| *** sacharya has quit IRC | 00:07 | |
| portdirect_ | SamYaple: i try i really do, but somtimes the crazy just demands to be free | 00:07 |
|---|---|---|
| kfox1111 | portdirect_: they other voices in my head agree. ;) | 00:08 |
| portdirect_ | kfox1111: you know the score :) | 00:08 |
| kfox1111 | hehe. | 00:09 |
| srwilkers | ;) | 00:11 |
| *** williamcaban has quit IRC | 00:14 | |
| *** williamc_ has joined #openstack-kolla | 00:14 | |
| *** zhangshuai has quit IRC | 00:15 | |
| *** sayantan_ has joined #openstack-kolla | 00:22 | |
| v1k0d3n | aren't we all a little crazy? | 00:24 |
| v1k0d3n | i mean...it's the weekend...and yet...here we are. | 00:24 |
| v1k0d3n | :) | 00:24 |
| *** n0isyn0ise has quit IRC | 00:26 | |
| *** n0isyn0ise has joined #openstack-kolla | 00:27 | |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: WIP: neutron service package https://review.openstack.org/412155 | 00:30 |
| *** eaguilar has quit IRC | 00:32 | |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: WIP: helm build server https://review.openstack.org/412100 | 00:32 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: WIP: helm build server https://review.openstack.org/412100 | 00:48 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: WIP: helm build server https://review.openstack.org/412100 | 01:00 |
| *** dave-mccowan has joined #openstack-kolla | 01:07 | |
| *** gomarivera has joined #openstack-kolla | 01:07 | |
| *** sacharya has joined #openstack-kolla | 01:08 | |
| *** brad[] has joined #openstack-kolla | 01:09 | |
| *** Pavo has quit IRC | 01:10 | |
| *** gomarivera has quit IRC | 01:12 | |
| *** sacharya has quit IRC | 01:13 | |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Fix rolling upgrades https://review.openstack.org/412162 | 01:15 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Fix rolling upgrades https://review.openstack.org/412162 | 01:42 |
| openstackgerrit | Merged openstack/kolla-kubernetes: Helm chart for glance api pod https://review.openstack.org/410481 | 01:47 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Fix rolling upgrades https://review.openstack.org/412162 | 01:50 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: WIP: helm build server https://review.openstack.org/412100 | 01:56 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: WIP: helm build server https://review.openstack.org/412100 | 02:01 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: WIP: helm build server https://review.openstack.org/412100 | 02:05 |
| *** sacharya has joined #openstack-kolla | 02:09 | |
| *** sacharya has quit IRC | 02:14 | |
| *** sayantan_ has quit IRC | 02:17 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Support multi key in merge_config module https://review.openstack.org/412101 | 02:27 |
| *** sayantan_ has joined #openstack-kolla | 02:28 | |
| *** zhubingbing_ has joined #openstack-kolla | 02:28 | |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: WIP: neutron service package https://review.openstack.org/412155 | 02:30 |
| openstackgerrit | Merged openstack/kolla-kubernetes: Privileged cleanup https://review.openstack.org/412099 | 02:32 |
| sbezverk | portdirect: ping | 02:33 |
| openstackgerrit | Steve Wilkerson proposed openstack/kolla-kubernetes: Helm chart for glance registry https://review.openstack.org/412154 | 02:41 |
| srwilkers | evening sbezverk | 02:41 |
| sbezverk | srwilkers: hey good evening | 02:43 |
| sbezverk | srwilkers: have you played with new storageclass? | 02:44 |
| srwilkers | sbezverk, i have not yet | 02:44 |
| sbezverk | srwilkers: I tried to use it, but fails with strange error. I heard portdirect has managed to get it working.. | 02:45 |
| *** zhubingbing_ has quit IRC | 02:48 | |
| *** zhubingbing has joined #openstack-kolla | 02:56 | |
| v1k0d3n | hey srwilkers :) you still working!? lol | 03:04 |
| srwilkers | v1k0d3n, hey man. off and on ;) | 03:04 |
| *** zhubingbing has quit IRC | 03:04 | |
| v1k0d3n | same. updating the rdb tools for the custom kubernetes controller in a minute. | 03:05 |
| v1k0d3n | for 1.5.1 | 03:05 |
| v1k0d3n | and 1.5.0 | 03:05 |
| srwilkers | oh nice | 03:05 |
| *** gomarivera has joined #openstack-kolla | 03:09 | |
| openstackgerrit | Steve Wilkerson proposed openstack/kolla-kubernetes: Helm charts for cinder keystone v1 endpoints https://review.openstack.org/410923 | 03:09 |
| openstackgerrit | Steve Wilkerson proposed openstack/kolla-kubernetes: WIP -- Helm charts for cinder keystone v1 endpoints https://review.openstack.org/410923 | 03:10 |
| *** dave-mccowan has quit IRC | 03:10 | |
| *** Pavo has joined #openstack-kolla | 03:12 | |
| *** Pavo has quit IRC | 03:12 | |
| *** Pavo has joined #openstack-kolla | 03:13 | |
| *** gomarivera has quit IRC | 03:13 | |
| *** Jeffrey4l has quit IRC | 03:22 | |
| *** zhubingbing_ has joined #openstack-kolla | 03:51 | |
| *** david-lyle_ has quit IRC | 03:54 | |
| *** david-lyle has joined #openstack-kolla | 03:56 | |
| *** sacharya has joined #openstack-kolla | 03:59 | |
| *** goldyfruit has joined #openstack-kolla | 04:00 | |
| *** sacharya has quit IRC | 04:01 | |
| *** sdake has joined #openstack-kolla | 04:06 | |
| *** Pavo has quit IRC | 04:14 | |
| *** david-lyle has quit IRC | 04:16 | |
| *** david-lyle has joined #openstack-kolla | 04:16 | |
| *** Pavo has joined #openstack-kolla | 04:18 | |
| zhubingbing_ | hi | 04:20 |
| zhubingbing_ | sdake | 04:20 |
| *** david-lyle has quit IRC | 04:22 | |
| *** sdake_ has joined #openstack-kolla | 04:23 | |
| *** david-lyle has joined #openstack-kolla | 04:24 | |
| *** sdake has quit IRC | 04:26 | |
| *** sdake_ has quit IRC | 04:27 | |
| *** awiddersheim has quit IRC | 04:31 | |
| openstackgerrit | zhubingbing proposed openstack/kolla-ansible: Fix aodh upgrade permission issue https://review.openstack.org/410046 | 04:35 |
| *** srwilkers has quit IRC | 04:54 | |
| *** goldyfruit has quit IRC | 04:56 | |
| openstackgerrit | Steve Wilkerson proposed openstack/kolla-kubernetes: WIP -- Helm charts for cinder keystone v1 endpoints https://review.openstack.org/410923 | 04:57 |
| *** sdake has joined #openstack-kolla | 05:00 | |
| portdirect_ | hey sbezverk | 05:03 |
| portdirect_ | sbezverk: i have a working impementation of ceph using the new stroageclass, based on the work of docker-ceph and aic-helm that uses a helm plugin for configuration here: https://github.com/portdirect/marina | 05:06 |
| *** v1k0d3n has quit IRC | 05:07 | |
| portdirect_ | to use this you need to both have acces to the k8s network from the host, and be using a controller manager image with rbd installed | 05:07 |
| *** sp_ has quit IRC | 05:09 | |
| portdirect_ | I'm going to put in a ps for kolla to make a set of images for use with kubeadm as they can be customised in the current master, but unfortuantly they need to share the same name as the google provided ones, so I will build and retag them and put them at the docker.io/kollakube namespace if appropriate? | 05:10 |
| *** gomarivera has joined #openstack-kolla | 05:10 | |
| *** gomarivera has quit IRC | 05:15 | |
| kfox1111 | portdirect_: lets call it something thats apropriate and just make it a documentation issue to retag locally until they get a better config option in place? | 05:19 |
| *** sp_ has joined #openstack-kolla | 05:25 | |
| portdirect_ | no probs, I'm back on an almost sensible uk schedule, so will put in a ps for the k8s images after breakfast :) | 05:30 |
| *** david-lyle has quit IRC | 05:30 | |
| *** david-lyle has joined #openstack-kolla | 05:32 | |
| kfox1111 | nice. thanks. :) | 05:37 |
| *** sdake has quit IRC | 05:39 | |
| openstackgerrit | Merged openstack/kolla-kubernetes: Helm chart for glance registry https://review.openstack.org/412154 | 05:40 |
| *** cinerama has quit IRC | 05:41 | |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Fix rolling upgrades https://review.openstack.org/412162 | 05:43 |
| *** v1k0d3n has joined #openstack-kolla | 05:53 | |
| *** Pavo has quit IRC | 06:14 | |
| *** Pavo has joined #openstack-kolla | 06:18 | |
| *** senk has joined #openstack-kolla | 06:38 | |
| *** gomarivera has joined #openstack-kolla | 07:12 | |
| *** gomarivera has quit IRC | 07:17 | |
| *** Jeffrey4l has joined #openstack-kolla | 07:30 | |
| *** l4yerffeJ_ has quit IRC | 07:36 | |
| *** l4yerffeJ_ has joined #openstack-kolla | 07:36 | |
| *** saneax is now known as saneax-_-|AFK | 08:04 | |
| *** Pavo has quit IRC | 08:14 | |
| *** Pavo has joined #openstack-kolla | 08:19 | |
| *** david-lyle has quit IRC | 08:30 | |
| *** david-lyle has joined #openstack-kolla | 08:32 | |
| *** zhangqiankun has quit IRC | 08:51 | |
| *** zhangqiankun has joined #openstack-kolla | 08:52 | |
| *** matrohon has joined #openstack-kolla | 08:55 | |
| *** Jeffrey4l has quit IRC | 09:08 | |
| *** jrich523 has quit IRC | 09:10 | |
| *** gomarivera has joined #openstack-kolla | 09:11 | |
| *** gomarivera has quit IRC | 09:16 | |
| *** senk has quit IRC | 09:31 | |
| *** senk has joined #openstack-kolla | 10:03 | |
| *** david-lyle has quit IRC | 10:05 | |
| *** matrohon has quit IRC | 10:05 | |
| *** david-lyle has joined #openstack-kolla | 10:06 | |
| *** Pavo has quit IRC | 10:14 | |
| *** Pavo has joined #openstack-kolla | 10:19 | |
| *** sayantan_ has quit IRC | 10:20 | |
| *** l4yerffeJ__ has joined #openstack-kolla | 10:31 | |
| *** l4yerffeJ_ has quit IRC | 10:31 | |
| *** sayantan_ has joined #openstack-kolla | 10:33 | |
| *** l4yerffeJ__ has quit IRC | 10:34 | |
| *** l4yerffeJ__ has joined #openstack-kolla | 10:34 | |
| *** zhubingbing_ has quit IRC | 10:35 | |
| openstackgerrit | Pete Birley proposed openstack/kolla: WIP: Add Kubernetes Images https://review.openstack.org/412205 | 11:04 |
| openstackgerrit | Pete Birley proposed openstack/kolla: WIP: Add Kubernetes Images https://review.openstack.org/412205 | 11:06 |
| openstackgerrit | Pete Birley proposed openstack/kolla: WIP: Add Kubernetes Images https://review.openstack.org/412205 | 11:08 |
| openstackgerrit | Pete Birley proposed openstack/kolla: Add Kubernetes Images for Kolla-Kubernetes https://review.openstack.org/412205 | 11:11 |
| *** gomarivera has joined #openstack-kolla | 11:13 | |
| *** senk has quit IRC | 11:16 | |
| *** gomarivera has quit IRC | 11:17 | |
| openstackgerrit | Pete Birley proposed openstack/kolla: Add Kubernetes Images for Kolla-Kubernetes https://review.openstack.org/412205 | 11:19 |
| *** portdirect_ is now known as portdirect_away | 11:28 | |
| *** n0isyn0ise has quit IRC | 11:29 | |
| *** n0isyn0ise has joined #openstack-kolla | 11:40 | |
| *** JRobinson__ has quit IRC | 11:40 | |
| *** JRobinson__ has joined #openstack-kolla | 11:41 | |
| *** cmceniry_ has joined #openstack-kolla | 11:48 | |
