 17:00 |
| evrardjp | I am probably tired. | 17:00 |
| SamYaple | haha been there | 17:00 |
| evrardjp | you have feature freeze in loci? | 17:00 |
| evrardjp | I will go for food and see how it goes | 17:01 |
| evrardjp | :) | 17:01 |
| SamYaple | we don't really do versions the same way as everything else, so the answer is "not really" and "its still early" | 17:01 |
| evrardjp | ok | 17:01 |
| evrardjp | good | 17:01 |
| SamYaple | basically loci builds N-HEAD of master | 17:01 |
| SamYaple | when we can no longer build N, we will need to bump x in x.y.z version | 17:01 |
| SamYaple | thats our current thoughts anyway | 17:01 |
| SamYaple | should help us mesh with this LTS attempt | 17:02 |
| *** e0ne has joined #openstack-loci | 17:20 | |
| *** e0ne has quit IRC | 17:37 | |
| evrardjp | hey | 18:21 |
| evrardjp | I am back from food | 18:21 |
| evrardjp | I found the issue | 18:21 |
| *** e0ne has joined #openstack-loci | 18:22 | |
| evrardjp | bindep 2.6.0 vs bindep.2.8.0 + distro 1.3.0 | 18:22 |
| evrardjp | I will check on the reasons of the freeze | 18:23 |
| SamYaple | evrardjp: reason for freeze was we were using features (the groups) that werent in bindep 2.5.0 | 18:24 |
| SamYaple | we patched them into 2.6.0 and pinned arbitrarily | 18:24 |
| SamYaple | you can remove the pinning | 18:24 |
| evrardjp | ok | 18:24 |
| evrardjp | will patch this | 18:24 |
| SamYaple | good news is, all the "historical" reasons for decisions like this are well remembered :) | 18:25 |
| evrardjp | yeah I checked the history it's clear | 18:26 |
| evrardjp | I am removing the pin, we can re-instate it if 2.8 is giving trouble. | 18:26 |
| SamYaple | that was the original goal for the pin, yes :) | 18:27 |
| SamYaple | we can leave it unpinned, and only pin when new issues occur. *if* they occur. bindep is relatively static | 18:27 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/loci master: Remove bindep pin https://review.openstack.org/586315 | 18:28 |
| evrardjp | SamYaple: that's exactly what I meant | 18:29 |
| evrardjp | SamYaple: you'll learn that my english is REALLY bad and I can't express myself with something else than code ;p | 18:29 |
| evrardjp | hahah | 18:29 |
| evrardjp | anyway, it would be good to bring this in ^ | 18:29 |
| evrardjp | https://review.openstack.org/586315 | 18:29 |
| SamYaple | already approved | 18:30 |
| SamYaple | well wait for gate to merge ;) | 18:30 |
| evrardjp | haha yeah ofc | 18:30 |
| evrardjp | I have the impression I am gonna hang out here for a while. | 18:31 |
| SamYaple | i have the impression LOCI will greatly benefit from you and hrw being here for a while :) | 18:32 |
| evrardjp | hahah | 18:32 |
| evrardjp | cores if you have urgent reviews, just ping me. | 18:33 |
| evrardjp | SamYaple: it's my pleasure to help. | 18:34 |
| SamYaple | happy to have you! | 18:36 |
| SamYaple | want to run an idea by you then | 18:36 |
| SamYaple | so the gates are likely to change so we can test changes to requirements consumed in the final images | 18:37 |
| SamYaple | its going to increase the gate time, but reduce the jobs | 18:37 |
| SamYaple | also allow for better testing of stable releases since we don't plan on doing stable/branch releases | 18:38 |
| SamYaple | any thoughts on that plan? or the reason for the desire? | 18:38 |
| evrardjp | let me parse that | 18:39 |
| evrardjp | I'd like more details on the implementation. You said earlier that it takes 30 minutes tops to build requirements, am I correct? So what's the increase? | 18:41 |
| evrardjp | I think I'd like to understand better with code :) | 18:42 |
| evrardjp | or at least eplain more about the testing of stable/ branches. | 18:42 |
| evrardjp | I think on my side having a stable reproducible build is the best, and tracking stable/rocky will be the deal. | 18:43 |
| evrardjp | (when this will exist) | 18:43 |
| SamYaple | rather than keystone, nova, glance, etc etc, we would have one build gate. -- https://github.com/openstack/loci/tree/master/.zuul.d | 18:43 |
| SamYaple | new services (like octavia is new) will stay non-voting jobs until they are solid | 18:43 |
| evrardjp | ok | 18:44 |
| SamYaple | then absorbed into the bigger gate job which builds rebuildsments first, then the other images | 18:44 |
