| *** kevzha01 has joined #openstack-kolla | 00:00 | |
| *** Pavo has joined #openstack-kolla | 00:00 | |
| *** newmember has quit IRC | 00:00 | |
| *** kevzha01_ has joined #openstack-kolla | 00:01 | |
| *** kevzha01 has quit IRC | 00:04 | |
| *** kevzha01_ has quit IRC | 00:09 | |
| *** yangyapeng has quit IRC | 00:15 | |
| *** yangyapeng has joined #openstack-kolla | 00:15 | |
| *** danardelean has quit IRC | 00:15 | |
| *** glycerine has quit IRC | 00:19 | |
| *** yangyapeng has quit IRC | 00:20 | |
| *** Pavo has quit IRC | 00:22 | |
| *** Pavo has joined #openstack-kolla | 00:25 | |
| *** mummer has quit IRC | 00:31 | |
| *** danardelean has joined #openstack-kolla | 00:31 | |
| *** danardelean has quit IRC | 00:35 | |
| *** kbaegis has joined #openstack-kolla | 00:41 | |
| *** trozet has quit IRC | 00:45 | |
| *** tovin07_ has joined #openstack-kolla | 00:51 | |
| *** heigao has joined #openstack-kolla | 00:54 | |
| *** kiennt26 has joined #openstack-kolla | 00:55 | |
| *** glycerine has joined #openstack-kolla | 00:56 | |
| *** Pavo has quit IRC | 00:56 | |
| *** Pavo has joined #openstack-kolla | 00:56 | |
| *** Pavo has quit IRC | 00:58 | |
| *** rwsu has joined #openstack-kolla | 01:02 | |
| *** dave-mccowan has quit IRC | 01:05 | |
| *** heigao has quit IRC | 01:09 | |
| *** dave-mccowan has joined #openstack-kolla | 01:12 | |
| *** lukepatrick has joined #openstack-kolla | 01:14 | |
| *** lxdong has quit IRC | 01:15 | |
| *** tovin07_ has quit IRC | 01:16 | |
| openstackgerrit | Heetae Ahn proposed openstack/kolla master: Add RHEL subscription registration https://review.openstack.org/495148 | 01:18 |
|---|---|---|
| *** salv-orl_ has joined #openstack-kolla | 01:18 | |
| *** salv-orlando has quit IRC | 01:21 | |
| *** kiennt26 has quit IRC | 01:23 | |
| *** huanxie has joined #openstack-kolla | 01:24 | |
| *** yangyapeng has joined #openstack-kolla | 01:26 | |
| *** yangyape_ has joined #openstack-kolla | 01:27 | |
| *** yangyapeng has quit IRC | 01:27 | |
| *** daidv has joined #openstack-kolla | 01:27 | |
| *** unicell has joined #openstack-kolla | 01:30 | |
| *** unicell has quit IRC | 01:34 | |
| *** kiennt26 has joined #openstack-kolla | 01:34 | |
| *** tovin07_ has joined #openstack-kolla | 01:41 | |
| *** itlinux has joined #openstack-kolla | 01:51 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible master: Optimize zuul v3 jobs https://review.openstack.org/513612 | 01:58 |
| *** lxdong has joined #openstack-kolla | 02:03 | |
| lxdong | Will session client reduce tokens growing in the keystone database? | 02:03 |
| korean101 | Hi guys | 02:08 |
| korean101 | http://paste.openstack.org/show/624651/ | 02:08 |
| korean101 | anyone knows MTU set? | 02:08 |
| korean101 | i use nested KVM and Docker | 02:08 |
| korean101 | i login to VM but can't reach outside | 02:09 |
| *** zhubingbing_ has joined #openstack-kolla | 02:10 | |
| *** rwsu has quit IRC | 02:11 | |
| *** lukepatrick has quit IRC | 02:13 | |
| *** unicell has joined #openstack-kolla | 02:13 | |
| *** rwsu has joined #openstack-kolla | 02:14 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible master: Remove test variable for service_name in panko.conf file https://review.openstack.org/513640 | 02:15 |
| *** unicell has quit IRC | 02:17 | |
| *** jamesbenson has joined #openstack-kolla | 02:19 | |
| *** vhosako__ has quit IRC | 02:22 | |
| *** vhosakot has joined #openstack-kolla | 02:22 | |
| openstackgerrit | Merged openstack/kolla stable/pike: Use upgrade rather than create_schema for wather database https://review.openstack.org/513609 | 02:23 |
| *** unicell has joined #openstack-kolla | 02:26 | |
| *** zhubingbing_ has quit IRC | 02:27 | |
| *** zhubingbing_ has joined #openstack-kolla | 02:27 | |
| *** itlinux has quit IRC | 02:29 | |
| *** unicell has quit IRC | 02:30 | |
| *** masuberu has quit IRC | 02:34 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible stable/pike: Fix prechecks fail with docker not in deployment host https://review.openstack.org/515267 | 02:34 |
| *** unicell has joined #openstack-kolla | 02:38 | |
| *** pratapagoutham has quit IRC | 02:38 | |
| *** unicell has quit IRC | 02:42 | |
| *** larsks has quit IRC | 02:42 | |
| *** larsks has joined #openstack-kolla | 02:44 | |
| *** itlinux has joined #openstack-kolla | 02:45 | |
| *** vhosakot_ has joined #openstack-kolla | 02:47 | |
| *** itlinux has quit IRC | 02:48 | |
| openstackgerrit | Duong Ha-Quang proposed openstack/kolla-ansible master: Revert "Remove show_multiple_locations in glance-api" https://review.openstack.org/509792 | 02:48 |
| *** vhosakot has quit IRC | 02:48 | |
| *** zhubingbing has joined #openstack-kolla | 02:50 | |
| *** unicell has joined #openstack-kolla | 02:50 | |
| *** zhubingbing_ has quit IRC | 02:51 | |
| *** lhx__ has joined #openstack-kolla | 02:53 | |
| *** itlinux has joined #openstack-kolla | 02:53 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/kolla-kubernetes master: Updated from global requirements https://review.openstack.org/509443 | 02:54 |
| *** rwsu has quit IRC | 02:54 | |
| *** unicell has quit IRC | 02:55 | |
| *** masber has joined #openstack-kolla | 02:55 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/kolla-kubernetes master: Updated from global requirements https://review.openstack.org/509443 | 02:55 |
| *** masber has quit IRC | 02:57 | |
| *** lukepatrick has joined #openstack-kolla | 02:57 | |
| *** masber has joined #openstack-kolla | 02:57 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/kolla-kubernetes master: Updated from global requirements https://review.openstack.org/509443 | 02:58 |
| *** dave-mccowan has quit IRC | 03:10 | |
| *** kbaegis has quit IRC | 03:12 | |
| *** caowei has joined #openstack-kolla | 03:16 | |
| *** masber has quit IRC | 03:19 | |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible master: Optimise the number of workers for each project https://review.openstack.org/515275 | 03:22 |
| *** itlinux has quit IRC | 03:22 | |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible master: Optimise the "openstack_freezer_auth" https://review.openstack.org/515276 | 03:28 |
| openstackgerrit | SongHao proposed openstack/kolla-ansible master: Modify the instructions for setting up vip. https://review.openstack.org/512087 | 03:38 |
| *** zhubingbing_ has joined #openstack-kolla | 03:40 | |
| *** zhubingbing has quit IRC | 03:41 | |
| *** gkadam has joined #openstack-kolla | 03:42 | |
| *** mdnadeem has joined #openstack-kolla | 03:54 | |
| *** jamesbenson has quit IRC | 04:00 | |
| *** masber has joined #openstack-kolla | 04:02 | |
| openstackgerrit | Merged openstack/kolla master: Add rsyslog image https://review.openstack.org/514632 | 04:02 |
| openstackgerrit | Merged openstack/kolla-ansible master: Update the manila shares link https://review.openstack.org/513004 | 04:02 |
| *** yangyape_ has quit IRC | 04:03 | |
| *** masber has quit IRC | 04:04 | |
| *** lukepatrick has quit IRC | 04:04 | |
| *** masber has joined #openstack-kolla | 04:04 | |
| masber | good afternoon, my cinder_api container is crashing (centos binary 4.0.0) this is the docker logs output http://paste.openstack.org/raw/624659/. Any thoughts? | 04:08 |
| *** jamesbenson has joined #openstack-kolla | 04:16 | |
| *** huanxie has quit IRC | 04:20 | |
| *** jamesbenson has quit IRC | 04:20 | |
| masber | question regarding the documentation page https://docs.openstack.org/kolla-ansible/latest/user/multinode.html should the docker service configuration be "/etc/systemd/system/docker.service" or "/etc/systemd/system/docker.service.d/kolla.conf" or doesn't matter? | 04:26 |
| *** danardelean has joined #openstack-kolla | 04:32 | |
| *** skramaja has joined #openstack-kolla | 04:33 | |
| *** heigao has joined #openstack-kolla | 04:34 | |
| *** genek has quit IRC | 04:34 | |
| *** danardelean has quit IRC | 04:36 | |
| *** kiennt26 has quit IRC | 04:36 | |
| *** classicsnail has quit IRC | 04:41 | |
| *** classicsnail has joined #openstack-kolla | 04:41 | |
| inc0 | masber: doesn't matter afair | 04:42 |
| spsurya | morning guys | 04:43 |
| inc0 | newer dockers also works with /etc/docker/daemon.json | 04:43 |
| inc0 | to confirm if your config works, you can call docker info | 04:43 |
| masber | true | 04:43 |
| masber | so when I do kolla-build --registry XXXXX will it replace the images already exists with the new ones? | 04:50 |
| *** genek has joined #openstack-kolla | 04:50 | |
| *** huanxie has joined #openstack-kolla | 04:59 | |
| *** vhosakot_ has quit IRC | 05:04 | |
| *** vhosakot has joined #openstack-kolla | 05:05 | |
| *** vhosakot_ has joined #openstack-kolla | 05:12 | |
| *** vhosakot has quit IRC | 05:14 | |
| *** yangyapeng has joined #openstack-kolla | 05:18 | |
| kolla-slack | <egonzalez> masber, need to add --push to upload them, --registry only set the registry ip in the image name | 05:20 |
| *** yangyapeng has quit IRC | 05:23 | |
| *** rwsu has joined #openstack-kolla | 05:24 | |
| *** lrensing has joined #openstack-kolla | 05:24 | |
| *** genek has quit IRC | 05:26 | |
| *** dasTor_ has joined #openstack-kolla | 05:27 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/kolla-kubernetes master: Updated from global requirements https://review.openstack.org/509443 | 05:27 |
| openstackgerrit | wangwei proposed openstack/kolla-ansible master: Optimize the configuration of ceph https://review.openstack.org/496608 | 05:28 |
| *** dasTor has quit IRC | 05:30 | |
| *** dasTor has joined #openstack-kolla | 05:31 | |
| *** dasTor_ has quit IRC | 05:31 | |
| *** jbadiapa_ has quit IRC | 05:32 | |
| *** genek has joined #openstack-kolla | 05:39 | |
| *** genek has quit IRC | 05:44 | |
| *** dasTor_ has joined #openstack-kolla | 05:46 | |
| *** lrensing has quit IRC | 05:48 | |
| *** dasTor has quit IRC | 05:49 | |
| *** egonzalez has joined #openstack-kolla | 06:00 | |
| *** salv-orlando has joined #openstack-kolla | 06:03 | |
| *** salv-orl_ has quit IRC | 06:04 | |
