| corvus | it *does* pass tests :) | 00:02 |
|---|---|---|
| *** harlowja has joined #zuul | 00:05 | |
| tristanC | corvus: with openshift, when you run oc cluster up, you get a deployment with default scc and roles with a developper account created | 00:09 |
| clarkb | I expect that with k8s nodepool would talk to a cluster that already exists? | 00:12 |
| tristanC | corvus: with k8s, it seems like i only get cluster-admin role by default | 00:12 |
| *** rcarrill1 has joined #zuul | 00:16 | |
| *** rcarrillocruz has quit IRC | 00:17 | |
| clarkb | tristanC: I half expect that nodepool would be given appropriate credentials for that cluster (whatever that means for that cluster and nodepool) and then it can provision pods as necessary? | 00:18 |
| clarkb | and not be in the business of managing access directly? | 00:18 |
| tristanC | clarkb: sure, let me rephrase the initial question: how are we going to test the k8s driver? | 00:20 |
| clarkb | tristanC: I imagine it could be done similar to how we test openstack. Fakes in the main test suite and an integration job that talks to an actual k8s. | 00:20 |
| tristanC | clarkb: nodepool do need to manage individual pod access using service account for the zuul executor | 00:20 |
| pabelanger | for integration testing, on nodepool, i think we could also do kubespray, like we use devstack | 00:21 |
| tristanC | clarkb: could we define an actual k8s then? | 00:21 |
| clarkb | tristanC: sure we deploy an actual openstack in the integration tests | 00:21 |
| tristanC | pabelanger: does kubespray setup rbac? is this what nodepool k8s user are going to use? | 00:22 |
| clarkb | minikube is the official thing iirc for this use case | 00:22 |
| pabelanger | tristanC: I am not sure to be honest, most I've done with kubespray is in https://review.openstack.org/549768/ | 00:22 |
| pabelanger | was a POC zuul job | 00:22 |
| pabelanger | also kubeup, honestly don't know the differences | 00:23 |
| clarkb | minikube is the devstack of k8s iirc | 00:23 |
| pabelanger | ah | 00:23 |
| clarkb | hrm though it apparently demands a Vm | 00:24 |
| clarkb | tristanC: anyway that is where I would start for testing. Deploy a local single node cluster and do integration testing on it | 00:24 |
| clarkb | use fakes in the test suite | 00:24 |
| clarkb | figure out what needs to be added from there | 00:25 |
| tristanC | clarkb: yes, that's what i started to do, but, i couldn't figure rbac and user access. so i think we should set expectation in the spec regarding k8s deployment configuration | 00:25 |
| pabelanger | https://kubernetes.io/docs/setup/pick-right-solution/ | 00:25 |
| clarkb | tristanC: when you say you couldn't figure it out you mean that you couldn't get it to work at all or to do something specific? | 00:25 |
| tristanC | clarkb: the initial k8s driver works with kubespray, e.g. with a cluster-admin role: https://review.openstack.org/535557 | 00:26 |
| clarkb | I see you are asking about creating access to the namespaces that are created so that one job doens't have access to another? | 00:27 |
| tristanC | clarkb: i don't know what nodepool k8s user context (rbac access) would be | 00:27 |
| clarkb | I think nodepool would have an admin like cluster role (not sure if full admn is necessary) then create namespaces and hand out credentials to ze for a role on that namespace? | 00:28 |
| clarkb | oh we just bind roles to users that maybe makes it easier | 00:29 |
| clarkb | we can document a role that works. have people install it then provision users? | 00:29 |
| tristanC | clarkb: that kind of technical details would be useful to know | 00:30 |
| clarkb | tristanC: I think we don't know them yet though | 00:30 |
| clarkb | (this is me sort of hand waving based on a high level understanding of k8s rbac | 00:31 |
| corvus | yeah, we're going to have to figure details like this out as a team as we go | 00:35 |
| tristanC | in the meantime, the openshift implementation is working with the default deployment developper account... | 00:37 |
