| *** rlandy is now known as rlandy|bbl | 00:02 | |
| jlk | Woo, good response on the github3.py bug. | 00:13 |
|---|---|---|
| * fungi cheers | 00:36 | |
| tristanC | corvus: ianw: the nodepool-web interface i proposed on zuul-web could be moved to nodepool when it grew the tooling to support it | 00:37 |
| tristanC | corvus: ianw: in the meantime, i added it to zuul-web to get the ball moving | 00:37 |
| tristanC | or maybe i misunderstood what you said :) | 00:55 |
| *** odyssey4me has quit IRC | 00:59 | |
| *** odyssey4me has joined #zuul | 00:59 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul-jobs master: DNM: negative test https://review.openstack.org/522438 | 01:16 |
| *** myoung|bbl is now known as myoung | 02:39 | |
| *** yolanda has quit IRC | 02:44 | |
| *** yolanda has joined #zuul | 02:45 | |
| *** yolanda has quit IRC | 02:53 | |
| *** yolanda has joined #zuul | 02:54 | |
| *** rlandy|bbl is now known as rlandy | 03:22 | |
| *** rlandy has quit IRC | 03:24 | |
| *** bhavik1 has joined #zuul | 03:28 | |
| *** bhavik1 has quit IRC | 03:53 | |
| *** myoung is now known as myoung|afk | 04:28 | |
| *** CrayZee has joined #zuul | 05:49 | |
| *** flepied has quit IRC | 06:27 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/nodepool master: Implement an Azure driver https://review.openstack.org/554432 | 07:07 |
| *** chrnils has joined #zuul | 07:11 | |
| *** flepied has joined #zuul | 07:31 | |
| *** hashar has joined #zuul | 08:05 | |
| *** ekan has joined #zuul | 08:08 | |
| *** jpena|off is now known as jpena | 08:11 | |
| *** fbouliane has quit IRC | 08:31 | |
| *** fbouliane has joined #zuul | 08:33 | |
| *** electrofelix has joined #zuul | 08:38 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/nodepool master: Implement a VMWare driver https://review.openstack.org/554463 | 09:03 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/nodepool master: Implement an Azure driver https://review.openstack.org/554432 | 09:05 |
| *** ekan has quit IRC | 10:35 | |
| *** sshnaidm|afk is now known as sshnaidm|sick | 10:35 | |
| *** jpena is now known as jpena|brb | 10:38 | |
| *** JasonCL has quit IRC | 10:44 | |
| *** JasonCL has joined #zuul | 10:45 | |
| *** JasonCL has quit IRC | 10:50 | |
| *** jpena|brb is now known as jpena | 11:37 | |
| *** odyssey4me has quit IRC | 12:11 | |
| *** odyssey4me has joined #zuul | 12:11 | |
| *** dkranz has quit IRC | 12:19 | |
| *** Wei_Liu has quit IRC | 12:27 | |
| *** Wei_Liu has joined #zuul | 12:28 | |
| *** openstackgerrit has quit IRC | 12:33 | |
| *** rlandy has joined #zuul | 12:33 | |
| *** jpena is now known as jpena|lunch | 12:44 | |
| *** myoung|afk is now known as myoung|rover | 12:55 | |
| *** myoung|rover is now known as myoung|rover|mtg | 13:00 | |
| *** JasonCL has joined #zuul | 13:04 | |
| *** JasonCL has quit IRC | 13:09 | |
| dmsimard | tobiash: you're running executors on openshift right ? | 13:18 |
| dmsimard | tobiash: is that the actual (long lived) executors themselves ? or are you spawning a new executor for every job ? | 13:19 |
| *** dkranz has joined #zuul | 13:26 | |
| *** Wei_Liu has quit IRC | 13:29 | |
| *** Wei_Liu has joined #zuul | 13:29 | |
| pabelanger | I thought they were long lived, IIRC | 13:40 |
| mordred | dmsimard: long lived | 13:40 |
| *** jpena|lunch is now known as jpena | 13:42 | |
| *** openstackgerrit has joined #zuul | 13:49 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: dashboard: add /{tenant}/job.html page to display job details https://review.openstack.org/535545 | 13:49 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: dashboard: add /{tenant}/projects.html web page https://review.openstack.org/537870 | 13:49 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Fix indentation and renable the eslint rule https://review.openstack.org/545671 | 13:49 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Shift html templates into components https://review.openstack.org/551327 | 13:49 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Upgrade to webpack 4 https://review.openstack.org/551987 | 13:49 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Upgrade from angularjs (v1) to angular (v5) https://review.openstack.org/551989 | 13:49 |
