| kfox1111 | it would be nice if patch 334115 was unnessisary. | 00:00 |
|---|---|---|
| patchbot | kfox1111: https://review.openstack.org/#/c/334115/ - kolla - Horizon Frontend - (for kolla-kubernetes) | 00:00 |
| *** Jeffrey4l_ has joined #openstack-kolla | 00:00 | |
| dcwangmit01_ | Still reading the patches. | 00:02 |
| kfox1111 | k. thanks for having a look. | 00:02 |
| wirehead_ | Yeah, like, I keep feeling like the way your patches are looking right now isn’t quite the right route, but I’m still at a loss for what the better way is. | 00:02 |
| dcwangmit01_ | It most definitely is a problem we need to solve. | 00:02 |
| kfox1111 | yeah. I get that. I am at a loss too for a better way. :/ | 00:02 |
| wirehead_ | Like, in theory, that’s what namespaces are for. | 00:02 |
| kfox1111 | kindof? | 00:03 |
| kfox1111 | namespaces are for discovery too, and putting it in a different namespace then breaks discovery. | 00:03 |
| kfox1111 | so it may or may not be easier in the end. | 00:03 |
| wirehead_ | Well, if you don’t use a namespace, then you also have to change the selector. | 00:04 |
| kfox1111 | either way though, haproxy in front is needed, and it needs access to the haproxy image variable. | 00:04 |
| wirehead_ | If you do use a namespace, you need to go from mariadb to mariadb.openstack | 00:04 |
| kfox1111 | right, which was done. the service for horizon also includes the version. | 00:05 |
| kfox1111 | and the haproxy does the selection. | 00:05 |
| kfox1111 | and keystone | 00:05 |
| kfox1111 | and if they start supporting more CORS calls, each of those endpoints? | 00:05 |
| kfox1111 | though, maybe keystone endpoints handle that case... | 00:05 |
| kfox1111 | I'd kind of rather have a mariadb per horizon version too though. | 00:06 |
| kfox1111 | so maybe the only thing it needs is a config back to keystone? | 00:06 |
| kfox1111 | I' | 00:06 |
| kfox1111 | I'd caution about using namespaces explicitly though like mariadb.openstack. | 00:07 |
| kfox1111 | I think it would be facinating to use kolla-kubernetes running on one cluster to deploy the control plains for multiple regions. | 00:07 |
| wirehead_ | yeah, it feels like we need a higher-level map. | 00:07 |
| kfox1111 | yeah. | 00:08 |
| kfox1111 | if everything was in one namespace, that isn't really an issue. | 00:08 |
| *** salv-orlando has joined #openstack-kolla | 00:08 | |
| *** harlowja has quit IRC | 00:10 | |
| dcwangmit01_ | kfox1111: So lets say you have 2 different versions of horizon running. How is versioning done on the API side? Does webclient1 access horizon/v1 and webclient2 access horizon/v2? | 00:11 |
| dcwangmit01_ | I mean, do the paths change between versions, or are they the same? | 00:11 |
| kfox1111 | dcwangmit01_: the first time you access horizon, the haproxy gives you a cookie with the default version specified in the config. | 00:11 |
| dcwangmit01_ | Oh, then the client hands back the cookie with every request | 00:12 |
| kfox1111 | then subsequent visits by the browser return the cookie, and haproxy directs that traffic to that version of the cluster. | 00:12 |
| kfox1111 | yeah. cause a client can disconnect between requests for images, html, js, etc. | 00:12 |
| dcwangmit01_ | Did you make the patch against horizon to do so, or is this purely in the frontend patch | 00:12 |
| kfox1111 | and if one page load gets spread across multiple horizon versions, things can break in odd ways. | 00:12 |
| kfox1111 | purely frontend. horizon needs no changes for it to work. | 00:12 |
| kfox1111 | https://review.openstack.org/#/c/334115/3/ansible/roles/horizon-frontend/templates/haproxy.cfg.j2 | 00:13 |
| patchbot | kfox1111: patch 334115 - kolla - Horizon Frontend - (for kolla-kubernetes) | 00:13 |
| kfox1111 | line 24/25 creates the pool of servers for all horizon versions configured, | 00:13 |
| kfox1111 | sets the default to the one specified when there is no cookie, and sets the rest to backups. | 00:14 |
| kfox1111 | which get traffic directed to them when the cookie is set. | 00:14 |
| kfox1111 | so, with this, you can deploy horizon 1, default 1, and horizon 2. | 00:14 |
| kfox1111 | all users will go to horizon 1. | 00:14 |
| kfox1111 | you can tweak the cookie in your browser to 2 for testing and see the new version. | 00:14 |
| kfox1111 | once you are ok with it, you can set the default to 2. | 00:15 |
| dcwangmit01_ | What about just pointing the browser to a new url or DNS name? | 00:15 |
| kfox1111 | all new sessions will go there, and all existing ones will still go to 1. | 00:15 |
| dcwangmit01_ | We could do that with Ingress | 00:15 |
| kfox1111 | because web sites are stateless. | 00:15 |
| kfox1111 | as soon as you update dns, a page loading that was starting with v1, | 00:15 |
| kfox1111 | would get parts of v2. | 00:16 |
| dcwangmit01_ | Is the goal to be able to sanity test the 2nd version before activating it. Or do you have other goals? | 00:16 |
| kfox1111 | thats part of it. | 00:16 |
| kfox1111 | say I have a user who logs into the system. | 00:16 |
| kfox1111 | the click on sahara. | 00:16 |
| kfox1111 | they now have a page of links that are somewhat specific to the horizon version they are on. | 00:16 |
| kfox1111 | if they click on a link, they expect it to work, instead of taking them to a 404 (say, if the new horizon version moved stuff around on the server) | 00:17 |
| wirehead_ | Yeah, like you can use a deployment and create metadata that the haproxy would be able to sort users on. | 00:17 |
| wirehead_ | except that doesn’t solve the other larger ‘I want a map’ problem; | 00:17 |
| kfox1111 | what this does, is allow you to run v2 for new connections, and v1 for old established users until their browsing session times out. | 00:18 |
| kfox1111 | then, you can delete the old version once drained. | 00:18 |
| dcwangmit01_ | Oh. So basically you want to support rolling update | 00:18 |
| *** britthouser has joined #openstack-kolla | 00:18 | |
| kfox1111 | yes. rolling upgrade. | 00:18 |
| kfox1111 | its much much harder with websites. | 00:18 |
| kfox1111 | with an api like nova, you can just use a k8s deployment. | 00:18 |
| kfox1111 | cause a rest api call is atomic. | 00:19 |
| kfox1111 | but with a website, lots of pieces are being loaded in parallel, statelessly. | 00:19 |
| dcwangmit01_ | Ok, I think that is different case than having 2 versions fully accessible. One for prod, and one for testing | 00:19 |
| kfox1111 | so this patch set adds a state cookie to direct you to a version that you started with. | 00:19 |
| wirehead_ | Well if you create a metadata key for ‘version = v1’ on the deployment, then do a rolling slow deployment of a new config that has a metadata key for ‘version = v2’ on the deployment, with the haproxy state cookie still works. | 00:19 |
| kfox1111 | kind of... | 00:19 |
| wirehead_ | But only for going to v1 to v2. | 00:19 |
| wirehead_ | Not for getting some users from v1 to v2 and others from v1 to v2b. | 00:20 |
| kfox1111 | You start with v1 being the default, and launching v2. then you test v2. once happy, you premote the default to v2. | 00:20 |
| kfox1111 | old users stay on v2, and new ones hit v2 right away. | 00:20 |
| kfox1111 | then you wait for all v1 sessions to time out (say, 24 hours) | 00:20 |
| kfox1111 | then you can safely delete v1. | 00:20 |
| kfox1111 | no user would hten know the rolling upgrade happened. | 00:20 |
| kfox1111 | other then maybe a login banner changes when they next login. | 00:20 |
| kfox1111 | wirehead_: right. this might be extendable to handle that case too, but not the way it is now. | 00:21 |
| wirehead_ | kfox1111: yeah, you can leave a deployment rollout paused for an extended time. | 00:22 |