| *** n0isyn0ise has quit IRC | 11:49 | |
| *** cmceniry has quit IRC | 11:50 | |
| *** dupengfei_ has joined #openstack-kolla | 11:58 | |
| *** dupengfei has quit IRC | 11:58 | |
| *** dupengfei_ is now known as dupengfei | 11:58 | |
| *** senk has joined #openstack-kolla | 11:59 | |
| *** l4yerffeJ__ has quit IRC | 12:04 | |
| *** l4yerffeJ__ has joined #openstack-kolla | 12:04 | |
| *** l4yerffeJ__ has quit IRC | 12:09 | |
| *** l4yerffeJ__ has joined #openstack-kolla | 12:09 | |
| *** l4yerffeJ__ has quit IRC | 12:11 | |
| *** l4yerffeJ__ has joined #openstack-kolla | 12:11 | |
| *** Pavo has quit IRC | 12:14 | |
| *** zhangshuai has joined #openstack-kolla | 12:18 | |
| *** Pavo has joined #openstack-kolla | 12:19 | |
| *** dupengfei has quit IRC | 12:19 | |
| *** dupengfei has joined #openstack-kolla | 12:20 | |
| *** sp_ has quit IRC | 12:21 | |
| *** dupengfei has quit IRC | 12:43 | |
| *** dupengfei has joined #openstack-kolla | 12:44 | |
| *** Jeffrey4l has joined #openstack-kolla | 13:01 | |
| *** gomarivera has joined #openstack-kolla | 13:14 | |
| *** gomarivera has quit IRC | 13:19 | |
| *** severion has joined #openstack-kolla | 13:20 | |
| *** v1k0d3n has quit IRC | 13:23 | |
| *** severion has quit IRC | 14:05 | |
| sbezverk | portdirect: ping | 14:08 |
| *** Pavo has quit IRC | 14:14 | |
| *** v1k0d3n has joined #openstack-kolla | 14:17 | |
| *** Pavo has joined #openstack-kolla | 14:18 | |
| SamYaple | morning | 14:20 |
| sbezverk | SamYaple: good morning | 14:26 |
| *** senk has quit IRC | 14:35 | |
| *** senk has joined #openstack-kolla | 14:36 | |
| openstackgerrit | Merged openstack/kolla-kubernetes: Fix rolling upgrades https://review.openstack.org/412162 | 14:39 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Optimized reconfigure action for glance https://review.openstack.org/406978 | 14:46 |
| *** awiddersheim has joined #openstack-kolla | 15:08 | |
| v1k0d3n | morning everyone | 15:12 |
| v1k0d3n | hey portdirect_away or sbezverk are you guys seeing issues with tiller deploying via helm init for k8s 1.5.1 and helm 2.1.0? | 15:13 |
| *** williamc_ has quit IRC | 15:13 | |
| v1k0d3n | new environment and i see a crashloop | 15:13 |
| *** williamcaban has joined #openstack-kolla | 15:13 | |
| *** saneax-_-|AFK is now known as saneax | 15:14 | |
| *** gomarivera has joined #openstack-kolla | 15:16 | |
| *** gomarivera has quit IRC | 15:20 | |
| sbezverk | v1k0d3n: morning.. you played with dynamic pvc and rbd storageclass right? | 15:22 |
| sbezverk | v1k0d3n: as per your helm question, it got installed fine in my test bed | 15:23 |
| v1k0d3n | sbezverk: some yes. we needed to build the container and add the appropriate tools to k8s-controller-manager for this. | 15:25 |
| v1k0d3n | sbezverk: when was the last time you installed? | 15:25 |
| v1k0d3n | and are you using the same versions? minikube/kubeadm? | 15:25 |
| *** senk has quit IRC | 15:26 | |
| *** zhubingbing_ has joined #openstack-kolla | 15:29 | |
| *** senk has joined #openstack-kolla | 15:32 | |
| *** senk has quit IRC | 15:33 | |
| sbezverk | v1k0d3n: I use baremetal and kubeadm helm 2.1.0 | 15:35 |
| sbezverk | v1k0d3n: for controller-manager have you pushed the fix upstream? I hit the same issue with missing rbd tool | 15:36 |
| *** williamc_ has joined #openstack-kolla | 15:38 | |
| v1k0d3n | ah, yes we did actually. just change your /etc/kubernetes/manifests/kube-controller-manager.json to use the following "image" | 15:39 |
| v1k0d3n | quay.io/attcomdev/kube-controller-manager:1.5.1 | 15:40 |
| v1k0d3n | give me one sec though sbezverk | 15:40 |
| v1k0d3n | let me make absolute sure there are no typos for you. | 15:40 |
| *** williamc_ has quit IRC | 15:40 | |
| sbezverk | v1k0d3n: sure thank you | 15:40 |
| *** kklimonda has quit IRC | 15:41 | |
| v1k0d3n | absolutely. quay.io/attcomdev/kube-controller-manager:1.5.1 | 15:41 |
| v1k0d3n | https://quay.io/repository/attcomdev/kube-controller-manager?tab=tags | 15:41 |
| v1k0d3n | let me know if you need a 1.5.0 tag, and i can build/push it for you. | 15:41 |
| v1k0d3n | we're using 1.5.1. i think you guys are too. | 15:41 |
| v1k0d3n | we needed to skip 1.5.0 | 15:41 |
| *** kklimonda has joined #openstack-kolla | 15:41 | |
| v1k0d3n | but plan on maintaining these. you're more than welcome to use them for the project. these will be long standing until a better (possibly native ceph tools container) comes along. | 15:42 |
| sbezverk | yeah, we use 1.5.1 too | 15:42 |
| sbezverk | I am surprised how kube community managed to miss this huge hick up with ceph | 15:43 |
| v1k0d3n | in fact, alan and i were just discussing the whole CI and maint issues last week. | 15:43 |
| v1k0d3n | i think they have other primary objects. | 15:43 |
| v1k0d3n | different storage etc | 15:43 |
| v1k0d3n | objectives | 15:43 |
| v1k0d3n | and this may have something to do with avoiding gpl ? | 15:44 |
| *** zhubingbing_ has quit IRC | 15:44 | |
| v1k0d3n | we're trying to work these out. portdirect_away is doing a lot in this area too. | 15:44 |
| *** zhubingbing_ has joined #openstack-kolla | 15:45 | |
| v1k0d3n | it just requires a lot of assistance, teamwork, etc outside of the openstack community. | 15:45 |
| v1k0d3n | (learning too in my case...haha) | 15:45 |
| sbezverk | v1k0d3n: based on their doc, ceph support comes at the same level as other storage solutions.. | 15:48 |
| v1k0d3n | havn't tried many of the others. so do users need to package up their own supporting tools, like we're doing? | 15:49 |
| v1k0d3n | for some reason, i was thinking some of the other tooling was included in controller-manager, whereas ceph tools are not. | 15:50 |
| *** goldyfruit has joined #openstack-kolla | 15:50 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Optimized reconfigure action for glance https://review.openstack.org/406978 | 15:50 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Optimiztion reconfiguration for keystone https://review.openstack.org/408933 | 15:50 |
| sbezverk | v1k0d3n: it is clearly just a miss, as some rbd related things were in the controller manager | 15:51 |
| v1k0d3n | hmmm. one sec, let me check something | 15:51 |
| v1k0d3n | oh you already submitted a ticket | 15:52 |
| v1k0d3n | i see | 15:52 |
| *** saneax is now known as saneax-_-|AFK | 15:54 | |
| v1k0d3n | commented on your ticket as a workaround, just in case someone else notices and feels "stuck". | 15:55 |
| *** n0isyn0ise has joined #openstack-kolla | 15:56 | |
| sbezverk | v1k0d3n: when I try to use your image. it is not starting controller manager pod hmmm | 15:59 |
| v1k0d3n | whoa, that's not good | 15:59 |
| v1k0d3n | make sure saved the GCR image link just in case? | 16:00 |
| v1k0d3n | i can look into this | 16:00 |
| v1k0d3n | do you have an error message? | 16:00 |
| v1k0d3n | and would you mine creating an issue? i can work it today. | 16:00 |
| v1k0d3n | you can create the issue here: https://github.com/att-comdev/aic-helm/issues | 16:01 |
| v1k0d3n | and just link the docker container here: https://github.com/att-comdev/dockerfiles/tree/master/kube-controller-manager | 16:01 |
| sbezverk | v1k0d3n: http://paste.openstack.org/show/592736/ | 16:03 |
| sbezverk | v1k0d3n: ok I know what happened. when controller-manager image gets changed in manifest, the cluster reboot is required | 16:08 |
| v1k0d3n | ah! | 16:09 |
| v1k0d3n | it's funny you say that because i did that anyway because of updates on my host. | 16:09 |
| v1k0d3n | i think that's why i missed it. | 16:09 |
| v1k0d3n | i need to document that. | 16:09 |
| v1k0d3n | thanks for catching that and letting me know sbezverk :) | 16:09 |
| sbezverk | v1k0d3n: yep.. would be helpful for future generations ;-) | 16:10 |
| v1k0d3n | absolutely. | 16:10 |
| v1k0d3n | what i'm going to do is submit the changes upstream in k8s (at least find out what it takes). | 16:10 |
| v1k0d3n | been hanging around that repo for two years. time i do something about it, and actually commit something. | 16:10 |
| v1k0d3n | let me find out what it will take. | 16:11 |
| v1k0d3n | so it's just "fixed". we need this just the same as kolla. | 16:11 |
| v1k0d3n | now i need to figure out what the heck is wrong with tiller | 16:12 |
| v1k0d3n | so sbezverk your environment sounds very much like the one i'm currently using. | 16:12 |
| v1k0d3n | bare metal, etc. | 16:12 |
| v1k0d3n | are you using CNI? | 16:13 |
| v1k0d3n | and what sdn? | 16:13 |
| *** wcb has joined #openstack-kolla | 16:13 | |
| v1k0d3n | also, when you started your cluster, what did you include with your init statements for kubeadm? did you specify api-server (i did), or anything specific to networking? | 16:13 |
| *** williamcaban has quit IRC | 16:14 | |
| *** Pavo has quit IRC | 16:14 | |
| v1k0d3n | the networking i left default, and i think this is what's biting me. nothing can talk back to 10.96.x.1 ;) | 16:14 |
| *** Pavo has joined #openstack-kolla | 16:14 | |
| sbezverk | kubeadm init --pod-network-cidr 10.57.0.0/19 --service-cidr 10.57.32.0/19 --service-dns-domain openstack.kolla. | 16:15 |
| sbezverk | v1k0d3n: missing binary issue is fixed with your image, but now I see another one :-( | 16:16 |
| v1k0d3n | uh oh. | 16:18 |
| v1k0d3n | ? | 16:18 |
| v1k0d3n | oh question too...how many interfaces are you using and what address range? | 16:19 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Optimize reconfigure action for glance https://review.openstack.org/406978 | 16:20 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Optimize reconfigure action for memcached https://review.openstack.org/412225 | 16:20 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Optimize reconfigure action for memcached https://review.openstack.org/412225 | 16:20 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Optimize reconfiguration for keystone https://review.openstack.org/408933 | 16:22 |
| SamYaple | Jeffrey4l: wow youre up late | 16:24 |
| *** bmace has quit IRC | 16:25 | |
| Jeffrey4l | SamYaple, it is my 00:25 am ;( | 16:25 |
| *** bmace has joined #openstack-kolla | 16:25 | |
| Jeffrey4l | SamYaple, could u review the optimize reconfiguration patch i made https://review.openstack.org/406978 ? | 16:26 |
| Jeffrey4l | made some big change and it is more optimal, imo. | 16:26 |
| sbezverk | v1k0d3n: I use single interface for management | 16:27 |
| SamYaple | Jeffrey4l: looks ok | 16:31 |
| Jeffrey4l | thanks. | 16:31 |
| SamYaple | Jeffrey4l: im mostly focused on salt at the moment though | 16:31 |
| SamYaple | Jeffrey4l: i havent tested anything you have | 16:32 |
| Jeffrey4l | i see. | 16:32 |
| Jeffrey4l | np. | 16:32 |
| SamYaple | Jeffrey4l: i need to get the static uid/gid implemented | 16:32 |