| masber | can I use /kolla/tools/cleanup-images to delete images from registry? | 06:05 |
| masber | if not how can I delete all images in local registry? I have too many rubish from previous attempts | 06:06 |
| *** genek has joined #openstack-kolla | 06:06 | |
| *** lrensing has joined #openstack-kolla | 06:09 | |
| *** janki has joined #openstack-kolla | 06:31 | |
| egonzalez | masber, images cannot be removed from registry, only method afik, rm -rf <path_to_registry_volume>/* | 06:32 |
| *** lrensing has quit IRC | 06:33 | |
| *** lpetrut has joined #openstack-kolla | 06:33 | |
| *** lpetrut has quit IRC | 06:34 | |
| kolla-slack | <jascott1> might check to see if `docker prune` could work there | 06:40 |
| *** kiennt26 has joined #openstack-kolla | 06:40 | |
| *** Zophar has joined #openstack-kolla | 06:40 | |
| *** danardelean has joined #openstack-kolla | 06:47 | |
| masber | egonzalez, there is a docker registry API but I could not find the way of using it to delete images, I can query tags and repositories tho | 06:47 |
| masber | egonzalez, how can I find the path to volumes? | 06:47 |
| masber | also fluentd container is failing, i am using kolla stable/pike branch. I am trying to find the error... could be this one? INFO:kolla.image.build.fluentd:/bin/sh: line 0: ulimit: open files: cannot modify limit: Operation not permitted | 06:49 |
| masber | centos binary | 06:50 |
| *** danardelean has quit IRC | 06:51 | |
| openstackgerrit | Juan Badia Payno proposed openstack/kolla master: WIP - fluentd rpm candidate https://review.openstack.org/515292 | 06:55 |
| *** jbadiapa_ has joined #openstack-kolla | 06:57 | |
| *** jbadiapa_ is now known as jbadiapa | 06:57 | |
| *** yangyapeng has joined #openstack-kolla | 07:05 | |
| *** salv-orlando has quit IRC | 07:12 | |
| *** salv-orlando has joined #openstack-kolla | 07:12 | |
| *** salv-orlando has quit IRC | 07:16 | |
| *** dciabrin has joined #openstack-kolla | 07:18 | |
| openstackgerrit | Martin André proposed openstack/kolla master: Add Apache packages to mistral-api container https://review.openstack.org/513089 | 07:21 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible stable/pike: Move to zuul v3 in project jobs https://review.openstack.org/513608 | 07:22 |
| *** hachi_ has joined #openstack-kolla | 07:23 | |
| *** hrw has quit IRC | 07:26 | |
| egonzalez | zhubingbing_, ping | 07:28 |
| zhubingbing_ | hi | 07:28 |
| openstackgerrit | Merged openstack/kolla-ansible master: Optimize zuul v3 jobs https://review.openstack.org/513612 | 07:30 |
| *** huanxie has quit IRC | 07:31 | |
| *** hrw has joined #openstack-kolla | 07:31 | |
| *** danardelean has joined #openstack-kolla | 07:32 | |
| *** huanxie has joined #openstack-kolla | 07:39 | |
| *** salv-orlando has joined #openstack-kolla | 07:42 | |
| *** egonzalez has quit IRC | 07:44 | |
| *** kiennt26 has quit IRC | 07:48 | |
| *** serlex has joined #openstack-kolla | 07:49 | |
| *** paramite has joined #openstack-kolla | 07:49 | |
| *** hachi__ has joined #openstack-kolla | 07:50 | |
| *** hachi_ has quit IRC | 07:50 | |
| *** kiennt26 has joined #openstack-kolla | 07:51 | |
| *** danardelean has quit IRC | 07:52 | |
| *** danardelean has joined #openstack-kolla | 07:53 | |
| *** dangtrinhnt has joined #openstack-kolla | 07:54 | |
| *** danardelean has quit IRC | 07:57 | |
| *** dardelean has joined #openstack-kolla | 07:59 | |
| *** dougsz has joined #openstack-kolla | 08:02 | |
| *** hachi__ has quit IRC | 08:07 | |
| *** hachi__ has joined #openstack-kolla | 08:07 | |
| *** ktibi has joined #openstack-kolla | 08:18 | |
| *** klindgren__ has joined #openstack-kolla | 08:22 | |
| *** danardelean has joined #openstack-kolla | 08:23 | |
| *** david-lyle has quit IRC | 08:24 | |
| *** klindgren_ has quit IRC | 08:24 | |
| *** david-lyle has joined #openstack-kolla | 08:25 | |
| *** danardelean has quit IRC | 08:27 | |
| *** huanxie has quit IRC | 08:29 | |
| *** shardy has joined #openstack-kolla | 08:29 | |
| *** huanxie has joined #openstack-kolla | 08:35 | |
| *** lpetrut has joined #openstack-kolla | 08:36 | |
| *** jiriprox has quit IRC | 08:39 | |
| openstackgerrit | Juan Badia Payno proposed openstack/kolla master: WIP - fluentd rpm candidate https://review.openstack.org/515292 | 08:41 |
| openstackgerrit | zhongshengping proposed openstack/kolla-ansible master: Add binary related jobs to check https://review.openstack.org/515325 | 08:47 |
| *** lhx__ has quit IRC | 08:49 | |
| *** lhx__ has joined #openstack-kolla | 08:49 | |
| *** heigao has quit IRC | 08:56 | |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible master: Optimise the "openstack_freezer_auth" https://review.openstack.org/515276 | 08:56 |
| masber | hi, could someone please give me a hand? cinder_api keeps crashing, these are the docker logs http://paste.openstack.org/raw/624681/ | 08:59 |
| masber | I am using 4.0.0 centos binary | 08:59 |
| *** danardelean has joined #openstack-kolla | 09:01 | |
| *** huanxie has quit IRC | 09:05 | |
| *** danardelean has quit IRC | 09:06 | |
| openstackgerrit | Martin André proposed openstack/kolla master: Add Apache packages to mistral-api container https://review.openstack.org/513089 | 09:08 |
| *** danardelean has joined #openstack-kolla | 09:12 | |
| hrw | masber: add setting permissions for /run/httpd | 09:19 |
| *** david-lyle has quit IRC | 09:26 | |
| *** danardelean has quit IRC | 09:26 | |
| *** dklyle has joined #openstack-kolla | 09:26 | |
| *** huanxie has joined #openstack-kolla | 09:27 | |
| openstackgerrit | Heetae Ahn proposed openstack/kolla master: Add RHEL subscription registration https://review.openstack.org/495148 | 09:35 |
| openstackgerrit | Duong Ha-Quang proposed openstack/kolla-ansible master: Add missing hostnames to fluentd log messages https://review.openstack.org/514593 | 09:38 |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible master: Optimise the "openstack_auth" for freezer and horizon https://review.openstack.org/515276 | 09:40 |
| *** nrado has joined #openstack-kolla | 09:40 | |
| masber | hrw, how could I do that is the container is constantly restarting? | 09:42 |
| *** nrado has quit IRC | 09:44 | |
| openstackgerrit | Duong Ha-Quang proposed openstack/kolla master: XenServer: Add packages for initial support https://review.openstack.org/510266 | 09:45 |
| *** danardelean has joined #openstack-kolla | 09:47 | |
| *** danardelean has quit IRC | 09:48 | |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible master: Remove the unnecessary horizon_auth https://review.openstack.org/515340 | 09:52 |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible master: Remove the unnecessary horizon_auth https://review.openstack.org/515340 | 09:55 |
| *** tovin07_ has quit IRC | 10:04 | |
| *** sambetts|afk is now known as sambetts | 10:09 | |
| *** masber has quit IRC | 10:09 | |
| hrw | https://review.openstack.org/#/c/514614/ - can someone approve new repo for stable/pike? without it pike is useless on aarch64/debian | 10:09 |
| openstackgerrit | zhangfei gao proposed openstack/kolla master: ironic-pxe: generate aarch64 bootfile grubaa64.efi https://review.openstack.org/512540 | 10:10 |
| *** caowei has quit IRC | 10:15 | |
| *** kiennt26 has quit IRC | 10:16 | |
| *** danardelean has joined #openstack-kolla | 10:25 | |
| *** danardelean has quit IRC | 10:25 | |
| *** gfidente|afk is now known as gfidente | 10:31 | |
| openstackgerrit | Merged openstack/kolla-ansible stable/pike: Fix prechecks fail with docker not in deployment host https://review.openstack.org/515267 | 10:34 |
| *** vhosakot_ has quit IRC | 10:34 | |
| zhubingbing_ | hi guys | 10:35 |
| *** serlex has quit IRC | 10:38 | |
| *** serlex has joined #openstack-kolla | 10:39 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla master: implement build jobs in kolla https://review.openstack.org/513971 | 10:40 |
| *** serlex has left #openstack-kolla | 10:41 | |
| *** pbourke has quit IRC | 10:42 | |
| *** dasTor_ has quit IRC | 10:42 | |
| *** pbourke has joined #openstack-kolla | 10:44 | |
| gema | guys, are we publishing the x86 images in docker hub? | 10:44 |
| gema | Jeffrey4l: ^ | 10:44 |
| zhubingbing_ | ocata ? | 10:46 |
| gema | any images | 10:46 |
| gema | we are trying to figure out where to put the arm64 ones | 10:46 |
| zhubingbing_ | yeah | 10:46 |
| hrw | ocata are present iirc | 10:46 |
| gema | hrw: where? | 10:46 |
| hrw | https://hub.docker.com/r/kolla/ubuntu-source-collectd/tags/ shows 4.0.0 ones as latest | 10:47 |
| gema | hrw: cannot we put ours there as well? | 10:47 |
| *** hachi__ has quit IRC | 10:48 | |
| hrw | gema: if we are kolla upstream then yes. if we as linaro then we lack access | 10:48 |
| *** hachi__ has joined #openstack-kolla | 10:48 | |
| gema | hrw: let's add the relevant bits to CI in openstack so that our images end up there then | 10:48 |
| gema | but it loks to me as if the publishing is rather sporadic | 10:48 |
| gema | I wonder if it was done manually or how | 10:48 |
| gema | or if that namespace is just abandoned | 10:49 |
| zhubingbing_ | Jeffrey4l know it | 10:49 |
| gema | ok, we'll wait for his answer | 10:49 |
| zhubingbing_ | I'll call him for you waiting | 10:50 |
| gema | thanks, it's ok though | 10:53 |
| gema | he'll come at some point :D | 10:53 |
| hrw | and then I have a question to him too ;D | 10:53 |
| hrw | we lack 6.0.0 and 5.0.1 tags on kolla ;D | 10:53 |
| Jeffrey4l | gema, before we pushed images to docker hub manually. now we are trying to push the images automatically , based on zuul v3. | 10:54 |
| Jeffrey4l | 5.0.0 images is never be pushed. and 5.0 for centos and oraclelinux should be dead duo to some bugs in centos 7.4 | 10:54 |
| hrw | Jeffrey4l: hi | 10:57 |
| Jeffrey4l | 5.0.1 tags should be tag soon and push the images to docker hub. | 11:02 |
| Jeffrey4l | based on the policy, 6.0.0 images won't be pushed until kolla release 6.0.0 image. | 11:02 |