| *** salv-orlando has quit IRC | 00:22 | |
| kfox1111 | hmm..... | 00:22 |
| kfox1111 | so your saying have the deployment paused switching from one to the other to get both? interesting.... | 00:22 |
| kfox1111 | but how would you funnel access to the old or the new? | 00:23 |
| wirehead_ | Metadata with the version number. | 00:23 |
| kfox1111 | youd have to replace the service at the same time, which I don't think a deployoment can do? | 00:23 |
| kfox1111 | interesting... | 00:23 |
| wirehead_ | You’d create a ‘-new’ service and an ‘-old’ service. | 00:23 |
| wirehead_ | I mean, for just the case of proper Horizon upgrades, that’s unambiguously the right way to do things… just a LOT of moving peices. | 00:24 |
| kfox1111 | it would be less steps on the deployment side, but more on the haproxy side. | 00:24 |
| kfox1111 | ok, what about this case... | 00:24 |
| dcwangmit01_ | I kind of want to support this without the frontend (without the need to run an HAProxy and set cookies). There must be some trick we can do with running both versions, each version exposed via External Service, and doing some tricks with changing DNS CNAMES. | 00:25 |
| kfox1111 | a horizon version is made up of the following pieces. "horizon, gallera cluster, memcached." | 00:25 |
| dcwangmit01_ | How often do the clients re-resolve the DNS | 00:25 |
| wirehead_ | never when you need them to, dcwangmit01_, that’s why k8s implements a VIP. | 00:25 |
| kfox1111 | dns changes suck. | 00:25 |
| kfox1111 | its like a minimum 24 hour ordeal here. :/ | 00:26 |
| wirehead_ | All of the way back to the days when home.netscape.com was silently magically round-robbined in the netscape client. :P | 00:26 |
| kfox1111 | I kind of don't like using a cookie, but you must have cookies enabled for horizon authentication to work at all, so its not so bad. | 00:27 |
| dcwangmit01_ | Or maybe we can patch against Kubernetes Ingress | 00:28 |
| kfox1111 | but back to my previous thing, if the version is done at the name level, "horizon, gallera cluster, memcached" can all be done as a set, with a bootstrap and service. | 00:28 |
| kfox1111 | while you could maybe make 3 deployments and upgrade/pause them together, that may get really ugly. | 00:28 |
| kfox1111 | dcwangmit01_: yeah, ingress might be a good long term solution. I think its quite a ways out though. | 00:29 |
| *** sogabe has joined #openstack-kolla | 00:29 | |
| dcwangmit01_ | This actually might be an ingress patch. Patch to say... don't break old connnections | 00:31 |
| dcwangmit01_ | It may not break old connections right now. Configs I belive ar reloaded without restarting the ingress controller. Maybe a SIGHUP | 00:31 |
| dcwangmit01_ | In that case, we update the Ingress configs to the new service | 00:31 |
| kfox1111 | the problem is there is no connections. | 00:31 |
| dcwangmit01_ | HTTP 1.1? | 00:31 |
| kfox1111 | every http call is a different connection by the browser. | 00:31 |
| wirehead_ | well, you can’t guarantee that keepalive will keep alive. | 00:32 |
| kfox1111 | so state must be kept somewhere else. | 00:32 |
| dcwangmit01_ | That was back in HTTP 1.0 | 00:32 |
| kfox1111 | wirehead_: right. | 00:32 |
| kfox1111 | and in fact, with lb's its preferable to keep keepalive very short. | 00:32 |
| kfox1111 | otherwise your servers get more unbalanced. | 00:32 |
| wirehead_ | I’d like to think this is a cross-kube issue. | 00:32 |
| kfox1111 | yeah. ansible has it too. | 00:33 |
| wirehead_ | It’s just that we can’t guarantee that it’ll happen when we want it. :) | 00:33 |
| kfox1111 | its just much easier to solve with the way kolla-kubernetes is layed out and how kubernetes works then with kolla-ansible. | 00:33 |
| kfox1111 | at the moment. | 00:33 |
| dcwangmit01_ | Sometimes we might want LBs to do session-affinity and rely on health checks. Would be better for caching. | 00:34 |
| kfox1111 | yeah. I think that would be good for k8s in general. | 00:35 |
| kfox1111 | I think horizon's probably the only openstack service that might beneifit from it though? | 00:35 |
| *** sogabe has quit IRC | 00:36 | |
| *** sogabe has joined #openstack-kolla | 00:37 | |
| wirehead_ | Yeah, this is fairly Horizon-specific. | 00:39 |
| *** daneyon has joined #openstack-kolla | 00:39 | |
| wirehead_ | Most everything else is just a REST service. | 00:39 |
| kfox1111 | yeah. | 00:39 |
| wirehead_ | But every single web app that you deploy atop Kubernetes is going to experience this. | 00:39 |
| kfox1111 | true. | 00:40 |
| kfox1111 | but its going to need something much more complex then a deployment to solve it. :/ | 00:40 |
| kfox1111 | so I think thats likely at least a year out. | 00:40 |
| wirehead_ | https://github.com/deis/router | 00:41 |
| kfox1111 | like in my horizon example, you might want a memcached, maraidb cluster and web app to all get rolling upgraded together to maintain that state. | 00:41 |
| *** jtriley has joined #openstack-kolla | 00:41 | |
| *** PyroMani has quit IRC | 00:42 | |
| kfox1111 | wirehead_: interesting. I'll have a look. | 00:42 |
| kfox1111 | wirehead_: did you see the new helm? its very interesting. | 00:42 |
| kfox1111 | https://github.com/kubernetes/helm | 00:42 |
| kfox1111 | I've got to head out. thanks for the discussion. | 00:42 |
| wirehead_ | kfox1111: I was playing with Helm before I started hacking on kolla-kubernetes. This is probably why I have a weird POV. | 00:43 |
| dcwangmit01_ | kfox1111: We'll talk later | 00:43 |
| wirehead_ | Thanks, kfox1111 :) | 00:43 |
| dcwangmit01_ | kfox1111: Are you East or West coast? | 00:43 |
| *** daneyon has quit IRC | 00:44 | |
| kfox1111 | west coast. | 00:44 |
| kfox1111 | yeah. sounds good. :) | 00:45 |
| kfox1111 | the new helm looks like it addresses all my concerns with the previous one. | 00:45 |
| kfox1111 | it would be interesting to see how much of openstack could be deployed with it. :) | 00:45 |
| *** diogogmt has quit IRC | 00:45 | |
| *** salv-orlando has joined #openstack-kolla | 00:48 | |
| rhallisey | wassup | 00:48 |
| *** PyroMani has joined #openstack-kolla | 00:52 | |
| rhallisey | kfox1111, still around? | 00:53 |
| *** salv-orlando has quit IRC | 01:03 | |
| *** aernhart has quit IRC | 01:04 | |
| *** huikang has joined #openstack-kolla | 01:08 | |
| *** harlowja has joined #openstack-kolla | 01:09 | |
| *** zhiwei has joined #openstack-kolla | 01:22 | |
| openstackgerrit | David Wang proposed openstack/kolla-kubernetes: POC (do no merge): Kolla-Kubernetes multi-node persistence pattern https://review.openstack.org/335215 | 01:27 |
| openstackgerrit | David Wang proposed openstack/kolla-kubernetes: Refactor, speed up, and fix bugs in jinja template dict generation https://review.openstack.org/336745 | 01:27 |
| *** salv-orlando has joined #openstack-kolla | 01:29 | |
| openstackgerrit | Hui Kang proposed openstack/kolla: Add etcd Docker container https://review.openstack.org/298451 | 01:30 |
| *** mark-casey has joined #openstack-kolla | 01:31 | |
| *** banix has joined #openstack-kolla | 01:31 | |
| mark-casey | is kolla-toolbox intended to be able to run build.py inside it? or just kolla-ansible? or have I fully misunderstood its purpose? | 01:36 |
| *** diogogmt has joined #openstack-kolla | 01:37 | |
| *** zhurong has joined #openstack-kolla | 01:39 | |
| *** salv-orlando has quit IRC | 01:45 | |
| *** ravig has quit IRC | 01:54 | |
| *** phuongnh has joined #openstack-kolla | 02:01 | |
| *** huikang_ has joined #openstack-kolla | 02:07 | |
| *** beekhof-ng-v2 has quit IRC | 02:10 | |
| *** huikang has quit IRC | 02:11 | |
| *** salv-orlando has joined #openstack-kolla | 02:11 | |
| *** rhallisey has quit IRC | 02:21 | |
