| Hallfonce | then I tried another kolla install with openvswitch: ovs-vsctl show would be br-ex with port enp0s31f6 | 00:00 |
|---|---|---|
| Hallfonce | no success either | 00:00 |
| SamYaple | i think youre a bit confused as to how this hsould be setup. with a single port the setup is the same, whether linuxbridge or openvsiwtch | 00:01 |
| Hallfonce | You're right, I got totally confused, I've tried so many setup from so many references... | 00:01 |
| SamYaple | let me see if i can find some link from when i wrote this | 00:01 |
| SamYaple | but again, i dont recommend doing this for the trouble youre having | 00:01 |
| Hallfonce | You mean setup an AIO kolla with a single port? | 00:02 |
| SamYaple | yea. its ok for testing, but its still a pain | 00:02 |
| Hallfonce | Sure, I understand | 00:03 |
| Hallfonce | I've found this out the hard way... | 00:03 |
| *** chas__ has quit IRC | 00:08 | |
| *** chas has joined #openstack-kolla | 00:08 | |
| SamYaple | Hallfonce: score! one of my pastebins from over a year ago | 00:12 |
| SamYaple | http://paste.openstack.org/show/478029/ | 00:12 |
| Hallfonce | Epic ! | 00:12 |
| SamYaple | Hallfonce: ip l a fake-br10 type veth peer name fake-ovs; brctl addif br10 fake-br10; ip l s fake-br10 up; ip l s fake-ovs up | 00:12 |
| SamYaple | where br10 can be replaced with enp0s31f6 | 00:13 |
| SamYaple | fake-ovs is neutron_external_interface | 00:13 |
| *** chas has quit IRC | 00:13 | |
| SamYaple | so anyway. the deal is it doesnt matter whether you are using ovs or linuxbridge, though i would recommend ovs | 00:15 |
| SamYaple | the bridge setup on the outside is the same | 00:15 |
| SamYaple | also, i dont think its talked about, but you can _only_ use flat networks | 00:15 |
| SamYaple | once you launch a vm, assign a floating ip and add the appropriate secgroup rules and youll be able to hit the floating ip | 00:16 |
| Hallfonce | Ok, i went a bit fast for me | 00:18 |
| Hallfonce | I assume you bridged enp0s31f6 into br10 ? | 00:19 |
| Hallfonce | so i would add brctl addif br10 enp0s31f6 | 00:20 |
| Hallfonce | If so, I've tried this setup at some point, and I understand well, I was missing the secgroup rules? | 00:22 |
| *** rhallisey has quit IRC | 00:25 | |
| *** JRobinson__ has joined #openstack-kolla | 00:26 | |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: kolla-common value get functions https://review.openstack.org/416069 | 00:27 |
| *** JRobinson__ has quit IRC | 00:29 | |
| SamYaple | Hallfonce: let me put it to you this way, i doubt you messed up the veth pair part, i never said it was complicated, just confusing | 00:30 |
| Hallfonce | sure | 00:30 |
| Hallfonce | I got confused with the neutron_plugin_agent | 00:31 |
| Hallfonce | as you said it doesnt matter | 00:31 |
| Hallfonce | as long as it's the same on the outside | 00:31 |
| SamYaple | i would stick with openvswitch, only because thats the piece that i wrote and is default because its most tested (its gated against) | 00:31 |
| Hallfonce | ok I will stick to it then | 00:33 |
| Hallfonce | about the secgroup rules, I sticked to what was in my /usr/share/kolla/init-runonce | 00:33 |
| Hallfonce | I only edited 1 line in there about the public subnet | 00:34 |
| Hallfonce | to match my external network subnet | 00:34 |
| Hallfonce | anyway, I'm going to give this another try with your advices | 00:35 |
| SamYaple | use a cirros image and setup a ping from the console. then find out where the icmp is dropping and youll figure out where your problem is | 00:36 |
| Hallfonce | are there default user/password for the cirros image ? | 00:38 |
| Hallfonce | otherwise I need to access port 22 to use a cert | 00:38 |
| SamYaple | Hallfonce: there are indeed default user and password | 00:38 |
| Hallfonce | cool, I'll google it | 00:40 |
| SamYaple | Hallfonce: it tells you when you try to log in, but googling will work too | 00:41 |
| Hallfonce | Ok good to know! | 00:41 |