| corvus | i think we can generally expect nodepool to "own" the kubernetes that it's using. my understanding is that kubernetes is still not really considered multi-tenant from a security standpoint, so i don't think we'd have to worry about folks colocating production with a hostile ci system as a use-case. so saying "give nodepool the cluster-admin role" sounds reasonable to me. | 00:38 |
| corvus | if we can further restrict that, great. | 00:38 |
| corvus | let's start with k8s first, then openshift later | 00:39 |
| corvus | clarkb: fortunately, we have a ci system that can give us virtual machines to install minikube :) | 00:41 |
| corvus | or, if kubespray is the thing, that. | 00:42 |
| corvus | tristanC: so can you move forward using the cluster-admin role for now? | 00:42 |
| tristanC | corvus: using minikube need a nested vm | 00:43 |
| corvus | oh, nested. ick. | 00:44 |
| corvus | kubespray doesn't have that requirement? | 00:44 |
| tristanC | corvus: i'll be on vacation until next week, i can have another look after using that new piece of information | 00:45 |
| corvus | tristanC: ok. have a nice vacation :) | 00:45 |
| tristanC | corvus: though the container native base jobs remains to be defined. | 00:48 |
| tristanC | i guess you'll want the job to use pbrx? | 00:48 |
| corvus | tristanC: not sure what you mean there -- what's "container native base jobs"? | 00:50 |
| tristanC | corvus: when the driver create a namespace resource | 00:50 |
| tristanC | "native container workflow" | 00:50 |
| tristanC | corvus: for the container as machine, we can easily test a linter can be executed on a vanilla image using the kubectl connection | 00:52 |
| tristanC | corvus: but for the native container workflow, maybe the base job just need to demonstrate it can access the namespace using the k8s_raw module? | 00:52 |
| corvus | tristanC: i guess i'm having trouble with what "base job" means in this context | 00:53 |
| tristanC | corvus: the sequence of action to build an image of the speculative source of the project, deploy the pod(s) and give access to the untrusted job | 00:54 |
| tristanC | isn't this the base job of the native container workflow? | 00:54 |
| corvus | tristanC: ah, that doesn't have to be done by a base job. that can be done entirely as an untrusted job. i think the action of building the image and deploying, etc, will depend on how the project is set up. we can write jobs in zuul-jobs that do that for certain patterns. | 00:55 |
| tristanC | corvus: note that base job doesn't have to trusted ;) | 00:55 |
| corvus | tristanC: so i think we're almost saying the same thing. except it doesn't need to be priviledged, or specifically be a base job. | 00:55 |
| corvus | it does actually | 00:56 |
| tristanC | corvus: hum, but then the implemnetation i proposed with openshift isn't openshift centric (beside the rolebinding action from nodepool) | 00:56 |
| corvus | https://zuul-ci.org/docs/zuul/user/config.html#job "A job with no parent is called a base job and may only be defined in a config-project." | 00:56 |
| tristanC | corvus: what clarkb found to be openshift centric was actually a zuul-jobs then | 00:57 |
| corvus | tristanC: maybe you meant "parent job" instead of "base job" | 00:57 |
| tristanC | corvus: well pabelanger started to move the base job (the default parent) in an untrusted project, and keep the trusted part in a base-minimal job (base has base-minimal as parent) | 00:58 |
| tristanC | corvus: my bad, it's the default parent that is called base job | 00:58 |
| corvus | yeah. the default value for default-parent is "base". but it doesn't have to be a base job. | 00:59 |
| tristanC | alright, then i'll remove the base-openshift-native job proposal from the zuul-base-jobs project | 00:59 |
| corvus | anyway, now that we have the terminology settled :) it seems reasonable to have multiple parent jobs for different circumstances. that will let us try out build systems easily. i can definitely see us having a k8s parent job and a openshift parent job, and possibly different variants of those based on different image builders, etc. | 01:00 |