| Jeffrey4l | but only want u check the implementation direction ;) | 16:32 |
| SamYaple | Jeffrey4l: I have new info from RDO packaging | 16:32 |
| Jeffrey4l | SamYaple, you will start this? | 16:32 |
| SamYaple | Jeffrey4l: i suppose I can | 16:32 |
| Jeffrey4l | SamYaple, what's it? | 16:32 |
| Jeffrey4l | nice | 16:33 |
| SamYaple | Jeffrey4l: so RDO is no longer assigning static uid/gid to any projects. however, the existing projects that have the static uid/gid stuff will stay | 16:33 |
| Jeffrey4l | So? | 16:33 |
| SamYaple | so we can match RDO for existing things, and do our own | 16:33 |
| Jeffrey4l | ;( | 16:33 |
| SamYaple | or just do entirely our own. i think community can be convinved of that | 16:33 |
| SamYaple | since RDO isn't pursuing it further, i dont have a strong interest in following them | 16:34 |
| Jeffrey4l | yep. since rdo think static uid in RDO is bad ( at least not good ) , why we need follow him? | 16:34 |
| Jeffrey4l | cool | 16:34 |
| SamYaple | this way we can just pick a range and work with that | 16:35 |
| Jeffrey4l | agree \o/ | 16:35 |
| Jeffrey4l | hrm, do we have other disagreement, now? SamYaple | 16:35 |
| Jeffrey4l | upgrade migration script? | 16:36 |
| Jeffrey4l | in the future. | 16:36 |
| SamYaple | Jeffrey4l: i would prefer a range of 400-500 rather than 10000 | 16:36 |
| Jeffrey4l | hrm, let me check, 1 sec | 16:36 |
| SamYaple | if we do 10,000 it does break any chance of decent namespace remapping (makes it much more difficult) | 16:36 |
| SamYaple | if we keep it lower we "future proof" | 16:37 |
| Jeffrey4l | OK. | 16:37 |
| SamYaple | even if we dont use user namespace remapping yet | 16:37 |
| Jeffrey4l | ( checking the default linux uid range ) | 16:37 |
| SamYaple | 500-60000 on centos family | 16:37 |
| SamYaple | 1000-60000 on ubuntu | 16:37 |
| SamYaple | /etc/login.defs | 16:37 |
| Jeffrey4l | got. | 16:38 |
| *** senk has joined #openstack-kolla | 16:38 | |
| Jeffrey4l | is there any define for 400-500? it seems OK. no one use it, in default. | 16:39 |
| Jeffrey4l | for < 500 | 16:39 |
| SamYaple | right. its not commonly used unless you have ~400+ system users | 16:39 |
| Jeffrey4l | or for < 500 | 16:39 |
| Jeffrey4l | cool. | 16:39 |
| Jeffrey4l | i am agree on that. let's take from 400-500 | 16:40 |
| SamYaple | that keeps it low so if we do userns remapping we dont have to remap 10,000+ uid/gid | 16:40 |
| Jeffrey4l | yep. ( i thought the UID_MAX=1000 ;( ) | 16:40 |
| Jeffrey4l | 10000* | 16:40 |
| SamYaple | haha nope | 16:40 |
| SamYaple | I think we should also make it configurable in the build. but that doesnt have to happen immediately | 16:41 |
| v1k0d3n | sbezverk: were you able to paste some info about the next issue? | 16:41 |
| Jeffrey4l | SamYaple, another thing is: will kolla support migration solution/script for user who change the user-uid mapping? | 16:42 |
| SamYaple | Jeffrey4l: also, docker-py is now called 'docker', so its 'pip install docker' | 16:42 |
| Jeffrey4l | thanks for the info. | 16:42 |
| SamYaple | Jeffrey4l: if we are writing the scripts to do it anyway (since we have to do an initial permissions change for upgrade) i dont see why we can't leave that mechanism in place | 16:42 |
| Pavo | SamYaple isn't that only the python dependences for docker and not docker itself? | 16:43 |
| Jeffrey4l | But in openstack requirements.txt. we still depends on docker-py=1.10.6 | 16:43 |
| Jeffrey4l | SamYaple, hrm. OK. lets keep this. | 16:43 |
| SamYaple | Pavo: docker-py, the python lib, was renamed to 'docker'. pypi now only updates docker | 16:43 |
| SamYaple | Jeffrey4l: it will be a new requirements.txt line | 16:43 |
| Pavo | oh ok | 16:43 |
| Pavo | thx for the info | 16:44 |
| SamYaple | Jeffrey4l: got to submit that to requirements repo | 16:44 |
| Pavo | SamYaple where do you work? Since you are only like an hour from me me a few of my colleagues might come up and see you one day | 16:44 |
| Jeffrey4l | we have keep the migration script at least for recent two cycle. | 16:44 |
| Jeffrey4l | we can talk this in the future. | 16:44 |
| Jeffrey4l | cool. | 16:45 |
| SamYaple | Pavo: I work remote for Rackspace | 16:45 |
| SamYaple | Pavo: id be up for that | 16:45 |
| SamYaple | Jeffrey4l: depending on how we do migration, i think it should be baked into our config script | 16:46 |
| SamYaple | Jeffrey4l: in which case we dont even need to get rid of it | 16:46 |
| Pavo | oh awesome then we really need to chat | 16:46 |
| SamYaple | im sure we will need it again | 16:46 |
| Jeffrey4l | we implement kinda migration in set_config.py check this https://github.com/openstack/kolla-ansible/blob/master/ansible/roles/nova/templates/nova-compute.json.j2#L24 | 16:47 |
| Jeffrey4l | SamYaple, | 16:47 |
| Jeffrey4l | but it only work for kolla-ansible. | 16:47 |
| SamYaple | Jeffrey4l: yea i think we can reuse that I mean. maybe extend it a bit more | 16:47 |
| SamYaple | Jeffrey4l: why? | 16:47 |
| Jeffrey4l | in kolla-ansible repo :) | 16:47 |
| Jeffrey4l | nova-compute.json.j2 | 16:47 |
| SamYaple | Jeffrey4l: https://github.com/SamYaple/kolla-salt/blob/master/kolla/zookeeper/files/config.json.j2 | 16:48 |
| SamYaple | im using the config.json as well :) | 16:48 |
| Jeffrey4l | hrm, OK it works. | 16:48 |
| SamYaple | i think thats a really good entry point | 16:48 |
| Jeffrey4l | yep. | 16:48 |
| Jeffrey4l | set_config can be extend for more things. | 16:48 |
| Jeffrey4l | i want to say, maintain migration script may be hard :). and after we implement fix uid, only very few person may use this kind of thing. | 16:50 |
| SamYaple | Jeffrey4l: are we agreed to implement all of the users inside the base container? | 16:50 |
| Jeffrey4l | SamYaple, hrm. no. i want to create the user when needed. rather than create all users in base image. | 16:50 |
| Pavo | so is there away to fix this qemu issue with Centos right now? | 16:50 |
| Pavo | assuming just use Centos < 7.3 correct? | 16:51 |
| SamYaple | Jeffrey4l: im not sure thats the best path. that could lead to sharing uid/gid again | 16:51 |
| Jeffrey4l | SamYaple, no share. when creating nova user, we chose 400, when creating glance user we choose 401. | 16:51 |
| Jeffrey4l | we may maintain a user/uid map in build.py | 16:52 |
| SamYaple | Jeffrey4l: but containers sometimes see other files. like log files | 16:52 |
| SamYaple | those will just be owned by numbers | 16:52 |
| SamYaple | If we have a static list of uid/gid, it should be in base container | 16:52 |
| SamYaple | so all contaienrs have a consistent view | 16:52 |
| Jeffrey4l | SamYaple, i think it is the same put the creation in base or each container. | 16:53 |
| openstackgerrit | Steve Wilkerson proposed openstack/kolla-kubernetes: WIP -- Helm charts for cinder keystone v1 endpoints https://review.openstack.org/410923 | 16:53 |
| Jeffrey4l | no effect. | 16:53 |
| SamYaple | putting it in the base is better for consistency | 16:54 |
| Jeffrey4l | for example? | 16:54 |
| SamYaple | /var/log/kolla | 16:54 |
| Jeffrey4l | /var/log/kolla is owner by <user>:Kolla | 16:55 |
| SamYaple | don't you see a problem with that? | 16:55 |
| Jeffrey4l | we can not solve the issue by creating all user in base. | 16:55 |
| Jeffrey4l | yep. it cause some issue. | 16:55 |
| SamYaple | no youre wrong | 16:55 |
| SamYaple | that would be fixed | 16:55 |
| Jeffrey4l | how? | 16:55 |
| SamYaple | because we log to /var/log/kolla/<service> | 16:56 |
| SamYaple | and not directly to /var/log/kolla | 16:56 |
| SamYaple | but that means all the container see all the log files. they should also see the users they are owned by | 16:57 |
| Jeffrey4l | currently, mariadb create log with 660, which break the heka to read. but others are find. | 16:57 |
| Jeffrey4l | fine* | 16:57 |
| SamYaple | instead of numbers | 16:57 |
| SamYaple | Jeffrey4l: https://github.com/openstack/kolla-ansible/blob/master/ansible/roles/nova/templates/nova.conf.j2#L5 | 16:57 |
| *** portdirects_lapt has joined #openstack-kolla | 16:57 | |
| Jeffrey4l | SamYaple, it is useless to see the users. | 16:57 |
| SamYaple | i disagree. this would be useful to operators | 16:57 |
| SamYaple | seeing "401:401" isnt as useful as seeing "glance:glance" | 16:58 |
| Jeffrey4l | let me think this. any other reason? | 16:58 |
| SamYaple | there is no reason not to do it in base | 16:58 |
| SamYaple | id like to think of one reason not to do it in base since we have static mappins | 16:58 |
| portdirects_lapt | o/ | 16:58 |
| *** senk has quit IRC | 16:59 | |
| Jeffrey4l | lol | 16:59 |
| SamYaple | portdirects_lapt: o/ | 16:59 |
| *** haplo37_ has quit IRC | 16:59 | |
| portdirects_lapt | sup SamYaple | 16:59 |
| *** fragatina has joined #openstack-kolla | 16:59 | |
| SamYaple | portdirects_lapt: since RDO is abandoning uid/gid mapping, i think kolla should do its own | 16:59 |
| *** portdirects_lapt is now known as portdirect_ | 16:59 | |
| SamYaple | portdirects_lapt: thats what we were tlaking about | 16:59 |
| *** fragatina has quit IRC | 17:00 | |
| SamYaple | i would normally say match RDO for what _does_ exist, but there are conflicts with that | 17:00 |
| portdirect_ | oh ok - will need to read scrollback, but if they are dumping it that kinda changes things | 17:00 |
| *** haplo37_ has joined #openstack-kolla | 17:00 | |
| SamYaple | thats why RDO is moving away in the first place | 17:00 |
| *** fragatina has joined #openstack-kolla | 17:00 | |
| portdirect_ | so uid/gid in 400-500 range? | 17:02 |
| SamYaple | portdirect_: yea. | 17:02 |
| portdirect_ | works for me :) | 17:02 |
| SamYaple | 500+ is used by centos as the start range for dynamic allocation | 17:03 |
| Jeffrey4l | OK. it seems OK. I am agree with that SamYaple | 17:03 |
| *** goldyfruit has quit IRC | 17:03 | |
| portdirect_ | Nice thanks for looking into it SamYaple and Jeffrey4l :) | 17:04 |
| sbezverk | portdirect_: hey man. I am struggling with ceph dynamic pvc | 17:04 |
| sbezverk | have you get it working? | 17:04 |
| portdirect_ | yeah - just noticed that :( | 17:04 |
| portdirect_ | yes | 17:04 |
| SamYaple | awesome portdirect_ and Jeffrey4l. glad we are in agreement. ill work on the patch post-haste | 17:04 |
| portdirect_ | sbezverk , whats not working? | 17:04 |
| sbezverk | portdirect_: first issue was fixed by using v1k0d3n image | 17:05 |
| Jeffrey4l | i am concerning that: if there any package creates the user again when the user exist ;( | 17:05 |
| Jeffrey4l | thanks SamYaple | 17:05 |
| portdirect_ | nice I have a ps in for kolla, which adds rbd tools to controller manager as well | 17:05 |
| portdirect_ | 2nd issue? | 17:05 |
| sbezverk | portdirect_: second issue it does not see secrets | 17:06 |