| Hallfonce | Thanks a lot for time, it's been very helpful! | 00:41 |
| Hallfonce | hope I won't have to bother you again about this! | 00:42 |
| SamYaple | its no bother, just hard to predict one of the 20 things that could be wrong when doing this. tcpdump is going to be your friend | 00:43 |
| Hallfonce | indeed | 00:45 |
| Hallfonce | I'll get there step by step I guess | 00:45 |
| kfox1111 | there we go. | 00:47 |
| *** david-lyle has joined #openstack-kolla | 00:51 | |
| *** jmccarthy has quit IRC | 00:51 | |
| *** jmccarthy has joined #openstack-kolla | 00:52 | |
| *** tovin07 has joined #openstack-kolla | 00:54 | |
| *** david-lyle has quit IRC | 00:55 | |
| *** tovin07 has quit IRC | 00:57 | |
| *** salv-orlando has joined #openstack-kolla | 01:00 | |
| *** salv-orlando has quit IRC | 01:02 | |
| *** salv-orlando has joined #openstack-kolla | 01:02 | |
| *** zhurong has joined #openstack-kolla | 01:05 | |
| *** salv-orlando has quit IRC | 01:07 | |
| *** Hallfonce has quit IRC | 01:11 | |
| *** tovin07 has joined #openstack-kolla | 01:16 | |
| *** yingjun has joined #openstack-kolla | 01:29 | |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Move trove precheck into its own role https://review.openstack.org/416032 | 01:36 |
| *** salv-orlando has joined #openstack-kolla | 02:03 | |
| openstackgerrit | jimmygc proposed openstack/kolla: Add sanity check to install_pip in macros.j2 https://review.openstack.org/414848 | 02:03 |
| *** clayton has quit IRC | 02:06 | |
| *** mdnadeem has joined #openstack-kolla | 02:07 | |
| *** salv-orlando has quit IRC | 02:08 | |
| *** n0isyn0ise has quit IRC | 02:11 | |
| *** sdake has joined #openstack-kolla | 02:12 | |
| *** n0isyn0ise has joined #openstack-kolla | 02:13 | |
| *** sdake has quit IRC | 02:15 | |
| *** sdake has joined #openstack-kolla | 02:16 | |
| sdake | evening folks | 02:16 |
| openstackgerrit | Wei Cao proposed openstack/kolla-ansible: Add freezer ansible role https://review.openstack.org/402224 | 02:23 |
| *** zhurong has quit IRC | 02:25 | |
| sdake | off to more tech support BBL | 02:25 |
| *** clayton has joined #openstack-kolla | 02:25 | |
| Jeffrey4l__ | morning | 02:26 |
| sdake | sup Jeffrey4l__ | 02:26 |
| Jeffrey4l__ | hey sdake, happy new year ;) | 02:26 |
| sdake | say - might try running if gate==rax_iad then set libvirt type to qemu | 02:26 |
| sdake | thanks you too :) | 02:26 |
| sdake | although yours is down the road a bit :) | 02:26 |
| sdake | ok gotta jet reworking my dad's computer | 02:27 |
| Jeffrey4l__ | sdake, libvirty_type=qemu all the time. | 02:27 |
| Jeffrey4l__ | in all gate. | 02:27 |
| Jeffrey4l__ | so it should not related. | 02:28 |
| *** Jeffrey4l__ is now known as Jeffrey4l | 02:28 | |
| *** sdake has quit IRC | 02:31 | |
| dupengfei | srwilkers： hello | 02:34 |
| openstackgerrit | Merged openstack/kolla-ansible: Move horizon precheck into its own role https://review.openstack.org/416068 | 02:40 |
| openstackgerrit | dupengfei proposed openstack/kolla: Not to pull the base images when building. https://review.openstack.org/408940 | 02:41 |
| Pavo | anyone know when the new murano and magma changes are goona get pushed to pip package of kolla? | 02:42 |
| openstackgerrit | dupengfei proposed openstack/kolla-kubernetes: Helm charts for keepalived pod https://review.openstack.org/413985 | 02:43 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Add enable_horizon_* parameter for dashboard https://review.openstack.org/413484 | 02:50 |
| openstackgerrit | dupengfei proposed openstack/kolla-kubernetes: Helm charts for heat-create-db and heat-manage-db https://review.openstack.org/414496 | 02:50 |
| *** david-lyle has joined #openstack-kolla | 02:52 | |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Move ironic precheck into its own role https://review.openstack.org/415481 | 02:56 |
| *** david-lyle has quit IRC | 02:57 | |