| hrw | Jeffrey4l: asking about 5.0.1 cause recent images have this tag but no tag on git repo | 11:02 |
| hrw | http://git.openstack.org/cgit/openstack/kolla/refs/tags | 11:03 |
| *** huanxie has quit IRC | 11:03 | |
| hrw | Jeffrey4l: and once 6.0.0 tag appear I can go and build queens images for tests without using "--tag queens" ;D | 11:03 |
| Jeffrey4l | sup hrw | 11:06 |
| * hrw wishes for --extra-image-prefix argument | 11:09 | |
| hrw | haerwu/kolla-is-omg-so-cool-what-it-deserves-extras-debian-source-base ;d | 11:09 |
| gema | Jeffrey4l: the reason is because I was asking to put images on linaro docker as linar/kolla-openstack-imagename | 11:11 |
| gema | linaro/kolla-openstack-imagename | 11:11 |
| gema | since we have other images at linaro as well, I wanted to distinguish between then | 11:12 |
| gema | them | 11:12 |
| Jeffrey4l | gema, i think that is cool. | 11:16 |
| Jeffrey4l | push arm images is a good idea. | 11:16 |
| Jeffrey4l | it will be helpful for users. | 11:17 |
| gema | Jeffrey4l: the question is where we push them | 11:20 |
| gema | official kolla repos | 11:20 |
| bjolo | lastlog mtu | 11:20 |
| gema | or linaro | 11:20 |
| *** masber has joined #openstack-kolla | 11:22 | |
| masber | ?msg NickServ identify masuberu | 11:23 |
| Jeffrey4l | does docker support multi arch now? | 11:23 |
| gema | Jeffrey4l: it does, you can have labels with the arch, but hrw can explain better | 11:24 |
| hrw | never built multiarch image yet. All main distros used by Kolla (centos, debian, ubuntu) provide now mutliarch images | 11:25 |
| Jeffrey4l | i think it is OK to push the arm64 images to kolla namespace | 11:26 |
| gema | Jeffrey4l: ok, so we'll need to figure out how to do that and help with the zuul v3 automation | 11:27 |
| hrw | gema: to kolla/ namespace we can push 5.0.1 ones. 6.0.0 for tests will need place anyway | 11:27 |
| gema | Jeffrey4l: I'd like to understand the images lifecycle, when do you push images to docker hub | 11:28 |
| *** jamesbenson has joined #openstack-kolla | 11:29 | |
| *** janki has quit IRC | 11:31 | |
| hrw | gema: I see kolla/ on docker hub as space for stable tested images. so up to pike ones for now | 11:33 |
| *** jamesbenson has quit IRC | 11:33 | |
| gema | hrw: ok | 11:35 |
| gema | hrw: maybe we should add the option to have an image prefix | 11:35 |
| gema | but before we go for that I'd like to know how that would affect kolla-ansible and kolla-k8s | 11:35 |
| gema | and make it superoptional | 11:36 |
| *** danardelean has joined #openstack-kolla | 11:41 | |
| *** danardelean has quit IRC | 11:41 | |
| *** danardelean has joined #openstack-kolla | 11:41 | |
| *** egonzalez has joined #openstack-kolla | 11:42 | |
| *** danardelean has quit IRC | 11:45 | |
| *** ansmith has quit IRC | 11:47 | |
| *** egonzalez has quit IRC | 11:51 | |
| *** dave-mccowan has joined #openstack-kolla | 11:51 | |
| *** jmccarthy has joined #openstack-kolla | 11:54 | |
| hrw | gema: good part is that image-prefix is set in one place so patch would not complicate much | 11:54 |
| *** magicboiz has quit IRC | 11:57 | |
| hrw | even more fun. | 12:02 |
| hrw | with local registry you can do whatever you want cause kolla_base_distro is set in ansible/group_vars/all.yml so namespace/whatever-you-want-distro-type-imagename is possible | 12:03 |
| hrw | extra note in documentation would solve that | 12:04 |
| *** salv-orl_ has joined #openstack-kolla | 12:06 | |
| hrw | nope. checks for kolla_base_distro exist | 12:06 |
| lhx__ | hi guys, I'm very appreciate it if you spare time to have a look at this: https://bugs.launchpad.net/kolla-kubernetes/+bug/1726784 | 12:07 |
| openstack | Launchpad bug 1726784 in kolla-kubernetes "can't get mirror with default repo url" [Undecided,New] - Assigned to Hanxi Liu (hanxi-liu) | 12:07 |
| kolla-slack | <egonzalez> As far i remember images are not pushed to dockerhub for various reasons | 12:07 |
| kolla-slack | <egonzalez> Extra work to do it manually for all distros/types | 12:07 |
| kolla-slack | <egonzalez> Not tested images | 12:07 |
| kolla-slack | <egonzalez> A decision in governance to not do it until we have an automatic method | 12:07 |
| *** salv-orlando has quit IRC | 12:09 | |
| kolla-slack | <masber> @egonzalez from where is docker-build getting the docker files? tarball? | 12:19 |
| *** shardy has quit IRC | 12:20 | |
| rwellum | lhx__: What is this vagrant environment you are using? | 12:21 |
| lhx__ | rwellum, thanks for your reply, I use Vagrant 1.8.1 | 12:23 |
| rwellum | gema: there's also the tarballs location - that's where I am getting pike and master images from now. While wait on zuulv3 | 12:24 |
| gema | rwellum: where are they? | 12:25 |
| rwellum | lhx__: I mean this halcyon dev environment has been abandoned for a long time AFAIK. | 12:26 |
| rwellum | gema: http://tarballs.openstack.org/kolla/images/ | 12:26 |
| lhx__ | rwellum, sorry, I'm the new for kolla, I just built the environment according to the guide: https://docs.openstack.org/kolla-kubernetes/latest/development-environment.html | 12:28 |
| rwellum | lhx__: there's a big warning at the top of that page :) | 12:28 |
| rwellum | lhx__: You should start here: https://docs.openstack.org/kolla-kubernetes/latest/deployment-guide.html | 12:29 |
| rwellum | lhx__: Also please try: https://github.com/RichWellum/k8s | 12:29 |
| rwellum | I'm committing this to kolla-kubernetes - it's in review. | 12:30 |
| lhx__ | rwellum, Oh, damn it | 12:30 |
| bjolo | hey is the newton branch removed from kolla repo? | 12:30 |
| gema | hrw: ^ | 12:30 |
| rwellum | lhx__: sorry! | 12:30 |
| kolla-slack | <egonzalez> masber, kolla-build renders dockerfiles from <repo>/docker folder and build from them | 12:30 |
| lhx__ | rwellum, It's my negligence, I will try the new guide | 12:32 |
| lhx__ | lol | 12:32 |
| gema | rwellum: is there a patch for your k8s new guide? | 12:32 |
| rwellum | np - I started there 1 year ago - when it was still relevant and it still didn't work. That's why we built the other guide. | 12:33 |
| gema | rwellum: I can put kevinz on the right path if there is | 12:33 |
| lhx__ | rwellum, btw, there is a trivial fix: https://review.openstack.org/#/c/514511/ | 12:33 |
| gema | rwellum: I did ask him to fix the docs for the bug he found in the guide | 12:33 |
| bjolo | http://paste.openstack.org/show/624703/ | 12:33 |
| bjolo | fresh git clone | 12:34 |
| rwellum | gema: https://review.openstack.org/#/c/487972/ - but I have like a 900 line change to it - haven't re-committed because I wanted to get Pike up and running. | 12:34 |
| gema | rwellum: ack | 12:34 |
| gema | will give this info to him anyway | 12:34 |
| gema | better to put effort on the new thing | 12:34 |
| gema | than on the obsolete one | 12:34 |
| rwellum | He can use my github version for now gema - it brings up Ocata like a champ. | 12:35 |
| gema | rwellum: ack | 12:35 |
| lhx__ | rwellum, thanks guy :) | 12:35 |
| rwellum | lhx__: I just gave it +W so it should merge soon. | 12:35 |
| lhx__ | I saw | 12:36 |
| lhx__ | :) | 12:36 |
| openstackgerrit | Juan Badia Payno proposed openstack/kolla-ansible master: WIP - fluentd rpm candidate test https://review.openstack.org/515389 | 12:36 |
| rwellum | Cool - and lhx__ glad to have you onboard and trying this out. | 12:36 |
| rwellum | gema: I got 'added' to a hangouts group for the Upstream Institute for Sydney, apparently I am supposed to talk about containers. And they woke me up 4:15am this morning - texting to the group.... | 12:37 |
| *** jinke_ has quit IRC | 12:37 | |
| gema | rwellum: omg, don't sleep next to your phone! | 12:38 |
| lhx__ | rwellum, sure, I will try it , hope everything goes well :-D | 12:38 |
| rwellum | I know right... :( | 12:38 |
| gema | rwellum: I have disabled most notifications when I was travelling and forgot to enable them back | 12:38 |
| gema | so I am having a good relaxed time xD | 12:38 |
| gema | phone-wise | 12:38 |
| rwellum | Yeah much better - until I ignore all the texts from my wife and kids... | 12:38 |
| *** jinke_ has joined #openstack-kolla | 12:38 | |
| rwellum | For too long | 12:38 |
| gema | haha | 12:39 |
| openstackgerrit | Merged openstack/kolla-kubernetes master: Remove extra characters https://review.openstack.org/514511 | 12:39 |
| *** ansmith has joined #openstack-kolla | 12:41 | |
| rwellum | gema: be cool if kevenz sees the same pike issues I am - there's definitely something messed up with libvirt. | 12:41 |
| gema | rwellum: yep | 12:46 |
| gema | rwellum: you'll have to talk to him on your evening time | 12:46 |
| gema | rwellum: he is off today though | 12:46 |
| rwellum | Not if I got up at 4:15am I won't. | 12:46 |
| gema | rwellum: I don't mean today, I mean in general x D | 12:48 |
| bjolo | hrw: around | 12:49 |
| bjolo | do you have info about newton branch and pip packages? | 12:50 |
| bjolo | they all seem to be removed | 12:50 |
| bjolo | this is not cool. we have production clouds runing newton | 12:50 |
| rwellum | Yeah I know... 😃 | 12:50 |
| *** goldyfruit has joined #openstack-kolla | 12:53 | |
| bjolo | pbourke: ^^ | 12:53 |
| *** skramaja has quit IRC | 12:53 | |
| genek | Hi, have anyone tried building images for arm64 architecture? | 12:56 |
| *** shardy has joined #openstack-kolla | 12:57 | |
| rwellum | bjolo: images or kolla itself? | 12:59 |
| rwellum | https://releases.openstack.org/teams/kolla.html - I still see Newton | 12:59 |
| *** rhallisey has joined #openstack-kolla | 13:00 | |
| *** lhx__ has quit IRC | 13:03 | |
| ktibi | Hi, I try to use ceph but libvirt have error : libvirtError: Secret not found: no secret with matching uuid 'XXXXXX' | 13:06 |
| ktibi | kolla configure secret for libvirt ? | 13:06 |
| bjolo | kolla itself | 13:08 |
| bjolo | http://paste.openstack.org/show/624703 | 13:09 |
| bjolo | pip install kolla-ansible also does not have any newton version | 13:10 |
| bjolo | but maybe thats in pip install kolla | 13:10 |