| *** JasonCL has joined #zuul | 13:50 | |
| *** dmellado has quit IRC | 13:55 | |
| *** JasonCL has quit IRC | 13:55 | |
| dmsimard | pabelanger, mordred: tristanC and I were brainstorming about short lived executors the other day.. I feel like it might be a good fit. Has anyone brought that up before ? | 13:56 |
| dmsimard | It would be a way to spin a precise version of Ansible and Zuul, for example | 13:56 |
| pabelanger | yah, SpamapS / jlk have talked about that before using k8s | 13:57 |
| dmsimard | Increased isolation, etc. | 13:57 |
| pabelanger | yah | 13:57 |
| pabelanger | I know a few people are interested in that | 13:57 |
| pabelanger | it might have even been discussed in the bwrap spec for zuulv3 | 13:58 |
| pabelanger | on review.o.o | 13:58 |
| pabelanger | I think of the items is how do we deal with git repos, since there would be a cost / overhead in building them each time | 13:59 |
| dmsimard | If we want to "offer" something like this for the OpenStack community (assuming we have the features we need to be able to do that, of course) it would also mean standing up a k8s/OpenShift cluster I guess. We could also make it available to the community for actual workloads ? e.g, kolla-k8s | 14:00 |
| dmsimard | pabelanger: building what each time ? The executors is just runtime code right ? There's no persistence like the scheduler ? | 14:01 |
| mordred | dmsimard: cache | 14:02 |
| pabelanger | yah, git cache | 14:02 |
| mordred | dmsimard: the hard part about dynamic executors is git cache with the mergers | 14:02 |
| dmsimard | I didn't realize there was a git cache on the executors.. it's for speeding up zuul merger ? | 14:03 |
| mordred | dmsimard: dynamic executors is certainly a thing that's been discussed - but it's a thing we need to dig in on the design of because there are some hard parts | 14:03 |
| mordred | dmsimard: yah - if we didn't have persistent cache on the executors zuul would completely not scale to openstack size | 14:04 |
| dmsimard | Is that an actual process ? Or is the cache simply "warmed up" by the fact that the executor just merges things all the time ? | 14:05 |
| mordred | dmsimard: the second | 14:06 |
| mordred | dmsimard: also - we've got a bunch of notes on zuul+k8s/openshift that need to be written up and shared out | 14:06 |
| mordred | dmsimard: let's circle up and get those shared first before you get too far down that path | 14:07 |
| mordred | dmsimard: since I think we're likley sharing some thoughts | 14:07 |
| *** JasonCL has joined #zuul | 14:07 | |
| *** JasonCL has quit IRC | 14:07 | |
| dmsimard | oh, I'm nowhere deep.. it's just an idea that I had and tristanC had also thought about it before | 14:07 |
| mordred | dmsimard: ++ | 14:07 |
| dmsimard | It's not like I have time to work on this or anything :( | 14:07 |
| *** JasonCL has joined #zuul | 14:08 | |
| mordred | dmsimard: :( | 14:08 |
| *** JasonCL has quit IRC | 14:11 | |
| *** JasonCL has joined #zuul | 14:12 | |
| *** elyezer has joined #zuul | 14:14 | |
| *** myoung|rover|mtg is now known as myoung|rover | 14:23 | |
| tristanC | we could probably uses a configmap per build to store all the git data, but we likely still need to spin a pod per playbook to isolate secrets from untrusted playbooks | 14:23 |
| dmsimard | Or maybe there could be a central pod for git cache? I dunno. I'll wait for the write up :) | 14:26 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul-jobs master: Uninstall and reinstall siblings one at a time https://review.openstack.org/554297 | 14:27 |