| openstackgerrit | dupengfei proposed openstack/kolla-kubernetes: Helm charts for keepalived pod https://review.openstack.org/413985 | 03:00 |
| *** Pavo has quit IRC | 03:03 | |
| *** salv-orlando has joined #openstack-kolla | 03:04 | |
| *** salv-orlando has quit IRC | 03:08 | |
| *** chas has joined #openstack-kolla | 03:10 | |
| *** chas has quit IRC | 03:15 | |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Move searchlight precheck into its own role https://review.openstack.org/415135 | 03:18 |
| *** dave-mccowan has joined #openstack-kolla | 03:23 | |
| *** goldyfruit has joined #openstack-kolla | 03:29 | |
| *** clayton has quit IRC | 03:34 | |
| *** tovin07 has quit IRC | 03:34 | |
| *** jmccarthy has quit IRC | 03:34 | |
| *** yingjun has quit IRC | 03:34 | |
| *** mdnadeem has quit IRC | 03:34 | |
| *** yingjun has joined #openstack-kolla | 03:35 | |
| *** clayton has joined #openstack-kolla | 03:35 | |
| *** tovin07 has joined #openstack-kolla | 03:35 | |
| *** jmccarthy has joined #openstack-kolla | 03:35 | |
| *** mdnadeem has joined #openstack-kolla | 03:35 | |
| *** mdnadeem has quit IRC | 03:35 | |
| *** prameswar has joined #openstack-kolla | 03:50 | |
| *** sdake has joined #openstack-kolla | 03:51 | |
| *** g3ek has quit IRC | 03:54 | |
| *** sdake has quit IRC | 03:56 | |
| *** g3ek has joined #openstack-kolla | 04:02 | |
| *** salv-orlando has joined #openstack-kolla | 04:05 | |
| *** yingjun has quit IRC | 04:06 | |
| *** Pavo has joined #openstack-kolla | 04:09 | |
| *** salv-orlando has quit IRC | 04:09 | |
| *** bmace has quit IRC | 04:14 | |
| *** bmace has joined #openstack-kolla | 04:14 | |
| *** goldyfruit has quit IRC | 04:15 | |
| *** sacharya has quit IRC | 04:15 | |
| *** sacharya has joined #openstack-kolla | 04:16 | |
| *** sacharya has quit IRC | 04:21 | |
| *** dave-mccowan has quit IRC | 04:22 | |
| *** masber has joined #openstack-kolla | 04:27 | |
| openstackgerrit | jimmygc proposed openstack/kolla: Add sanity check to install_pip in macros.j2 https://review.openstack.org/414848 | 04:27 |
| *** mdnadeem has joined #openstack-kolla | 04:27 | |
| masber | hi | 04:38 |
| masber | does kolla provide high availability? | 04:38 |
| masber | I assume that with the infrastructure in place (e.g. kubernetes, docker swarm, mesos + marathon, etc) someone could just deploy the controller openstack components on containers and monitor them so options like galera or pacemaker wouldn't be needed. Am I crazy saying that if is it true? | 04:42 |
| *** Pavo has quit IRC | 04:51 | |
| *** prameswar has quit IRC | 04:54 | |
| *** sdake has joined #openstack-kolla | 04:56 | |
| sdake | Jeffrey4l i looked over the code and don't see where qemu is being used | 04:57 |
| sdake | mind offering a pointer for my sanity? :) | 04:57 |
| Jeffrey4l | sdake, kolla project re-user kolla-ansible's deploy_aio.sh script. | 04:58 |
| Jeffrey4l | https://github.com/openstack/kolla-ansible/blob/master/tools/deploy_aio.sh#L86 | 04:58 |
| sdake | kolla repo | 04:58 |
| sdake | ? | 04:58 |
| sdake | that looks good | 04:58 |
| Jeffrey4l | kolla repo reuse kolla-ansible's deploy_aio.sh script. | 04:59 |
| Jeffrey4l | yep. | 04:59 |
| sdake | i forgot about the refactor | 04:59 |
| sdake | you did on that file | 04:59 |
| sdake | even though I'm pretty sure i reviewed it :) | 04:59 |
| Jeffrey4l | ;) | 04:59 |
| sdake | 12 hrs of tech support for parents | 04:59 |
| sdake | = painful | 04:59 |
| sdake | the yearly reinstall of windows | 05:00 |
| sdake | actually pristine backup restore :) | 05:00 |
| Jeffrey4l | i will try to enable libvirt-debuginfo to catch more info. | 05:00 |
| Jeffrey4l | haha, buy a mac ;) | 05:00 |
| sdake | i have a mac | 05:00 |
| Jeffrey4l | for your parent | 05:00 |
| sdake | 5 in fact :) | 05:00 |
| sdake | they dont like laptops | 05:00 |
| Jeffrey4l | wow | 05:00 |
| sdake | they want a computer they can sit at | 05:00 |
| Jeffrey4l | ok.. | 05:01 |
| sdake | i have my work 15", my personal 15" my personal 13" and a personal 12" | 05:01 |