| bjolo | ok pip kolla newton is still there | 13:11 |
| ktibi | ok need to restart libvirt container >< | 13:12 |
| bjolo | but why is the branch removed? | 13:12 |
| rwellum | bjolo - I don't see it as a branch either https://www.irccloud.com/pastebin/7nMlSfah/ | 13:15 |
| *** jaosorior has quit IRC | 13:16 | |
| *** mdnadeem has quit IRC | 13:17 | |
| genek | Got error creating aufs mount to /var/lib/docker/aufs/mnt/ : invalid argument when building neutron-server on arm64 | 13:17 |
| genek | Any ideas of possible causes to this error? | 13:18 |
| genek | other images build fine | 13:18 |
| bjolo | rwellum: nope its gone afaict | 13:18 |
| bjolo | rwellum: it was there just few weeks ago | 13:19 |
| bjolo | i did a build on the latest newton tarballs | 13:19 |
| bjolo | anyone knows when and who? most importantly why? | 13:20 |
| *** sbezverk has joined #openstack-kolla | 13:20 | |
| rwellum | It's a question to inc0 or some other cores - I don't know. | 13:20 |
| bjolo | its one thing to not have support for an old version, but removing the branch is straight up a bad idea | 13:20 |
| rwellum | Yeah there's no point to it either - not like it takes up much space etc. | 13:21 |
| rwellum | mark it as unsupported maybe. | 13:21 |
| bjolo | inc0: should be online in a few | 13:22 |
| bjolo | sdake: are you online yet. Usually up early :) | 13:23 |
| sdake | sup bjolo | 13:24 |
| bjolo | hey sdake | 13:24 |
| bjolo | i knew you where up :) | 13:24 |
| bjolo | it seems like the newton and mitak branch is removed from kolla repo | 13:24 |
| bjolo | you know why? | 13:25 |
| sdake | openstack supports upstream n and n-1 | 13:25 |
| bjolo | https://www.irccloud.com/pastebin/7nMlSfah | 13:25 |
| openstackgerrit | Mark Duggan proposed openstack/kolla-ansible master: Fixing missing copy of iscsi container config files https://review.openstack.org/515406 | 13:25 |
| sdake | that is pike and ocata | 13:25 |
| sdake | i am not sure why newton was removed | 13:25 |
| sdake | as pike hasn't been released yet afaik :) | 13:25 |
| bjolo | yes support is one thing but removing the branches? | 13:25 |
| *** gkadam_ has joined #openstack-kolla | 13:26 | |
| sdake | yes the branches are removed because the stable team thinks that having them results in a situation where the branch must be maintained | 13:26 |
| sdake | personally I see no harm in keeping branched around and labeling them as unsupported | 13:26 |
| sdake | however the fundamental purpose of a branch is to create something supported for backports | 13:26 |
| bjolo | is this true for all openstack projects or just kolla* | 13:26 |
| *** gkadam has quit IRC | 13:27 | |
| sdake | if you need an older version hyou could try the tarballs | 13:27 |
| sdake | bjolo yes true for all os projects | 13:27 |
| bjolo | ok | 13:27 |
| sdake | this is the value of buying openstack from a vendor | 13:27 |
| rwellum | sdake: but this is the kolla repo - not the images | 13:27 |
| bjolo | i know | 13:27 |
| bjolo | but kolla just had one repo for newton | 13:27 |
| *** rmart04 has joined #openstack-kolla | 13:27 | |
| bjolo | kolla, kolla-ansible, kolla-kub shared it | 13:28 |
| sdake | i dont agree with it but understand why it is there | 13:28 |
| bjolo | i dont agree either | 13:28 |
| sdake | so if you are asking me to defend the decision, your asking the wrong dude | 13:28 |
| sdake | if your asking me to fight it, your asking the wrong dude :) | 13:28 |
| bjolo | no not asking you | 13:28 |
| rwellum | But anyway - until Pike is released we should get Newton branch back. Who does this? | 13:28 |
| bjolo | i did not know this is how it was. just noticed that kolla newton branch was gone and we run a prod cloud on that | 13:29 |
| bjolo | dont feel comfortable not have the source tree for something i have in prod | 13:29 |
| sdake | right | 13:30 |
| bjolo | in my world you never delete your history | 13:30 |
| sdake | i agree - deleting anything from git is terrible :) | 13:30 |
| sdake | the original model of openstack was to be an upstream for downstream software distributors | 13:30 |
| sdake | some people in the community (sounds like you) want openstack to be a community supported LTS type producvt | 13:31 |
| ktibi | Hi all I found bug with glance + ceph with snapshot nova. Need to change show_image_direct_url to show_multiple_locations https://review.openstack.org/#/c/279630/ | 13:31 |
| sdake | openstack doesn't have LTS | 13:31 |
| sdake | I don't knwo why - I think mainy because the cost is very high | 13:31 |
| ktibi | the change is here https://github.com/openstack/kolla-ansible/blob/2e6fa4fe15025eaa1d4110b882e7522cd814491b/ansible/roles/glance/templates/glance-api.conf.j2#L15 | 13:31 |
| ktibi | I test on fresh install and it works | 13:31 |
| sdake | ktibi can you submit a patch via gerrit plz | 13:32 |
| ktibi | yes | 13:32 |
| ktibi | sdake do you have tuto ? | 13:32 |
| sdake | https://docs.openstack.org/infra/manual/developers.html | 13:33 |
| sdake | bjolo i'd recommend upgrading from newton prior to the removal of the newton branch :) | 13:34 |
| bjolo | sdake: we are working on that | 13:34 |
| sdake | cool | 13:34 |
| bjolo | i needed the newton branch in the lab exactly for that reason. testing the upgrade | 13:35 |
| sdake | tarballs? | 13:35 |
| bjolo | knowing this we need to setup our own git mirrors | 13:35 |
| sdake | we don't recommend working from branches anyway - only tarballs | 13:36 |
| *** gkadam_ has quit IRC | 13:37 | |
| bjolo | well... old habit. cherrypick options, etc | 13:37 |
| *** danardelean has joined #openstack-kolla | 13:38 | |
| bjolo | but OS internally must have a golden version of the git tree right? | 13:39 |
| bjolo | i mean, what if tarballs.openstack got hacked. how are they able to recreate/verify the tarballs otherwise | 13:40 |
| *** pratapagoutham has joined #openstack-kolla | 13:45 | |
| masber | hi all, I used pike images from tarbal as I could not make the kolla-build to work, deployment is fine, I can create an instance and ssh through the router namespace but connection to outside world does not work, any thoughts? | 13:47 |
| *** magicboiz has joined #openstack-kolla | 13:48 | |
| *** dosaboy has quit IRC | 13:48 | |
| ktibi | sdake ok ubuntu.com down gerrit next time ^^ | 13:49 |
| *** jaosorior has joined #openstack-kolla | 13:53 | |
| *** goldyfruit1 has joined #openstack-kolla | 13:54 | |
| *** egonzalez has joined #openstack-kolla | 13:55 | |
| *** goldyfruit has quit IRC | 13:55 | |
| *** hongbin has joined #openstack-kolla | 13:56 | |
| *** jinke_ has quit IRC | 13:56 | |
| *** jinke_ has joined #openstack-kolla | 13:57 | |
| egonzalez | ktibi, what release deploying? In master this is the revert for the issue which will need to be backported to stable branch | 14:00 |
| egonzalez | https://review.openstack.org/#/c/509792/ | 14:00 |
| *** dosaboy has joined #openstack-kolla | 14:00 | |
| ktibi | 5.0.0 | 14:01 |
| ktibi | ok patch already exist | 14:01 |
| *** jtriley has joined #openstack-kolla | 14:03 | |
| egonzalez | Pike is released already, newton is being removed now | 14:03 |
| egonzalez | bjolo, you can download latest change in git in newton-eol branch | 14:03 |
| *** goldyfruit1 has quit IRC | 14:03 | |
| ktibi | do you have plan for a 5.1.0 or 5.0.1 ? with all minor fix like the qemu.conf or glance-api.conf ? | 14:03 |
| egonzalez | ktibi, yes, is in release calendar | 14:05 |
| egonzalez | but looks like we missed the date Oct 16 - Oct 20 | 14:06 |
| ktibi | ok ^^, do you have link for calendar ? | 14:06 |
| egonzalez | ktibi, queens https://releases.openstack.org/queens/schedule.html , all https://releases.openstack.org/index.html | 14:06 |
| ktibi | hum ok this is the official calender for openstack | 14:07 |
| *** jbadiapa has quit IRC | 14:07 | |
| egonzalez | ktibi, yep, there is a schedule ics in "combined release calendar" link to apply to your calendar and be always updated | 14:08 |
| *** dklyle has quit IRC | 14:15 | |
| masber | egonzalez, did suggested me to add kolla/centos-source-manila-data in order to make manila to work? | 14:18 |
| egonzalez | masber, not remember that but maybe I did, who knows :) | 14:22 |
| masber | mmm just wondering because it is not in the documentation so someone may have told me to put it | 14:23 |
| *** newmember has joined #openstack-kolla | 14:25 | |
| egonzalez | masber, not sure if manila-data is needed in ocata, in pike binary package exists for sure, but in ocata rpm package was missing in the image | 14:25 |
| masber | the thing is that I have this in my globals for ocata manila_data_image_full: kolla/centos-source-manila-data:4.0.0, now I wanted to do same for pike, like manila_data_image_full: kolla/centos-source-manila-data:5.0.1 but ansible complains about NotFound: 404 Client Error: Not Found (\"manifest for kolla/centos-source-manila-data:5.0.1 not found | 14:26 |
| masber | egonzalez, ^^ | 14:26 |
| egonzalez | more than missing, installed along manila-common in rpm packages | 14:26 |
| egonzalez | masber, in pike not need to use source image anymore | 14:26 |
| egonzalez | masber, images with tag 5.0.1 are not pushed to dockerhub | 14:27 |
| masber | yes, kolla is my local repo | 14:27 |
| masber | I got the images from tarbal | 14:28 |
| egonzalez | masber, did you build also source images? | 14:28 |
| masber | only binaries | 14:28 |
| egonzalez | masber, in pike, remove the manila_data_image_full from globals and try again | 14:28 |
| egonzalez | image exists in binary | 14:28 |
| masber | ok | 14:28 |
| sdake | bjolo i dont have the answer to your question about doomsday scenarios with tarballs.oo | 14:29 |
| masber | just enable_manila: "yes" | 14:29 |
| masber | and enable_manila_backend_generic: "yes" | 14:29 |
| masber | right? | 14:29 |
| sdake | bjolo I do assume it is properly archived in multipel places :) | 14:29 |
| sdake | bad to make assumptions :) | 14:29 |
| sdake | II can find out if you like | 14:30 |