| *** zhubingbing_ has quit IRC | 17:06 | |
| Jeffrey4l | but we can do nothing before meet such issue. so ignore it. | 17:06 |
| portdirect_ | sbezverk: could you check this, https://review.openstack.org/#/c/412205/ | 17:06 |
| sbezverk | http://paste.openstack.org/show/592738/ | 17:06 |
| portdirect_ | sbezverk: was just going to ask :) | 17:06 |
| portdirect_ | sbezverk: that from controller manager yes? | 17:07 |
| portdirect_ | sbezverk: did you follow/use my helm package or aic's or start from scratch? | 17:08 |
| sbezverk | portdirect_: it is all in one bare metal installed by kubeadm | 17:08 |
| v1k0d3n | portdirect_: and this is why ceph needs to be ridiculously easy. | 17:09 |
| v1k0d3n | :-/ | 17:09 |
| portdirect_ | try the setup I have up on git :) | 17:09 |
| v1k0d3n | looking forward to seeing how you can clean this up a bit. | 17:09 |
| portdirect_ | but not for single node yet :( | 17:09 |
| sbezverk | portdirect_: the cluster itself is a separate cluster which is working fine for other applications | 17:09 |
| portdirect_ | sweet - that makes things easier :) | 17:10 |
| *** zhugaoxiao has quit IRC | 17:10 | |
| portdirect_ | 2 sec and I'll log into my workstation (at the folks) | 17:10 |
| sbezverk | the issue is with controller-manager which by some reason do not see secret whihc DOES exist in kube-system namespace | 17:10 |
| *** zhugaoxiao has joined #openstack-kolla | 17:10 | |
| portdirect_ | can you paste your storageclass yaml? | 17:12 |
| *** fragatina has quit IRC | 17:13 | |
| sbezverk | portdirect_: http://paste.openstack.org/show/592739/ | 17:13 |
| portdirect_ | sbezverk: and your secret looks somthing like this? http://paste.openstack.org/show/592740/ | 17:15 |
| SamYaple | portdirect_: Jeffrey4l you think a range of 100 is enough? | 17:15 |
| portdirect_ | SamYaple: I would have hoped so | 17:16 |
| SamYaple | portdirect_: i suppose because they are static mappings we can always expand it to like 350 if needed | 17:16 |
| Jeffrey4l | we have 69 docker service now. i am afraid may be no :( | 17:16 |
| sbezverk | portdirect_: http://paste.openstack.org/show/592741/ | 17:16 |
| Jeffrey4l | SamYaple, yep. | 17:16 |
| SamYaple | Jeffrey4l: no we have 69 _containers_. most of those share the same users | 17:16 |
| SamYaple | Jeffrey4l: im counting ~30 users | 17:16 |
| Jeffrey4l | 69 folders in docker | 17:17 |
| SamYaple | ah right. i wasnt counting non-openstack service users | 17:17 |
| Jeffrey4l | and they won't share user. | 17:17 |
| *** gomarivera has joined #openstack-kolla | 17:18 | |
| SamYaple | like mysql and rabbitmq | 17:18 |
| SamYaple | yea | 17:18 |
| Jeffrey4l | yep. | 17:18 |
| SamYaple | good news is we can remove all of the usermod -a -G kolla stuff now | 17:21 |
| SamYaple | do it all in one place, one for loop | 17:21 |
| Jeffrey4l | haha. | 17:22 |
| *** gomarivera has quit IRC | 17:22 | |
| Jeffrey4l | if you create all user in base, i think we can use a loop. and maintain the parameters in build.py dict. | 17:22 |
| *** sayantan_ has joined #openstack-kolla | 17:23 | |
| SamYaple | Jeffrey4l: thats what i was thinking | 17:23 |
| Jeffrey4l | ;) | 17:23 |
| portdirect_ | sbezverk: ok so everything looks ok with that :( could you post a bit more scrollback from the controller-manager pod | 17:23 |
| Jeffrey4l | need get some sleep. bye buys | 17:25 |
| sbezverk | portdirect_: can you give me your email? will send you log by email | 17:25 |
| portdirect_ | sbezverk: pete at port.direct | 17:25 |
| portdirect_ | Jeffrey4l : have a good one :) | 17:26 |
| Jeffrey4l | thansk. | 17:26 |
| portdirect_ | sbezverk: what version of k8s api you using 1.5? | 17:27 |
| sbezverk | 1.5.1 | 17:27 |
| SamYaple | ubuntu mistakenly packages with 'memcache' user rather than 'memcached'. since memcache doesnt use any files..... i say we standardize on memcached | 17:27 |
| SamYaple | do you think anyone will care or object? | 17:27 |
| SamYaple | as an ubuntu user, I certainly don't object | 17:27 |
| sbezverk | portdirect_: just sent you the log | 17:27 |
| sbezverk | portdirect_ one more strange thing is that it tried to get "auth: unable to find a keyring on /etc/ceph/ceph.client.admin.keyring" | 17:29 |
| sbezverk | keyring file, but it should not since we use secret object for authentrication.. | 17:29 |
| portdirect_ | sbezverk thats normal behaviour, you could load the keyring there and it will fallback to that if there is no secret | 17:29 |
| portdirect_ | sbezverk - got it! | 17:31 |
| portdirect_ | I think | 17:31 |
| SamYaple | secret object for authentication? | 17:31 |
| portdirect_ | so you need to create a ceph-secret, with the same content as your admin-ceph-secret | 17:32 |
| SamYaple | are you talking about cephfs stuff? | 17:32 |
| SamYaple | or nova secret.xml? | 17:32 |
| portdirect_ | SamYaple: neither | 17:32 |
| portdirect_ | SamYaple: the secret is a k8s object container a base64 representation of the ceph.client.admin.keyring | 17:33 |
| portdirect_ | *containing | 17:33 |
| SamYaple | ah this a k8s specific thing | 17:34 |
| sbezverk | portdirect_: I have both | 17:35 |
| portdirect_ | ok - in the kube-system namespace? | 17:35 |
| sbezverk | kubectl get secret --namespace kube-system | 17:35 |
| sbezverk | NAME TYPE DATA AGE | 17:35 |
| sbezverk | admin-ceph-secret Opaque 1 1h | 17:35 |
| sbezverk | ceph-secret Opaque 1 1d | 17:35 |
| sbezverk | I am not a complete dork ;-) | 17:35 |
| portdirect_ | ok - what namespace is trying to consume the pvc? | 17:36 |
| portdirect_ | sbezverk: dork is not a word i would ever use to describe you :) | 17:36 |
| sbezverk | portdirect_: the thing is, error message gets generated even when pvc request is not initiated | 17:37 |
| sbezverk | but pvc would use default namespace | 17:37 |
| sbezverk | portdirect_: scratch that pvc request is also in kube-system namespace | 17:38 |
| portdirect_ | the ceph-secret needs to exist in that ns as well, but the error you are getting in me in that case.. especially with what you are saying, in my setups it doen't try and access ceph config untill i have something that needs it. | 17:38 |
| portdirect_ | sbezverk: was typing that as you replied | 17:38 |
| portdirect_ | give me a min and I'll look through this properly | 17:39 |
| sbezverk | portdirect_: need to step out, snow shoveling time. will be back in 1 hour | 17:39 |
| portdirect_ | np - I wish we had snow here, just loads of rain!... catch you soon - hopfully I'll have though of somthing other than the obvious that you've already done by then :) | 17:40 |
| *** fragatina has joined #openstack-kolla | 17:51 | |
| portdirect_ | sbezverk: think I may have it, on line 12 of the secret you sent me the type is listed as "Opaque" this should be changed to "kubernetes.io/rbd" as I see the same logs as you when I set the secret type to this | 17:56 |
| *** sayantan_ has quit IRC | 17:56 | |
| portdirect_ | sbezverk: logs from my test: http://paste.openstack.org/show/592743/ | 17:58 |
| *** sayantan_ has joined #openstack-kolla | 17:59 | |
| SamYaple | whats the command to gen kolla config | 17:59 |
| portdirect_ | SamYaple: kolla-ansible genconfig? | 18:00 |
| SamYaple | nah. buildconf | 18:01 |
| SamYaple | i got it | 18:01 |
| SamYaple | tox -e genconfig | 18:01 |
| SamYaple | our python code has just grown and grown :/ no pruning or refactoring | 18:02 |
| SamYaple | its a bit gangly | 18:02 |
| portdirect_ | SamYaple: yeah it terrifies me a little bit, I'm sure that at some point it will become sentient | 18:04 |
| SamYaple | it could use a good refactoring for sure | 18:04 |
| SamYaple | its sorta setup like it can be imported.... but good luck importing it | 18:04 |
| *** pbourke has quit IRC | 18:06 | |
| *** pbourke has joined #openstack-kolla | 18:06 | |
| SamYaple | esh im about to add a whole buncha sections to the config files :P | 18:06 |
| SamYaple | nova-user, keystone-user, mysql-user.... | 18:07 |
| SamYaple | dont know another way to make this configurable | 18:07 |
| portdirect_ | SamYaple, could do it by adding a file to the base image containing the mappings and a macro to do it? | 18:08 |
| SamYaple | portdirect_: yea but then will it be user configurable? | 18:09 |
| SamYaple | shouldnt it be in the conf file if its... configuration? | 18:09 |
| portdirect_ | if you make it possible to override. then the user could just replace it, though your last point is spot on | 18:10 |
| SamYaple | this would be an exceedingly rare thing to change I think portdirect_. might be best to do that | 18:11 |
| *** fragatina has quit IRC | 18:12 | |
| SamYaple | IM HERE TO CLAIM YOUR SOUL | 18:12 |
| SamYaple | kolla-conf sent me | 18:13 |
| *** senk has joined #openstack-kolla | 18:13 | |
| portdirect_ | Too late dude, I sold it for these *AWSEOME* magic beans I've got | 18:13 |
| SamYaple | did.. did you just eat those. | 18:13 |
| SamYaple | dude. | 18:13 |
| *** Pavo has quit IRC | 18:14 | |
| portdirect_ | no comment. | 18:14 |
| *** Pavo has joined #openstack-kolla | 18:14 | |
| portdirect_ | anyway better get back to dealing with the folks diy capers :) catch you soon | 18:15 |
| SamYaple | adidas | 18:15 |
| portdirect_ | oh actaully was it you who was asking about the difference between Persistant Volumes and Persistant Volume Claims the other day? | 18:17 |
| SamYaple | PV vs PVC? if so yes | 18:17 |
| portdirect_ | ok so some stoarge backends cannot create a volume automatically (eg NFS), so an admin needs to create a PV. A user can create a PVC that then claims this acloacted storage for them. If the storage type supports dynamic allocation (AWS/Ceph etc) then the user just needs to create a PVC and the PV is created for them. | 18:20 |
| portdirect_ | so the first case is kinda similar to glusterfs in manila, where the admin needs to create glusterfs volumes to be mapped to shares | 18:21 |
| portdirect_ | in k8s it seems a bit overkll at the moment so split them out into two objects when most clusters are single tennant, but will make much more sense when multitenancy takes off. | 18:22 |
| SamYaple | hmm. ok. that is more info | 18:22 |
| SamYaple | ill keep commiting this info to memory until this clicks | 18:22 |
| SamYaple | this is why list comprehenssion is bad | 18:23 |
| SamYaple | for plugin in [match.group(0) for match in | 18:23 |
| SamYaple | (re.search('^{}-plugin-.+'.format(image.name), | 18:23 |
| SamYaple | section) for section in | 18:23 |
| SamYaple | all_sections) if match]: | 18:23 |
| SamYaple | like. what. | 18:23 |
| SamYaple | no one is reading and understanding that in one go | 18:23 |
| portdirect_ | :/ | 18:25 |
| kfox1111 | morning. | 18:26 |