| sdake | the 13" is my wifes, and my wife also has a 12" | 05:01 |
| *** Pavo has joined #openstack-kolla | 05:01 | |
| Jeffrey4l | cool. | 05:01 |
| Pavo | sdake you have any idea when the lastest murano and magma patches are gonna get pushed to kolla pip package? | 05:04 |
| sdake | pavo i've been on pto for 2 weeks bro | 05:05 |
| *** salv-orlando has joined #openstack-kolla | 05:05 | |
| Pavo | just wondering an expected time frame | 05:06 |
| Pavo | I would just do a git clone of master but then I have to re build all the images | 05:06 |
| Pavo | guess I can do that to test upgrade functionality | 05:07 |
| *** salv-orlando has quit IRC | 05:10 | |
| *** chas has joined #openstack-kolla | 05:11 | |
| *** prameswar has joined #openstack-kolla | 05:12 | |
| sdake | Pavo inc0 is probably the person to talk to | 05:13 |
| sdake | pavo i really have no visibility into kolla-ansible at a detailed level atm | 05:13 |
| sdake | pavo focused on kolla-kubernetes atm. | 05:13 |
| Pavo | ok | 05:14 |
| *** mdnadeem has quit IRC | 05:15 | |
| sdake | pavo and kolla repo of course | 05:15 |
| *** chas has quit IRC | 05:16 | |
| sdake | pavo have hard time reviewing 3 deliverables at high output | 05:16 |
| Pavo | ? | 05:16 |
| sdake | kolla = deliverable, kolla-ansible = deliverable, kolla-kubernetes = deliverable | 05:16 |
| Pavo | ah | 05:16 |
| *** mdnadeem has joined #openstack-kolla | 05:16 | |
| sdake | we have a slew of people reviewing kolla-ansible - suspect one of them would know | 05:17 |
| sdake | pavo the real issue is there is no backports team that i am aware of | 05:17 |
| sdake | as in no stable maintenance team | 05:17 |
| sdake | inc0 can probably fix that up | 05:17 |
| sdake | something to be added to the agenda i think | 05:17 |
| sdake | portdirect you about | 05:17 |
| *** sacharya has joined #openstack-kolla | 05:17 | |
| Pavo | wished I knew what that means lol but ok | 05:17 |
| sdake | pavo a stable maintenance team does backports on a consistent basis and releases z streams consistently | 05:18 |
| sdake | pavo fwiw, I believe Jeffrey4l did release a new z stream | 05:18 |
| Pavo | ah ok | 05:18 |
| sdake | Pavo although whatyou mention sound like features | 05:18 |
| sdake | pavo you might try pbourke | 05:18 |
| Pavo | how would I know when the features would be added to the pip package install of kolla? | 05:19 |
| *** sacharya has quit IRC | 05:22 | |
| *** skramaja has joined #openstack-kolla | 05:22 | |
| Jeffrey4l | sdake, i didn't release next z stream still now. we need fix the selinux issue, in mitaka branch Or annoucement that mitaka do not support selinux very well. | 05:23 |
| Pavo | anything I can help with that Jeffrey4l | 05:23 |
| sdake | Jeffrey4l bummer - just a heads up - i had indicated iwould be off pto on jan 4th | 05:24 |
| sdake | Jeffrey4l i have to travel 3rd-6th | 05:24 |
| sdake | and wont make wednesday meeting | 05:24 |
| Jeffrey4l | roger. | 05:24 |
| sdake | i'll also mostly be unavailable on irc as well | 05:24 |
| Jeffrey4l | i will talked with inc0 about this. | 05:25 |
| Jeffrey4l | Pavo, mitaka stable branch gate failed unless disable selinux. but it is OK in other branch and locally test. | 05:25 |
| Pavo | hmmm | 05:26 |
| Jeffrey4l | I have debugged this, but got nothing. really no idea what to do next. | 05:26 |
| Pavo | ok I guess I am little ocnfused because newton 3.0.1 does the same thing | 05:26 |
| Jeffrey4l | "does the same thing" means? | 05:27 |
| Pavo | instances do not run, horizon doesn't allow uploading of images in dahboard with selinx enabled | 05:27 |
| Jeffrey4l | Pavo, yep. have u ever gOt the audit logs? | 05:28 |
| Pavo | but selinux disabled works perfectly fine | 05:28 |
| Pavo | yeah I posted them in the bug report | 05:28 |
| Pavo | but didn't see anything interesting | 05:28 |
| * Jeffrey4l is checking. | 05:29 | |
| Pavo | https://www.dropbox.com/s/bbzj85er47to68a/audit.log?dl=0 | 05:30 |