| evrardjp | so not all of them in a pipeline, but all of them in ONE job | 18:44 |
| SamYaple | we can then have a 'master' job, 'queens' job, etc etc | 18:44 |
| SamYaple | thats correct | 18:44 |
| SamYaple | to help support what we would *like* to do with our versioning and testing | 18:44 |
| evrardjp | could you explain more? | 18:45 |
| evrardjp | what would you like to do in terms of versioning | 18:45 |
| SamYaple | we don't plan on having a stable/queens version of LOCI, rather currently HEAD of LOCI can build Newton and up | 18:45 |
| SamYaple | we want jobs to ensure the above statement is accurate | 18:45 |
| evrardjp | (I am also thinking of pre jobs for having another way to do pipeline things) | 18:45 |
| SamYaple | but i dont want 70 jobs per patch | 18:46 |
| evrardjp | ok | 18:46 |
| evrardjp | opinions here, not based on loci, but general experience: | 18:46 |
| evrardjp | 1) do not branch if you can avoid it and the code is simple. | 18:47 |
| evrardjp | loci code is suffienctly simple to handle branches with no spaghetti code | 18:47 |
| evrardjp | so I'd say that's a noble goal to not create the branches | 18:47 |
| evrardjp | 2) 70 jobs have pros and cons | 18:48 |
| evrardjp | pros: you can make some jobs easily voting or non voting | 18:48 |
| evrardjp | (be more surgical) | 18:48 |
| evrardjp | cons: you will have trouble gating if they are all voting -- infra issues happen. | 18:48 |
| evrardjp | for dealing with large jobs,remember there are jobs templates | 18:49 |
| evrardjp | large amount of jobs* | 18:49 |
| SamYaple | oh its more about not *wanting* large amounts of jobs | 18:49 |
| SamYaple | not because itll be more code to maintain | 18:49 |
| SamYaple | infra issues happen :) | 18:49 |
| SamYaple | but also, it would be seventy 3-7 minute jobs, or 6 30 minute jobes | 18:50 |
| evrardjp | SamYaple: yeah that was the "cons" part | 18:50 |
| SamYaple | 30m+ | 18:50 |
| evrardjp | SamYaple: it makes sense for me | 18:50 |
| SamYaple | ok cool. passes the sanity test from the surface | 18:51 |
| evrardjp | I'd say 30 minutes is fine. | 18:51 |
| SamYaple | ah also, for pulling in a new service to an old version of openstack (Newton for example) the jobs would need to "backport" through the stable gates, merging with the larger gate job when they are stable | 18:52 |
| evrardjp | SamYaple: I am wondering if it's possible to have pre-jobs for building the requirements though | 18:52 |
| SamYaple | evrardjp: no :( | 18:52 |
| SamYaple | thats job artifact sharing, it doesnt exist yet | 18:52 |
| SamYaple | i follow it | 18:52 |
| evrardjp | ok | 18:53 |
| SamYaple | when it does, thats 100% the solution | 18:53 |
| evrardjp | k | 18:53 |
| SamYaple | but zuulv3 was right around the corner for a while..... so im moving forward now haha | 18:53 |
| evrardjp | I am not sure about what you just said about thew new service for Newton | 18:53 |
| evrardjp | it would be master branch of loci, with different "stable branches" jobs, right? | 18:54 |
| SamYaple | ah, so if we add murano for example, it would need to be non-voting master until its stable, then merge with master job. then be non-voting for rocky, then queens, etc etc | 18:54 |
| SamYaple | so how far back do you want to pull it is mroe of the question | 18:54 |
| evrardjp | I think to not become hairy you need to properly write the job names | 18:55 |
| SamYaple | correct, there would be a format just like now | 18:55 |
| evrardjp | I think it's worth spec-ing to make sure it's clear | 18:55 |
| evrardjp | or a patch | 18:55 |
| SamYaple | the non-voting jobs will follow basically the same templates we have now, and the larger test-everything-thats-stable jobs will be the voting ones | 18:55 |
| evrardjp | yeah | 18:56 |
| SamYaple | ya patch is the plan. we don't spec alot | 18:56 |
| evrardjp | k | 18:56 |
| SamYaple | not yet anyway | 18:56 |
| evrardjp | yeah I can definitely understand a job named: murano-queens (non-voting) | 18:56 |
| evrardjp | and that job would only created when murano-rocky (non-voting) is ok | 18:56 |
| evrardjp | which would be created when murano-master is ok | 18:57 |
| evrardjp | (voting or non voting) | 18:57 |
| SamYaple | yep thats the plan anyway. we will see how it goes | 18:57 |