| corvus | tristanC: for functional testing, i'm not sure we really need to worry about all that right now. your idea of running a job that talks to the k8s api is probably enough to validate the code. the rest we can do inside of running zuuls by defining jobs. | 01:01 |
| tristanC | corvus: alright, then a single task parent job that demonstrate usage of the k8s_raw module | 01:02 |
| corvus | sounds like a plan | 01:02 |
| tristanC | and for the container as machine, an inventory node that can be accessed using the kubectl plugin | 01:02 |
| corvus | ++ | 01:02 |
| tristanC | assuming a kubespray pre-run | 01:03 |
| tristanC | thanks for the detail, i should be able to work on that next week | 01:04 |
| corvus | tristanC: thank you! | 01:04 |
| *** swest has quit IRC | 01:14 | |
| *** harlowja has quit IRC | 01:17 | |
| *** swest has joined #zuul | 01:29 | |
| *** myoung has quit IRC | 02:39 | |
| SpamapS | mmmmm old patch :) | 03:27 |
| SpamapS | glad dequeue is finally getting some love. | 03:27 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: Add a dequeue command to zuul client https://review.openstack.org/95035 | 04:21 |
| *** pcaruana has joined #zuul | 04:28 | |
| *** pcaruana has quit IRC | 04:30 | |
| *** nchakrab has joined #zuul | 04:45 | |
| openstackgerrit | Merged openstack-infra/zuul master: Add a dequeue command to zuul client https://review.openstack.org/95035 | 05:03 |
| *** zigo_ has joined #zuul | 05:53 | |
| *** zigo has quit IRC | 05:53 | |
| *** hashar has joined #zuul | 06:15 | |
| *** nchakrab has quit IRC | 07:38 | |
| *** nchakrab has joined #zuul | 07:50 | |
| *** nchakrab has quit IRC | 09:15 | |
| *** nchakrab has joined #zuul | 09:16 | |
| *** rcarrill1 is now known as rcarrillocruz | 09:41 | |
| *** nchakrab has quit IRC | 09:43 | |
| *** sshnaidm|afk has quit IRC | 10:09 | |
| *** nchakrab has joined #zuul | 10:45 | |
| *** nchakrab has quit IRC | 11:41 | |
| *** jtanner has quit IRC | 11:43 | |
| *** jtanner has joined #zuul | 11:46 | |
| ianw | SpamapS: what's that joke about how when I tell you I'll fix it, I'll fix it, you don't have to keep reminding me every 3 years :) | 11:55 |
| SpamapS | :-D | 11:55 |
| *** pawelzny has joined #zuul | 12:00 | |
| *** pawelzny has left #zuul | 12:05 | |
| *** pawelzny has joined #zuul | 12:05 | |
| *** sshnaidm has joined #zuul | 12:08 | |
| *** pawelzny has quit IRC | 12:08 | |
| *** sshnaidm is now known as sshnaidm|off | 12:08 | |
| *** rlandy has quit IRC | 12:48 | |
| *** rlandy has joined #zuul | 12:48 | |
| *** rcarrill1 has joined #zuul | 13:01 | |
| *** rcarrill2 has joined #zuul | 13:03 | |
| *** rcarrillocruz has quit IRC | 13:03 | |
| *** gchenuet has joined #zuul | 13:05 | |
| *** rcarrill2 is now known as rcarrillocruz | 13:05 | |
| *** rcarrill1 has quit IRC | 13:06 | |
| gchenuet | Hi guys ! I'm currently setup a Zuul v3 stack and i have a question about merger nodes, do I need to setup a reverse proxy (like zuul v2) to serve git repository ? | 13:07 |
| *** samccann has joined #zuul | 13:07 | |
| *** rcarrill1 has joined #zuul | 13:08 | |
| mordred | gchenuet: no, you do not. in v3 the git repos are prepared directly on the executors - and then in the base job we push the code from the executor to the node(s) | 13:08 |
| openstackgerrit | Matthieu Huin proposed openstack-infra/zuul master: scheduler: return project_canonical in status page https://review.openstack.org/582451 | 13:08 |
| gchenuet | ok, perfect! thanks mordred | 13:09 |
| *** rcarrillocruz has quit IRC | 13:10 | |
| *** bbayszczak has joined #zuul | 13:14 | |
| *** rcarrillocruz has joined #zuul | 13:19 | |
| *** rcarrill1 has quit IRC | 13:21 | |
| *** rcarrill1 has joined #zuul | 13:22 | |
| *** rcarrillocruz has quit IRC | 13:25 | |
| bbayszczak | hi everyone I'm working at leboncoin with gchenuet and we are bootstrapping zuul v3 | 13:29 |
| mordred | \o/ | 13:30 |