| SamYaple | morning Mr Fox | 18:26 |
| portdirect_ | kfox1111 morning! (though I need to go - back in an hour) | 18:27 |
| *** portdirect_ has quit IRC | 18:27 | |
| sbezverk | portdirect: where do you change the type of secret? | 18:27 |
| sbezverk | kfox1111: good morning | 18:27 |
| sbezverk | portdirect_: man that was it!!!!! I have not seen in docs that we need to specify type of secret, we always used opaque... | 18:32 |
| kfox1111 | sbezverk: https://review.openstack.org/#/c/412004 wf please. | 18:35 |
| kfox1111 | nice. https://review.openstack.org/#/c/412159/ merged. | 18:36 |
| *** portdirects_lapt has joined #openstack-kolla | 18:36 | |
| kfox1111 | so we have a check-experimental now for testing helm-entrypoint service packages. | 18:37 |
| portdirects_lapt | sbezverk: so my folks are till debating where to install the new sockets :) / :( | 18:37 |
| portdirects_lapt | sbezverk unfortunatly you will need to delete the secret and create it again, kubectl edit will not work here :( | 18:38 |
| *** portdirects_lapt is now known as portdirect_ | 18:38 | |
| sbezverk | portdirects_lapt: it is not a problem. I am experimenting so rebuilding everything is not big deal. What is important is the recipe | 18:39 |
| openstackgerrit | Merged openstack/kolla-kubernetes: Cleanup ceph_workflow https://review.openstack.org/412004 | 18:40 |
| kfox1111 | sbezverk: thanks. | 18:40 |
| sbezverk | kfox1111: link to track https://github.com/kubernetes/kubernetes/issues/38923#issuecomment-267821690 | 18:41 |
| kfox1111 | sbezverk: did you see the nested global patched merged to h | 18:41 |
| kfox1111 | helm? | 18:41 |
| sbezverk | kfox1111: it is supposed to be in canary | 18:42 |
| sbezverk | no kubernetes ceph issue | 18:42 |
| portdirect_ | sbezverk: I've got some images in ps here that have rbd https://review.openstack.org/#/c/412205/ | 18:42 |
| kfox1111 | sbezverk: I havent' seen any issues with rbd like that except for dynamic provisioning. | 18:43 |
| sbezverk | portdirect_: yeah I saw but I use kubeadm and I am sure tons of other folks will use it as well | 18:43 |
| *** rsjethani has joined #openstack-kolla | 18:43 | |
| sbezverk | so these images are only useful if you use kolla to install kube cluster | 18:43 |
| kfox1111 | sbezverk: I think the images are only for if you want to use dynamic pvc's with ceph, and don't have a differnet solution handy. | 18:44 |
| kfox1111 | so, batteries included option. | 18:44 |
| portdirect_ | sbezverk, actualy kubeadm now allows you to specify the images used - only namespace at the moment, but should allow full path. If you can get rbd in the upstream images though thats a total win :) | 18:44 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: WIP: helm build server https://review.openstack.org/412100 | 18:46 |
| sbezverk | portdirect_: for now we could use att image for dynamic pvc, but since other folks hit the same issue I think kube community fix it | 18:46 |
| kfox1111 | sbezverk: +1 for kube community to provide maintined images. | 18:47 |
| portdirect_ | yeah if they do thats great - I'll still continue with adding the k8s images though as a workaround for now though? | 18:47 |
| *** senk has quit IRC | 18:48 | |
| kfox1111 | sbezverk: is att committed to keeping an up to date image? | 18:48 |
| kfox1111 | sbezverk: otherwise, I think we should have kolla repo should provide the build stuff. so users can keep it up to date. | 18:48 |
| portdirect_ | kfox1111: they are the same boat as us on that front, so I'd see no reason why not | 18:48 |
| sbezverk | portdirect_: sure I was not aware about kubeadm option to use custom image.. | 18:48 |
| *** senk has joined #openstack-kolla | 18:49 | |
| portdirect_ | sbezverk: only in master at the moment, but should come through pretty quick, its not possible to use in china at the moment... | 18:49 |
| sbezverk | portdirect_: are you serious about your last comment? or it was a joke about china? | 18:50 |
| kfox1111 | sbezverk: that may be an enterprise use case. | 18:50 |
| kfox1111 | firewall blocks pulling from some places. | 18:50 |
| portdirect_ | sbezverk I think serious, I'm reporting what people have told me | 18:50 |
| portdirect_ | google servers? | 18:51 |
| kfox1111 | probably. | 18:51 |
| sbezverk | portdirect_: wow | 18:51 |
| portdirect_ | :( | 18:51 |
| kfox1111 | portdirect_: tengentially, I was thinking maybe we provide a job that builds gainst trunk, | 18:51 |
| kfox1111 | then does a docker save at the very end of a successfull job of all containers on the system. | 18:51 |
| kfox1111 | would make a good cache for subsequent runs. | 18:52 |
| kfox1111 | other jobs could then docker load that tarball and not have to pull any containers. | 18:52 |
| kfox1111 | same trick could work for the enterprise use case. | 18:52 |
| v1k0d3n | sbezverk: we're using kubeadm in our lab as well. this is why i asked the question earlier today. | 18:53 |
| portdirect_ | kfox1111: that sounds good - be good to have a bleeding edge gate | 18:53 |
| sbezverk | v1k0d3n: right, but you change manifest after kubeadm finish right? to point to your image | 18:54 |
| v1k0d3n | and kfox1111 yes, in our case currently ceph is a requirement...so we will definitely be maintaining those. | 18:54 |
| kfox1111 | portdirect_: +1. would be harder then the rest, but totally doable. | 18:54 |
| v1k0d3n | welcome to call me out to answer some of these questions. | 18:54 |
| portdirect_ | sbezverk: or preload the image and retag it - this is the current recomendation from upstream | 18:54 |
| kfox1111 | portdirect_: +1 for preload/retag the image solution until image can be specified. | 18:55 |
| kfox1111 | v1k0d3n: k. | 18:55 |
| v1k0d3n | in fact, this week i am going to add to an automatic CI build for each release. the only thing i'm going a little bit around about, is storing the compiled bin in the repo. i really don't like carrying the extra weight in git. | 18:55 |
| portdirect_ | v1k0d3n i dont get why you do that - why not just curl it in? | 18:56 |
| portdirect_ | or are you using a cutomised binary? | 18:56 |
| kfox1111 | v1k0d3n: yeah. I don't think goog'es going to drop the binaries so just linking to it is probably pretty safe. | 18:56 |
| v1k0d3n | sbezverk: "sure I was not aware about kubeadm option to use custom image" this is also critical functionality to us. one of the reasons why we're throwing support in kubeadm...it just seems to be the path for k8s going forward, even though it still lacks some features today (like self-hosted, which we want to move to). | 18:57 |
| v1k0d3n | portdirect_away: yup, we pull, retag, use. that's the way to go. | 18:58 |
| v1k0d3n | portdirect_away: totally fine with curling in. just need to set up the external server and slap an att domain name on it :) | 18:59 |
| kfox1111 | I do kind of like the idea of putting the container in the kolla container build system for the reason of, | 18:59 |
| kfox1111 | the rbd can be made to match the release. | 18:59 |
| v1k0d3n | portdirect_away: no, not customized, just built for each release. | 18:59 |
| kfox1111 | so 2.0.2 ceph hamer, 3.0.1 ceph jewel. | 18:59 |
| kfox1111 | though I don't think that will matter much longer. | 18:59 |
| sbezverk | bb in 1 hour | 18:59 |
| SamYaple | at the rate its going, kolla 6.0 ceph kraken | 18:59 |
| v1k0d3n | LOL | 19:00 |
| kfox1111 | SamYaple: not sure we dont want to stick to lts. | 19:00 |
| SamYaple | might even be 7.0 | 19:00 |
| kfox1111 | so jewel would be around for while. | 19:00 |
| SamYaple | kfox1111: you _dont_ want lts? | 19:00 |
| kfox1111 | SamYaple: sorry. worded badly. I think we mgiht want to stick to lts? | 19:00 |
| SamYaple | yea id agree | 19:01 |
| SamYaple | kraken has just taken a long time | 19:01 |
| kfox1111 | yeah. now that jewel is out, they get to play with a bunch of stuff. so they are in play mode I think. | 19:01 |
| v1k0d3n | are you guys using ceph the same way; as a requirement for multi-node shared store in the kube (by default)? | 19:03 |
| v1k0d3n | been a little while since i've looked honestly. | 19:04 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: WIP: neutron service package https://review.openstack.org/412155 | 19:04 |
| kfox1111 | v1k0d3n: yeah. | 19:04 |
| v1k0d3n | cool | 19:04 |
| kfox1111 | nothings blocking someone else from implementing a different solution though. | 19:04 |
| kfox1111 | I'm just most familior with ceph. | 19:04 |
| v1k0d3n | so they just use a different helm chart of their own? that simple? | 19:05 |
| kfox1111 | the pv/pvc charts are tepmlated out so you can select different backends. | 19:05 |
| v1k0d3n | cool | 19:05 |
| kfox1111 | you could use aws block storage for example. | 19:05 |
| openstackgerrit | Pete Birley proposed openstack/kolla: Add Kubernetes Images for Kolla-Kubernetes https://review.openstack.org/412205 | 19:06 |
| kfox1111 | portdirect_: don't you have to specify the binary to the command? | 19:06 |
| v1k0d3n | be really interested to see those usecase and documented. | 19:07 |
| v1k0d3n | i feel like that could get really messy | 19:07 |
| kfox1111 | yean. | 19:07 |
| portdirect_ | kfox1111, no these are the same as the upstreams | 19:07 |
| kfox1111 | upstream just sets them blank? then always overrides them on launch? | 19:07 |
| portdirect_ | kfox1111 yup | 19:08 |
| kfox1111 | k. +1 then. :) | 19:08 |
| portdirect_ | weird i know | 19:08 |
| kfox1111 | makes some sence. it may be pretty unusable without at least some args. | 19:08 |
| kfox1111 | they really should at least specify the binary though. | 19:09 |
| portdirect_ | yeah thou i'd kind expect them to have the binary as entrypoint so you could just provide the args, but thats how they roll | 19:10 |
| portdirect_ | they used to do it with hyperkube, but dont think they ever had for the broken out images | 19:10 |
| kfox1111 | ah. | 19:12 |
| SamYaple | does anyone know if this static uid/gid stuff has a blueprint? | 19:13 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: [WIP] static uid/gid in Dockerfiles https://review.openstack.org/412231 | 19:15 |
| portdirect_ | v1k0d3n: we are using ceph slightly differently from aic at the moment, as we were supporting 1.4, but i want us to move over to dynamic pvc by default now we are on k8s 1.5, but leave the manual creation in for users who want the extra controll | 19:15 |
| SamYaple | portdirect_: would appreciate a once over https://review.openstack.org/#/c/412231/1 | 19:16 |
| SamYaple | ended up using config file | 19:16 |
| v1k0d3n | portdirect_away: that's what i thought actually. that's one of the reasons we moved to 1.5, so i figured there were differences. | 19:16 |
| v1k0d3n | *that those were some of /the differences | 19:16 |