| *** salv-orlando has quit IRC | 02:26 | |
| *** unicell has quit IRC | 02:27 | |
| *** daneyon has joined #openstack-kolla | 02:27 | |
| *** daneyon has quit IRC | 02:32 | |
| *** hanchao has joined #openstack-kolla | 02:45 | |
| *** jtriley has quit IRC | 02:46 | |
| *** yuanying has quit IRC | 02:50 | |
| *** salv-orlando has joined #openstack-kolla | 02:52 | |
| *** banix has quit IRC | 03:00 | |
| hanchao | Hello, anyone suffered such an issue: when using `kolla-ansible prechecks`, in `TASK [prechecks | checking ansible version]` got failed... | 03:01 |
| hanchao | Never experienced this case before. The error message is helpless, so would expect suggestions here. | 03:02 |
| openstackgerrit | zhongshengping proposed openstack/kolla: Fix repo system https://review.openstack.org/337971 | 03:03 |
| *** salv-orlando has quit IRC | 03:07 | |
| *** banix has joined #openstack-kolla | 03:10 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Use more graceful wait service method https://review.openstack.org/337973 | 03:10 |
| hanchao | I followed the `stable/liberty` version suggestion, using ansible version 1.9.4. it worked for me even yesterday, but not now... and after i upgraded the ansible version to 2.1.0.0, no exception raised anymore. but it does not respect the quickstart... | 03:16 |
| *** mark-casey has quit IRC | 03:18 | |
| *** mark-casey has joined #openstack-kolla | 03:19 | |
| *** ayoung has quit IRC | 03:30 | |
| *** salv-orlando has joined #openstack-kolla | 03:33 | |
| mark-casey | hanchao: as far as I can tell this is the commit that changed it https://github.com/openstack/kolla/commit/3b1b6abf5b0133ed79581d2970c8a9abaacd0a37 and I do not think that commit is in stable/liberty. Is it possible you've started out with kolla stable/liberty and now have a newer version checked out/installed? | 03:35 |
| *** zhurong has quit IRC | 03:35 | |
| *** zhurong has joined #openstack-kolla | 03:36 | |
| *** yuanying has joined #openstack-kolla | 03:49 | |
| *** salv-orlando has quit IRC | 03:49 | |
| *** banix has quit IRC | 03:55 | |
| *** banix has joined #openstack-kolla | 03:56 | |
| *** huikang_ has quit IRC | 04:00 | |
| *** banix has quit IRC | 04:03 | |
| hanchao | mark-casey: probably you are right, i was in the master branch. as a result, the requirement is different... thanks. | 04:10 |
| mark-casey | hanchao: welcome | 04:12 |
| *** salv-orlando has joined #openstack-kolla | 04:15 | |
| *** daneyon has joined #openstack-kolla | 04:16 | |
| *** daneyon has quit IRC | 04:20 | |
| *** zhurong has quit IRC | 04:25 | |
| *** mark-casey has quit IRC | 04:28 | |
| hanchao | it's actually weird. i used the master branch and upgraded the ansible, which does not obey the stable/liberty requirement. however, i successfully deployed a liberty openstack... | 04:29 |
| *** salv-orlando has quit IRC | 04:30 | |
| *** zhurong has joined #openstack-kolla | 04:45 | |
| *** salv-orlando has joined #openstack-kolla | 04:57 | |
| *** salv-orlando has quit IRC | 05:01 | |
| *** salv-orlando has joined #openstack-kolla | 05:06 | |
| *** unicell has joined #openstack-kolla | 05:13 | |
| *** kaushikc has joined #openstack-kolla | 05:29 | |
| *** kaushikc has quit IRC | 05:47 | |
| *** unicell has quit IRC | 06:04 | |
| *** unicell has joined #openstack-kolla | 06:04 | |
| *** daneyon has joined #openstack-kolla | 06:04 | |
| *** fragatina has joined #openstack-kolla | 06:06 | |
| *** fragatina has quit IRC | 06:06 | |
| *** fragatina has joined #openstack-kolla | 06:07 | |
| *** daneyon has quit IRC | 06:08 | |
| -openstackstatus- NOTICE: All python 3.5 jobs are failing today, we need to build new xenial images first. | 06:27 | |
| *** harlowja has quit IRC | 06:34 | |
| *** Serlex has joined #openstack-kolla | 06:37 | |
| *** athomas has joined #openstack-kolla | 07:05 | |
| *** salv-orlando has quit IRC | 07:10 | |
| *** phuongnh has quit IRC | 07:15 | |
| *** gfidente has joined #openstack-kolla | 07:21 | |
| *** gfidente has joined #openstack-kolla | 07:21 | |
| *** shardy has joined #openstack-kolla | 07:26 | |
| *** zhiwei has quit IRC | 07:30 | |
| *** zhiwei has joined #openstack-kolla | 07:36 | |
| *** zhiwei has quit IRC | 07:44 | |
| *** mewald1 has joined #openstack-kolla | 07:48 | |
| mewald1 | Our openstack installation with Kolla seems to be a bit slow responding to API calls. Especially in Horizon it is noticable. Any ideas where this might come from? | 07:49 |
| *** daneyon has joined #openstack-kolla | 07:52 | |
| *** zhiwei has joined #openstack-kolla | 07:56 | |
| *** daneyon has quit IRC | 07:57 | |
| *** zhongshengping has joined #openstack-kolla | 08:09 | |
| *** zhongshengping has quit IRC | 08:09 | |
| *** godleon has quit IRC | 08:19 | |
| *** tyrola has joined #openstack-kolla | 08:29 | |
| mewald1 | Did anyone deploy with separate networks for internal and external on the controller? I am currently having the problem that my default gateway goes through the internal network's router and therefore the external IPs are not reachable at all. Traffic reaches the external VIP but responses are routed through internal as that's where the default gateway sits | 08:45 |
| *** pbourke has quit IRC | 08:45 | |
| *** pbourke has joined #openstack-kolla | 08:46 | |
| Daviey | mewald1: should you not add a route? | 09:21 |
| *** daneyon has joined #openstack-kolla | 09:41 | |
| *** tyrola has quit IRC | 09:41 | |
| *** zhurong has quit IRC | 09:44 | |
| *** daneyon has quit IRC | 09:45 | |
| *** athomas has quit IRC | 09:48 | |
| *** athomas has joined #openstack-kolla | 09:55 | |
| *** salv-orlando has joined #openstack-kolla | 10:31 | |
| *** pmisiak has joined #openstack-kolla | 10:39 | |
| pmisiak | Hi guys | 10:39 |
| pmisiak | anyone tried to configure internal and admin endpoints to use HTTPS ? | 10:39 |
| pmisiak | I see that haproxy.cfg template needs some changes and set internal_protocol and admin_protocol variables to https | 10:41 |
| pmisiak | anything more? | 10:41 |
| *** salv-orlando has quit IRC | 10:42 | |
| openstackgerrit | Paul Bourke proposed openstack/kolla: Read GPT label using sgdisk rather than udev https://review.openstack.org/326609 | 10:43 |
| *** salv-orlando has joined #openstack-kolla | 10:45 | |
| mewald1 | We have two network nodes running neutron-dhcp-agent. Both dhcp agents register successfully via rabbitmq and show in "neutron agent-list". After a while, one of the drop s out. Neutron-server logs that there where no heartbeats for a while. Any ideas? | 10:51 |
| *** coolsvap has joined #openstack-kolla | 11:05 | |
| mewald1 | my Linux bridge agents hang executing commands to create vxlan interfaces and others: https://gist.github.com/mewald1/d5ac2cd37ca645ba02efe4c34bf42782 Any suggestions? | 11:20 |
| *** jtriley has joined #openstack-kolla | 11:21 | |
| *** jtriley has quit IRC | 11:28 | |
| *** daneyon has joined #openstack-kolla | 11:29 | |
| *** zhiwei has quit IRC | 11:30 | |
| *** zhiwei has joined #openstack-kolla | 11:33 | |
| *** daneyon has quit IRC | 11:33 | |
| *** zhiwei has quit IRC | 11:37 | |
| *** zhiwei has joined #openstack-kolla | 11:43 | |
| mewald1 | https://gist.github.com/mewald1/44d4d85fe7f28977680595c6cd945417 Is this a bug? We write a file to /etc/sudoers.d/ but don't include those files in /etc/sudoers? | 11:45 |
| coolsvap | mewald1, files in sudoers.d directory are default included | 11:47 |
| mewald1 | coolsvap: Ok just found out "#includedir" is the expected syntax is sudoers :D | 11:48 |
| *** zhiwei has quit IRC | 11:48 | |
| mewald1 | I am hunting a problem where neutron-openvswitch-agent and others cannot run sudo to execute commands | 11:48 |
| *** ppowell has joined #openstack-kolla | 11:51 | |
