| sfbender | Tristan de Cacqueray created www.softwarefactory-project.io master: Add Kubernetes Nodepool Driver blog post https://softwarefactory-project.io/r/13725 | 05:08 |
|---|---|---|
| tristanC | shanemcd: here is a documentation about using kubernetes nodepool driver: https://softwarefactory-project.io/logs/25/13725/1/check/build-pages/43169a0/pages/kubernetes-nodepool-driver-tech-preview.html | 05:10 |
| *** chkumar|ruck has joined #softwarefactory | 06:40 | |
| sfbender | Tristan de Cacqueray created scl/zuul-distgit master: Add React web interface https://softwarefactory-project.io/r/13726 | 06:53 |
| sfbender | Tristan de Cacqueray created software-factory/sf-config master: zuul: adapt gateway rewrite for React interface https://softwarefactory-project.io/r/13727 | 06:59 |
| *** jpena|off is now known as jpena | 07:49 | |
| sfbender | Tristan de Cacqueray created software-factory/sf-ci master: zuul: fix gateway test for React interface https://softwarefactory-project.io/r/13728 | 08:03 |
| sfbender | Merged software-factory/sf-config master: grafana: support missing providers https://softwarefactory-project.io/r/13526 | 09:00 |
| *** zoli is now known as zoli|lunch | 09:32 | |
| *** zoli|lunch is now known as zoli | 09:32 | |
| sfbender | Merged software-factory/sf-config master: resources: use config_location to clone the config repository https://softwarefactory-project.io/r/12844 | 11:00 |
| *** jpena is now known as jpena|lunch | 11:10 | |
| *** jpena|lunch is now known as jpena | 12:14 | |
| shanemcd | tristanC: just now seeing this link. ty. i'll take a look after i get these static nodes provisioned for you. | 13:31 |
| tristanC | shanemcd: nice, if you can also setup a k8s cluster and give me the kube/config, i think we could try using it right away | 13:32 |
| shanemcd | oh sure. | 13:33 |
| tristanC | it seems like the openshift-base job "oc rsync" task works with kube pod | 13:33 |
| matburt | ^^ nice | 13:50 |
| *** chkumar|ruck has quit IRC | 14:08 | |
| *** chandankumar has joined #softwarefactory | 14:14 | |
| *** chandankumar is now known as chkumar|ruck | 14:15 | |
| matburt | where is the ansible tenant configuration from the perspective of the config repo? | 14:31 |
| matburt | https://softwarefactory-project.io/cgit/config/commit/?id=d1ce9b01b9d164abfc8d03fd7e95feaf342e401f | 14:32 |
| tristanC | matburt: it's https://github.com/ansible/zuul-config/blob/master/resources/ansible.yaml | 14:34 |
| matburt | dig it | 14:35 |
| pabelanger | https://ansible-network.softwarefactory-project.io/zuul/api/status: Proxy Error | 14:38 |
| tristanC | pabelanger: seems like rdo-cloud network issue... | 14:39 |
| pabelanger | yah | 14:39 |
| tristanC | pabelanger: it's back | 14:40 |
| tristanC | matburt: the sf-project.io config-update job generate zuul configuration from that ansible/zuul-config/resources directory, you can check the final zuul tenant configuration using: "curl https://softwarefactory-project.io/manage/v2/configurations/zuul" | 14:41 |
| tristanC | (it's a bit slow to reply) | 14:41 |
| matburt | np, I'm just looking at what config goes into building up the ansible tenants jobs and pipeline configurations | 14:42 |
| matburt | I see some base jobs defined in that zuul config and was wondering how that diverges from whatever the default for SF or zuul's normal base | 14:42 |
| tristanC | matburt: ansible only has one config project (as seen here: https://ansible.softwarefactory-project.io/zuul/projects.html ) | 14:43 |
| tristanC | that projects hold pipeline and base jobs (jobs which have no parent) | 14:43 |
| tristanC | the jobs are generated similarly to the SF normal base | 14:44 |
| matburt | gotcha | 14:44 |
| tristanC | matburt: one thing to note is that config projects are considered trusted, that means they are not evaluated speculatively, only merged content is effective | 14:46 |
| matburt | yep that makes sense | 14:46 |
| matburt | https://github.com/ansible/awx/pull/2309 | 14:56 |
| matburt | that'll be the initial pipeline / job usage | 14:56 |
| pabelanger | matburt: you'll want to drop gate, since zuul isn't merging commits. Will be okay to just have check | 14:57 |
| matburt | I mean, we do want zuul to merge commits | 14:57 |
| pabelanger | right, but that means zuul would merge all PRs | 14:58 |
| matburt | that's right | 14:58 |
| matburt | is there some reason we shouldn't do that? | 14:58 |
| pabelanger | mostly rdocloud | 14:59 |
| pabelanger | just need to be away of the issues it has and if awx wants to be blocked by that | 14:59 |
| matburt | Well, I've been using this for Runner... and it's been fine? Should we start looking at setting up our own zuul infrastructure? | 15:00 |