| dmsimard | a read-only git cache volume shared on all pods? | 14:29 |
| * dmsimard stops thinking about this | 14:29 | |
| tristanC | dmsimard: let's not distract #zuul with non 3.0 release topic :-) | 14:30 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Upgrade from angularjs (v1) to angular (v5) https://review.openstack.org/551989 | 14:37 |
| mordred | tristanC: I don't know if you've gotten a chance to look at ^^ - it's also a non-3.0 release topic, so no rush, but it's definitely a patch I'd want your review on before landing | 14:38 |
| mordred | tristanC: it'll require an update to how you build the js stack for softwarefactory (if, that is, the latest version of the patch actually works) | 14:39 |
| tristanC | mordred: i'll give it a try then. note that the zuul-webui package we use is built using the same command as zuul-build-dashboard | 14:43 |
| mordred | tristanC: awesome - so the main difference for softwarefactory is that you'll need to add a --define ZUUL_BASE_HREF="/zuul" to the build:dist command | 14:44 |
| *** dmellado has joined #zuul | 14:54 | |
| *** electrofelix has quit IRC | 14:54 | |
| *** dmellado has quit IRC | 14:56 | |
| *** JasonCL has quit IRC | 14:59 | |
| *** JasonCL has joined #zuul | 15:00 | |
| *** dmellado has joined #zuul | 15:01 | |
| *** dmellado has quit IRC | 15:04 | |
| *** dmellado has joined #zuul | 15:05 | |
| *** flepied_ has joined #zuul | 15:12 | |
| *** JasonCL has quit IRC | 15:13 | |
| *** JasonCL has joined #zuul | 15:14 | |
| *** flepied has quit IRC | 15:15 | |
| *** dmellado has quit IRC | 15:16 | |
| *** JasonCL has quit IRC | 15:20 | |
| *** JasonCL has joined #zuul | 15:20 | |
| *** JasonCL has quit IRC | 15:28 | |
| *** JasonCL has joined #zuul | 15:28 | |
| *** JasonCL has quit IRC | 15:29 | |
| *** JasonCL has joined #zuul | 15:30 | |
| *** JasonCL has quit IRC | 15:32 | |
| *** JasonCL has joined #zuul | 15:32 | |
| *** JasonCL has quit IRC | 15:33 | |
| *** bhavik1 has joined #zuul | 15:34 | |
| *** JasonCL has joined #zuul | 15:34 | |
| *** JasonCL has quit IRC | 15:46 | |
| *** bhavik1 has quit IRC | 15:46 | |
| *** JasonCL has joined #zuul | 15:48 | |
| clarkb | Yesterday when debugging zuul's config loading I noticed this exception http://paste.openstack.org/show/706233/ I thought I could quickly debug it this morning but as I dig into it more I'm just more confused | 15:49 |
| clarkb | the code explicitly checks if that value is None but seems to move ahead and lookup the None value anyways (based on the log message saying it is looking for a hold for None) | 15:50 |
| clarkb | I'm left thinking either that None valid is being stringified before the is not None check or we are running older scheduler code and this was somehow fixed along the way? | 15:50 |
| clarkb | oh hey | 15:50 |
| clarkb | git log -p says we are running older scheduler code | 15:51 |
| clarkb | except the traceback shows the check for is not None in it | 15:51 |
| clarkb | corvus: Shrews ^ any ideas on that? | 15:51 |
| corvus | clarkb: i think that's fixed in master | 15:52 |
| clarkb | there is 876d55fd5420944a5169f32d247984cc411d87c2 | 15:53 |
| clarkb | which added the is not None check | 15:53 |
| corvus | clarkb: i believe the traceback shows the code on the filesystem, not the code that's running | 15:53 |
| clarkb | corvus: aha | 15:53 |
| clarkb | that also explains why that line appears to be off by one | 15:53 |
| corvus | yup | 15:54 |
| clarkb | ok I guess I'll want to restart zuul (and reboot the host) sometime soon to make that annoying exception go away (it makes it hard to filter for exceptions when there are many "expected" exceptions) | 15:54 |
| *** JasonCL has quit IRC | 15:55 | |