| mewald1 | I changed neutron_sudoers to "neutron ALL = (root) NOPASSWD: /usr/bin/neutron-rootwrap /etc/neutron/rootwrap.conf *" and now it seems to work | 11:53 |
| *** zhiwei has joined #openstack-kolla | 11:55 | |
| *** salv-orl_ has joined #openstack-kolla | 11:59 | |
| *** zhiwei has quit IRC | 11:59 | |
| *** salv-orlando has quit IRC | 12:01 | |
| mewald1 | I just filed a bug report: https://bugs.launchpad.net/kolla/+bug/1599489 | 12:02 |
| openstack | Launchpad bug 1599489 in kolla "Neutron Open vSwitch Agent hangs because of Sudo" [Undecided,New] | 12:02 |
| *** banix has joined #openstack-kolla | 12:03 | |
| *** salv-orl_ has quit IRC | 12:03 | |
| *** banix has quit IRC | 12:05 | |
| *** zhiwei has joined #openstack-kolla | 12:06 | |
| *** jtriley has joined #openstack-kolla | 12:06 | |
| *** zhiwei has quit IRC | 12:10 | |
| *** jtriley has quit IRC | 12:13 | |
| *** zhiwei has joined #openstack-kolla | 12:13 | |
| *** fragatina has quit IRC | 12:13 | |
| *** fragatina has joined #openstack-kolla | 12:14 | |
| *** rhallisey has joined #openstack-kolla | 12:16 | |
| *** jtriley has joined #openstack-kolla | 12:25 | |
| *** berendt has joined #openstack-kolla | 12:27 | |
| *** jtriley has quit IRC | 12:32 | |
| *** williamcaban has joined #openstack-kolla | 12:37 | |
| *** mhayden has quit IRC | 12:46 | |
| *** dwalsh has joined #openstack-kolla | 12:51 | |
| *** mhayden has joined #openstack-kolla | 13:03 | |
| *** mhayden has quit IRC | 13:03 | |
| *** mhayden has joined #openstack-kolla | 13:04 | |
| *** mhayden has quit IRC | 13:07 | |
| *** jtriley has joined #openstack-kolla | 13:08 | |
| *** mhayden has joined #openstack-kolla | 13:09 | |
| *** daneyon has joined #openstack-kolla | 13:18 | |
| *** zhurong has joined #openstack-kolla | 13:18 | |
| *** inc0 has joined #openstack-kolla | 13:20 | |
| *** inc0_ has joined #openstack-kolla | 13:22 | |
| *** daneyon has quit IRC | 13:22 | |
| *** diogogmt has quit IRC | 13:23 | |
| *** banix has joined #openstack-kolla | 13:25 | |
| *** inc0 has quit IRC | 13:25 | |
| *** diogogmt has joined #openstack-kolla | 13:26 | |
| *** banix has quit IRC | 13:27 | |
| *** rhallisey has quit IRC | 13:29 | |
| *** rhallisey has joined #openstack-kolla | 13:30 | |
| *** ayoung has joined #openstack-kolla | 13:38 | |
| *** diogogmt has quit IRC | 13:41 | |
| *** Serlex has left #openstack-kolla | 13:41 | |
| *** Serlex has joined #openstack-kolla | 13:41 | |
| *** stvnoyes1 has quit IRC | 13:46 | |
| *** michauds has joined #openstack-kolla | 13:47 | |
| mewald1 | Can somebody look at this line in the code: https://github.com/openstack/kolla/blob/master/ansible/roles/prechecks/tasks/package_checks.yml#L10 Why are we checking the ansible version _locally_ for every host? | 13:47 |
| inc0_ | mewald1, good catch,we shouldn't | 13:48 |
| mewald1 | :) | 13:51 |
| mewald1 | I will use this to practise using git and gerrit and all :D | 13:52 |
| *** huikang has joined #openstack-kolla | 13:55 | |
| *** kangh_ has joined #openstack-kolla | 13:55 | |
| *** kangh_ has quit IRC | 13:55 | |
| *** williamcaban has quit IRC | 13:58 | |
| *** williamcaban has joined #openstack-kolla | 13:59 | |
| *** huikang_ has joined #openstack-kolla | 14:02 | |
| *** stvnoyes1 has joined #openstack-kolla | 14:03 | |
| *** williamcaban has quit IRC | 14:04 | |
| *** michauds has quit IRC | 14:08 | |
| *** michauds has joined #openstack-kolla | 14:15 | |
| *** ppowell has quit IRC | 14:24 | |
| *** zhurong has quit IRC | 14:24 | |
| *** pbourke_ has quit IRC | 14:26 | |
| *** stvnoyes1 has quit IRC | 14:30 | |
| *** ravig has joined #openstack-kolla | 14:31 | |
| *** pbourke has quit IRC | 14:31 | |
| *** sdake has joined #openstack-kolla | 14:33 | |
| sdake | morning | 14:34 |
| sdake | btw I am on pto this week | 14:34 |
| sdake | in mammoth lakes atm | 14:34 |
| *** pbourke has joined #openstack-kolla | 14:34 | |
| sdake | rhallisey inc0_ can oone of our un th meetin toay pls | 14:34 |
| rhallisey | yar | 14:34 |
| sdake | its nie to get some vacation :) | 14:35 |
| rhallisey | sdake, nice dude | 14:35 |
| rhallisey | should jump off the computer and go for a swim | 14:35 |
| rhallisey | :) | 14:35 |
| sdake | it lik 60 derees outside | 14:36 |
| sdake | not into polar bear club activiie ;) | 14:36 |
| inc0_ | sdake, you're in AZ? | 14:37 |
| inc0_ | because in TX is scorching... | 14:37 |
| mag009_ | morning | 14:42 |
| mewald1 | Shouldn't this line https://github.com/openstack/kolla/blob/stable/mitaka/docker/neutron/neutron-base/Dockerfile.j2#L65 be in the if-condition right above it? The sudoers file references /var/lib/kolla/venv which only exists when we're doing a source installation | 14:42 |
| *** ppowell has joined #openstack-kolla | 14:43 | |
| *** hongbin has joined #openstack-kolla | 14:43 | |
| *** hongbin has left #openstack-kolla | 14:43 | |
| *** stvnoyes has joined #openstack-kolla | 14:49 | |
| *** dwalsh has quit IRC | 14:53 | |
| *** diogogmt has joined #openstack-kolla | 14:54 | |
| *** huikang_ has quit IRC | 14:57 | |
| *** huikang_ has joined #openstack-kolla | 14:57 | |
| *** thumpba has joined #openstack-kolla | 15:00 | |
| *** huikang_ has quit IRC | 15:02 | |
| *** huikang_ has joined #openstack-kolla | 15:05 | |
| *** daneyon has joined #openstack-kolla | 15:06 | |
| *** banix has joined #openstack-kolla | 15:06 | |
| *** harlowja has joined #openstack-kolla | 15:07 | |
| *** daneyon has quit IRC | 15:10 | |
| *** stvnoyes has quit IRC | 15:16 | |
| *** sbezverk has joined #openstack-kolla | 15:20 | |
| *** inc0_ has quit IRC | 15:25 | |
| *** inc0 has joined #openstack-kolla | 15:25 | |
| *** zhiwei has quit IRC | 15:26 | |
| *** ppowell has quit IRC | 15:26 | |
| *** rhallisey has quit IRC | 15:27 | |
| mewald1 | Guys, I need you to have a look at something. I am trying to deploy an environment here with the images from docker hub. I checked the openvswitch agent image for sudo configuration like this https://gist.github.com/mewald1/809f9bf771db64623a07ddb97316ffa6 You can see that there are two files in /etc/sudoers.d: neutron and neutron_sudoers. After deploying the image to my nodes, the /etc/sudoers.d/neutron file is gone: https://gist.github.com/mewald1 | 15:28 |
| mewald1 | What am I doing wrong? | 15:28 |
| *** ccesario has quit IRC | 15:33 | |
| *** zhiwei has joined #openstack-kolla | 15:34 | |
| *** salv-orlando has joined #openstack-kolla | 15:34 | |
| *** ccesario has joined #openstack-kolla | 15:35 | |
| openstackgerrit | Greg Herlein proposed openstack/kolla-kubernetes: Documentation cleanup consisting mostly of re-ordering sections and links to make the docs more readable for new developers. Some small clarifications. https://review.openstack.org/338347 | 15:37 |
| *** zhiwei has quit IRC | 15:39 | |
| *** vhosakot has joined #openstack-kolla | 15:39 | |
| *** ppowell has joined #openstack-kolla | 15:39 | |
| *** rhallisey has joined #openstack-kolla | 15:41 | |
| *** stvnoyes1 has joined #openstack-kolla | 15:43 | |
| *** zhiwei has joined #openstack-kolla | 15:45 | |
| *** huikang_ has quit IRC | 15:49 | |
| *** zhiwei has quit IRC | 15:50 | |
| *** huikang_ has joined #openstack-kolla | 15:50 | |
| *** jmccarthy has joined #openstack-kolla | 15:50 | |
| *** jmccarthy has left #openstack-kolla | 15:51 | |
| *** dwalsh has joined #openstack-kolla | 15:51 | |
| *** hrito has joined #openstack-kolla | 15:54 | |
| *** zhiwei has joined #openstack-kolla | 15:56 | |
| *** ayoung has quit IRC | 16:00 | |
| Jeffrey4l_ | meeting time? sdake inc0 | 16:01 |
| coolsvap | rhallisey, ^^ | 16:01 |
| inc0 | right | 16:01 |
| *** zhiwei has quit IRC | 16:01 | |
| *** michauds has quit IRC | 16:03 | |
| *** athomas has quit IRC | 16:04 | |