| pabelanger | well, running own has lots of other issues to discuss | 15:00 |
| matburt | I mean... the cloud goes down occasionally but it generally comes back up. | 15:01 |
| *** chkumar|ruck is now known as chkumar|off | 15:02 | |
| pabelanger | my may concern, is if we start gating things, and not understand how fragile the cloud is we are running jobs, we'll give zuul a bad impression | 15:02 |
| pabelanger | then we'll stop using zuul | 15:02 |
| pabelanger | I'd love to see us start gating, but want to make sure everybody understands what that means | 15:02 |
| pabelanger | right now, we don't have an SLA's in place, so this is just best effort | 15:02 |
| pabelanger | and, rdocloud is scheduled to be replaced with another cloud, upshift public | 15:03 |
| pabelanger | so, we'll need to deal with migration issues in the next month | 15:03 |
| pabelanger | or too | 15:03 |
| tristanC | pabelanger: have you found a way to make zuul the only user able to merge PR? | 15:08 |
| tristanC | i think that in the meantime, it's fine to add gate, and use it as best effort | 15:08 |
| pabelanger | tristanC: looking, I think it is either something in branch protection, or remove merge permissions for all users on repo (adding them back when something is wedged) | 15:09 |
| pabelanger | right, I want to make sure people 100% understand this is best effort, and not get upset and frustrated at zuul if it doesn;t work | 15:10 |
| pabelanger | Ansible is in a great play to start iterating on zuul as CI, but also don't want to get under water is external issues not related to zuul (eg: infrastructure) | 15:11 |
| tristanC | matburt: no event was sent for awx/pull/2309, maybe the project is missing github app installation? | 15:12 |
| matburt | it is, I need to get a certain person to add that to the config | 15:13 |
| tristanC | matburt: i just submitted the exact same change to an awx copy in my zuul sandbox :) | 15:14 |
| pabelanger | remote: https://softwarefactory-project.io/r/13734 Create ansible-fedora-27 in vexxhost | 15:23 |
| pabelanger | tristanC: nhicher: fbo: mhu: could I get a review of ^ please | 15:23 |
| matburt | zuul configuration added, recheck running: https://github.com/ansible/awx/pull/2309 | 15:35 |
| pabelanger | remote: https://softwarefactory-project.io/r/13735 Add haveged to ansible-base element | 15:39 |
| pabelanger | tristanC: mhu: nhicher: fbo: another review^ adds haveged to diskimages | 15:40 |
| spredzy | matburt: tristanC what's the way to log in https://ansible.softwarefactory-project.io ? | 16:02 |
| *** zoli is now known as zoli|gone | 16:07 | |
| *** zoli|gone is now known as zoli | 16:07 | |
| tristanC | spredzy: there is no authentication configured because there are no authenticated service, zuul doesn't support rest action (yet) | 16:11 |
| tristanC | spredzy: why would you want to log in? | 16:11 |
| spredzy | tristanC: old habbits :D | 16:13 |
| spredzy | I thought zuul-jobs were on Gerrit | 16:13 |
| spredzy | (and zuul-config) but no I just saw that they are on Github | 16:14 |
| pabelanger | remote: https://softwarefactory-project.io/r/13736 Add mising ansible-fedora-27 label for nodepool | 16:16 |
| pabelanger | nhicher: ^missed the label again | 16:16 |
| pabelanger | nhicher: I approved | 16:20 |
| nhicher | pabelanger: ok | 16:21 |
| spredzy | tristanC: I think that adding ansible_python_version below ansible_version here https://github.com/openstack-infra/zuul-jobs/blob/master/roles/emit-job-header/tasks/main.yaml#L12 might be useful | 16:51 |
| spredzy | pabelanger: ^ | 16:52 |
| spredzy | wanted to hear from you guys if you think its a good idea and hence I should submit a rview | 16:52 |
| spredzy | Or if there's another way to get the current python version in the runner | 16:53 |
| pabelanger | spredzy: it is in inventory file | 16:53 |
| pabelanger | https://softwarefactory-project.io/logs/36/13736/1/check/config-check/d668a80/zuul-info/inventory.yaml | 16:53 |
| spredzy | Ack, then nvm what I said above | 16:56 |
| spredzy | Thanks for the pointer | 16:56 |
| pabelanger | np! | 16:56 |
| pabelanger | spredzy: but zuul will hardcode it to python2 right now | 16:57 |
| pabelanger | long term, plan is to remove that | 16:57 |
| spredzy | Actually I also can get it from here https://ansible.softwarefactory-project.io/logs/09/2309/1699fb63f48de856bdff8b2d4e6a71d502cd0ba3/check/flake8/6eda380/zuul-info/host-info.container.yaml | 16:58 |
| spredzy | This isn't hardcoded right ^ ? | 16:58 |
| spredzy | ansible_python_version: 2.7.14 | 16:58 |
| pabelanger | the python version isn't hardcoded, but to use python2 is | 16:58 |