| openstackgerrit | Andreas Jaeger proposed openstack-infra/zuul-jobs master: Uninstall and reinstall siblings one at a time https://review.openstack.org/554297 | 16:01 |
| *** jamielennox has quit IRC | 16:01 | |
| clarkb | not to jinx it but we haven't had to restart zuul scheduler in a while | 16:03 |
| *** myoung|rover is now known as myoung|rover|mtg | 16:05 | |
| *** JasonCL has joined #zuul | 16:09 | |
| *** JasonCL has quit IRC | 16:13 | |
| *** hashar is now known as hasharAway | 16:15 | |
| *** JasonCL has joined #zuul | 16:20 | |
| SpamapS | dmsimard: pabelanger mordred yes yes yes to k8s for executors. And the git cache is actually pretty easy to solve with volumes or some other methods. | 16:22 |
| SpamapS | It would be kind of weird to use bwrap inside k8s though... so that might be one valid use case for nullwrap. | 16:22 |
| clarkb | fwiw tobiash does run his executors under k8s | 16:23 |
| clarkb | so I'm not sure there is more that needs to be solved directly? | 16:23 |
| clarkb | SpamapS: in tobiash apparently a new openshift had to be spun up that allowed the executors to run in privileged containers | 16:23 |
| clarkb | *in tobiash's case | 16:23 |
| clarkb | so it is doable but yes clunky | 16:23 |
| mordred | clarkb: I think there are two takes - one is "run zuul executors in k8s as a human like tobiash is currently doing" - and "have zuul manage its executors in an elastic fashion" | 16:24 |
| *** flepied_ has quit IRC | 16:24 | |
| clarkb | I see, that to me is less about containers and k8s in particular and more make zuul scale up and down executors | 16:25 |
| mordred | clarkb: yes, that's right | 16:25 |
| clarkb | since the containers and k8s part (mostly) works | 16:25 |
| *** elyezer has quit IRC | 16:25 | |
| mordred | clarkb: using k8s is one way in which zuul could implementing zuul-managed-elastic-executors, but is certainly not the only way | 16:25 |
| *** JasonCL has quit IRC | 16:26 | |
| clarkb | mordred: in particular nodepool is currently a required component and happens to scale up and down based on demand | 16:26 |
| pabelanger | auto-scaling of zuul services, would be so freaking awesome! :D | 16:26 |
| pabelanger | just saying | 16:26 |
| clarkb | (and then you can point your nodepool at k8s or openstack or whatever | 16:26 |
| clarkb | but I might be overthinking that | 16:26 |
| mordred | given that one can already, as clarkb points out, run executors in k8s successfully, I'd argue that we should focus attention more on k8s build resources and distributed scheduler - since those are existing pain points - before we tackle managed-elastic-executors | 16:26 |
| *** JasonCL has joined #zuul | 16:27 | |
| pabelanger | wfm | 16:27 |
| corvus | these conversations will be much more productive if we can have them in a structured manner. once we get the release out of the door, i'll write up a summary of all the discussions we've had about this so far in spec form, and we can iterate from there | 16:27 |
| mordred | corvus: ++ | 16:27 |
| mordred | clarkb, pabelanger: have a second to be a second set of eyes on a build failure real quick? | 16:28 |
| clarkb | mordred: sur | 16:29 |
| mordred | clarkb, pabelanger: nevermind- prepping the links to paste to you made me see the issue differently - so thanks! | 16:29 |
| *** JasonCL has quit IRC | 16:31 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Upgrade from angularjs (v1) to angular (v5) https://review.openstack.org/551989 | 16:33 |
| *** jamielennox has joined #zuul | 16:38 | |
| *** elyezer has joined #zuul | 16:38 | |
| mordred | tristanC: you have a patch somewhere to add swagger docs for the api yeah? | 16:38 |
| *** JasonCL has joined #zuul | 16:39 | |
| mordred | tristanC: found it - also, nice use of depends-on a gh pr :) https://review.openstack.org/#/c/535541/ | 16:39 |
| *** JasonCL has quit IRC | 16:43 | |
| *** JasonCL has joined #zuul | 16:45 | |