| bbayszczak | I'm facing issues while running the web component. I can see in the log that the server in running and I can reach an api endpoint (/api/info). But it's impossible to display the dashboard, am I missing some steps ? | 13:30 |
| gchenuet | FYI: we are using the OpenStack ansible roles to setup Zuul | 13:31 |
| *** rcarrill1 is now known as rcarrillocruz | 13:31 | |
| mordred | bbayszczak: possibly - if you're installing from source you need to have yarn installed to build the javascript things | 13:31 |
| mordred | gchenuet: the windmill ones? | 13:31 |
| gchenuet | morded: this one -> https://github.com/openstack/ansible-role-zuul | 13:32 |
| mordred | ok. cool | 13:33 |
| mordred | pabelanger: ^^ is that a thing you can help with ? | 13:33 |
| pabelanger | mordred: gchenuet: sure, what is the issue | 13:34 |
| *** myoung has joined #zuul | 13:35 | |
| *** rcarrill1 has joined #zuul | 13:35 | |
| gchenuet | After running the playbook to setup a web node, we faced some errors when browsing the dashboard | 13:37 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul-jobs master: Add a yarn role https://review.openstack.org/586550 | 13:37 |
| gchenuet | should we need to run a script or something else ? | 13:37 |
| *** rcarrillocruz has quit IRC | 13:37 | |
| pabelanger | ah, zuul web is still missing from the role. Need to add support to build it from source | 13:37 |
| mordred | well - it SHOULD all just work as long as you have yarn installed | 13:38 |
| mordred | pabelanger: sweet - good to know - | 13:38 |
| * mordred couldn't see anything, but isn't familiar with those yet | 13:38 | |
| pabelanger | I'll poke around it today | 13:39 |
| pabelanger | you might be able just to use http://tarballs.openstack.org/zuul/zuul-content-latest.tar.gz for now | 13:39 |
| pabelanger | until role is updated | 13:39 |
| gchenuet | ok, just need to serve these files with a webserver and it should be ok ? | 13:41 |
| pabelanger | yah | 13:41 |
| openstackgerrit | Joshua Hesketh proposed openstack-infra/zuul master: Add instructions for deploying zuul with openSUSE https://review.openstack.org/581255 | 13:41 |
| *** rcarrillocruz has joined #zuul | 13:41 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Change npm reference to yarn https://review.openstack.org/586552 | 13:41 |
| gchenuet | pbelanger: thanks ! and setup yarn too ? | 13:43 |
| mordred | you don't need yarn if you grab those files | 13:43 |
| *** rcarrill1 has quit IRC | 13:43 | |
| mordred | and if you unpack them into the zuul/web/static dir in the zuul python installation, zuul should just find them and serve them out for you | 13:44 |
| mordred | (if you install zuul from source with yarn installed, they should be built and put into zuul/web/static) | 13:44 |
| mordred | BUT - you can also just set up a webserver to serve those files ... | 13:45 |
| pabelanger | I'll try using yarn this weekend, haven't yet. | 13:46 |
| mordred | cool. it's hopefully pretty easy - but let me know if you run in to issues | 13:46 |
| gchenuet | Cool ! thanks guys ! | 13:47 |
| *** rcarrill1 has joined #zuul | 13:48 | |
| bbayszczak | thanks ! | 13:49 |
| *** rcarrillocruz has quit IRC | 13:50 | |
| *** rcarrillocruz has joined #zuul | 13:52 | |
| *** rcarrill1 has quit IRC | 13:54 | |
| *** ianychoi has joined #zuul | 13:56 | |
| *** ianychoi_ has quit IRC | 14:00 | |
| *** jlvacation is now known as jlvillal | 14:18 | |
| *** bbayszczak has quit IRC | 14:47 | |
| *** maeca has joined #zuul | 15:04 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: web: add /{tenant}/job/{job_name} route https://review.openstack.org/550978 | 15:12 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: web: add /{tenant}/projects and /{tenant}/project/{project} routes https://review.openstack.org/550979 | 15:12 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: web: add /{tenant}/pipelines route https://review.openstack.org/541521 | 15:12 |
| -openstackstatus- NOTICE: A zuul config error slipped through and caused a pile of job failures with retry_limit - a fix is being applied and should be back up in a few minutes | 15:41 | |