| portdirect_ | SamYaple: so every user in every container? | 19:17 |
| SamYaple | portdirect_: yep | 19:17 |
| kfox1111 | v1k0d3n: they are wrapped up in the pvc abstraction we're using. | 19:17 |
| kfox1111 | so adding ceph rbd dynamic provisioning support is pretty trivial. | 19:18 |
| SamYaple | portdirect_: remember, they share a common log dir /var/log/kolla. so they do "share" files. its worth having proper accounts attached to those files | 19:18 |
| SamYaple | instead of seeing it is owned by 401:401, its owned by nova:nova, no matter the contaienr youre in | 19:18 |
| portdirect_ | good point, wfm | 19:18 |
| kfox1111 | v1k0d3n: see: https://github.com/openstack/kolla-kubernetes/blob/master/helm/kolla-common/templates/_common_pvc.yaml | 19:19 |
| *** gomarivera has joined #openstack-kolla | 19:20 | |
| portdirect_ | kfox1111: so in that file it's only https://github.com/openstack/kolla-kubernetes/blob/master/helm/kolla-common/templates/_common_pvc.yaml#L8, we need to update, to 'beta' and 'general' | 19:20 |
| kfox1111 | portdirect_: yup. | 19:21 |
| SamYaple | portdirect_: also baking it into the config allows user defined 'users' for whatever they are doing to extend these files should they wany | 19:21 |
| kfox1111 | and make the storage class specified by variable. | 19:21 |
| portdirect_ | kfox1111: yup - though I'm not sure (never tried) how helm will reposnd to an empty pv.yaml? | 19:22 |
| kfox1111 | portdirect_: I think we probably should leave the pv's out of helm super packages. | 19:22 |
| kfox1111 | maybe just the pvc's. | 19:22 |
| portdirect_ | right, being fed, back soon :) | 19:22 |
| kfox1111 | k. :) | 19:23 |
| portdirect_ | kfox1111: yeah that makes sense | 19:23 |
| *** portdirect_ is now known as portdirect_away2 | 19:23 | |
| kfox1111 | I've always seen the pv's as an admin thing. | 19:23 |
| kfox1111 | so having pv scripts are just batteries. | 19:23 |
| kfox1111 | http://logs.openstack.org/00/412100/14/check/gate-kolla-kubernetes-deploy-centos-binary-external-ovs-nv/a575d01/console.html is really confusing me. :/ | 19:24 |
| *** gomarivera has quit IRC | 19:24 | |
| kfox1111 | helm search # shows the package. | 19:24 |
| kfox1111 | but helm install cant find it. :/ | 19:24 |
| kfox1111 | very weird... | 19:28 |
| *** portdirect_away2 has quit IRC | 19:33 | |
| openstackgerrit | Sam Yaple proposed openstack/kolla: [WIP] static uid/gid in Dockerfiles https://review.openstack.org/412231 | 19:38 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: WIP: neutron service package https://review.openstack.org/412155 | 19:41 |
| *** jmccarthy has quit IRC | 19:41 | |
| v1k0d3n | portdirect_away: have you played with some of the other charts, like jenkins upstream? | 19:42 |
| *** jmccarthy has joined #openstack-kolla | 19:42 | |
| v1k0d3n | this imagePullPolicy nonsense is really annoying stuff. | 19:42 |
| kfox1111 | yeah. I don't know why they started enforcing it on init containers but not the rest. | 19:45 |
| kfox1111 | seems silly. | 19:45 |
| v1k0d3n | well, referring to the bugs for not having a default ipp | 19:52 |
| v1k0d3n | frustrating to work around it. | 19:52 |
| kfox1111 | yeah. | 19:53 |
| kfox1111 | had to touch almost all of the kolla-kubernetes tempaltes to fix for 1.5. | 19:53 |
| v1k0d3n | same here | 19:55 |
| kfox1111 | seems silly not to require bumping to v2 and adding imagepullpolicy to all the other containers too. | 19:55 |
| v1k0d3n | and caught up on a strange failure on the k8s side too | 19:55 |
| kfox1111 | if your goint to touch them all, might as well do them all. :/ | 19:55 |
| v1k0d3n | hahaha | 19:55 |
| v1k0d3n | so everything breaks | 19:55 |
| kfox1111 | yeah. break them all if your going to basically break all the pods anyway. get it over with. just pul the bandade off all at once. :) | 19:56 |
| v1k0d3n | @dims hey man | 19:56 |
| v1k0d3n | you around? | 19:56 |
| v1k0d3n | i noticed you're a core member over there? is there any way to get some eyes on this one? https://github.com/kubernetes/kubernetes/pull/38574 | 19:57 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: WIP: neutron service package https://review.openstack.org/412155 | 20:01 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: WIP: helm build server https://review.openstack.org/412100 | 20:03 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: [WIP] static uid/gid in Dockerfiles https://review.openstack.org/412231 | 20:04 |
| *** fragatina has joined #openstack-kolla | 20:05 | |
| *** Pavo has quit IRC | 20:14 | |
| *** portdirects_lapt has joined #openstack-kolla | 20:15 | |
| *** matrohon has joined #openstack-kolla | 20:16 | |
| *** fragatina has quit IRC | 20:17 | |
| *** Pavo has joined #openstack-kolla | 20:18 | |
| v1k0d3n | kfox1111 i missed your message earlier about the binaries. you mean just download from the release (all binaries), unpack them, and throw out what you don't need inside the container? | 20:19 |
| kfox1111 | v1k0d3n: yeah. I think you were debating putting the binary for the contianer in git, or just curling and throwing out what you don't need during build? | 20:20 |
| kfox1111 | was saying the latter seems relatively safe to me. | 20:20 |
| v1k0d3n | i think that's fine, just makes the container a bit more....more. be better for us to just CI them, and add a single bin. | 20:20 |
| v1k0d3n | faster, less clutter. | 20:20 |
| kfox1111 | that works too. | 20:21 |
| kfox1111 | sticking binaries in git just always felt grungy to me. :/ | 20:21 |
| kfox1111 | I can see some use cases that might fit though. | 20:21 |
| v1k0d3n | we make the container much smaller by just pre-building, and packing the bin in there. | 20:21 |
| v1k0d3n | there are different ways of doing it. none really wrong. | 20:21 |
| kfox1111 | yeah. | 20:21 |
| v1k0d3n | eventually we want to be extremely (ultra) efficient. | 20:22 |
| kfox1111 | +1 | 20:22 |
| kfox1111 | same with kolla. | 20:22 |
| v1k0d3n | for the size of our cloud...size really matters and adds up. | 20:22 |
| v1k0d3n | ok | 20:22 |
| v1k0d3n | not sure kolla can exactly do that. different case. | 20:22 |
| kfox1111 | portdirects: wants to try a alpine based kolla distro too at some point. | 20:23 |
| kfox1111 | thin it down as much as possible. | 20:23 |
| v1k0d3n | that got brought up | 20:23 |
| v1k0d3n | in BCN | 20:23 |
| v1k0d3n | we want to do it without having to go to alpine. | 20:23 |
| v1k0d3n | it can be done, but it requires a different method than what kolla is doing. | 20:23 |
| kfox1111 | yeah. I think super thin containers are orthoganal to alpine. | 20:23 |
| v1k0d3n | for 70-80% out there, alpine would probably be just fine :) | 20:24 |
| kfox1111 | both can/should be done I think. | 20:24 |
| v1k0d3n | for audits or anyone concerned about CVE....no way. | 20:24 |
| kfox1111 | yeah. I get that. :) | 20:24 |
| v1k0d3n | alpine is a total mess. | 20:24 |
| kfox1111 | I prefer centos myself. | 20:24 |
| kfox1111 | so a thinned out centos woudl be awesome. | 20:24 |
| v1k0d3n | yeah, that's tough with kolla today. not sure how you'd get there exactly while keeping the keystone > keystone-base > openstack-base model. | 20:25 |
| v1k0d3n | i see both sides entirely | 20:25 |
| sbezverk | what could be better than fresh 40 cm of snow :-) | 20:25 |
| v1k0d3n | lol | 20:25 |
| v1k0d3n | car won't start AND 40 cm of snow? | 20:25 |
| sbezverk | sorry for interrupting your highly technical talk ;-) | 20:26 |
| v1k0d3n | "highly" ha! | 20:26 |
| v1k0d3n | please do :) | 20:26 |
| kfox1111 | v1k0d3n: I think kolla shoudl provide all the above options. it fits in kolla's mission to provide usable openstack containers for all interested in it. | 20:26 |
| v1k0d3n | ok. that's cool. | 20:26 |
| *** portdirects_lapt has quit IRC | 20:27 | |
| sbezverk | v1k0d3n: entrypoint can block on pvc state? | 20:29 |
| kfox1111 | sbezverk: no. :/ | 20:29 |
| kfox1111 | sbezverk: entnrypoint needs to run in a container. pvc's have no container. | 20:30 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: WIP: neutron service package https://review.openstack.org/412155 | 20:30 |
| kfox1111 | helm treats the 'local' repo differently then the rest of repo's for some reason. :/ | 20:30 |
| kfox1111 | its "magic" | 20:30 |
| sbezverk | kfox1111: I do not see a difference, if it can check for a job to be present , why it cannot check for pvc state? am I missing something? | 20:30 |
| kfox1111 | sbezverk: oh. you mean, check if a pvc is created? | 20:31 |
| sbezverk | like when it block of mariadb service or mariadb bootstrap job to be completed | 20:31 |
| sbezverk | why not to check for a state of specific pvc? | 20:31 |
| kfox1111 | sbezverk: its not in the list here: https://github.com/stackanetes/kubernetes-entrypoint | 20:32 |
| kfox1111 | should probably put in a feature request. | 20:32 |
| kfox1111 | on the other hand, | 20:32 |
| kfox1111 | I think the pod will naturally block while its not in existence? | 20:32 |
| kfox1111 | same with configmaps/secrets. | 20:33 |
| sbezverk | kfox1111: not saure if it is sufficient I got pvc in pending state | 20:33 |
| kfox1111 | sbezverk: https://review.openstack.org/#/c/412100/ looks like its getting close. can you please have a look and let me konw what you think? | 20:33 |
| sbezverk | sure thing | 20:34 |
| kfox1111 | sbezverk: it does a more helm native build. with all the infrastructure to do so. | 20:34 |
| sbezverk | kfox1111: what builds requirements.yaml? | 20:35 |
| kfox1111 | sbezverk: nothing. they are checked in. | 20:35 |
| kfox1111 | it doesnt' change microservice building. at least for now. | 20:35 |
| sbezverk | I am just curious the port used you picked it up or what is the deal | 20:35 |
| kfox1111 | completely random. | 20:36 |
| kfox1111 | I really hate that part. :/ | 20:36 |
| kfox1111 | feels like it will be a bit fragile. | 20:36 |
| kfox1111 | but helm doesn't support sane building there yet. :/ | 20:36 |
| kfox1111 | but once that localpath dep feature makes it in, we can get rid of a lot of it and get back to something more sane. :/ | 20:37 |
| sbezverk | yeah would be much safer to pull packages frm the path | 20:39 |
| kfox1111 | v1k0d3n: if you can help push that through faster, that would be great. | 20:39 |
| kfox1111 | v1k0d3n: this: https://github.com/kubernetes/helm/issues/1639 | 20:40 |
| kfox1111 | sbezverk: an exaple of a package with requirements here: | 20:41 |
| kfox1111 | https://review.openstack.org/#/c/412155/ | 20:41 |
| v1k0d3n | we don't think we have the same issue. | 20:41 |
| v1k0d3n | we're building charts differently. | 20:41 |
| v1k0d3n | not that it wouldn't be useful. | 20:42 |
| v1k0d3n | just saying... | 20:42 |