| Jeffrey4l | roger. | 05:30 |
| Pavo | Jeffrey4l if there is any testing you need let me know | 05:31 |
| Pavo | I am not good at coding but am good at testing things out | 05:32 |
| Jeffrey4l | Pavo, are u pretty sure that the audit.log has the logs when issue occur? | 05:32 |
| Jeffrey4l | i do not found any interesting log, too. ;( | 05:33 |
| Pavo | yeah I turned selinux back on, rebooted tried to upload an image in horizon and it failed as usual | 05:33 |
| Pavo | uploaded the log to dropbox | 05:33 |
| *** yingjun has joined #openstack-kolla | 05:33 | |
| Pavo | and turned selinux back off | 05:33 |
| Pavo | then rebooted | 05:33 |
| Pavo | now everything is working perfect | 05:33 |
| Jeffrey4l | Pavo, hrm. could u try again. and post all /var/lib/docker/volumes/kolla_logs file out? | 05:34 |
| Jeffrey4l | with the audit.log. | 05:34 |
| Pavo | can tomorrow, already in bed now | 05:34 |
| Jeffrey4l | if possible ;) | 05:34 |
| Pavo | but will do | 05:34 |
| Jeffrey4l | cool. thanks. | 05:34 |
| Pavo | no problem | 05:35 |
| *** sdake has quit IRC | 05:42 | |
| *** satyar has joined #openstack-kolla | 05:44 | |
| *** yingjun has quit IRC | 05:51 | |
| openstackgerrit | Merged openstack/kolla-ansible: Move sahara precheck into its own role https://review.openstack.org/416051 | 05:53 |
| *** Pavo has quit IRC | 06:01 | |
| *** salv-orlando has joined #openstack-kolla | 06:06 | |
| *** Pavo has joined #openstack-kolla | 06:06 | |
| openstackgerrit | Merged openstack/kolla-ansible: Move memcached precheck into its own role https://review.openstack.org/416059 | 06:09 |
| *** salv-orlando has quit IRC | 06:11 | |
| *** duonghq has joined #openstack-kolla | 06:14 | |
| duonghq | happy new year guys | 06:14 |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Update the horizon port into all.yml https://review.openstack.org/415495 | 06:19 |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Update the horizon port into all.yml https://review.openstack.org/415495 | 06:21 |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Update the horizon port into all.yml https://review.openstack.org/415495 | 06:24 |
| *** caowei has joined #openstack-kolla | 06:33 | |
| *** Pavo has quit IRC | 06:41 | |
| *** msimonin has joined #openstack-kolla | 06:41 | |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Move nova precheck into its own role https://review.openstack.org/415564 | 06:43 |
| *** msimonin1 has joined #openstack-kolla | 06:44 | |
| *** msimonin has quit IRC | 06:44 | |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Move nova precheck into its own role https://review.openstack.org/415564 | 06:44 |
| *** sdake has joined #openstack-kolla | 06:54 | |
| *** salv-orlando has joined #openstack-kolla | 06:55 | |
| *** sdake has quit IRC | 07:06 | |
| *** msimonin1 has quit IRC | 07:10 | |
| *** msimonin has joined #openstack-kolla | 07:10 | |
| *** jianyi has joined #openstack-kolla | 07:10 | |
| *** chas has joined #openstack-kolla | 07:12 | |
| *** salv-orl_ has joined #openstack-kolla | 07:13 | |
| *** msimonin has quit IRC | 07:14 | |
| *** salv-orlando has quit IRC | 07:16 | |
| *** chas has quit IRC | 07:17 | |
| *** sacharya has joined #openstack-kolla | 07:18 | |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Move neutron precheck into its own role https://review.openstack.org/415552 | 07:19 |
| *** matrohon has quit IRC | 07:19 | |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Move cinder precheck into its own role https://review.openstack.org/415492 | 07:22 |
| *** sacharya has quit IRC | 07:22 | |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Move cinder precheck into its own role https://review.openstack.org/415492 | 07:24 |
| *** chas has joined #openstack-kolla | 07:25 | |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Move gnocchi precheck into its own role https://review.openstack.org/415326 | 07:30 |
| *** salv-orl_ has quit IRC | 07:37 | |