| Zophar | Quick question: Could we re-trigger CI on review (with our "poor" permissions") or i should ask someone to do that? | 14:30 |
| masber | also manila client is not installed by default | 14:30 |
| masber | needs to be installed manually | 14:30 |
| Zophar | https://review.openstack.org/#/c/514676/ | 14:30 |
| egonzalez | masber, you can cherry-pick this change to stable/ocata, forgot to do that https://review.openstack.org/#/c/465018/ | 14:30 |
| kolla-slack | <inc0> Bjolo you can access old code via git tags | 14:31 |
| egonzalez | masber, client is user task not deployers | 14:31 |
| masber | understood | 14:31 |
| kolla-slack | <inc0> There is also discussion on summit about lts | 14:31 |
| egonzalez | bjolo, sdake, git repo has a tag with newton-eol with latest state of the newton branch | 14:31 |
| egonzalez | https://github.com/openstack/kolla/tree/newton-eol | 14:32 |
| *** genek has quit IRC | 14:32 | |
| egonzalez | Zophar, just put recheck in comment box | 14:32 |
| Zophar | ok i did... so i wait ;) | 14:33 |
| Zophar | thanks | 14:33 |
| *** lrensing has joined #openstack-kolla | 14:46 | |
| *** zhubingbing__ has joined #openstack-kolla | 14:46 | |
| *** david-lyle has joined #openstack-kolla | 14:48 | |
| hrw | bjolo: look at newton-eol branch | 14:49 |
| *** lrensing has quit IRC | 14:50 | |
| *** kbaegis has joined #openstack-kolla | 14:55 | |
| *** kbaegis has joined #openstack-kolla | 14:55 | |
| *** kbaegis has joined #openstack-kolla | 14:55 | |
| *** kbaegis has quit IRC | 14:55 | |
| *** kbaegis has joined #openstack-kolla | 14:56 | |
| *** kbaegis has quit IRC | 14:56 | |
| *** kbaegis has joined #openstack-kolla | 14:56 | |
| *** kbaegis has joined #openstack-kolla | 14:57 | |
| *** jamesbenson has joined #openstack-kolla | 14:57 | |
| *** rmart04 has quit IRC | 14:58 | |
| *** zhubingbing__ has quit IRC | 14:58 | |
| *** zhubingbing has joined #openstack-kolla | 14:58 | |
| *** rich1iu has joined #openstack-kolla | 14:59 | |
| *** genek has joined #openstack-kolla | 14:59 | |
| *** genek has quit IRC | 14:59 | |
| *** genek has joined #openstack-kolla | 15:00 | |
| *** trozet has joined #openstack-kolla | 15:00 | |
| *** gkadam_ has joined #openstack-kolla | 15:02 | |
| *** jaosorior has quit IRC | 15:02 | |
| *** jbrooks has quit IRC | 15:04 | |
| hrw | gema: ? | 15:06 |
| *** egonzalez has quit IRC | 15:09 | |
| masber | egonzalez, so to have ephemeral in local storage I need to have this in globals | 15:09 |
| masber | nova_backend_ceph: "no" | 15:09 |
| masber | nova_backend: "{{ 'rbd' if nova_backend_ceph | bool else 'default' }}" | 15:09 |
| masber | is that right? | 15:09 |
| gema | hrw: sorry on a meeting now | 15:09 |
| hrw | gema: np | 15:09 |
| gema | hrw: what was the question? | 15:10 |
| hrw | gema: you pinged me 3h ago | 15:10 |
| gema | hrw: irrelevant, I cannot remember anymore :D | 15:10 |
| hrw | ;) | 15:10 |
| gema | sorry! | 15:11 |
| *** kbaegis has quit IRC | 15:11 | |
| *** genek has quit IRC | 15:12 | |
| *** kbaegis has joined #openstack-kolla | 15:13 | |
| *** kbaegis has quit IRC | 15:13 | |
| *** kbaegis has joined #openstack-kolla | 15:13 | |
| *** jbrooks has joined #openstack-kolla | 15:14 | |
| *** lukepatrick has joined #openstack-kolla | 15:15 | |
| kolla-slack | <egonzalez> masber, yep | 15:16 |
| *** lrensing has joined #openstack-kolla | 15:18 | |
| * hrw off | 15:19 | |
| *** newmember has quit IRC | 15:19 | |
| *** newmember has joined #openstack-kolla | 15:19 | |
| *** hachi__ has quit IRC | 15:23 | |
| kfox1111 | morning. | 15:27 |
| ktibi | When I try to re-deploy ceph, Checking free port for RadosGW ==> FAILED! => {"changed": false, "elapsed": 1, "failed": true, "msg": "Timeout when waiting for 172.16.7.2:6780 to stop."} | 15:31 |
| ktibi | I guess kolla doesn't try to stop container before no ? | 15:32 |
| openstackgerrit | Michal Jastrzebski (inc0) proposed openstack/kolla-ansible master: Enable multinode scenarios in zuulv3 https://review.openstack.org/512779 | 15:39 |
| *** itlinux has joined #openstack-kolla | 15:39 | |
| *** trozet has quit IRC | 15:45 | |
| *** klindgren_ has joined #openstack-kolla | 15:48 | |
| *** klindgren__ has quit IRC | 15:50 | |
| rich1iu | Hi, one question : Can kolla run on ARM64 platform ? | 15:51 |
| *** iniaziaway is now known as iniazi | 15:55 | |
| *** athomas has quit IRC | 15:57 | |
| kolla-slack | <masber> @egonzalez ok, I think I got the ephemeral on local disks setup, I have created a flavor and an instance... however the ephemeral disk should live in the /var/lib/nova/instances/<instance_id> on the same compute node hosting the instance but the folder is missing... | 16:01 |
| *** danardelean has quit IRC | 16:05 | |
| ktibi | masber with local storage you can't use live migration, ok for you ? | 16:07 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla master: implement build jobs in kolla https://review.openstack.org/513971 | 16:07 |
| masber | ktibi, yes no problem, this is for ephemeral disks only, I can always create volumes on ceph and attach it to the instance | 16:08 |
| masber | ktibi, and live migration is not a problem for me. I don't run 24h services but more of a batch jobs | 16:09 |
| masber | so if the node dies it can be recreated | 16:09 |
| ktibi | ok ;) | 16:09 |
| masber | ephemeral is just for dirty IOps operations | 16:09 |
| masber | means scratch data | 16:09 |
| ktibi | for your pb, with /var/lib/nova/instances/, you check in the compute container ? | 16:09 |
| masber | mmm not on the container | 16:10 |
| masber | which container? | 16:10 |
| masber | nova_compute? | 16:10 |
| ktibi | nova compute | 16:10 |
| ktibi | or maybe libvirt | 16:10 |
| masber | ktibi, are ephemeral disks supposed to be thin provisionning? | 16:12 |
| masber | I think I found it but the size doesn't match | 16:13 |
| masber | 46M /var/lib/docker/volumes/nova_compute/_data/instances/98c0618a-b3bc-476d-9d35-93b6581d793e | 16:13 |
| masber | I specified 512Gigs of ephemeral but the instance folder only has 46M? | 16:14 |
| ktibi | yes thin | 16:14 |
| masber | ok | 16:14 |
| masber | thanks! | 16:14 |
| ktibi | check Id of instance | 16:15 |
| ktibi | masber Maybe libvirt use link clone, 46M is just the delta between image and VM | 16:17 |
| masber | ktibi, yes the folder name matches the instance id | 16:17 |
| *** zhubingbing has quit IRC | 16:17 | |
| *** zhubingbing has joined #openstack-kolla | 16:18 | |
| masber | ktibi, what does it means --? "libvirt use link clone, 46M is just the delta between image and VM"? | 16:18 |
| masber | you mean that delta is the boot drive? | 16:18 |
| bjolo | if i do a pip install kolla, how do i generate kolla-build.conf? | 16:18 |
| masber | bjolo, pip install tox then tox -e genconfig | 16:19 |
| bjolo | yes | 16:19 |
| bjolo | but in what dir? | 16:19 |
| masber | kolla dir | 16:20 |
| rwellum | https://docs.openstack.org/kolla/latest/admin/image-building.html bjolo | 16:20 |
| masber | root kolla dir | 16:20 |
| bjolo | yes but pip install kolla is all over the place | 16:20 |
| masber | I don't know where pip installs kolla I clone from github | 16:20 |
| bjolo | i usually do to | 16:20 |
| bjolo | but since newton branch is removed now... and i need to build newton | 16:21 |
| masber | bjolo, look for a tox.ini? | 16:21 |
| *** mummer has joined #openstack-kolla | 16:23 | |
| bjolo | not found | 16:23 |
| bjolo | updatedb; locate tox.ini | 16:23 |
| rwellum | https://www.irccloud.com/pastebin/SW6OqBGR/ | 16:24 |
| bjolo | the docs are missing some steps | 16:24 |
| bjolo | rwellum: i dont have the git source | 16:24 |
| bjolo | no newton branch right | 16:24 |
| bjolo | i have done a pip install kolla | 16:24 |
| bjolo | no tox.ini afaict | 16:25 |
| rwellum | bjolo: what about the the git tags mentioned above? | 16:25 |
| bjolo | ah ok saw that now | 16:25 |
| bjolo | kewl | 16:25 |
| bjolo | but anyhow, the docs .... | 16:26 |
| rwellum | The docs need some work - I agree. But they are expecting the source | 16:26 |
| ktibi | masber I'm not sure but I guess if you deploy 50VM with one image, nova don't provide 50 disk with image size | 16:26 |
| bjolo | why do we even have pip install kolla then for ocata and pike. newton i understand since kolla-ansible is in there | 16:27 |
| ktibi | I add designate to my kolla and I kill horizon T_T | 16:27 |
| ktibi | centos-binary-horizon:5.0.0 "kolla_start" 6 minutes ago Restarting (1) Less than a second ago horizon | 16:27 |
| masber | ktibi, I am testing that now, I am going to put some data into the ephemeral disks and see if the instance folder grows | 16:27 |
| ktibi | Need to rebuild horizon image for add designate plugin ? | 16:27 |
| rwellum | bjolo - what I do: https://www.irccloud.com/pastebin/ASdyUWS9/ | 16:28 |
| ktibi | ok horizon fail.. cp: cannot stat '/usr/lib/python2.7/site-packages/designatedashboard/enabled/_*[^__].py': No such file or directory | 16:29 |
| *** lukepatrick has quit IRC | 16:31 | |
| bjolo | what branch should i checkout the newton-eol tag on? | 16:31 |
| bjolo | git checkout tags/newton-eol -b <branch> | 16:32 |
| bjolo | should i be on master branch? | 16:32 |
| *** zhubingbing has quit IRC | 16:32 | |
| ktibi | How Can I add designatedashboard in horizon container ??? | 16:36 |
| *** magicboiz has quit IRC | 16:37 | |
| masber | ktibi, yes the disk grows so it is thin provisioning | 16:37 |
| masber | thanks for your help! | 16:37 |
| masber | :) | 16:37 |
| ktibi | found https://review.openstack.org/#/c/480325/1/docker/horizon/Dockerfile.j2 Thx for centos ;) | 16:40 |
| rwellum | bjolo: ni actually - what options do you have? | 16:42 |
| *** lpetrut has quit IRC | 16:43 | |
| *** mwynne has joined #openstack-kolla | 16:50 | |
| *** lukepatrick has joined #openstack-kolla | 16:52 | |
| openstackgerrit | Mark Duggan proposed openstack/kolla-ansible master: Fixing missing copy of iscsi container config files https://review.openstack.org/515406 | 16:52 |