| pabelanger | we don't yet support python3 as default with zuul | 16:58 |
| spredzy | ah, oki gotcha | 16:59 |
| pabelanger | https://git.zuul-ci.org/cgit/zuul/tree/zuul/executor/server.py#n1522 | 16:59 |
| pabelanger | err | 16:59 |
| pabelanger | https://git.zuul-ci.org/cgit/zuul/tree/zuul/executor/server.py#n1523 | 16:59 |
| *** sshnaidm has joined #softwarefactory | 17:11 | |
| *** sshnaidm is now known as sshnaidm|pto | 17:13 | |
| *** sshnaidm|pto has quit IRC | 17:20 | |
| *** jpena is now known as jpena|off | 17:21 | |
| matburt | it does look like it picked up python 3 actually for that flake8 test run | 17:23 |
| *** sshnaidm|pto has joined #softwarefactory | 17:33 | |
| *** sshnaidm|pto has quit IRC | 18:36 | |
| pabelanger | remote: https://softwarefactory-project.io/r/13741 Add ntp / ntpdate to ansible-network images | 19:45 |
| pabelanger | nhicher: do you mind a review^ adds NTP support to our images | 19:46 |
| nhicher | done pabelanger | 19:50 |
| pabelanger | nhicher: thanks, do you mind deleted the latest fedora-28 DIB in nodepool-builder, to kick off a rebuild once that lands? | 19:51 |
| pabelanger | still trying to debug why glean isn't working properly on fedora / vexxhost | 19:51 |
| nhicher | pabelanger: ok, I will | 19:53 |
| pabelanger | ty | 19:54 |
| pabelanger | nhicher: can you confirm which version of shade is installed on nodepool-launcher? | 20:03 |
| nhicher | pabelanger: shade-1.28.0-1 | 20:04 |
| nhicher | pabelanger: nodepool is building the image | 20:05 |
| pabelanger | thanks | 20:07 |
| nhicher | pabelanger: image is ready | 20:12 |
| *** sshnaidm|pto has joined #softwarefactory | 20:34 | |
| pabelanger | https://softwarefactory-project.io/r/13744 | 20:45 |
| pabelanger | nhicher: thanks! Do you mind ^ too, we'll also need an image rebuild please | 20:45 |
| nhicher | pabelanger: ok | 20:50 |
| pabelanger | nhicher: thank you | 21:03 |
| nhicher | pabelanger: you're faster than me, image is ready =) | 21:03 |
| pabelanger | yay | 21:04 |
| nhicher | pabelanger: we need to add your user to sf-core team, it will be easier for you to debug your stuff. ansible-network team need a sf-core user | 21:09 |
| *** sshnaidm|pto has quit IRC | 21:10 | |
| pabelanger | nhicher: what does that allow me to review? eg: gerrit? | 21:10 |
| nhicher | ssh access to sf instances | 21:10 |
| nhicher | managesf, zuul-scheduler, nodepool ... | 21:11 |
| pabelanger | nhicher: okay, so the plan will be to give ansible-network person access to the control plane in SF | 21:11 |
| pabelanger | just want to make sure I understand | 21:11 |
| nhicher | yes, that's the plan | 21:12 |
| pabelanger | okay, I don't mind that being me. I'm mostly reaching out to SF member today, most to keep you in the loop of issues. But yes, access to servers will be easier to debug | 21:13 |
| nhicher | to be able to check executor logs, rebuild image ... | 21:13 |
| pabelanger | yah | 21:13 |
| nhicher | it's not an issue for me when you ping me for a review, but it will be easier for you | 21:14 |
| nhicher | I will discuss with the team tomorow | 21:14 |
| pabelanger | sure | 21:17 |
| pabelanger | also, issue with glean and fedora is hostname | 21:17 |
| pabelanger | it is too long | 21:17 |
| pabelanger | ansible-fedora-28-vexxhost-ansible-networking-ca-ymq-1-0000188812 | 21:17 |
| pabelanger | I think tomorrow, we'll rename ansible-networking tenant in nodepool. It really should be ansible-network | 21:18 |
| pabelanger | in the mean time, we can setup the hostname via nodepool.yaml | 21:19 |
| nhicher | the issue is the hostname ? | 21:21 |
| nhicher | pabelanger: ^ | 21:21 |
| pabelanger | nhicher: yup | 21:21 |
| pabelanger | cap is 64 | 21:21 |
| pabelanger | but we setting 65 | 21:21 |
| nhicher | that's weird | 21:21 |
| pabelanger | working on patch | 21:21 |
| nhicher | pabelanger: ok | 21:22 |
| pabelanger | remote: https://softwarefactory-project.io/r/13745 Fix hostname limit with ansible-network tenant | 21:23 |
| pabelanger | nhicher: ^ | 21:23 |
| nhicher | done pabelanger | 21:25 |
| pabelanger | cool, will try again once merged, and tomorrow we can look to roll back some of the debug stuff we added to glean | 21:25 |
| pabelanger | and I'll update glean upstream to just do that by default | 21:25 |
| nhicher | pabelanger: ok, work for me | 21:28 |
| *** sshnaidm|pto has joined #softwarefactory | 21:40 | |
| *** shanemcd- has joined #softwarefactory | 22:59 | |
| *** shanemcd has quit IRC | 23:00 | |
| *** matburt has quit IRC | 23:00 | |
| *** sshnaidm|pto has quit IRC | 23:04 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!