| *** pcaruana has joined #openstack-kolla | 07:41 | |
| *** hawi has quit IRC | 07:47 | |
| *** hawi has joined #openstack-kolla | 07:47 | |
| *** shardy has joined #openstack-kolla | 07:58 | |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Move murano precheck into its own role https://review.openstack.org/415113 | 08:05 |
| *** saneax-_-|AFK is now known as saneax | 08:12 | |
| *** msimonin has joined #openstack-kolla | 08:26 | |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Move senlin precheck into its own role https://review.openstack.org/415486 | 08:37 |
| *** salv-orlando has joined #openstack-kolla | 08:38 | |
| *** salv-orlando has quit IRC | 08:42 | |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Move trove precheck into its own role https://review.openstack.org/416032 | 08:46 |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Move mistral precheck into its own role https://review.openstack.org/415515 | 08:51 |
| openstackgerrit | jianyi proposed openstack/kolla-ansible: Add zun ansible role https://review.openstack.org/416179 | 08:54 |
| *** david-lyle has joined #openstack-kolla | 08:57 | |
| *** david-lyle has quit IRC | 09:01 | |
| *** prameswar has quit IRC | 09:04 | |
| *** absubram has joined #openstack-kolla | 09:05 | |
| *** zhugaoxiao has joined #openstack-kolla | 09:17 | |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Move searchlight precheck into its own role https://review.openstack.org/415135 | 09:18 |
| *** sacharya has joined #openstack-kolla | 09:19 | |
| *** sacharya has quit IRC | 09:23 | |
| *** msimonin has quit IRC | 09:24 | |
| *** mnasiadka has joined #openstack-kolla | 09:28 | |
| *** msimonin has joined #openstack-kolla | 09:29 | |
| *** papacz has joined #openstack-kolla | 09:30 | |
| *** salv-orlando has joined #openstack-kolla | 09:35 | |
| *** Serlex has joined #openstack-kolla | 09:38 | |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Move magnum precheck into its own role https://review.openstack.org/415513 | 09:39 |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Update the place of Watcher HAProxy by alphabet https://review.openstack.org/416194 | 09:45 |
| *** pbourke has joined #openstack-kolla | 09:53 | |
| *** salv-orlando has quit IRC | 09:53 | |
| *** salv-orl_ has joined #openstack-kolla | 09:54 | |
| openstackgerrit | jianyi proposed openstack/kolla-ansible: Add zun ansible role https://review.openstack.org/416179 | 09:57 |
| *** david-lyle has joined #openstack-kolla | 09:58 | |
| *** tovin07 has quit IRC | 09:59 | |
| *** david-lyle has quit IRC | 10:02 | |
| satyar | Hi Jeffrey4l | 10:06 |
| satyar | Hi SamYaple | 10:06 |
| *** matrohon has joined #openstack-kolla | 10:07 | |
| *** prameswar has joined #openstack-kolla | 10:08 | |
| *** athomas has joined #openstack-kolla | 10:09 | |
| *** wagiel has joined #openstack-kolla | 10:11 | |
| *** gfidente has joined #openstack-kolla | 10:12 | |
| *** gfidente has joined #openstack-kolla | 10:12 | |
| openstackgerrit | jianyi proposed openstack/kolla-ansible: Add zun ansible role https://review.openstack.org/416179 | 10:14 |
| *** rmart04 has joined #openstack-kolla | 10:17 | |
| *** hieulq has quit IRC | 10:17 | |
| *** hieulq has joined #openstack-kolla | 10:18 | |
| Jeffrey4l | sup satyar | 10:26 |
| *** JoseMello has joined #openstack-kolla | 10:28 | |
| *** klindgren_ has joined #openstack-kolla | 10:29 | |
| *** gfidente^2nd has joined #openstack-kolla | 10:30 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/kolla-kubernetes: Updated from global requirements https://review.openstack.org/416199 | 10:31 |
| *** matrohon has quit IRC | 10:31 | |
| *** klindgren has quit IRC | 10:32 | |
| *** gfidente has quit IRC | 10:32 | |
| *** giulivo^3rd has joined #openstack-kolla | 10:33 | |
| openstackgerrit | jianyi proposed openstack/kolla-ansible: Add zun ansible role https://review.openstack.org/416179 | 10:33 |
| *** gfidente^2nd has quit IRC | 10:35 | |