| *** jmccarthy has left #openstack-kolla | 16:53 | |
| *** ntpttr_laptop has joined #openstack-kolla | 16:53 | |
| *** sambetts is now known as sambetts|afk | 16:58 | |
| *** danardelean has joined #openstack-kolla | 16:58 | |
| *** danardelean has quit IRC | 16:59 | |
| openstackgerrit | Martin André proposed openstack/kolla-ansible master: [WIP] Use local cirros image if available https://review.openstack.org/498449 | 17:00 |
| *** rwsu has quit IRC | 17:13 | |
| *** ntpttr_laptop has quit IRC | 17:18 | |
| *** ntpttr_laptop has joined #openstack-kolla | 17:21 | |
| glycerine | is there any sort of workaround for using ceph with ubuntu pike images? I get "no active mgr" and 0 pools using binary or source. | 17:22 |
| SamYaple | glycerine: kolla-ansible doesnt have support for luminous, and jewel doesnt work with the Pike images to my understanding | 17:23 |
| SamYaple | glycerine: so you have to go external ceph on the situation | 17:23 |
| hrw | glycerine: you need ceph luminous support. it is not present even in kolla/master yet | 17:24 |
| *** rwsu has joined #openstack-kolla | 17:26 | |
| glycerine | alright thanks for the help | 17:26 |
| hrw | rwellum: can you take a look at https://review.openstack.org/#/c/514614/ one? It adds repo for pike/debian which we need to be able to use pike on aarch64 at all. packages for x86-64 are provided as well. | 17:27 |
| kolla-slack | <carlp> Greetings! I'm trying to build and push the kolla images for pike based on centos to a local registry and it's not working. Many images do not build, and the ones that do build give an error like: ```INFO:kolla.image.build.opendaylight:Trying to push the image ERROR:kolla.image.build.opendaylight:Unknown error when pushing Traceback (most recent call last): File | 17:29 |
| kolla-slack | "/home/carperry/.virtualenvs/py3-kolla/lib/python3.5/site-packages/kolla/image/build.py", line 217, in run self.push_image(image) File "/home/carperry/.virtualenvs/py3-kolla/lib/python3.5/site-packages/kolla/image/build.py", line 238, in push_image stream = json.loads(response) File "/usr/lib/python3.5/json/__init__.py", line 312, in loads s.__class__.__name__)) TypeError: the JSON object must be str, not 'bytes' | 17:29 |
| kolla-slack | INFO:kolla.image.build:Attempt number: 2 to run task: PushTask(opendaylight) ``` any ideas? | 17:29 |
| *** dougsz has quit IRC | 17:29 | |
| *** newmember has quit IRC | 17:33 | |
| hrw | https://bugs.launchpad.net/kolla-ansible/+bug/1727795 - low priority for now but will have to be sorted out | 17:34 |
| openstack | Launchpad bug 1727795 in kolla-ansible "Cirros image is hardcoded to x86-64" [Undecided,New] | 17:34 |
| *** gkadam_ has quit IRC | 17:40 | |
| *** newmember has joined #openstack-kolla | 17:50 | |
| pratapagoutham | anybody knows how to fix this error?? "pip install -U pip when i do this why is it throwing some locale error i tried google and tried almost all nothin worked " | 17:51 |
| kolla-slack | <carlp> Fixed my problem, I installed the registry cert but forgot to restart the docker service. Sorry! | 18:05 |
| *** kbaegis has quit IRC | 18:11 | |
| *** lpetrut has joined #openstack-kolla | 18:18 | |
| *** danardelean has joined #openstack-kolla | 18:19 | |
| SamYaple | hrw: deploying cirros is not a requirement for kolla-ansible. if i remember, it runs in init-runonce, which ive never used on my production clusters | 18:20 |
| *** hachi__ has joined #openstack-kolla | 18:20 | |
| SamYaple | hrw: but just do a bit of an ENV update there and youll be fine | 18:20 |
| *** paramite has quit IRC | 18:20 | |
| *** hachi__ has quit IRC | 18:25 | |
| *** jamesbenson has quit IRC | 18:33 | |
| *** lrensing has quit IRC | 18:36 | |
| *** jamesbenson has joined #openstack-kolla | 18:39 | |
| rwellum | hrw: done | 18:48 |
| sdake | SamYaple hrw - luminous is new vrsion of ceph? | 18:49 |
| *** ntpttr_laptop has quit IRC | 18:51 | |
| SamYaple | ya. latest lts | 18:51 |
| jamesbenson | sdake: 12.2.1 L. | 18:58 |
| SamYaple | jamesbenson: 12.2.2 youre behind a version! | 18:59 |
| SamYaple | jamesbenson: oh nevermind. your right, i forgot i was running devel | 18:59 |
| SamYaple | heh | 18:59 |
| jamesbenson | samyaple: ha! ;-) | 19:00 |
| *** hachi__ has joined #openstack-kolla | 19:00 | |
| jamesbenson | samyaple: have you gotten this message before on the gui: POOL_APP_NOT_ENABLED: application not enabled on 2 pool(s) | 19:02 |
| *** lpetrut has quit IRC | 19:04 | |
| *** lpetrut has joined #openstack-kolla | 19:04 | |
| SamYaple | jamesbenson: yup | 19:05 |
| *** danardelean has quit IRC | 19:05 | |
| SamYaple | in luminous you should set teh application for the pools | 19:05 |
| SamYaple | so `ceph osd pool set application glance rbd | 19:05 |
| SamYaple | somethign like that | 19:05 |
| SamYaple | dont have t odo it for cache layers and stuff | 19:05 |
| *** hachi__ has quit IRC | 19:14 | |
| *** danardelean has joined #openstack-kolla | 19:15 | |
| *** lukepatrick has quit IRC | 19:15 | |
| *** danardelean has quit IRC | 19:15 | |
| *** hachi__ has joined #openstack-kolla | 19:19 | |
| rwellum | jamesbenson: do you use your single NIC concept much? | 19:20 |
| *** gfidente has quit IRC | 19:20 | |
| jamesbenson | We primarily have 2 nics here... one public, one private with no internet access | 19:22 |
| jamesbenson | rwellum^ | 19:22 |
| rwellum | Ok - just curious because I have a case where I might need to deploy without that second nic | 19:23 |
| rwellum | https://review.openstack.org/#/c/484006/5/doc/network-interface.rst - will give this a go | 19:24 |
| jamesbenson | yeah, I think I tested that on our dev cluster, so it should work... | 19:26 |
| *** kbyrne has quit IRC | 19:27 | |
| *** ntpttr_laptop has joined #openstack-kolla | 19:28 | |
| rwellum | You use/setup br0 - but br-ex would be a bit better for the documentation because that's the default in the kolla documentation. | 19:28 |
| rwellum | imo | 19:29 |
| *** kbyrne has joined #openstack-kolla | 19:30 | |
| jamesbenson | samyaple, should gnocchi be rbd? | 19:31 |
| jamesbenson | true, good point. | 19:31 |
| jamesbenson | samyaple: FYI: sudo ceph osd pool application enable volumes rbd | 19:32 |
| SamYaple | jamesbenson: yea its backed into my provisioning so i dont remember the exact syntax :) | 19:33 |
| SamYaple | what does gnocchi store its data as? | 19:33 |
| jamesbenson | samyaple: I guess the automatic tagging doesn't work right off the bat... http://ceph.com/community/new-luminous-pool-tags/#Automatic_Tagging | 19:33 |
| jamesbenson | the command: sudo ceph osd pool application set images rbd. Does work, but I needed to enable them first, so that other command does it all for me. | 19:34 |
| SamYaple | jamesbenson: automatic-tagging is for upgraded clusters, not newly create pools to my understanding | 19:34 |
| jamesbenson | ah, okay, makes sense... | 19:34 |
| jamesbenson | I was going to say, all of those volumes exist here and none are tagged... | 19:34 |
| SamYaple | jamesbenson: anyway if cephfs, rbd, or rgw doesnt fit your application, you can name it something free-form (for example if gnocchi is just injecting raw objects into librados, it wouldnt fit any of that) | 19:35 |
| SamYaple | it needs *a* name, but the only builtin optimizations exist for those 3 above | 19:36 |
| *** lukepatrick has joined #openstack-kolla | 19:38 | |
| SamYaple | jamesbenson: ah even better. as of now the application tags do nothing. its for future use. so you can name them whatever without affecting anything | 19:38 |
| jamesbenson | samyaple: I don't personally used gnocchi at all... | 19:38 |
| *** hrw has quit IRC | 19:38 | |
| jamesbenson | :-\ | 19:39 |
| *** hrw has joined #openstack-kolla | 19:41 | |
| hrw | SamYaple: (cirros) it may not be a requirement but I opened bug as kind of reminder for me/linaro to make it handled | 19:42 |
| hrw | sdake: luminous adds new service: ceph-mgr and so far no one wrote image for it | 19:43 |
| jamesbenson | SamYaple hrw I made my own little script to go and deploy public/private network/router/cirros and SSH into it to verify basics work... | 19:43 |
| hrw | jamesbenson: init-runonce does that | 19:43 |
| hrw | or kind of it | 19:44 |
| jamesbenson | yeah I found that out after the fact... but at least this is tailored for our env. | 19:44 |
| jamesbenson | hrw are you trying to get Luminous on kolla-ansible or k8s? | 19:44 |
| hrw | jamesbenson: I am more just kolla user. other people in team do -ansible -k8s ones | 19:45 |
| rwellum | init-runonce is very tied to the gate architecture though | 19:45 |
| *** harlowja has quit IRC | 19:46 | |
| SamYaple | yea i wouldnt run init-runonce directly. its from like the initial kolla days. honestly we shouldnt even be using it in the gates anymore | 19:46 |
| SamYaple | i didnt realize it still existed | 19:46 |
| *** lrensing has joined #openstack-kolla | 19:52 | |
| rwellum | Why's that SamYaple ? | 19:56 |
| SamYaple | its a non-idempotent script that is hardcoded to do legacy things still | 19:57 |
| * hrw off | 19:57 | |
| rwellum | Right it does call nova, neutron etc - instead of openstack api | 19:57 |
| SamYaple | rwellum: and if you run it more than once you break stuff :) | 19:58 |
| rwellum | Yeah when I was first playing with stuff I was told 'just run init-runonce' - not realizing it was somewhat dangerous and very gate-proprietary. | 19:59 |
| SamYaple | i used to tell people "look at what init-runonce does" but yea | 20:00 |
| *** lpetrut has quit IRC | 20:02 | |
| *** Pavo has joined #openstack-kolla | 20:10 | |
| *** kbaegis has joined #openstack-kolla | 20:14 | |
| *** kbaegis has joined #openstack-kolla | 20:14 | |
| *** kbaegis has quit IRC | 20:18 | |
| *** danardelean has joined #openstack-kolla | 20:19 | |
| *** danardelean has quit IRC | 20:20 | |
| *** danardelean has joined #openstack-kolla | 20:25 | |
| inc0 | now I think we have lock in place that won't allow you to run it more than once | 20:26 |