| openstackgerrit | Greg Herlein proposed openstack/kolla-kubernetes: Added details on exactly how to do documentation patches. https://review.openstack.org/338375 | 16:07 |
| *** ppowell has quit IRC | 16:07 | |
| *** zhiwei has joined #openstack-kolla | 16:07 | |
| *** dwalsh has quit IRC | 16:08 | |
| *** unicell has quit IRC | 16:09 | |
| pmisiak | anyone tried to configure internal and admin endpoints to use HTTPS ? | 16:09 |
| pmisiak | I see that haproxy.cfg template needs some changes and needs set internal_protocol and admin_protocol variables to https | 16:10 |
| pmisiak | anything more? | 16:10 |
| inc0 | pmisiak, issue is certificates | 16:11 |
| *** mark-casey has joined #openstack-kolla | 16:11 | |
| *** zhiwei has quit IRC | 16:12 | |
| *** ayoung has joined #openstack-kolla | 16:15 | |
| pmisiak | inc0: why? | 16:18 |
| pmisiak | aa ok | 16:18 |
| pmisiak | domain name :) | 16:18 |
| pmisiak | but i can create internal endpoints with domain name... | 16:19 |
| pmisiak | and inject second certificate to haproxy container | 16:19 |
| *** huikang_ has quit IRC | 16:21 | |
| *** huikang_ has joined #openstack-kolla | 16:22 | |
| *** ccesario has quit IRC | 16:22 | |
| *** zhiwei has joined #openstack-kolla | 16:24 | |
| openstackgerrit | Merged openstack/kolla: Fix the native ovsdb_interace failed https://review.openstack.org/336352 | 16:24 |
| *** huikang_ has quit IRC | 16:27 | |
| *** zhiwei has quit IRC | 16:28 | |
| *** david-lyle_ has joined #openstack-kolla | 16:30 | |
| *** david-lyle_ is now known as david-lyle | 16:32 | |
| *** michauds has joined #openstack-kolla | 16:34 | |
| *** vhosakot has quit IRC | 16:35 | |
| *** vhosakot has joined #openstack-kolla | 16:36 | |
| *** david-lyle has quit IRC | 16:39 | |
| wirehead_ | Someone’s got some really nice smelling pastries on the train. | 16:40 |
| inc0 | so, as for docker cache, we can still use it with less of a change | 16:40 |
| Jeffrey4l_ | i think currently, the real issue is customization. Not only for the openstack project, like adding more horizon-plugins. But also add the ability to add private projects. | 16:41 |
| *** zhiwei has joined #openstack-kolla | 16:41 | |
| inc0 | so customization mechanisms will allow most of it | 16:41 |
| inc0 | if not all | 16:41 |
| Jeffrey4l_ | inc0, how? i really do not see the docker cache works as expected. | 16:41 |
| inc0 | Jeffrey4l_, what's the issue you're referring to? | 16:42 |
| Jeffrey4l_ | inc0, hmm, how to add the lbaas-dashboard in the current implementation? | 16:42 |
| Jeffrey4l_ | to the horizon image? | 16:42 |
| inc0 | is this apt package? | 16:42 |
| Jeffrey4l_ | one requirement: i need enable lbaas, murano dashboard in the horizon image. | 16:43 |
| inc0 | if so, {{ set horizon_apt_packages_append = ['lbaas-dashboard'] }} | 16:43 |
| Jeffrey4l_ | whatever. | 16:43 |
| inc0 | once we make horizon customizable | 16:43 |
| *** sogabe has quit IRC | 16:43 | |
| Jeffrey4l_ | inc0, how about the source install type? iirc, need copy some file to the horizon folder. | 16:43 |
| inc0 | you can add block and put stuff in it | 16:44 |
| *** ppowell has joined #openstack-kolla | 16:44 | |
| inc0 | or we can create yet another macro for this stuff | 16:44 |
| Jeffrey4l_ | ok.. | 16:44 |
| inc0 | but block would work | 16:44 |
| Jeffrey4l_ | inc0, how to change the repo url? | 16:45 |
| inc0 | macro I created is working with lists | 16:45 |
| inc0 | so we need to turn repo into list | 16:45 |
| inc0 | and then _append, _remove, _override works | 16:45 |
| *** zhiwei has quit IRC | 16:46 | |
| Jeffrey4l_ | is these in the code now? | 16:46 |
| inc0 | we need to put customizations into dockerfiles themselves | 16:46 |
| inc0 | mechanism is in code | 16:46 |
| inc0 | but we need to work with Dockerfiles | 16:46 |
| Jeffrey4l_ | inc0, hmm. could u explain more about this ? | 16:47 |
| inc0 | https://review.openstack.org/#/c/329651/2/docker/heat/heat-api/Dockerfile.j2 | 16:47 |
| patchbot | inc0: patch 329651 - kolla - Customizations for heat | 16:47 |
| inc0 | look instead of calling apt-get install heat-api | 16:47 |
| Jeffrey4l_ | i want to know how to change the repo url? | 16:48 |
| *** vhosakot has quit IRC | 16:48 | |
| Jeffrey4l_ | i know how the heat-api works and how to make customization for it, too. :p | 16:48 |
| *** vhosakot has joined #openstack-kolla | 16:48 | |
| inc0 | repos is just the same | 16:48 |
| inc0 | we need to make isntallation of repos from list | 16:49 |
| Jeffrey4l_ | OK. understand. | 16:49 |
| inc0 | and then you call list | customizable(repos) | 16:49 |
| inc0 | then heat_api_repos_append will add and such | 16:49 |
| inc0 | magic happens with filter | customizable | 16:49 |
| rahuls | Hi All, I need some help with the networking in openstack kolla. | 16:50 |
| inc0 | if you run custmomizable("repos") it will look for <<container_name>>_repos_append, <<container_name>>_repos_remove and <<container_name>>_repos_override | 16:50 |
| inc0 | and modify list accordingly | 16:51 |
| *** hrito has quit IRC | 16:51 | |
| rahuls | I did multinode install of openstack using kolla. Now, when I launch an instance, it fails because its not able to bring up the port. I can't see any openvswitch-agent in "neutron agent-list". Plus, running "ovs-vsctl show" in vswitchd container shows only br-ex. There is no br-tun or br-int. So, how does vm's communicate with controller/network node? | 16:51 |
| inc0 | rahuls, did playbooks ran all the way? | 16:52 |
| rahuls | yup, plus I used stable/mitaka branch | 16:52 |
| inc0 | if ovs client can't communicate to neutron-server something is really wrong | 16:52 |
| Jeffrey4l_ | inc0, how could i add a block only to the horizon image? | 16:52 |
| inc0 | Jeffrey4l_, put empty blocks with unique names | 16:53 |
| inc0 | like {% horizon_pre_install %}{% endblock %} | 16:53 |
| *** michauds has quit IRC | 16:53 | |
| Jeffrey4l_ | inc0, should i change the kolla's code? | 16:53 |
| inc0 | so bulk of our work is to add bunch of these blocks into dockerfiles | 16:53 |
| inc0 | hooks | 16:53 |
| rahuls | @inc0 so does this mean that networking should work the same way as it works in normal openstack deployment? | 16:53 |
| inc0 | for people to use | 16:53 |
| Jeffrey4l_ | ok | 16:54 |
| *** daneyon has joined #openstack-kolla | 16:54 | |
| inc0 | rahuls, yeah, no magic there | 16:54 |
| rahuls | ok | 16:54 |
| rahuls | one more question: when I login to neutron-server container, it logs in as neutron user. If I try to do sudo, it asks for password and I am not sure what the password would be | 16:54 |
| rahuls | is it something standard? | 16:55 |
| inc0 | rahuls, well, it's security | 16:55 |
| inc0 | yeah it's standard | 16:55 |
| inc0 | if someone breaks into container, they won't be able to do much outside of it | 16:55 |
| Jeffrey4l_ | rahuls, you can try `docker exec -u root neutron bash` | 16:55 |
| inc0 | brb | 16:56 |
| *** inc0 has quit IRC | 16:56 | |
| rahuls | Jeffrey4l_: Thanks | 16:56 |
| *** inc0 has joined #openstack-kolla | 16:57 | |
| mewald1 | Now that everyone is online, I would like to draw your attention to my previous message: | 16:58 |
| *** daneyon has quit IRC | 16:59 | |
| mewald1 | Guys, I need you to have a look at something. I am trying to deploy an environment here with the images from docker hub. I checked the openvswitch agent image for sudo configuration like this https://gist.github.com/mewald1/809f9bf771db64623a07ddb97316ffa6 You can see that there are two files in /etc/sudoers.d: neutron and neutron_sudoers. After deploying the image to my nodes, the /etc/sudoers.d/neutron file is gone: https://gist.github.com/mewald | 16:59 |
| *** zhiwei has joined #openstack-kolla | 17:03 | |