| kfox1111 | v1k0d3n: I talked to alen. he's doing somethign similar to the review. | 20:42 |
| kfox1111 | v1k0d3n: thats why I'm proposing that as a solutionk rather then something far less complicated but less helm native. | 20:42 |
| kfox1111 | v1k0d3n: to keep the two solutions more similar. | 20:43 |
| v1k0d3n | gotcha | 20:43 |
| sbezverk | kfox1111: you add base distro at the end of globals.yaml, what if before there is a conflicting setting, how does it get parsed? latest setting wins? | 20:43 |
| SamYaple | esh. huge security problem | 20:44 |
| kfox1111 | sbezverk: in which file/review? | 20:44 |
| *** portdirect has joined #openstack-kolla | 20:45 | |
| SamYaple | GAHH inc0 and pbourke | 20:45 |
| kfox1111 | portdirect: this is looking pretty close to ready: https://review.openstack.org/#/c/412100/ can you please weighin on the aproach? | 20:45 |
| SamYaple | they broke all security | 20:45 |
| v1k0d3n | i think it's already slated for 2.2.0, but the impact for us isn't quite as great. we worked around it. just different ways to solving issue (which is totally fine). | 20:45 |
| SamYaple | https://review.openstack.org/#/c/241438/9/docker/base/sudoers | 20:46 |
| SamYaple | anyone see a problem with that? | 20:46 |
| v1k0d3n | our biggest thing was 2.1.0 and 1.5, just a bit different stages and solving different ways. | 20:46 |
| sbezverk | kfox1111: line 7 https://review.openstack.org/#/c/412100/15/tests/bin/setup_config.sh | 20:46 |
| kfox1111 | v1k0d3n: says 2.3.0 which is further out. | 20:46 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: [WIP] static uid/gid in Dockerfiles https://review.openstack.org/412231 | 20:46 |
| v1k0d3n | whoa! SamYaple | 20:46 |
| kfox1111 | sbezverk: ah. thats just letting it checkout as it normaly does, rather then rename kolla-ansible to kolla. | 20:47 |
| kfox1111 | it was a quick hack origionally to work around the repo split. | 20:47 |
| kfox1111 | I thought it was breaking helm though. | 20:47 |
| kfox1111 | but not sure anymore. | 20:47 |
| portdirect | v1k0d3n: biggest thing i've noticed re helm charts repo vs aic so far is that most charts in the k8s repo use generic whereas the suff in your repo uses general for the default storage class | 20:47 |
| kfox1111 | I could undo that part. | 20:48 |
| v1k0d3n | i thought it was 2.2.0 triage | 20:48 |
| kfox1111 | says 2.3.0 triage for me. | 20:48 |
| SamYaple | who is online that is a core member? | 20:48 |
| v1k0d3n | really? that's strange: technosophos added this to the 2.2.0-Triage milestone 12 days ago | 20:49 |
| v1k0d3n | in that ticket you posted. | 20:49 |
| portdirect | kfox1111: looking now, but cant test for a few hours | 20:49 |
| kfox1111 | portdirect: ok. coo. thanks. | 20:49 |
| kfox1111 | v1k0d3n: oh. I see that in the log, yeah. but on the sidebar it still says 2.3.0. weird. | 20:50 |
| *** lrensing has joined #openstack-kolla | 20:50 | |
| v1k0d3n | 2.3.0 roadmap, triage for 2.2.0 is how i took that :) | 20:50 |
| kfox1111 | ah. ok. thanks. | 20:50 |
| v1k0d3n | who knows. maybe not. | 20:50 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Remove /usr/bin/install from allowed commands https://review.openstack.org/412235 | 20:50 |
| v1k0d3n | haven't spoke directly to matt about it. | 20:51 |
| SamYaple | people should review ^ that | 20:51 |
| v1k0d3n | perhaps when we get the calls started...we can bring it up. | 20:51 |
| v1k0d3n | i am supposed to talk to lachie about this on monday. | 20:51 |
| kfox1111 | v1k0d3n: +1. | 20:51 |
| kfox1111 | v1k0d3n: I just filed this one too: https://github.com/kubernetes/helm/issues/1719 | 20:51 |
| kfox1111 | v1k0d3n: which I think would let us get rid of most of the prebuild stuff we've been doing. | 20:52 |
| sbezverk | kfox1111: look really nice!! I like the idea of runnign helm serve in the background.. | 20:53 |
| kfox1111 | sbezverk: ok. cool. | 20:54 |
| sbezverk | kfox1111: it could probably get merged as is and then fine tuned later | 20:55 |
| sbezverk | there is one failure though with cents.. | 20:55 |
| portdirect | kfox1111: so helm_build_all.sh, builds all microservices, then reindexes the repo and builds all services? then helm_buildrepo.sh takes all those packages and throws them into the final repo? | 20:55 |
| kfox1111 | sbezverk:I'll pull the wip off of it and then recheck. | 20:56 |
| sbezverk | kfox1111: sounds good, thanks | 20:56 |
| kfox1111 | portdirect: essentially yeah. it builds the microservices and the services. | 20:56 |
| v1k0d3n | kfox1111: cool. once we get an agenda going, we can bring that up as well. | 20:57 |
| kfox1111 | v1k0d3n: sounds good. :) | 20:57 |
| v1k0d3n | it would be nice to get everyone on the same page as far as design, but i've thought about this a lot lately...having differences is cool. | 20:58 |
| portdirect | kfox1111, the hard deps still scare me :) but lgtm dude :) | 20:58 |
| v1k0d3n | just opens up opportunity to make the used project stronger (in this case helm). | 20:58 |
| kfox1111 | portdirect: you and me both. :/ but its a stopgap. :/ | 20:58 |
| kfox1111 | portdirect: should be much better as soon as helm gets a better sollution in place. | 20:58 |
| v1k0d3n | btw...to SamYaple point...why would we ever commit something like that to sudo'ers? | 20:59 |
| SamYaple | v1k0d3n: it was buried in a very very large patch | 20:59 |
| SamYaple | and clearly not reviewed well enough | 20:59 |
| portdirect | kfox1111 : well whatta we gonna do? :D | 20:59 |
| v1k0d3n | noticed that. maybe an oversight? | 20:59 |
| SamYaple | yea | 20:59 |
| portdirect | kfox1111: in summary: LGTM | 20:59 |
| v1k0d3n | make sense. | 20:59 |
| SamYaple | v1k0d3n: kolla-mesos, blame that | 21:00 |
| v1k0d3n | LOL | 21:00 |
| v1k0d3n | that's always my default | 21:00 |
| v1k0d3n | :) | 21:00 |
| kfox1111 | portdirect: ok. cool. thanks for reviewing. | 21:01 |
| v1k0d3n | it'll be interesting to see how helm folks take some of thoughts. | 21:01 |
| v1k0d3n | atm there are three different helm-openstack based projects, all approaching them differently. | 21:02 |
| kfox1111 | v1k0d3n: yeah. and I think each prpoject aproaching it from different angles will help show helm where its dificiencies are. | 21:02 |
| v1k0d3n | maybe | 21:03 |
| kfox1111 | each project seems ahead on one way or another. so if they merge features that each project is concerned with, | 21:03 |
| kfox1111 | helm should be much more flexible. | 21:03 |
| v1k0d3n | they could always say "have you tried this" | 21:03 |
| v1k0d3n | which is a perfectly reasonable response. | 21:03 |
| kfox1111 | true. and thats ok. so long as they wil llisten when someone says, yeah, but .... | 21:03 |
| v1k0d3n | i mean, they are very detailed in talking through how to set up very useful, consumable charts. | 21:04 |
| v1k0d3n | that's the whole point of the SIG, and why i wanted it. | 21:04 |
| kfox1111 | yeah. my goal is both consumable and modular, | 21:04 |
| kfox1111 | which they haven't seem to have put too much thought into yet. | 21:04 |
| kfox1111 | +1 | 21:04 |
| v1k0d3n | the break down between the approaches wasn't healthy. meeting in helm is the best solution...talking directly with the authors. | 21:05 |
| v1k0d3n | kfox1111: everyone's goal | 21:05 |
| v1k0d3n | different approaches still. | 21:05 |
| v1k0d3n | and honestly, sometimes it's ok not to be everything to everyone. | 21:05 |
| v1k0d3n | at our sides...bite size is the only thing we can do. | 21:06 |
| v1k0d3n | otherwise we don't know how to fit an 8x8' box into a 3x3' hole. | 21:06 |
| v1k0d3n | *size | 21:07 |
| kfox1111 | yeah. | 21:07 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Helm build server https://review.openstack.org/412100 | 21:08 |
| *** portdirect has quit IRC | 21:13 | |
| sbezverk | kfox1111: will ack as soon as the gate results will show up | 21:14 |
| kfox1111 | sbezverk: cool. thanks. :) | 21:14 |
| *** matrohon has quit IRC | 21:16 | |
| *** portdirect has joined #openstack-kolla | 21:16 | |
| *** senk has quit IRC | 21:16 | |
| *** goldyfruit has joined #openstack-kolla | 21:22 | |
| *** goldyfruit has quit IRC | 21:22 | |
| v1k0d3n | portdirect_away: are you using the direct chart for jenkins in helm? | 21:24 |
| v1k0d3n | it's a great upstream chart. | 21:24 |
| v1k0d3n | esp for kuberntes builds. | 21:25 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Helm entrypoint hooks https://review.openstack.org/412155 | 21:25 |
| *** goldyfruit has joined #openstack-kolla | 21:29 | |
| *** goldyfruit has quit IRC | 21:30 | |
| *** goldyfruit has joined #openstack-kolla | 21:30 | |
| *** flaper87 has quit IRC | 21:32 | |
| *** l4yerffeJ has joined #openstack-kolla | 21:32 | |
| *** portdirect has quit IRC | 21:33 | |
| *** l4yerffeJ__ has quit IRC | 21:34 | |
| *** Jeffrey4l has quit IRC | 21:35 | |
| *** Jeffrey4l has joined #openstack-kolla | 21:36 | |
| sbezverk | kfox1111: in prep to your building tools I am pushing mariadb service, which is less complex and will be easy to test.. | 21:37 |
| kfox1111 | sbezverk: https://review.openstack.org/#/c/412100/ looks to be ready. | 21:38 |
| kfox1111 | oh. either way. | 21:39 |
| sbezverk | kfox1111: nice !!!!! | 21:39 |
| openstackgerrit | Merged openstack/kolla-kubernetes: Helm build server https://review.openstack.org/412100 | 21:41 |
| sbezverk | kfox1111: so for testing service deployment we need to build values.yaml file with parameters of the gate instance and feed it to service in command line? | 21:41 |
| kfox1111 | yeah. | 21:41 |
| v1k0d3n | kfox1111: green gates look good :) | 21:42 |
| kfox1111 | sbezverk: https://review.openstack.org/#/c/412155/9/tests/bin/ceph_workflow.sh has an example. | 21:43 |
| kfox1111 | sbezverk: once I get a little further along, I wana try that helm trunk build with globals and see if we can use your nested globals idea. | 21:44 |
| kfox1111 | would make it a bit easier. | 21:44 |
| kfox1111 | sbezverk: there's a check experimental for helm-entrypoint now so should be able to start testing a bit like in that review. | 21:44 |
| openstackgerrit | Serguei Bezverkhi proposed openstack/kolla-kubernetes: Mariadb service deployment https://review.openstack.org/412237 | 21:48 |
| sbezverk | kfox1111: sounds good | 21:49 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Helm entrypoint hooks https://review.openstack.org/412155 | 21:54 |
| openstackgerrit | Serguei Bezverkhi proposed openstack/kolla-kubernetes: Mariadb service deployment https://review.openstack.org/412237 | 21:57 |
| *** msimonin has joined #openstack-kolla | 22:02 | |
| *** srwilkers has joined #openstack-kolla | 22:04 | |
| srwilkers | afternoon o/ | 22:05 |
| SamYaple | o/ | 22:05 |