| *** JasonCL has quit IRC | 16:51 | |
| *** JasonCL has joined #zuul | 16:58 | |
| *** JasonCL has joined #zuul | 16:58 | |
| *** myoung|rover|mtg is now known as myoung|rover | 17:01 | |
| *** flepied_ has joined #zuul | 17:11 | |
| *** JasonCL has quit IRC | 17:26 | |
| *** JasonCL has joined #zuul | 17:27 | |
| *** JasonCL has quit IRC | 17:27 | |
| *** JasonCL has joined #zuul | 17:28 | |
| *** JasonCL has quit IRC | 17:31 | |
| tobiash | I also like the idea of running an executor per job but have to second mordred that build resources and scale out scheduler are probably more important atm | 17:31 |
| *** pwhalen has quit IRC | 17:34 | |
| *** JasonCL has joined #zuul | 17:34 | |
| *** pwhalen has joined #zuul | 17:38 | |
| *** pwhalen has joined #zuul | 17:38 | |
| jlk | executor per job sounds cool, but what does it really buy us? | 17:39 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Upgrade from angularjs (v1) to angular (v5) https://review.openstack.org/551989 | 17:41 |
| tobiash | right, let me rephrase, an ansible container per job | 17:42 |
| tobiash | is there currently some v3.0 stuff to review? | 17:43 |
| *** jimi|ansible has quit IRC | 17:46 | |
| corvus | tobiash: i don't think so -- next things we're waiting on is the security patch, and i'm going to write some remote tests for the current log stream system | 17:47 |
| corvus | (and possibly a change from mordred related to js stuff if he determines we need it before 3.0) | 17:47 |
| tobiash | corvus: I'll be working on this tomorrow, it's too late today for doing such kind of things ;) | 17:48 |
| corvus | yeah, don't patch vulnerabilities when tired :) | 17:48 |
| *** JasonCL has quit IRC | 17:53 | |
| SpamapS | jlk: since the cache for an executor is critical to reducing job latency and repeated work, not much. But if one can leverage a merger's cache, it makes the executor more elastic. | 17:54 |
| SpamapS | Def a good thing to get into that spec. :) | 17:54 |
| jlk | So, this exists https://github.com/pricing/business-hosted it's GitHub hosted for business. An upgrade for an organization, which adds SAML sign on and stuff. I'm not sure if it changes anything with the API, but I should probably dig into it a bit more. | 17:54 |
| SpamapS | For me, at small scale, it doesn't help at all. :) | 17:54 |
| jlk | SpamapS: but what is the benefit of the elasticity? | 17:55 |
| SpamapS | jlk: public cloud cost | 17:55 |
| SpamapS | if you are having a big day and running a lot of concurrent jobs.. spend money.. if not.. spend nothing. | 17:55 |
| SpamapS | can do it with a nodepool of executors too ;) | 17:55 |
| jlk | how far are we from that now though? What does it take to spin up an executor? | 17:56 |
| SpamapS | Not much | 17:56 |
| jlk | I thought you could spin one up with out having to "notify" or touch other parts of the system, like it just plugs into the pool | 17:56 |
| SpamapS | especially if you have containerized it and have a kubernetes | 17:56 |
| jlk | and if it drops off, same, that's just one less worker | 17:56 |
| jlk | yeah, or at least have a good "boot this image that self-configures" system | 17:57 |
| SpamapS | another thing though, for jobs that can run without VM | 17:57 |
| SpamapS | spin single use executor.. run job local.. executor drops. | 17:57 |
| jlk | yeah... I'm somewhat less comfortable with those kind of jobs | 17:57 |
| SpamapS | unittest jobs would fall in that category | 17:57 |
| jlk | execution on MY infra vs execution on the runtime capacity I provide | 17:57 |
| jlk | I wouldn't necessarily mix the pool I use to RUN zuul with the pool I use to run zuul JOBS | 17:58 |
| SpamapS | agreed | 17:58 |
| jlk | zuul jobs may run in somebody else's infra | 17:58 |
| SpamapS | single use executors would be runtime | 17:58 |
| jlk | not mine | 17:58 |