| *** jianyi has quit IRC | 10:36 | |
| satyar | jeffrey4l | 10:38 |
| *** duonghq has quit IRC | 10:38 | |
| satyar | there is an issue with boot from volume | 10:38 |
| satyar | TypeError: Argument must be bytes or unicode, got 'NoneType' | 10:38 |
| Jeffrey4l | satyar, any more log? where did u get such a error? | 10:39 |
| satyar | http://paste.openstack.org/show/593772/ | 10:40 |
| Jeffrey4l | satyar, are u using ceph backedn? | 10:41 |
| satyar | yes | 10:41 |
| *** caowei has quit IRC | 10:41 | |
| Jeffrey4l | satyar, post the nova.conf file. | 10:42 |
| satyar | for nova and cinder both | 10:42 |
| Jeffrey4l | check whether [libvirt]rbd_secret_uuid has value. if yes, i suppose it should be a bug in nova/cinder ;( | 10:43 |
| *** giulivo^3rd has quit IRC | 10:43 | |
| satyar | rbd_secret_uuid = bbc5b4d5-6fca-407d-807d-06a4f4a7bccb | 10:45 |
| satyar | it have | 10:45 |
| *** giulivo^3rd has joined #openstack-kolla | 10:46 | |
| satyar | jeffrey4l is the secret uuid required in cinder also? | 10:47 |
| Jeffrey4l | satyar, no. cinder do not need that secret. | 10:49 |
| Jeffrey4l | satyar, does boot from image work? | 10:49 |
| satyar | yes | 10:50 |
| satyar | only boot from volume is failing | 10:50 |
| Jeffrey4l | which branch are u using? | 10:50 |
| satyar | the code from 05dec | 10:55 |
| Jeffrey4l | ok. i will test this in my env, which is the latest code. | 10:56 |
| Jeffrey4l | i guess it should be a bug in nova/cinder. | 10:57 |
| *** msimonin has quit IRC | 11:01 | |
| *** msimonin has joined #openstack-kolla | 11:03 | |
| *** msimonin has quit IRC | 11:03 | |
| *** msimonin has joined #openstack-kolla | 11:07 | |
| satyar | hmm | 11:07 |
| satyar | please check and let me know how it goes at your end | 11:07 |
| satyar | jeffrey4l | 11:08 |
| *** mliima has joined #openstack-kolla | 11:16 | |
| *** athomas has quit IRC | 11:18 | |
| *** sacharya has joined #openstack-kolla | 11:20 | |
| mliima | morning guys, i'm back | 11:21 |
| mliima | :) | 11:21 |
| *** sacharya has quit IRC | 11:24 | |
| *** athomas has joined #openstack-kolla | 11:26 | |
| Jeffrey4l | np | 11:33 |
| *** wagiel has quit IRC | 11:35 | |
| *** david-lyle has joined #openstack-kolla | 11:59 | |
| *** david-lyle has quit IRC | 12:04 | |
| *** rhallisey has joined #openstack-kolla | 12:07 | |
| *** NachoDuck has quit IRC | 12:19 | |
| *** serverascode has quit IRC | 12:19 | |
| *** kong_ has quit IRC | 12:19 | |
| *** portdirect has quit IRC | 12:19 | |
| *** alanmeadows has quit IRC | 12:19 | |
| *** portdirect has joined #openstack-kolla | 12:21 | |
| *** kong_ has joined #openstack-kolla | 12:21 | |
| *** NachoDuck has joined #openstack-kolla | 12:23 | |
| *** alanmeadows has joined #openstack-kolla | 12:23 | |
| *** serverascode has joined #openstack-kolla | 12:24 | |
| *** giulivo^3rd is now known as gfidente | 12:31 | |
| *** gfidente has quit IRC | 12:31 | |
| *** gfidente has joined #openstack-kolla | 12:31 | |
| *** salv-orl_ has quit IRC | 12:32 | |
| openstackgerrit | Merged openstack/kolla: Disable networking during bootstrapping https://review.openstack.org/415908 | 12:34 |
| *** matrohon has joined #openstack-kolla | 12:39 | |
| *** gfidente has quit IRC | 12:43 | |
| *** gfidente has joined #openstack-kolla | 12:45 | |
| *** msimonin has quit IRC | 12:48 | |
| *** saneax is now known as saneax-_-|AFK | 12:49 | |
| *** msimonin has joined #openstack-kolla | 12:50 | |
| openstackgerrit | Merged openstack/kolla: Remove unneeded oslo-rootwrap in swift-base https://review.openstack.org/413392 | 12:50 |
| *** msimonin has quit IRC | 12:50 | |
| openstackgerrit | Merged openstack/kolla: Clean up kolla-ansible related files from Kolla https://review.openstack.org/406429 | 12:50 |
| openstackgerrit | Merged openstack/kolla: Quickstart guide: disable libvirt apparmor https://review.openstack.org/415501 | 12:51 |
| *** srwilkers has joined #openstack-kolla | 12:51 | |
| srwilkers | morning | 12:51 |