| *** portdirect_away is now known as portdirect | 22:06 | |
| portdirect | evening all o/ | 22:07 |
| SamYaple | Jeffrey4l: you aroung? | 22:07 |
| openstackgerrit | Steve Wilkerson proposed openstack/kolla-kubernetes: Helm charts for cinder keystone v1 endpoints https://review.openstack.org/410923 | 22:08 |
| srwilkers | hey SamYaple portdirect, hows it going gents? | 22:08 |
| SamYaple | alright, yourself? | 22:08 |
| srwilkers | eh. lazy sunday | 22:08 |
| *** matrohon has joined #openstack-kolla | 22:09 | |
| portdirect | been diy'ing down at the folks... back in the pad now - living the dream, not sure who's dream, but living it :) | 22:10 |
| kfox1111 | portdirect: hehe | 22:12 |
| srwilkers | hey kfox1111 :) | 22:12 |
| kfox1111 | srwilkers: hey. :) | 22:12 |
| kfox1111 | srwilkers: that ps looks good to me. | 22:12 |
| v1k0d3n | hey portdirect where you ever away...seems like you're always here. | 22:12 |
| v1k0d3n | and this is pretty much the kfox1111 and portdirect chan :) | 22:13 |
| v1k0d3n | like 24x7 | 22:13 |
| v1k0d3n | i don't even know when you sleep portdirect | 22:13 |
| *** Pavo has quit IRC | 22:14 | |
| srwilkers | v1k0d3n, he doesnt | 22:15 |
| portdirect | lol - feels that way - I'm pretty light on sleep, generally try and consume 31 hours of media a day mobile comms are a dangerous thing - I've basicly reverted into a coffee fueled moody teenager with a keyboard, looking forward to some d/t over the next few weeks and reconnecting with reality | 22:15 |
| srwilkers | i think this is the first day i havent had my laptop running all day within an arms distance, and i feel weird about it | 22:15 |
| *** goldyfruit has quit IRC | 22:15 | |
| SamYaple | freakin adicts | 22:16 |
| *** goldyfruit has joined #openstack-kolla | 22:16 | |
| srwilkers | can confirm, am junkie | 22:16 |
| *** matrohon has quit IRC | 22:16 | |
| *** Pavo has joined #openstack-kolla | 22:18 | |
| *** goldyfruit has quit IRC | 22:24 | |
| *** msimonin has quit IRC | 22:25 | |
| kfox1111 | sbezverk: I fixed the macro thing, and the tests are all green. so should be good to go? https://review.openstack.org/#/c/412155/ | 22:27 |
| sbezverk | kfox1111: will do. can you check this link. It looks like Chart.yaml does not get generated | 22:28 |
| sbezverk | http://logs.openstack.org/37/412237/2/check/gate-kolla-kubernetes-deploy-centos-binary-external-ovs-nv/c2c8b4f/console.html#_2016-12-18_22_09_30_012258 | 22:28 |
| kfox1111 | thx. | 22:28 |
| kfox1111 | k. will do | 22:28 |
| kfox1111 | ah. yeah. Chart.yaml isn't generated. like with the microservices. so gota check one in. | 22:29 |
| kfox1111 | gota head out for a few min. be back in ~ 20 min. | 22:30 |
| sbezverk | kfox1111: ok got it byt some reason I thought it gets generated .. | 22:30 |
| openstackgerrit | Merged openstack/kolla-kubernetes: Helm entrypoint hooks https://review.openstack.org/412155 | 22:31 |
| *** msimonin has joined #openstack-kolla | 22:32 | |
| *** sayantan_ has quit IRC | 22:34 | |
| *** awiddersheim has quit IRC | 22:35 | |
| v1k0d3n | SamYaple: you're here quite a bit too again. | 22:35 |
| SamYaple | im not here | 22:36 |
| v1k0d3n | i remember seeing you a ton a while back when kolla kicked off. | 22:36 |
| v1k0d3n | but you left for a while. now you're back all the time. | 22:36 |
| openstackgerrit | Serguei Bezverkhi proposed openstack/kolla-kubernetes: Mariadb service deployment https://review.openstack.org/412237 | 22:36 |
| srwilkers | sbezverk, mariadb looking good :) going to start looking at some of the other services this next week | 22:43 |
| kfox1111 | back. | 22:47 |
| srwilkers | wb kfox1111 | 22:47 |
| kfox1111 | thx. :) | 22:49 |
| kfox1111 | srwilkers: +1 for looking at other services. I think we should play around with a bunch in parallel but not try and merge yet. | 22:58 |
| kfox1111 | not sure some common patterns we should be adopting, and they might fall out once we have enough in revew. | 22:58 |
| srwilkers | kfox1111, i agree | 22:59 |
| kfox1111 | I'm going to try and push all the infra from the one I've been working on down into trunk though, so we can easily play in parallel. | 23:00 |
| kfox1111 | most is merged now. but still needs a bit to make it easy to hook into the experimental gate. | 23:00 |
| kfox1111 | then we can test in gate. | 23:00 |
| srwilkers | awesome | 23:02 |
| openstackgerrit | Serguei Bezverkhi proposed openstack/kolla-kubernetes: Mariadb service deployment https://review.openstack.org/412237 | 23:03 |
| *** awiddersheim has joined #openstack-kolla | 23:11 | |
| sbezverk | kfox1111: I did not get your comment about init-element, could you please clarify.. | 23:11 |
| kfox1111 | sbezverk: saying, can you make the changes in helm/microservices/mariadb/.... instead of in the copied subchart? | 23:14 |
| kfox1111 | I think we can avoid forking the files at this point. | 23:15 |
| kfox1111 | makes it easier to review the changes. | 23:15 |
| sbezverk | kfox1111: got it | 23:19 |
| *** dave-mccowan has joined #openstack-kolla | 23:22 | |
| openstackgerrit | Sam Yaple proposed openstack/kolla: [WIP] static uid/gid in Dockerfiles https://review.openstack.org/412231 | 23:22 |
| *** sayantan_ has joined #openstack-kolla | 23:23 | |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Helm-entrypoint git hooks https://review.openstack.org/412238 | 23:25 |
| *** dave-mcc_ has joined #openstack-kolla | 23:25 | |
| kfox1111 | k. that should be the last of the infra that can be shared I think. | 23:25 |
| *** msimonin has quit IRC | 23:25 | |
| *** dave-mccowan has quit IRC | 23:27 | |
| *** dave-mc__ has joined #openstack-kolla | 23:27 | |
| *** sayantan_ has quit IRC | 23:27 | |
| *** dave-mcc_ has quit IRC | 23:29 | |
| sbezverk | kfox1111: yes yes yes.. nested in globals are working now :-) | 23:31 |
| sbezverk | {{- define "common_macro" }} | 23:31 |
| sbezverk | {{- $microservice := index .Values.global .microserviceName }} | 23:31 |
| sbezverk | {{- $variable_1 := index $microservice "variable_1" }} | 23:31 |
| sbezverk | name_global: {{ $variable_1 }} | 23:31 |
| sbezverk | {{- end }} | 23:31 |
| kfox1111 | awesome. | 23:31 |
| kfox1111 | that with their test release? | 23:31 |
| sbezverk | yep canary image | 23:32 |
| srwilkers | sbezverk, :) | 23:32 |
| kfox1111 | cool. | 23:32 |
| srwilkers | thats awesome | 23:32 |
| kfox1111 | we should snapshot that somewhere until the 2.1.1 release cuts. | 23:32 |
| kfox1111 | that way we dont have it break on us suddenly. | 23:32 |
| kfox1111 | is it just a client thing, or a server thing too? | 23:33 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: [WIP] static uid/gid in Dockerfiles https://review.openstack.org/412231 | 23:33 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: [WIP] static uid/gid in Dockerfiles https://review.openstack.org/412231 | 23:34 |
| *** mattmceuen has joined #openstack-kolla | 23:36 | |
| srwilkers | hey mattmceuen o/ | 23:36 |
| mattmceuen | yo srwilkers! | 23:36 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: WIP: neutron service package https://review.openstack.org/412239 | 23:36 |
| SamYaple | i wish nova-libvirt was just named 'libvirtd' | 23:36 |
| kfox1111 | there we go... taht would be a little more reviewable.. | 23:37 |
| kfox1111 | SamYaple: +1. :/ | 23:37 |
| kfox1111 | SamYaple: and +1 for optional. :/ | 23:37 |
| kfox1111 | there's a few places where I'd like to run kolla, but have libvirt alrady on the host. | 23:38 |
| SamYaple | at one point it was options :P | 23:38 |
| SamYaple | optional* | 23:38 |
| kfox1111 | lack of gate testing probably. :/ | 23:38 |
| SamYaple | but since ansible didnt bind to the host... eh | 23:38 |
| SamYaple | no | 23:38 |
| kfox1111 | easy to break that kind of thing. | 23:38 |
| SamYaple | its just not really usable unless you bind /var/lib/nova to the host | 23:38 |
| kfox1111 | thats an option though. | 23:39 |
| SamYaple | something kolla didnt really want to do | 23:39 |
| SamYaple | yea but it wasnt when libvirt was also optional lol | 23:39 |
| SamYaple | anyway. we can rename nova-libvirt by doing it with tags and a deprecation cycle | 23:39 |
| *** cliles has quit IRC | 23:39 | |
| kfox1111 | yeah. | 23:40 |
| SamYaple | basically produce ubuntu-source-libvirtd:4.0.0 and ubuntu-source-nova-libvirt:4.0.0 | 23:40 |
| kfox1111 | srwilkers: nice. :) | 23:42 |
| SamYaple | btw, if anyone wants to comment on my implementation of the static uid/gid... nows the time https://review.openstack.org/#/c/412231/ | 23:43 |
| kfox1111 | SamYaple: yeah. that would work. | 23:43 |
| SamYaple | if i need to change it I would like to know sooner rather than later | 23:43 |
| kfox1111 | sounds good to me. | 23:44 |
| sbezverk | kfox1111: I did both | 23:44 |
| openstackgerrit | Merged openstack/kolla-kubernetes: Helm charts for cinder keystone v1 endpoints https://review.openstack.org/410923 | 23:44 |
| kfox1111 | sbezverk: ah. would be much more complicated cloning the docker container too. :/ | 23:45 |
| kfox1111 | sbezverk: would be nice if it was a client only thing. | 23:46 |
| kfox1111 | sbezverk: but am kind of thinking vars are processed on the server. | 23:46 |
| *** dave-mc__ has quit IRC | 23:46 | |
| mattmceuen | multinode kolla-ansible question -- one of my nodes is not like the others w.r.t. interface names. I'm trying to override its interface names in my inventory like so: | 23:47 |
| mattmceuen | 192.168.1.3 network_interface=enp1s0f1 neutron_external_interface=enp1s0f0 kolla_external_vip_interface=enp1s0f0 api_interface=enp1s0f0 storage_interface=enp1s0f1 tunnel_interface=enp1s0f1 | 23:47 |
| mattmceuen | 1 | 23:47 |
| mattmceuen | but I'm still getting an error during deployment that that node doesn't have the "default" api_interface (eno1 in my case). Can anyone tell me if that's the correct way to specify per-host interface names? | 23:48 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Helm-entrypoint git hooks https://review.openstack.org/412238 | 23:49 |
| *** l4yerffeJ has quit IRC | 23:51 | |
| *** l4yerffeJ has joined #openstack-kolla | 23:51 | |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: WIP: neutron service package https://review.openstack.org/412239 | 23:51 |
| openstackgerrit | Serguei Bezverkhi proposed openstack/kolla-kubernetes: Mariadb service deployment https://review.openstack.org/412237 | 23:55 |
| sbezverk | kfox1111: not really because I always run --dry-run and I doubt tiller is participating here | 23:56 |
| sbezverk | kfox1111: I did both just in case ;-) | 23:57 |
| openstackgerrit | Serguei Bezverkhi proposed openstack/kolla-kubernetes: Mariadb service deployment https://review.openstack.org/412237 | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!