| SpamapS | but they might just run inside a kubernetes | 17:58 |
| jlk | sortof? zuul executors have credentials | 17:58 |
| jlk | which means I may want them on MY infra | 17:58 |
| jlk | and away from the infra where rando code runs | 17:59 |
| SpamapS | Yeah, I am with you. | 17:59 |
| SpamapS | Just falling into devil's advocate | 18:00 |
| SpamapS | The reality is, the thing we need a spec for, and that I would like, is a way to run a job without a VM and without making it trusted. | 18:00 |
| SpamapS | I don't much like the idea of spinning up sshd's in containers, but it does sort of achieve that. | 18:00 |
| clarkb | SpamapS: in theory something like nodepool + k8s and ansible k8s connector does that for you right? | 18:01 |
| jlk | SpamapS: right, I don't want ssh in the container either | 18:03 |
| jlk | I think the container shouldn't expose any ports, just boot. We should use k8s API to cause things to happen inside the container. | 18:03 |
| SpamapS | clarkb: nodepool + k8s? | 18:03 |
| clarkb | SpamapS: nodepool assigning compute resources out of a k8s cluster | 18:04 |
| openstackgerrit | Fabien Boucher proposed openstack-infra/zuul master: Add bootstrap alert message when conf loading errors exists https://review.openstack.org/554652 | 18:04 |
| jlk | SpamapS: we met at PTG to talk through what it would look like to have Zuul + Nodepool work with k8s as a capacity source | 18:04 |
| SpamapS | clarkb: eh, that's not really how k8s works. k8s is pretty much a nodepool. | 18:04 |
| jlk | Somebody has the notes on it online I think. | 18:04 |
| jlk | Nodepool would ask for a pod of certain types. | 18:04 |
| clarkb | jlk: corvus mentioned earlier that there would be a writeup, I don't thik that has been done yet | 18:04 |
| corvus | yep. i'll do it soon. | 18:05 |
| corvus | or if someone wants to write those remote tests for log streaming, i could write up the spec now. :) | 18:05 |
| jlk | hahah | 18:05 |
| SpamapS | One thought I have is to just have jobs that have a k8s-object instead of a nodeset. | 18:05 |
| jlk | I suggested a pod, with at least 2 containers | 18:05 |
| *** myoung|rover is now known as myoung|biab | 18:05 | |
| jlk | one for git content (sidecar) and one for the job itself | 18:05 |
| jlk | the sidecar so that we can push up the git content and the job container would have access to the source code | 18:06 |
| SpamapS | Teach either nodepool or zuul-scheduler to talk to k8s. Hand jobs the kubeconfig they need to access the deployment they asked for. | 18:06 |
| jlk | correct, nodepool would talk to k8s | 18:06 |
| jlk | But we also talked about two different use cases | 18:06 |
| SpamapS | yeah we're thinking similarly. | 18:06 |
| jlk | 1) a job wants to natively talk k8s | 18:06 |
| jlk | and 2) a job just wants to run something and doesn't care where | 18:06 |
| SpamapS | I care a lot more about 2) | 18:07 |
| jlk | handing jobs direct access to k8s is... dangerous | 18:07 |
| SpamapS | but really 2) is enabled by a parent that is 1) | 18:07 |
| *** JasonCL has joined #zuul | 18:07 | |
| jlk | k8s isn't well set up for multi-tenant | 18:07 |
| SpamapS | yeah, untrusted would only be able to do a k8s_exec or something like that. | 18:08 |
| jlk | SpamapS: well, we can teach Zuul's ansible how to run tasks on a "host" that is a k8s pod/container. That gets you to 2 without necessarily exposing 1 | 18:08 |
| SpamapS | gah | 18:08 |
| SpamapS | this rat hole | 18:08 |
| SpamapS | it's SO NICE | 18:08 |
| SpamapS | like I want to stay here | 18:08 |
| SpamapS | but | 18:08 |
| SpamapS | the rats | 18:08 |
| mordred | yah | 18:08 |
| * SpamapS has to go back to work ;) | 18:08 | |
| clarkb | and if you use gke every pod has root! | 18:08 |
| jlk | that's totally fine.. | 18:10 |