| inc0 | guys take on it ;) https://blueprints.launchpad.net/kolla/+spec/dockerfile-customizations | 17:03 |
| inc0 | rhallisey, do we need to add bp per item? or just use whiteboard of this bp ? | 17:04 |
| *** michauds has joined #openstack-kolla | 17:04 | |
| *** ravig has quit IRC | 17:04 | |
| *** ravig has joined #openstack-kolla | 17:04 | |
| *** ayoung has quit IRC | 17:06 | |
| *** zhiwei has quit IRC | 17:07 | |
| Mech422 | pbourke: ping? | 17:07 |
| Mech422 | Jeffrey4l_: I thought I replied last night about why the partition stuff didn't use a regex - it didnt need one | 17:09 |
| Mech422 | Jeffrey4l_: also, I made sure it had to be an exact match so that the code would fail fast and hard | 17:09 |
| Jeffrey4l_ | Mech422, you are using regex, why you say `it didn't need one`? | 17:11 |
| Mech422 | Jeffrey4l_: because what was there worked ? | 17:11 |
| Mech422 | Jeffrey4l_: why are we spinning cycles changing stuff that works for no benefit ? | 17:12 |
| Jeffrey4l_ | in the old code, you are using a pipe and grep. in the new code you are using regexp. You were changing what i asked. | 17:13 |
| Mech422 | Jeffrey4l_: And what is the advantage of that ? it does the same thing is and simply busy work | 17:14 |
| Jeffrey4l_ | doesn't it besome simple? the number of line reduced. | 17:15 |
| Jeffrey4l_ | Mech422, | 17:15 |
| *** papacz has quit IRC | 17:16 | |
| inc0 | Jeffrey4l_, can we split this into different patch then? | 17:16 |
| inc0 | merge fix and do refactoring later | 17:16 |
| Jeffrey4l_ | inc0, which one? | 17:17 |
| inc0 | ceph | 17:17 |
| inc0 | Mech422's patch | 17:17 |
| Mech422 | Jeffrey4l_: I think it increased the line counte by 1 line actually | 17:17 |
| Mech422 | Jeffrey4l_: byt thatas not the point... | 17:17 |
| *** unicell has joined #openstack-kolla | 17:17 | |
| Mech422 | Jeffrey4l_: if it works, and is meets the criteria for the problem - you shouldn't be requesting changes just because your 'style' is different | 17:18 |
| Jeffrey4l_ | first of all are u talking this change https://review.openstack.org/#/c/326609/7..8/docker/kolla-toolbox/find_disks.py Mech422 | 17:19 |
| patchbot | Jeffrey4l_: patch 326609 - kolla - Read GPT label using sgdisk rather than udev | 17:19 |
| Jeffrey4l_ | right? | 17:19 |
| *** ayoung has joined #openstack-kolla | 17:19 | |
| Mech422 | Jeffrey4l_: yeah - that it - though I was sure I commented on it last night before pbourke last iteration | 17:20 |
| *** zhiwei has joined #openstack-kolla | 17:20 | |
| Jeffrey4l_ | inc0, which one is better you think. https://review.openstack.org/#/c/326609/7..8/docker/kolla-toolbox/find_disks.py | 17:20 |
| patchbot | Jeffrey4l_: patch 326609 - kolla - Read GPT label using sgdisk rather than udev | 17:20 |
| Jeffrey4l_ | Mech422, where is you comment? | 17:21 |
| Mech422 | Jeffrey4l_: I don't know - can't find the dam thing | 17:21 |
| Jeffrey4l_ | Mech422, :( | 17:21 |
| Jeffrey4l_ | I see nothing | 17:21 |
| Mech422 | Jeffrey4l_: and I freely admit a regex is generally 'better' | 17:21 |
| Mech422 | Jeffrey4l_: but I wanted an exact match with fail fast/fail hard | 17:22 |
| Mech422 | Jeffrey4l_: so if the output format changes, we know | 17:22 |
| Mech422 | Jeffrey4l_: you put a regex in there and people will be tempted to make it 'generic' | 17:22 |
| Jeffrey4l_ | the old code has such issue too. | 17:22 |
| *** huikang_ has joined #openstack-kolla | 17:22 | |
| inc0 | Mech422, ok, I must agree with Jeffrey4l_ there | 17:22 |
| Mech422 | Jeffrey4l_: then it will fail silently, and inc0 will give me grief about how screen scrapign sucks | 17:22 |
| inc0 | use regex, it does provide good error handling too | 17:23 |
| Mech422 | inc0: ^^^ | 17:23 |
| inc0 | you do regex with grep;) | 17:23 |
| inc0 | if you want exact match, use python regex that's not too generic | 17:24 |
| Jeffrey4l_ | Mech422, regexp can do more the work than grep+split | 17:24 |
| Mech422 | inc0: not without extra flags and such | 17:24 |
| *** zhiwei has quit IRC | 17:24 | |
| *** Jeffrey4l_ has left #openstack-kolla | 17:25 | |
| *** Jeffrey4l_ has joined #openstack-kolla | 17:25 | |
| Mech422 | inc0: anyway, I built something solid that works - we've gone thru 8 iterations of a tiny patch for no real benefit | 17:25 |
| Jeffrey4l_ | Mech422, what kind of change you old code can change , but the regexp not? | 17:25 |
| inc0 | Mech422, but Paul uploaded working patch right? | 17:25 |
| inc0 | I mean gates are red, but that's another matter | 17:26 |
| Mech422 | inc0: Dunno - I'm still rebuilding my networking stuff here - havent rebuilt lately | 17:26 |
| inc0 | well, patch is upstream | 17:27 |
| inc0 | https://review.openstack.org/#/c/326609 | 17:27 |
| Mech422 | inc0: my point being though, that we're just spinning iterating over probably 'throw away' code once the 'new' partition scheme is decided upon | 17:27 |
| inc0 | Mech422, so if we won't change partition scheme anytime soon, it's better to have good code | 17:28 |
| inc0 | if we throw it away, cool | 17:28 |
| *** berendt has quit IRC | 17:28 | |
| inc0 | but we don't have exact plan for it now | 17:28 |
| *** berendt has joined #openstack-kolla | 17:29 | |
| Jeffrey4l_ | even if the code will be thrown. I do know think we can write it bad. | 17:29 |
| Mech422 | inc0: again, functionally its the same... except now, people are tempted to try 'fancey' regex's to match multiple versions of the string.. | 17:29 |
| *** unicell1 has joined #openstack-kolla | 17:29 | |
| *** unicell has quit IRC | 17:29 | |
| *** huikang_ has quit IRC | 17:30 | |
| Mech422 | also, you can't just cut/paste to the command line to verify the output IF the sgdisk output changes | 17:30 |
| *** unicell1 has quit IRC | 17:30 | |
| *** unicell has joined #openstack-kolla | 17:30 | |
| inc0 | Mech422, it is cleaner with regexp, and people getting fancy can be easily caught in review | 17:30 |
| Mech422 | Jeffrey4l_: I don't consider this 'better' code in this case - just 'different' | 17:30 |
| Mech422 | inc0: ^^^ | 17:30 |
| inc0 | but you can do docker exec -it kolla_toolbox find_disks | 17:31 |
| Jeffrey4l_ | Mech422, let's judge by other reviewers/cores | 17:31 |
| Mech422 | inc0: that doesn't show the raw output | 17:31 |
| *** unicell1 has joined #openstack-kolla | 17:31 | |
| Jeffrey4l_ | Mech422, if you want a raw output, you can add a log there. | 17:32 |
| *** unicell1 has joined #openstack-kolla | 17:32 | |
| *** unicell has quit IRC | 17:32 | |
| inc0 | Jeffrey4l_, issue with ansible - logging is pretty bad:P | 17:32 |
| inc0 | but what we can do is to include output of this command in one of returned variables | 17:32 |
| Mech422 | inc0: I specifically built this to be super simple and fail fast/hard because everyone hates screen scraping - if this is what you guys want fine, but don't blame screen scraping when people break it | 17:32 |
| Jeffrey4l_ | just formatted to one line with '\n' char. But still useful. | 17:33 |
| Mech422 | anyway - I've made my points | 17:33 |
| Mech422 | if thats what you want - pbourke did it already | 17:33 |
| *** huikang_ has joined #openstack-kolla | 17:33 | |
| inc0 | and they're taken, we will make sure that people won't go fancy about it | 17:33 |
| *** berendt has quit IRC | 17:34 | |
| inc0 | these red gates tho | 17:34 |
| Mech422 | inc0: fair enough | 17:34 |
| inc0 | doesn't seem to be connected to this change | 17:35 |
| inc0 | but it's bad | 17:35 |
| inc0 | let me dig into this | 17:35 |
| sdake | hey folks | 17:36 |
| sdake | heads up i am on pto this week | 17:36 |
| inc0 | sdake, we need to submit talk | 17:36 |
| inc0 | this week | 17:36 |
| sdake | what is deadline | 17:36 |
| inc0 | 14th | 17:37 |