| *** ChanServ changes topic to "A zuul config error slipped through and caused a pile of job failures with retry_limit - a fix is being applied and should be back up in a few minutes" | 15:41 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul-jobs master: Add upload-logs-swift role https://review.openstack.org/584541 | 15:51 |
| *** openstackgerrit has quit IRC | 16:04 | |
| *** gchenuet has quit IRC | 16:17 | |
| *** gchenuet has joined #zuul | 16:18 | |
| *** gchenuet has quit IRC | 16:18 | |
| *** openstackgerrit has joined #zuul | 16:21 | |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Add role to ensure per-node output dirs exist https://review.openstack.org/511823 | 16:21 |
| *** yolanda has quit IRC | 16:22 | |
| *** harlowja has joined #zuul | 16:27 | |
| *** yolanda has joined #zuul | 16:29 | |
| Shrews | just a reminder for folks: i'm having wrist surgery on Tuesday and will be out for an indeterminate amount of time (taking at least next week off). will return when i can use my fingers again. | 16:34 |
| Shrews | i'll likely be around a bit early monday | 16:35 |
| clarkb | Shrews: here's to quick healing wrists | 16:51 |
| Shrews | clarkb: yeah srsly. thx. def not looking fwd to this | 16:55 |
| *** panda|rover is now known as panda|rover|off | 17:03 | |
| *** hashar has quit IRC | 17:19 | |
| pabelanger | is there any reason not to have zuul-scheduler retry forever if it cannot connect to zookeeper on first start? | 17:21 |
| pabelanger | eg: http://logs.openstack.org/88/586588/1/check/ansible-role-zuul-fedora-27/7d2f82a/logs/var/log/zuul/scheduler-debug.log | 17:21 |
| pabelanger | 2018-07-27 15:52:16,317 WARNING kazoo.client: Failed connecting to Zookeeper within the connection retry policy. | 17:21 |
| Shrews | pabelanger: i thought tristanC put up a change for that? | 17:22 |
| pabelanger | ah, haven't looked | 17:22 |
| Shrews | commit f306a690a1bfd611eed7248c929c3f75891f4a91 | 17:23 |
| Shrews | https://review.openstack.org/576048 | 17:23 |
| pabelanger | ah, you know what, this is released version of zuul | 17:23 |
| pabelanger | let me check other job using zuul from master | 17:24 |
| pabelanger | http://logs.openstack.org/88/586588/1/check/ansible-role-zuul-src-fedora-27/6e344d1/logs/var/log/zuul/scheduler-debug.log | 17:24 |
| pabelanger | cool, works as expected | 17:24 |
| pabelanger | Shrews: thanks! | 17:24 |
| *** pwhalen has quit IRC | 17:26 | |
| clarkb | corvus: do you have a plan for how you want to add the swift logs changes to zuul-jobs? Just approve as reviewers get through them or does it need to be a bit more organized than that? | 17:29 |
| *** pwhalen has joined #zuul | 17:29 | |
| *** pwhalen has joined #zuul | 17:29 | |
| *** harlowja has quit IRC | 17:31 | |
| *** myoung is now known as myoung|lunch | 18:09 | |
| *** harlowja has joined #zuul | 18:15 | |
| *** tranzemc has quit IRC | 18:26 | |
| *** sshnaidm|off has quit IRC | 18:32 | |
| corvus | clarkb: re swift logs: there should be no interference with anything else so folks can approve at will. | 18:34 |
| tobiash | corvus: added a minor comment to it | 18:35 |
| tobiash | and ++ for approving and iterating on it | 18:35 |
| tobiash | there are enough disclaimers to prevent folks from using it yet :) | 18:35 |
| corvus | tobiash: re your comment: agreed. currently the icons are all broken anyway :) | 18:36 |
| tobiash | corvus: I got them working via apache config :) | 18:36 |
| openstackgerrit | Merged openstack-infra/zuul master: Add debug message to job freezing https://review.openstack.org/582750 | 18:36 |
| tobiash | corvus: but I have the apache in front of it anyway | 18:36 |
| corvus | heh. we'll have to do something like upload them on container creation since we're going to try avoidinga p roxy | 18:36 |
| tobiash | makes sense | 18:37 |
| tobiash | mordred: I got the swift proxy working with binary streaming :) | 18:38 |
| tobiash | mordred: with a simple custom request via conn.object_store.get() | 18:39 |
| clarkb | sort of related to this apparently s3 uses a global namespace | 18:40 |