| jlk | it's not like somebody could write a job that spawns a outbound connection to create a reverse proxy into the container, granting shell access to the nefarious user... | 18:10 |
| clarkb | jlk: clsuter root/admin I mean | 18:10 |
| jlk | right | 18:11 |
| clarkb | the default in gke for ease of use is that all pods have cluster admin aiui and all gce VMs (that run gke containers) have account admin in the cloud | 18:11 |
| jlk | I guess you could just write jobs to poke at the k8s API from within the container | 18:11 |
| clarkb | jlk: yes that | 18:11 |
| clarkb | and potentially to your cloud account | 18:11 |
| jlk | thisisfine.jpg | 18:12 |
| *** JasonCL has quit IRC | 18:12 | |
| *** JasonCL has joined #zuul | 18:13 | |
| fungi | dunno if anybody noticed yet (i'm not caught up on scrollback) but as of today zuul is a free agent and officially no longer under openstack tc governance | 18:15 |
| *** JasonCL has quit IRC | 18:17 | |
| *** JasonCL has joined #zuul | 18:17 | |
| fungi | there is no openstack, only zuul | 18:17 |
| jlk | freebird.mp3 | 18:17 |
| SpamapS | fungi: oh my | 18:18 |
| *** jpena is now known as jpena|away | 18:18 | |
| *** JasonCL has quit IRC | 18:19 | |
| *** JasonCL has joined #zuul | 18:20 | |
| *** JasonCL_ has joined #zuul | 18:22 | |
| *** harlowja has joined #zuul | 18:23 | |
| *** JasonCL has quit IRC | 18:25 | |
| *** JasonCL_ has quit IRC | 18:26 | |
| *** chrnils has quit IRC | 18:51 | |
| *** JasonCL has joined #zuul | 18:58 | |
| *** JasonCL has quit IRC | 18:59 | |
| *** JasonCL has joined #zuul | 19:01 | |
| *** JasonCL has quit IRC | 19:03 | |
| *** JasonCL has joined #zuul | 19:03 | |
| openstackgerrit | Fabien Boucher proposed openstack-infra/zuul master: Add a loading_errors info to {tenant}/info endpoint https://review.openstack.org/553873 | 19:04 |
| *** JasonCL has quit IRC | 19:07 | |
| *** myoung|biab is now known as myoung|rover | 19:14 | |
| *** JasonCL has joined #zuul | 19:43 | |
| *** JasonCL has quit IRC | 19:47 | |
| *** smyers has quit IRC | 20:02 | |
| *** smyers has joined #zuul | 20:03 | |
| *** smyers has quit IRC | 20:18 | |
| *** smyers has joined #zuul | 20:19 | |
| *** JasonCL has joined #zuul | 20:21 | |
| *** JasonCL has quit IRC | 20:30 | |
| *** openstackgerrit has quit IRC | 20:33 | |
| *** JasonCL has joined #zuul | 20:55 | |
| *** dmellado has joined #zuul | 20:56 | |
| *** dmellado has quit IRC | 20:58 | |
| *** JasonCL has quit IRC | 21:00 | |
| *** JasonCL has joined #zuul | 21:00 | |
| *** dmellado has joined #zuul | 21:02 | |
| *** JasonCL has quit IRC | 21:02 | |
| *** JasonCL has joined #zuul | 21:04 | |
| *** dmellado has quit IRC | 21:08 | |
| *** dmellado has joined #zuul | 21:12 | |
| *** dmellado has quit IRC | 21:13 | |
| *** myoung|rover is now known as myoung|off | 21:13 | |
| *** myoung|off is now known as myoung|bbl | 21:13 | |
| *** dkranz has quit IRC | 21:14 | |
| *** dmellado has joined #zuul | 21:15 | |
| *** myoung|bbl is now known as myoung|rover | 21:16 | |
| *** JasonCL has quit IRC | 21:16 | |
| *** JasonCL has joined #zuul | 21:16 | |
| *** dmellado has quit IRC | 21:32 | |
| *** ssbarnea has quit IRC | 21:46 | |
| *** myoung|rover is now known as myoung|afk | 21:46 | |
| *** openstackgerrit has joined #zuul | 21:48 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Add zuul-stream remote tests https://review.openstack.org/554714 | 21:48 |
| *** ssbarnea has joined #zuul | 21:52 | |
| *** dmellado has joined #zuul | 22:04 | |
| *** CrayZee has quit IRC | 22:41 | |
| *** harlowja has quit IRC | 22:57 | |
| *** hasharAway has quit IRC | 23:01 | |
| *** rlandy is now known as rlandy|bbl | 23:03 | |
| *** jpena|away is now known as jpena|off | 23:21 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Add zuul-stream remote tests https://review.openstack.org/554714 | 23:23 |
| *** harlowja has joined #zuul | 23:49 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!