| evrardjp | that's what you meant, right? | 18:57 |
| SamYaple | it seems managable and would be hard to lose track of where the projects are in terms of "stable" | 18:57 |
| evrardjp | I think it's organic | 18:59 |
| evrardjp | you'll see how it goes | 18:59 |
| evrardjp | change it in the future if you see that doesn't scale | 18:59 |
| evrardjp | just document things well and it's gonna be fine :) | 18:59 |
| evrardjp | the reason has to be well explained in the commit message, and it's gonna last forever! | 19:00 |
| evrardjp | (more when git is put into the blockchain) | 19:00 |
| evrardjp | :p | 19:00 |
| SamYaple | more than forever you say... | 19:01 |
| hrw | SamYaple: Newton is DEAD. you should not mention that word. Especially when I listen ;d | 19:11 |
| hrw | at Linaro we did Liberty on hacks, then Newton on smaller amount of hacks. Now we migrate to Queens/Rocky. | 19:13 |
| hrw | I more or less officially forgot that there was something older than Queens ;D | 19:13 |
| SamYaple | ya thats cool but no reason to *not* build it unless its actually broken | 19:13 |
| SamYaple | it follows the same code path after all | 19:13 |
| portdirect | hrw: i wish newton was dead | 19:14 |
| SamYaple | im not saying we are about to add a newton specific patch for the most part | 19:14 |
| SamYaple | also him | 19:14 |
| hrw | SamYaple: still - it is EOL. | 19:14 |
| SamYaple | EOL is such a loose term | 19:14 |
| hrw | :D | 19:15 |
| portdirect | Enterprise Old reLease? | 19:15 |
| hrw | I am trying to convince our admin to forget that he wanted queens ;D | 19:15 |
| hrw | portdirect: sort of ;D | 19:15 |
| SamYaple | oh no, it stands for end of life.... its just big companies seem to be able to keep dead thigns alive | 19:15 |
| hrw | SamYaple: RHOSP is queens iirc | 19:16 |
| SamYaple | portdirect: whats ATT on these days? | 19:16 |
| evrardjp | portdirect: hahah enterprise | 19:18 |
| openstackgerrit | Paul Bourke (pbourke) proposed openstack/loci master: [WIP] Add a profile for kolla images https://review.openstack.org/583255 | 19:19 |
| portdirect | SamYaple: we are planning for ocata, but still have a lot on kilo | 19:21 |
| evrardjp | portdirect: wow | 19:22 |
| SamYaple | saying wow like we werent running 3 years behind on RPC at one point | 19:22 |
| evrardjp | portdirect: you're not leaping to N? | 19:22 |
| portdirect | evrardjp: no - we're gonna keep the good stuff about for a bit longer yet ;) | 19:23 |
| SamYaple | we skipped from M->O with great result, N is a wierd release | 19:23 |
| portdirect | yeah, its the infection point for a lot of things | 19:24 |
| portdirect | cellsv2 was really messed up ;) | 19:24 |
| openstackgerrit | Paul Bourke (pbourke) proposed openstack/loci master: [WIP] Add a profile for kolla images https://review.openstack.org/583255 | 19:26 |
| evrardjp | cellsv2 is not N | 19:28 |
| evrardjp | or is it? | 19:28 |
| evrardjp | I thought it was O | 19:28 |
| portdirect | its barely in n | 19:28 |
| portdirect | you can set up a single v2 cell | 19:28 |
| portdirect | but I'd recommend against it - as it makes upgrade to O a pain... | 19:29 |
| portdirect | this and a few other things were there, but frankly half baked | 19:29 |
| evrardjp | yeah it was not really a thing yet -- O made it mandatory IIRC | 19:29 |
| portdirect | yup | 19:29 |
| evrardjp | I wanted to say "good old times" but I don't really see the "good" :) | 19:30 |
| evrardjp | anyway I think it's time for me to stop speaking, as I am now obviously tired. -- tty tomorrow! | 19:32 |
| *** e0ne has quit IRC | 20:13 | |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/loci master: [WIP] Add OpenSUSE Leap 15 support for Requirements https://review.openstack.org/586295 | 20:25 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/loci master: [WIP] Add OpenSUSE Leap 15 support for Requirements https://review.openstack.org/586295 | 20:25 |
| *** e0ne has joined #openstack-loci | 20:59 | |
| *** e0ne has quit IRC | 21:00 | |
| openstackgerrit | Merged openstack/loci master: Remove bindep pin https://review.openstack.org/586315 | 22:53 |
| openstackgerrit | Merged openstack/loci master: Update centos example in README to use WHEELS https://review.openstack.org/583411 | 22:53 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!