| sdake | ok monday then | 17:37 |
| inc0 | Monday I'm flying... | 17:37 |
| sdake | my internet here is shakey and i am tryng to r&R :) | 17:37 |
| sdake | i am back sunday late | 17:37 |
| sdake | satruday late i mean | 17:37 |
| sdake | could do sunday? | 17:37 |
| inc0 | duh | 17:38 |
| *** Jeffrey4l_ has quit IRC | 17:39 | |
| inc0 | I'll write up an abstract | 17:40 |
| *** lyncos has joined #openstack-kolla | 17:40 | |
| *** ravig has quit IRC | 17:40 | |
| Mech422 | inc0: if you want, I can run a build of master real quick... | 17:42 |
| Mech422 | inc0: I think everything is working atm (waiting on ubiquiti gear tommorrow) | 17:43 |
| *** Jeffrey4l_ has joined #openstack-kolla | 17:44 | |
| *** unicell1 has quit IRC | 17:46 | |
| *** unicell has joined #openstack-kolla | 17:46 | |
| *** unicell1 has joined #openstack-kolla | 17:47 | |
| *** unicell has quit IRC | 17:47 | |
| *** mark-casey has quit IRC | 17:50 | |
| *** mark-casey has joined #openstack-kolla | 17:51 | |
| *** aNupoisc has joined #openstack-kolla | 17:54 | |
| *** huikang_ has quit IRC | 17:55 | |
| *** sdake has quit IRC | 17:55 | |
| *** huikang_ has joined #openstack-kolla | 17:55 | |
| *** sdake has joined #openstack-kolla | 17:56 | |
| *** shardy has quit IRC | 17:58 | |
| *** bootsha has joined #openstack-kolla | 17:59 | |
| *** huikang_ has quit IRC | 18:00 | |
| *** huikang_ has joined #openstack-kolla | 18:02 | |
| *** sdake has quit IRC | 18:02 | |
| *** aNupoisc has quit IRC | 18:04 | |
| *** ravig has joined #openstack-kolla | 18:05 | |
| *** sdake has joined #openstack-kolla | 18:08 | |
| *** sdake has quit IRC | 18:08 | |
| *** sdake has joined #openstack-kolla | 18:08 | |
| *** ravig has quit IRC | 18:09 | |
| *** ravig has joined #openstack-kolla | 18:10 | |
| *** salv-orlando has quit IRC | 18:12 | |
| *** aNupoisc has joined #openstack-kolla | 18:13 | |
| *** mliima_ has quit IRC | 18:14 | |
| *** mewald1 has quit IRC | 18:14 | |
| *** huikang_ has quit IRC | 18:16 | |
| *** huikang_ has joined #openstack-kolla | 18:16 | |
| *** huikang_ has quit IRC | 18:17 | |
| *** huikang_ has joined #openstack-kolla | 18:17 | |
| *** inc0 has quit IRC | 18:18 | |
| *** inc0 has joined #openstack-kolla | 18:18 | |
| *** Jeffrey4l_ has quit IRC | 18:19 | |
| *** pmisiak has quit IRC | 18:21 | |
| *** lyncos has quit IRC | 18:22 | |
| *** zhiwei has joined #openstack-kolla | 18:23 | |
| Mech422 | inc0: is there a way to get a 'raw' text of a code diff in gerrit? | 18:25 |
| Mech422 | inc0: I'll paste in the diff for the current version of the ceph stuff and build | 18:25 |
| inc0 | Mech422, you can cherrypick a commit if that's what you're asking | 18:27 |
| Mech422 | inc0: nah - I use diff to patch the files... | 18:27 |
| *** huikang_ has quit IRC | 18:28 | |
| *** zhiwei has quit IRC | 18:28 | |
| Mech422 | inc0: this code hasn't been merged to master yet has it? the 'strategy' is 'merge if required' ? | 18:28 |
| *** huikang_ has joined #openstack-kolla | 18:28 | |
| inc0 | no | 18:29 |
| inc0 | hold on | 18:29 |
| inc0 | it's still not on master | 18:29 |
| *** bootsha has quit IRC | 18:30 | |
| inc0 | if you cherrypick this patchset, it will do new branch with it on top of master | 18:31 |
| *** huikang_ has quit IRC | 18:32 | |
| Mech422 | inc0: lemme figure out how to add that to my workflow | 18:34 |
| *** ayoung has quit IRC | 18:35 | |
| *** shmcfarl has joined #openstack-kolla | 18:36 | |
| *** zhiwei has joined #openstack-kolla | 18:40 | |
| *** ravig has quit IRC | 18:40 | |
| *** daneyon has joined #openstack-kolla | 18:42 | |
| *** zhiwei has quit IRC | 18:44 | |
| *** banix has quit IRC | 18:45 | |
| *** daneyon has quit IRC | 18:47 | |
| *** stvnoyes1 has quit IRC | 18:49 | |
| *** shmcfarl has quit IRC | 18:49 | |
| *** bootsha has joined #openstack-kolla | 18:53 | |
| *** Serlex has quit IRC | 18:54 | |
| *** huikang_ has joined #openstack-kolla | 18:57 | |
| *** zhiwei has joined #openstack-kolla | 19:02 | |
| *** gfidente has quit IRC | 19:03 | |
| *** mliima has joined #openstack-kolla | 19:04 | |
| *** stvnoyes has joined #openstack-kolla | 19:04 | |
| *** zhiwei has quit IRC | 19:06 | |
| *** inc0 has quit IRC | 19:07 | |
| *** godleon has joined #openstack-kolla | 19:07 | |
| *** inc0 has joined #openstack-kolla | 19:08 | |
| openstackgerrit | Merged openstack/kolla: Improve task names for Ceph https://review.openstack.org/337718 | 19:12 |
| *** ppowell has quit IRC | 19:13 | |
| openstackgerrit | Merged openstack/kolla: Fix the overwrite sudoers file issue https://review.openstack.org/336789 | 19:18 |
| *** zhiwei has joined #openstack-kolla | 19:19 | |
| *** fragatina has quit IRC | 19:21 | |
| *** fragatina has joined #openstack-kolla | 19:22 | |
| *** zhiwei has quit IRC | 19:23 | |
| *** bootsha has quit IRC | 19:27 | |
| *** huikang_ has quit IRC | 19:30 | |
| *** huikang_ has joined #openstack-kolla | 19:31 | |
| *** vhosakot has quit IRC | 19:31 | |
| *** ppowell has joined #openstack-kolla | 19:31 | |
| *** aNupoisc has quit IRC | 19:32 | |
| *** huikang_ has quit IRC | 19:35 | |
| Mech422 | inc0: How do I cherry pick the ceph change - its not in master yet, so what repo do I get it from ? | 19:40 |
| *** mewald1 has joined #openstack-kolla | 19:40 | |
| inc0 | https://review.openstack.org/#/c/326609/ | 19:41 |
| patchbot | inc0: patch 326609 - kolla - Read GPT label using sgdisk rather than udev | 19:41 |
| inc0 | top right corner there is a Download button | 19:41 |
| mewald1 | Does nobody have any advice on the sudo problem I am seeing? | 19:41 |
| inc0 | click it and you'll see a command for cherry pick | 19:41 |
| inc0 | mewald1, not atm...I need to understand it better:/ did you publish a bug? | 19:41 |
| mewald1 | inc0: no I assume I made some mistake because if not none of you should be able to launch openstack with neutron via Kolla and aktually be able to use it :D | 19:42 |
| inc0 | mewald1, you should and I haven't seen this issue before | 19:43 |
| inc0 | however that doesn't mean there is no issue | 19:43 |
| Mech422 | inc0: ohh - thats what I was looking for - thanks - all the online stuff seems to be for different vesions of gerrit | 19:43 |
| mewald1 | ok, I will file a bug report then | 19:43 |
| inc0 | please publish a bug with all the configs and outputs | 19:43 |
| inc0 | if it's misconfiguration, it's ok, we'll just close a bug | 19:43 |
| mewald1 | ok | 19:44 |
| *** vhosakot has joined #openstack-kolla | 19:44 | |
| mewald1 | inc0: actually I did file a report, but we did some more research which I postet in IRC. I will update the report with that information | 19:45 |
| inc0 | thank you | 19:45 |
| inc0 | I'll take a look at it later on | 19:45 |
| *** huikang_ has joined #openstack-kolla | 19:50 | |
| *** ayoung has joined #openstack-kolla | 19:56 | |
| *** zhiwei has joined #openstack-kolla | 19:58 | |
| Mech422 | inc0: building the containers now | 19:59 |
| Mech422 | inc0: should have a result in 30 minutes or so | 19:59 |
| mewald1 | inc0: https://bugs.launchpad.net/kolla/+bug/1599489 writing it all down make it all much clearer :D looks like the ubuntu images are broken. But take a look and tell me what you think | 20:00 |
| openstack | Launchpad bug 1599489 in kolla "Neutron Open vSwitch Agent hangs because of Sudo" [Undecided,New] | 20:00 |
| *** huikang_ has quit IRC | 20:01 | |
| inc0 | mewald1, I assigned it to myself | 20:01 |
| inc0 | will take a close look tomorrow probably | 20:01 |
| inc0 | thanks a lot | 20:01 |
| *** huikang_ has joined #openstack-kolla | 20:01 | |
| *** zhiwei has quit IRC | 20:02 | |
| Mech422 | mewald1: FWIW, building ubuntu from source works | 20:03 |
| Mech422 | mewald1: might get you over this hump | 20:03 |