| clarkb | er global namespace for container names | 18:40 |
| clarkb | corvus: one thought, should we put a few bytes into the test files for the role? | 18:41 |
| tobiash | clarkb: in our cloud the ceph-radosgw also has a global namespace for container names which is a little bit confusing if you don't know that yet... | 18:41 |
| tobiash | ++ for test files with a little bit of content | 18:42 |
| clarkb | tobiash: interesting. It seems like a bad constraint particularly if you are running a massive public cloud service like amazon does | 18:42 |
| corvus | well, the test doesn't actually upload them. it only uses them to generate index files. | 18:42 |
| clarkb | corvus: ah | 18:43 |
| tobiash | corvus: I tested uploaded that for testing :) but actually I don't mind | 18:43 |
| clarkb | I'm just now getting to the test script | 18:43 |
| corvus | (basically, the parts of the library which scans directories to decide what to upload, and generates indexes are unit tested. actual uploading is not) | 18:43 |
| corvus | tobiash: ah, ok, yeah, if it'll help for manual testing too, that's fine | 18:43 |
| corvus | (of course, this being zuul, we *should* be able to add a job that spins up a swift and tests it for real :) | 18:44 |
| tobiash | clarkb: but that also could well be an integration into openstack issue with our ceph-radosgw | 18:44 |
| tobiash | hrm, first try of swift upload in ansible: 2018-07-27 19:10:10.451891 | localhost | MODULE FAILURE | 19:11 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul-jobs master: Copy inventory file first for validate-host https://review.openstack.org/586639 | 19:11 |
| clarkb | tobiash: corvus I was just going to say I did a more in depth review and +2'd :P I noted some things we should probably update but we don't need all of that in the first pass | 19:12 |
| clarkb | that review has made me hungry too | 19:13 |
| tobiash | clarkb: might be a deployment issue: http://paste.openstack.org/show/726763/ | 19:14 |
| clarkb | tobiash: oh ya you need to install openstacksdk | 19:15 |
| clarkb | and get it into your bwrap envs | 19:15 |
| tobiash | that should then be added to zuul's requirements.txt | 19:15 |
| clarkb | possibly, though this role isn't required to run zuul | 19:16 |
| clarkb | maybe we add it as a setuptools optional dep list (I forget the actual name for these) | 19:17 |
| tobiash | clarkb: yes, I guess but it will become one of the standard roles for log upload so that should be supported out of the box imho | 19:17 |
| *** myoung|lunch is now known as myoung | 19:18 | |
| openstackgerrit | Merged openstack-infra/zuul master: fix zuul from scratch user and group creation https://review.openstack.org/581254 | 19:37 |
| openstackgerrit | Merged openstack-infra/zuul master: Add instructions for deploying zuul with openSUSE https://review.openstack.org/581255 | 19:37 |
| tobiash | yay, it worked within ansible :) | 19:44 |
| *** rlandy is now known as rlandy|brb | 19:55 | |
| *** samccann has quit IRC | 19:57 | |
| tobiash | corvus: it adds a superfluous logs/ dir between the prefix and the actual content | 20:13 |
| corvus | mordred, clarkb, jhesketh: i had a comment on PS2 of https://review.openstack.org/581255 which i don't beleieve was resolved satisfactorily | 20:13 |
| tobiash | apart from that it seems to work nicely :) | 20:13 |
| corvus | mordred, clarkb, jhesketh: is there no bigtop repo for opensuse? | 20:13 |
| clarkb | I am not sure, but I know the method jhesketh documented works and works well as it is what I use | 20:14 |
| clarkb | I'm having a hard time finding a list of the bigtop package repos | 20:15 |
| corvus | lot's of things work, but bigtop is an official apache project, so it seems like a really good place to get packages of apache software. and it's what we use elsewhere in the instructions, so i think it should be our first choice, and at least warrants some investigation. | 20:16 |
| *** rlandy|brb is now known as rlandy | 20:17 | |
| corvus | Bigtop support many Operating Systems, including Debian, Ubuntu, CentOS, Fedora, openSUSE and many others. | 20:18 |