| *** lukepatrick has quit IRC | 20:26 | |
| kolla-slack | <jascott1> neat https://github.com/nghialv/promviz | 20:27 |
| *** lukepatrick has joined #openstack-kolla | 20:28 | |
| *** newmember has quit IRC | 20:28 | |
| *** manjeets has joined #openstack-kolla | 20:30 | |
| *** harlowja has joined #openstack-kolla | 20:33 | |
| *** ansmith has quit IRC | 20:38 | |
| kfox1111 | jascott1: interesting. | 20:39 |
| kfox1111 | looks similar to the ui weave does. but potentialy much more indepth. | 20:39 |
| *** trozet has joined #openstack-kolla | 20:40 | |
| *** danardelean has quit IRC | 20:43 | |
| *** danardelean has joined #openstack-kolla | 20:43 | |
| *** danardelean has quit IRC | 20:48 | |
| *** hachi__ has quit IRC | 20:51 | |
| *** jtriley has quit IRC | 20:51 | |
| kolla-slack | <jascott1> ah… write up by our friend Matt “The” Butcher https://open.microsoft.com/2017/10/23/announcing-brigade-event-driven-scripting-kubernetes/ | 20:51 |
| kfox1111 | jascott1: is there a unique set in golang? | 20:52 |
| kolla-slack | <jascott1> not in stdlib afaik | 20:52 |
| kfox1111 | :( | 20:52 |
| kolla-slack | <jascott1> lemme check | 20:52 |
| kolla-slack | <jascott1> ya dont think so | 20:53 |
| kfox1111 | ok. I'll just use a map[string]bool then. | 20:53 |
| kfox1111 | thanks. | 20:54 |
| *** ktibi has quit IRC | 20:54 | |
| kolla-slack | <jascott1> btw I got the ceph multinode deploy working and into playbook | 20:58 |
| kolla-slack | <jascott1> felt like this at 1:15am | 20:58 |
| kolla-slack | <jascott1> \o/ | 20:58 |
| kfox1111 | cool. | 20:58 |
| *** mummer has quit IRC | 21:10 | |
| *** trozet has quit IRC | 21:11 | |
| SamYaple | i know this is salt code, but this is makes all ceph-mgr ceph-osd and ceph-mon come up at the same time, even on a fresh cluster, and it works https://github.com/SamYaple/home-salt/tree/master/salt/ceph | 21:11 |
| SamYaple | might be somethign useful to kolla-k8s | 21:11 |
| kfox1111 | SamYaple: I think long term, rook is a much more natural fit. | 21:13 |
| kfox1111 | as it follows the k8s operator pattern. | 21:13 |
| kfox1111 | you helm install rook-operator | 21:13 |
| kfox1111 | and kubectl create -f mycephcluster.yaml | 21:13 |
| *** ntpttr_laptop__ has joined #openstack-kolla | 21:13 | |
| kfox1111 | kubectl create -f cephfs-foo.yaml, etc. | 21:13 |
| kfox1111 | its a bit green yet. but developing fast. | 21:14 |
| kfox1111 | see rook.io for details. | 21:14 |
| *** ntpttr_laptop has quit IRC | 21:14 | |
| SamYaple | kfox1111: what is rook replacing in the scenario i described? | 21:16 |
| kolla-slack | <carlp> After fixing my registry cert issue, I am still unable to build a complete set of images (which I don't really need but a core set would be nice). Lots of the same JSON errors I posted earlier. Any ideas? | 21:16 |
| SamYaple | kfox1111: `kubectl create -f mycephcluster.yaml` should bring up the cluster. period | 21:16 |
| SamYaple | why bring in anything from the outside? | 21:16 |
| kfox1111 | SamYaple: it deals with ceph deployment/management in a k8s way. | 21:16 |
| kfox1111 | mycephcluster is an abstract declariative request for a ceph cluster. | 21:17 |
| SamYaple | yes. but why bring it in when ceph can deal with it itself? | 21:17 |
| SamYaple | im confused here | 21:17 |
| kfox1111 | but ceph cant. | 21:17 |
| SamYaple | it can | 21:17 |
| SamYaple | i just posted code showing that | 21:17 |
| *** Pavo has quit IRC | 21:18 | |
| kfox1111 | SamYaple: who runs that salt stack? where does it run? | 21:19 |
| kfox1111 | rook follows the operator pattern. | 21:19 |
| SamYaple | kfox1111: who runs kubectl, where does that run? | 21:19 |
| SamYaple | it doesnt need to be salt | 21:19 |
| kfox1111 | so, it runs in k8s. it constantly looks for declarative k8s objects to inform it what to do. | 21:19 |
| SamYaple | its what the code is doing | 21:19 |
| kfox1111 | operators are the idea of using k8s to extend k8s to orchestrate more complex things. | 21:20 |
| kfox1111 | so deploying a whole ceph cluster works like deploying a pod. | 21:20 |
| SamYaple | my point is you can remove the need for orchestration for ceph specifically... im making no comment on rook in genreal | 21:20 |
| kfox1111 | yeah, no. what I'm sayuing though, is in a k8s system, doing things with operators are much preffered to doing it manually. | 21:21 |
| SamYaple | im confused as to what manual action you are talking about | 21:21 |
| SamYaple | at some point an operator has to say "run ceph", but thats it | 21:21 |
| kfox1111 | maybe I'm miswording things... | 21:21 |
| kfox1111 | let me try a different angle. | 21:22 |
| kfox1111 | in k8s, using k8s is the preffered way of interacting with k8s then running an external tool that runs k8s for you. | 21:22 |
| kfox1111 | ie, kubectl create ceph is better then salt deploy ceph in k8s. | 21:22 |
| SamYaple | yes i feel you are definitely not understanding what im saying | 21:22 |
| kfox1111 | entirely possible. | 21:23 |
| SamYaple | im saying the logic in salt can be turned into mycephcluster.yaml which is just k8s pods and secrets and stuff and ceph will form on its own | 21:23 |
| SamYaple | im not saying you should use salt | 21:23 |
| kfox1111 | no. thats not quite right. | 21:23 |
| SamYaple | im saying look at that logic to make your ceph not need outside orchestration | 21:23 |
| kfox1111 | mycephcluster.yaml would be the values you feed your salt logic. | 21:23 |
| SamYaple | or rook | 21:23 |
| * inc0 grabs popcorn | 21:23 | |
| SamYaple | oh jesus. listen im nto talking about salt | 21:24 |
| kfox1111 | you have a few things: | 21:24 |
| kfox1111 | 1. declarative description of request. | 21:24 |
| kfox1111 | 2. logic to take in declarative request and realize it. | 21:24 |
| SamYaple | im saying i coded some logic in salt that orchestrationlessly starts a ceph cluster. | 21:24 |
| kfox1111 | 3. engine that performs #2. | 21:24 |
| SamYaple | we shoudl put that logic in k8s | 21:24 |
| kfox1111 | 4. env on which 3 executes. | 21:24 |
| SamYaple | kfox1111: do you think we need rook to start memcached containers? | 21:25 |
| kfox1111 | rook doesn't deal with memcached. | 21:25 |
| kfox1111 | it deals with deploying ceph using k8s as a k8s extention. | 21:26 |
| kfox1111 | rook covers 2 and 3. using k8s as #4, | 21:26 |
| kfox1111 | and #1 comes from the user via k8s crds. | 21:26 |
| SamYaple | ok. let me rephrase because you are going down the wrong path here | 21:26 |
| SamYaple | do you think we need an operator to deploy memcached | 21:27 |
| *** ansmith has joined #openstack-kolla | 21:27 | |
| kfox1111 | unlikely. memcached's so stateless and its orchestration is very primitive. k8s shoudl be able to handle it with its built in primatives. | 21:27 |
| kfox1111 | galera though, probably needs an operator. | 21:27 |
| SamYaple | yay. here we are. we dont need an operator for ceph then | 21:28 |
| SamYaple | twas my point | 21:28 |
| SamYaple | just like we dont need one for rabbitmq with autocluster+etcd | 21:28 |
| * sdake grabs inc0's popccorn | 21:28 | |
| kfox1111 | we do need an operator for ceph. as its a non trivial orchestration that k8s itself will have a hard time handling. | 21:28 |
| SamYaple | kfox1111: and weve finally drilled down to the issue. that is the part im saying ive solved. it is trival to "orchestrate" now | 21:29 |
| kfox1111 | you gotta bring up mons in the right order, generate creds, pass them around, then register osd's, etc. | 21:29 |
| *** danardelean has joined #openstack-kolla | 21:29 | |
| SamYaple | no you dont | 21:29 |
| sdake | kfox1111 there is always openshift-ansible | 21:29 |
| SamYaple | thats.. my.. point.. | 21:29 |
| sdake | or openshift-service-broker I mean | 21:29 |
| kfox1111 | SamYaple: your missing my point I think. rook is solving this use caase already. no need to reinvent that wheel. | 21:29 |
| SamYaple | kfox1111: see this isnt reinvention, this is removing that layer | 21:29 |
| SamYaple | its simplifying because thats not needed | 21:29 |
| kfox1111 | I don't think you understand operators. | 21:30 |
| SamYaple | but man this was a pain to talk about. im just going to stay away from any k8s talks in this channel. they are never fun | 21:30 |
| inc0 | kfox1111: I think what Sam is saying is, besides secret generation, key distribution and stuff like that, there is little orchestration | 21:30 |
| kfox1111 | hah. | 21:30 |
| kfox1111 | "besides all the hard stuff, there is little orchestration" | 21:30 |
| SamYaple | this wasnt even about k8s lol. this was about ceph | 21:31 |
| inc0 | so if you can pre-generate ceph secrets and keyrings, you can just load them and start mons | 21:31 |
| *** danardelean has quit IRC | 21:31 | |
| kfox1111 | yeah, bbut you said for kolla-kubernetes, we should deploy ceph like X. | 21:31 |
| kfox1111 | I'm saying, for kolla-kubernetes we shoudl suggesst deploying ceph like Y. | 21:31 |
| kfox1111 | where I am stating Y is more k8s natural. | 21:31 |
| kfox1111 | just my $0.02 | 21:31 |
| SamYaple | im saying dont complicate somethign that doesnt need complication | 21:32 |
| SamYaple | like dont use operaters to deploy memcached | 21:32 |
| inc0 | I'd say we shouldn't deploy ceph at all, but I think what's Sam is saying is that rook is essentialy redundant if you can pre-generate keyrings | 21:32 |
| kfox1111 | I never suggested we shoud deploy memcached with operators. | 21:32 |
| kfox1111 | but deploying system services using k8s and not k8s is more complicated then just deploying everythign with k8s. | 21:33 |
| SamYaple | im all about code reuse, but im more about simplifying to get rid of the need for more code | 21:33 |
| kfox1111 | thats I think where we differ. | 21:33 |
| SamYaple | but see, in this case it isnt | 21:33 |
| SamYaple | for ceph specifically | 21:33 |
| SamYaple | fire off all the containers at once and it can work, as ive done | 21:33 |
| inc0 | SamYaple: but you did pre-load secrets right | 21:34 |
| inc0 | ? | 21:34 |