| Mech422 | mewald1: btw - your not using openvswitch for the host networking are you? | 20:04 |
| *** huikang_ has quit IRC | 20:06 | |
| mewald1 | Mech422: yes I am, but I also tried with linuxbridge and saw the same symptoms | 20:06 |
| *** huikang_ has joined #openstack-kolla | 20:10 | |
| Mech422 | mewald1: ahh - well, just a heads up - if you want to use OVS for host networking, you need to stop kolla from starting the ovsdb container on the host, or it blows up host networking | 20:10 |
| *** sdake has quit IRC | 20:12 | |
| *** zhiwei has joined #openstack-kolla | 20:15 | |
| *** sdake has joined #openstack-kolla | 20:19 | |
| *** zhiwei has quit IRC | 20:19 | |
| *** williamcaban has joined #openstack-kolla | 20:21 | |
| mewald1 | Mech422: interesting. the container is running on compute and network nodes and all is working. are you referring to Xenial? We had some issues that our network nodes suddently showed extremely high cpu usage and swapping and ultimately crashed on Xenial. We downgraded network nodes to Trusty (compute nodes still von xenial) and now everything seems good | 20:23 |
| *** aNupoisc has joined #openstack-kolla | 20:23 | |
| *** huikang_ has quit IRC | 20:23 | |
| *** huikang_ has joined #openstack-kolla | 20:24 | |
| *** vhosakot has quit IRC | 20:26 | |
| Mech422 | mewald1: no - I'm referring to the container starting up an empty db, and blowing away any config the host was already using | 20:27 |
| Mech422 | mewald1: my boxes use OVS for all connectivity, and I had to stop the container | 20:27 |
| *** vhosakot has joined #openstack-kolla | 20:27 | |
| *** huikang_ has quit IRC | 20:27 | |
| Mech422 | mewald1: maybe you don't use OVS for native connectivity - just the opestack stuff ? | 20:27 |
| *** huikang_ has joined #openstack-kolla | 20:28 | |
| dcwangmit01_ | inc0: Would you mind commenting on: https://review.openstack.org/#/c/327925/ ? Want to know exactly what you want? | 20:28 |
| patchbot | dcwangmit01_: patch 327925 - kolla - Enable kolla k8s to override bind api bind address... | 20:28 |
| dcwangmit01_ | ^ I mean, need to know exactly what you want regarding your -1 | 20:28 |
| openstackgerrit | Merged openstack/kolla-kubernetes: Refactor, speed up, and fix bugs in jinja template dict generation https://review.openstack.org/336745 | 20:30 |
| *** daneyon has joined #openstack-kolla | 20:30 | |
| Mech422 | inc0: build blew up deploying mariadb - backtracking now | 20:30 |
| mewald1 | Mech422: right, I didnt know if ovs on top of ovs was a good idea, so I am using bonding and vlan for native connectivity | 20:31 |
| *** huikang has quit IRC | 20:31 | |
| Mech422 | mewald1: you can use OVS throughout - just need to stop the db from getting wiped | 20:32 |
| Mech422 | mewald1: if you decide you want to try, holler and I'll dig up my patch | 20:32 |
| *** huikang_ has quit IRC | 20:33 | |
| mewald1 | yeah would love to try! | 20:34 |
| *** daneyon has quit IRC | 20:35 | |
| *** jtriley has quit IRC | 20:36 | |
| *** ppowell has quit IRC | 20:36 | |
| *** zhiwei has joined #openstack-kolla | 20:37 | |
| *** bootsha has joined #openstack-kolla | 20:37 | |
| *** zhiwei has quit IRC | 20:41 | |
| openstackgerrit | Merged openstack/kolla-kubernetes: updated hyperkube version to 1.3 https://review.openstack.org/337903 | 20:42 |
| *** jeh has quit IRC | 20:43 | |
| *** ravig has joined #openstack-kolla | 20:43 | |
| *** mewald1 has quit IRC | 20:43 | |
| *** mliima has quit IRC | 20:46 | |
| *** jeh has joined #openstack-kolla | 20:46 | |
| *** jeh has quit IRC | 20:46 | |
| *** jeh has joined #openstack-kolla | 20:46 | |
| *** thumpba has quit IRC | 20:46 | |
| *** ravig has quit IRC | 20:48 | |
| *** sdake has quit IRC | 20:53 | |
| *** zhiwei has joined #openstack-kolla | 20:54 | |
| *** banix has joined #openstack-kolla | 20:55 | |
| *** zhiwei has quit IRC | 20:58 | |
| *** sdake has joined #openstack-kolla | 21:03 | |
| *** bootsha has quit IRC | 21:10 | |
| *** bootsha has joined #openstack-kolla | 21:11 | |
| *** ravig has joined #openstack-kolla | 21:13 | |
| *** bootsha has quit IRC | 21:16 | |
| *** dmsimard is now known as dmsimard|afk | 21:16 | |
| *** rhallisey has quit IRC | 21:16 | |
| *** Lyncos has joined #openstack-kolla | 21:16 | |
| *** michauds has quit IRC | 21:20 | |
| *** zhiwei has joined #openstack-kolla | 21:22 | |
| *** zhiwei has quit IRC | 21:26 | |
| *** ravig has quit IRC | 21:28 | |
| *** ravig has joined #openstack-kolla | 21:38 | |
| *** zhiwei has joined #openstack-kolla | 21:44 | |
| Mech422 | inc0: looks like we added an 'ansible' user somewhere... | 21:45 |
| *** bootsha has joined #openstack-kolla | 21:46 | |
| inc0 | Mech422, we did in kolla-toolbox | 21:46 |
| Mech422 | inc0: I'm getting permission errors about /home/ansible/.ansible | 21:46 |
| inc0 | ahh | 21:46 |
| *** banix has quit IRC | 21:46 | |
| inc0 | you have pretty old code right? | 21:46 |
| Mech422 | inc0: fresh pull from master an hour ago | 21:46 |
| inc0 | strange...we fixed this | 21:47 |
| Mech422 | inc0: I also got an error trying to create the mysql haproxy user | 21:47 |
| Mech422 | inc0: and deploy mysql | 21:47 |
| Mech422 | I'm trying to track those down now | 21:47 |
| *** banix has joined #openstack-kolla | 21:47 | |
| Mech422 | inc0: I'm pulling from github - maybe the fix hasn't made it there yet ? | 21:48 |
| inc0 | nah, github is up to date | 21:48 |
| inc0 | something is wrong | 21:48 |
| *** zhiwei has quit IRC | 21:49 | |
| Mech422 | inc0: I'm rebuilding from bare metal again - I'll log the deploy | 21:49 |
| Mech422 | inc0: if it helps - I'm doing an ubuntu source build from master | 21:50 |
| *** ravig has quit IRC | 21:51 | |
| *** vhosakot has quit IRC | 21:53 | |
| *** salv-orlando has joined #openstack-kolla | 21:57 | |
| *** zhiwei has joined #openstack-kolla | 22:01 | |
| *** zhiwei has quit IRC | 22:06 | |
| *** banix has quit IRC | 22:08 | |
| *** david-lyle_ has joined #openstack-kolla | 22:21 | |
| Mech422 | inc0: I'm curious - why do we have gate issues and other projects appear not too? Don't we all have basically the same requirements (OpenStack)? | 22:21 |
| *** ravig has joined #openstack-kolla | 22:22 | |
| *** david-lyle__ has joined #openstack-kolla | 22:26 | |
| *** zhiwei has joined #openstack-kolla | 22:29 | |
| *** zhiwei has quit IRC | 22:33 | |
| *** david-lyle_ is now known as david-lyle | 22:35 | |
| *** david-lyle__ has quit IRC | 22:36 | |
| *** fragatina has quit IRC | 22:40 | |
| *** salv-orl_ has joined #openstack-kolla | 22:43 | |
| *** salv-orlando has quit IRC | 22:46 | |
| *** zhiwei has joined #openstack-kolla | 22:51 | |
| *** fragatina has joined #openstack-kolla | 22:52 | |
| *** zhiwei has quit IRC | 22:55 | |
| *** bootsha has quit IRC | 23:06 | |
| *** zhiwei has joined #openstack-kolla | 23:08 | |
| *** zhiwei has quit IRC | 23:12 | |
| *** daneyon has joined #openstack-kolla | 23:13 | |
| *** williamcaban has quit IRC | 23:13 | |
| *** williamcaban has joined #openstack-kolla | 23:17 | |
| *** daneyon has quit IRC | 23:18 | |
| *** aNupoisc has quit IRC | 23:24 | |
| kfox1111 | fyi, new doc: https://github.com/kubernetes/kubernetes.github.io/blob/release-1.3/docs/user-guide/petset.md | 23:27 |
| *** zhiwei has joined #openstack-kolla | 23:30 | |
| *** zhiwei has quit IRC | 23:34 | |
| *** williamcaban has quit IRC | 23:34 | |
| *** sdake has quit IRC | 23:43 | |
| *** mark-casey has quit IRC | 23:44 | |
| *** williamcaban has joined #openstack-kolla | 23:45 | |
| *** zhiwei has joined #openstack-kolla | 23:47 | |
| *** beekhof has joined #openstack-kolla | 23:50 | |
| *** zhiwei has quit IRC | 23:51 | |
| *** Lyncos has quit IRC | 23:51 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!