| clarkb | the centos url isn't browsable either so having a hard time finding where suse might be at | 20:18 |
| clarkb | corvus: http://repos.bigtop.apache.org/ doesn't seem to have suse/leap/opensuse listed | 20:19 |
| clarkb | corvus: fwiw I'm happy to use official packaging if it exists. I ended up doing what jhesketh documented a while back because I couldn't find such a thing when I looked | 20:20 |
| corvus | i'm really uncomfortable recommending folks install packaging from random places | 20:21 |
| clarkb | corvus: I don't think we are recommending that? | 20:21 |
| corvus | what *are* we recommending? | 20:21 |
| clarkb | corvus: in the change that merged we are recommending that we use official zookeeper tarballs from apache/zookeeper | 20:22 |
| corvus | what's amaze.com.au? | 20:22 |
| clarkb | a mirror for apache's artifacts | 20:22 |
| clarkb | we can update the url to be the non mirror location | 20:22 |
| *** dtruong_ has quit IRC | 20:23 | |
| clarkb | ah they don't appera to have public download outside of mirrors | 20:23 |
| corvus | i guess at least we recommend verifying the checksum? | 20:23 |
| corvus | clarkb: thanks for looking into bigtop | 20:25 |
| pabelanger | https://archive.apache.org/dist/zookeeper/ might be good place | 20:25 |
| clarkb | pabelanger: the top of that page asks that you not download from it | 20:26 |
| pabelanger | yah, that seems... odd | 20:26 |
| *** dtruong_ has joined #zuul | 20:26 | |
| clarkb | pabelanger: its fairly common if bw is a concern. Easier on them to farm that out and distribute it | 20:27 |
| clarkb | another option is to see if there are any unofficial packages on obs; however, I'm personally more comfortable using official release artifacts (even if mirrored) than random obs repositories | 20:27 |
| corvus | we could just point at apache's release page | 20:27 |
| clarkb | corvus: ya that would work too | 20:28 |
| pabelanger | TIL: bigtop, going to try that for centos | 20:29 |
| corvus | pabelanger: it's in the zuul-from-scratch document :) | 20:29 |
| *** dmsimard has quit IRC | 20:29 | |
| pabelanger | indeed | 20:31 |
| pabelanger | clarkb: corvus: I'm ready to help with zuul now, if others are. I forgot I'll be leaving for sportsball in an hour | 20:34 |
| clarkb | ya I can help with a restart, quickly writing a patch to update the zookeeper download url | 20:34 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul-jobs master: Add upload-logs-swift role https://review.openstack.org/584541 | 20:38 |
| rcarrillocruz | I used bigtop for my Zuul ansible container app | 20:39 |
| rcarrillocruz | Worked fine | 20:39 |
| openstackgerrit | Clark Boylan proposed openstack-infra/zuul master: Point Suse users are zookeeper releases page https://review.openstack.org/586682 | 20:39 |
| clarkb | rcarrillocruz: the question is whether or not such packages exist for (open)suse | 20:40 |
| rcarrillocruz | For zookeeper dep that is | 20:40 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul-jobs master: Add upload-logs-swift role https://review.openstack.org/584541 | 20:40 |
| corvus | clarkb, tobiash: ^ replied and addressed | 20:40 |
| clarkb | I am unable to find them if they do | 20:40 |
| clarkb | corvus: does https://review.openstack.org/586682 look a bit better for suse? | 20:40 |
| corvus | clarkb: wfm | 20:41 |
| corvus | thx | 20:41 |
| pabelanger | not that I have logs now, but seems an outage in rdocloud wedge nodepool somehow. When rdocloud was fixed, nodepool didn't self heal | 21:36 |
| pabelanger | WARNING nodepool.driver.openstack.OpenStackProvider: Cannot find provider pool for node... was about the only thing I could see that might be related | 21:36 |
| clarkb | pabelanger: that is the type of message you get when a node in zk has its cloud removed from the yaml config iirc | 21:37 |
| clarkb | and I want to say that can wedge things because nodepool doesn't know how to handle that node | 21:37 |
| pabelanger | yah, that's what I thought too | 21:38 |