| SamYaple | inc0: if you look at secrets like passwords, then yes i did put in the passwords | 21:35 |
| kfox1111 | sounds like the root of discussion is a philisophical difference of opinion. thats ok. but I think we are at an impass. thats ok. | 21:35 |
| SamYaple | do you autogenerate keystone passwords? | 21:35 |
| inc0 | and you did put monmap, initial at least | 21:35 |
| SamYaple | inc0: nope | 21:35 |
| SamYaple | inc0: just the secrets | 21:35 |
| kfox1111 | kolla-kubernetes won't provide a ceph solution. so everyone can feel free to do whats best for them. | 21:35 |
| kfox1111 | ceph-deploy, salt, rook, all should work. | 21:36 |
| SamYaple | yea im just talking about ceph. im not going to talk about k8s anymore | 21:37 |
| kfox1111 | yeah. which orchestration tool you use is orthoganal to the ceph software. | 21:46 |
| SamYaple | again, no orchestration tool needed to deploy ceph. this is what ive done | 21:50 |
| openstackgerrit | Christian Berendt proposed openstack/kolla-ansible master: Remove trailing whitespace in tools/init-runonce https://review.openstack.org/515532 | 21:50 |
| SamYaple | that was the message i was bring the channel | 21:50 |
| kfox1111 | thats where I'm confused. salt is an orchestration tool? | 21:51 |
| SamYaple | remove salt from the equation here. i just-so-happened to be using salt. thats where implemented the stuff | 21:52 |
| openstackgerrit | Christian Berendt proposed openstack/kolla-ansible master: Remove trailing whitespaces https://review.openstack.org/515532 | 21:52 |
| kfox1111 | ah. so you are trying to discuss the algorithm in the salt logic more then the actual implementation? | 21:52 |
| SamYaple | yup | 21:53 |
| kfox1111 | ah. ok. | 21:53 |
| SamYaple | perhaps were mixing terms here. but my point is the same way you would deploy memcached or rabbitmq+autocluster is just start all the containers, and they work themselves out | 21:53 |
| SamYaple | now you can do that with ceph | 21:53 |
| *** dardelean has quit IRC | 21:53 | |
| kfox1111 | conceptually yes, but you still need to actually realize the implmentation, like you did in salt. its code to maintain. | 21:54 |
| SamYaple | i mean sure, so is telling k8s to start memcached containers | 21:54 |
| kfox1111 | for memcache, the procedure is something like this: | 21:55 |
| kfox1111 | user provides values and a chart to helm. helm uses templates and values to generate k8s objects. | 21:55 |
| *** jascott1 has joined #openstack-kolla | 21:55 | |
| kfox1111 | k8s objects are uploaded to k8s which uses its built in orchestration to deploy memcached in pods. | 21:55 |
| *** danardelean has joined #openstack-kolla | 21:56 | |
| kfox1111 | k8s's level of orchestration is very simplistic. it knows how to deploy pods basically in parallel but with no interconnectivity. its up to the pods themselves to orchestrate things, or something driving k8s to orchestrate. | 21:57 |
| kfox1111 | if k8s's built in primatives are not up to the task. | 21:57 |
| jascott1 | SamYapple where is this ceph stuff you made? Im working with ceph atm... | 21:57 |
| SamYaple | right thats what i mean by mixing terms. all we need are k8s primatives to deploy ceph now. no outside or advanced orchestration needed | 21:58 |
| SamYaple | jascott1: nah i just worked on a new flow for ceph deploying. happened to be using salt. can deploy ceph-mon ceph-osd ceph-mgr etc all in parallel without the orcehstration dance https://github.com/SamYaple/home-salt/tree/master/salt/ceph | 21:58 |
| jascott1 | thanks | 21:59 |
| *** dciabrin has quit IRC | 22:00 | |
| *** jascott1 has quit IRC | 22:01 | |
| kfox1111 | SamYaple: thats where i'm still confused. are you saying we should be using your salt logic, or implementing it in <what>? | 22:02 |
| kfox1111 | ie, what is the action you are proposing. it doesn't 'just work' todayu, without some change. what is the change? | 22:03 |
| SamYaple | kfox1111: im saying there is a massive reduction in "logic" to the point that there isnt a need to coordinate between containers. every container has all the info it needs to do its job | 22:03 |
| kfox1111 | ok. your arguing for 'internal orchestration' | 22:03 |
| kfox1111 | similar to the kubernetes-entrypoint idea. | 22:03 |
| kfox1111 | make the pods aware of k8s and each other enough to self orchestrate. | 22:04 |
| SamYaple | possibly? if init containers count as internal orechstartions | 22:04 |
| kfox1111 | yeah. its internal to the orchestration that k8s provides. | 22:04 |
| kfox1111 | external orchestration is driving k8s's orchestration. | 22:04 |
| kfox1111 | both are also possible. | 22:04 |
| kfox1111 | external orchestrator -> k8s orchestration -> internal pod orchestration. | 22:04 |
| kfox1111 | though you usually do one or the other. | 22:05 |
| *** jamesbenson has quit IRC | 22:05 | |
| kfox1111 | I am not against deploying ceph that way. but, I'm more against kolla-kubernetes implementing ceph at all. | 22:05 |
| kfox1111 | helm-ceph or something should. | 22:05 |
| kfox1111 | ie, helm-ceph would be a competitor of rook. | 22:06 |
| kfox1111 | I don't think kolla-kubernetes should compete with rook. | 22:06 |
| kfox1111 | the scope of just running openstack on k8s is huge enough. :/ | 22:06 |
| kfox1111 | thats not to say kolla-kubernetes shouldn't by default require a ceph. I think that is a good thing. | 22:07 |
| kfox1111 | openstack works best with ceph IMO. | 22:07 |
| kfox1111 | but more like, ceph & k8s are base requirements of kolla-kubernetes. you bring those two things before starting on kolla-kubernetes deployment. | 22:08 |
| *** genek has joined #openstack-kolla | 22:09 | |
| kfox1111 | does that kind of help explain my push back? | 22:12 |
| *** genek has quit IRC | 22:13 | |
| *** dciabrin has joined #openstack-kolla | 22:15 | |
| SamYaple | i was unaware that 1: kolla-k8s didnt deploy ceph currently and 2: kolla-k8s didnt want to deploy ceph | 22:15 |
| SamYaple | either of those facts would have helped, yes | 22:15 |
| kfox1111 | ah. ok. sorry for the confusion. | 22:16 |
| SamYaple | its cool. as always kfox1111, we leave on the same page agreeing with each other (mostly). we just take the long way to get tehre | 22:16 |
| kfox1111 | heh. I know, right? :) | 22:17 |
| SamYaple | its cool, i still like talking to you | 22:17 |
| kfox1111 | me too. :) | 22:17 |
| SamYaple | you talk to yourself? | 22:17 |
| SamYaple | ;) | 22:18 |
| kfox1111 | usually the voices do all the talking. ;) | 22:18 |
| *** spiette has quit IRC | 22:25 | |
| *** itlinux has quit IRC | 22:39 | |
| rwellum | kfox1111: the issue I think is that kolla-ansible is a deployer or orchestrator and kolla-kubernetes is not? | 22:42 |
| rwellum | But they look like a 1-1 comparison but they are not. | 22:42 |
| kfox1111 | yeah. thats definitely part of it I think. | 22:42 |
| kfox1111 | there is definately overlap between say ansible and kolla, | 22:43 |
| kfox1111 | but there is also some fundimentally differnet things each. | 22:43 |
| kfox1111 | ie, k8s does self healing. ansible doesnt so much. | 22:43 |
| rwellum | If anything helm is closer to ansible | 22:44 |
| kfox1111 | ansible provides primitives more at a node level. | 22:44 |
| kfox1111 | k8s is more cluster level primiatives. | 22:44 |
| kfox1111 | yeah. ansible does templating. k8s doesn't. | 22:44 |
| kfox1111 | helm + k8s is maybe closer to ansible. but still a bit disjoint. | 22:45 |
| kfox1111 | I tend to think of k8s in a much different way though. from this angle.... | 22:45 |
| kfox1111 | k8s is closer to a cluster os. | 22:45 |
| kfox1111 | a primative one, but on which more complex things can be built. | 22:45 |
| kfox1111 | ansible is more of a node management tool with a bit of workflow and resource management in it. | 22:46 |
| kfox1111 | good for different things. | 22:46 |
| kfox1111 | when you think of k8s as an os, then you deploy stuff in the os, rather then going under the hood and deploying things. | 22:47 |
| kfox1111 | (this is very similar to coreos's phylosophy) | 22:47 |
| kfox1111 | sshing into a node is kind of like attaching a debugger to a kernel and poking around. should be needed almost never. | 22:50 |
| rwellum | Good description/analogy ty | 22:53 |
| *** salv-orl_ has quit IRC | 22:53 | |
| *** salv-orlando has joined #openstack-kolla | 22:54 | |
| rwellum | so based on that kfox1111 - how SHOULD one orchestrate kolla-kubernetes? | 22:55 |
| kfox1111 | the jury's still out on that one. but I still kind of think we should be doing an operator similar to rook. | 22:55 |
| kfox1111 | openstacks orchestration is fairly complex. especially during upgrades. | 22:56 |
| kfox1111 | doing it as an operator allows for the greatest flexability and the easiest use I think. | 22:56 |
| rwellum | Cool. Let me know when the jury gets called in with the decision. :) | 22:56 |
| kfox1111 | part of that is its the communities role to decide, or implement both. | 22:56 |
| kfox1111 | I don't really have the cycles unfortunately to do too much with implementing orchestration. | 22:57 |
| *** salv-orlando has quit IRC | 22:59 | |
| kfox1111 | I guess what I'm saying is, those who write the code get to have the most say in that? | 23:00 |
| rwellum | I guess yeah. | 23:00 |
| *** trozet has joined #openstack-kolla | 23:04 | |
| *** lrensing has quit IRC | 23:08 | |
| openstackgerrit | Merged openstack/kolla master: XenServer: Add packages for initial support https://review.openstack.org/510266 | 23:10 |
| *** lukepatrick has quit IRC | 23:14 | |
| *** trozet has quit IRC | 23:17 | |
| *** danardelean has quit IRC | 23:19 | |
| *** jascott1 has joined #openstack-kolla | 23:28 | |
| *** danardelean has joined #openstack-kolla | 23:35 | |
| *** danardelean has quit IRC | 23:39 | |
| *** salv-orlando has joined #openstack-kolla | 23:54 | |
| *** gkadam_ has joined #openstack-kolla | 23:56 | |
| *** salv-orlando has quit IRC | 23:59 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!