| pabelanger | but I am not sure any cloud was removed from yaml | 21:38 |
| pabelanger | I cannot look now, but will dive deeper | 21:39 |
| clarkb | pabelanger: you can always cross check the contents of that node in zk against what you have for launcher config | 21:39 |
| pabelanger | yah | 21:40 |
| openstackgerrit | Merged openstack-infra/zuul master: Point Suse users are zookeeper releases page https://review.openstack.org/586682 | 21:43 |
| pabelanger | 2018-07-18 05:48:32,644 WARNING nodepool.CleanupWorker: Marking for delete leaked instance upstream-centos-7-rdo-cloud-tripleo-0000063539 (04413a49-bbff-4de5-9892-3dcb8e9338e5) in rdo-cloud-tripleo (unknown node id 0000063539) | 21:45 |
| pabelanger | I see that oo | 21:45 |
| pabelanger | too* | 21:45 |
| pabelanger | okay, now I run, will debug more later | 21:45 |
| clarkb | corvus: did you see my note about os.walk and symlinks in the swift logs change? (I am going to approve that change now unless you think we should wait, but wanted to make sure you had seen that comment first as its probably something to be aware of as we move forward) | 21:48 |
| corvus | clarkb: oh yes, i suspect you're right about that -- we probably do want to handle symlinks. | 21:50 |
| clarkb | corvus: I don't think w eneed to fix that right now, but ya we should followup on it | 21:50 |
| clarkb | corvus: any object to approving that change as is? | 21:50 |
| corvus | clarkb: nope. we can do symlinks in a followup. | 21:51 |
| corvus | i'm rechecking https://review.openstack.org/585830 now | 21:51 |
| *** sshnaidm|off has joined #zuul | 22:04 | |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Add upload-logs-swift role https://review.openstack.org/584541 | 22:17 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Add openstacksdk to executor extras https://review.openstack.org/586717 | 22:20 |
| *** sambetts_ has quit IRC | 22:24 | |
| *** sambetts_ has joined #zuul | 22:26 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: WIP: Support line comments in Gerrit https://review.openstack.org/577035 | 22:48 |
| *** harlowja has quit IRC | 23:09 | |
| mordred | corvus: re: symlinks - the zuul multi-tenant dashboard draft job is a good trial ballon - since it uses directory symlinks to simulate the apache rewrite rule for multi-tenants | 23:16 |
| corvus | mordred: thanks, i thought that sounded familiar :) | 23:17 |
| mordred | yah - it's ... it's a tricky one the way it's done right now, becuase it would basically be a recursive symlink | 23:19 |
| mordred | (it's two symlinks to '.') | 23:19 |
| mordred | so ... might be a good thought exercise to think about how else we might solve it other than symlinks in a swift world | 23:20 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Support line comments in Gerrit https://review.openstack.org/577035 | 23:20 |
| corvus | ^ not WIP :) | 23:20 |
| mordred | \o/ | 23:20 |
| *** myoung has quit IRC | 23:22 | |
| mordred | corvus: 577027 has 2x+2 - you wanna hold off on it for any reason? | 23:31 |
| mordred | corvus: also - in 577035, lines 703 through 708 looks like perhaps it wants to be after line 642 instead? | 23:34 |
| openstackgerrit | Merged openstack-infra/zuul master: scheduler: return project_canonical in status page https://review.openstack.org/582451 | 23:34 |
| *** rlandy has quit IRC | 23:37 | |
| corvus | mordred: 027 should be fine | 23:50 |
| corvus | mordred: and yep, that's a rebase gone awry. | 23:50 |
| corvus | that pgraph was moved, so it should just be deleted from my patch | 23:52 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Support line comments in Gerrit https://review.openstack.org/577035 | 23:53 |
| mordred | corvus: +2 - and +A to 027 | 23:54 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Support line comments in Gerrit https://review.openstack.org/577035 | 23:56 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul-jobs master: Add role to fetch zuul logs from nodes https://review.openstack.org/583346 | 23:58 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!