| srwilkers | happy new year and such | 12:51 |
| *** msimonin has joined #openstack-kolla | 12:52 | |
| *** sdake has joined #openstack-kolla | 13:04 | |
| sbezverk | srwilkers: morning, same to you :) | 13:05 |
| sbezverk | srwilkers: when you have time please check https://review.openstack.org/#/c/414348/ | 13:06 |
| srwilkers | sbezverk, taking a gander now :) | 13:12 |
| *** mdnadeem has quit IRC | 13:18 | |
| *** sacharya has joined #openstack-kolla | 13:20 | |
| *** sacharya has quit IRC | 13:25 | |
| *** salv-orlando has joined #openstack-kolla | 13:28 | |
| *** dave-mccowan has joined #openstack-kolla | 13:30 | |
| *** lamt has joined #openstack-kolla | 13:31 | |
| portdirect | happy new year kollas :D | 13:31 |
| srwilkers | hey mate :) | 13:33 |
| srwilkers | happy new year | 13:33 |
| *** jheroux has joined #openstack-kolla | 13:34 | |
| sbezverk | portdirect: morning!! | 13:34 |
| portdirect | hey man :) just had a look through the lm work you've been doing looks good - but we should really be mounting cgroups read only if we need to read them in a pod i think? | 13:36 |
| *** caowei has joined #openstack-kolla | 13:36 | |
| sbezverk | portdirect: unfortunately iscsiadm needs it for not just read only | 13:36 |
| sbezverk | portdirect: it stores there targets information | 13:37 |
| *** salv-orlando has quit IRC | 13:37 | |
| sbezverk | behavior by design (not mine ;) ) | 13:37 |
| portdirect | ah ok - need to check whats going on with my stuff - but do we need to use the hosts cgroups in that case - I'm using the one in the docker container (I think, unless its mounting the hosts...) | 13:39 |
| *** apuimedo is now known as apuimedo|away | 13:39 | |
| portdirect | https://github.com/portdirect/harbor/blob/latest/kubernetes/templates/cinder/controllers.yaml#L603-L611 | 13:39 |
| *** salv-orlando has joined #openstack-kolla | 13:40 | |
| *** sdake has quit IRC | 13:41 | |
| sbezverk | portdirect: targetcli tool uses that | 13:43 |
| sbezverk | I think | 13:43 |
| sbezverk | portdirect: and we do mount storage/lvm volumes on the host | 13:44 |
| portdirect | ah - whoops see what your doing | 13:44 |
| *** gfidente has quit IRC | 13:45 | |
| *** yee37913 has joined #openstack-kolla | 13:46 | |
| *** yee379 has quit IRC | 13:46 | |
| *** l4yerffeJ__ has joined #openstack-kolla | 13:46 | |
| *** PyroMani_ has quit IRC | 13:47 | |
| portdirect | was thinking about the way my containers are built - where I use a fat container for this | 13:47 |
| portdirect | https://github.com/portdirect/harbor/blob/latest/docker/openstack/openstack-cinder/openstack-cinder-volume-iscsi/assets/etc/supervisord.conf | 13:47 |
| *** aimeeu has quit IRC | 13:47 | |
| *** gfidente has joined #openstack-kolla | 13:48 | |
| *** skramaja has quit IRC | 13:48 | |
| *** l4yerffeJ_ has quit IRC | 13:48 | |
| *** PyroMani has joined #openstack-kolla | 13:48 | |
| *** skramaja has joined #openstack-kolla | 13:48 | |
| *** Jeffrey4l has quit IRC | 13:49 | |
| *** l4yerffeJ__ has quit IRC | 13:54 | |
| *** l4yerffeJ__ has joined #openstack-kolla | 13:54 | |
| *** aimeeu has joined #openstack-kolla | 13:54 | |
| *** Jeffrey4l has joined #openstack-kolla | 13:59 | |
| *** zhugaoxiao has quit IRC | 14:00 | |
| *** david-lyle has joined #openstack-kolla | 14:01 | |
| *** zhugaoxiao has joined #openstack-kolla | 14:01 | |
| *** david-lyle has quit IRC | 14:05 | |
| openstackgerrit | Merged openstack/kolla-kubernetes: Adding iscsi support to the gate. https://review.openstack.org/414713 | 14:08 |
| openstackgerrit | Merged openstack/kolla-kubernetes: Adds cinder volume lvm microservice https://review.openstack.org/414348 | 14:08 |
| *** eaguilar has joined #openstack-kolla | 14:08 | |
| *** caowei has quit IRC | 14:14 | |
| *** fguillot has joined #openstack-kolla | 14:18 | |
| *** msimonin has quit IRC | 14:31 | |
| *** caowei has joined #openstack-kolla | 14:33 | |
| *** erhudy has joined #openstack-kolla | 14:34 | |
| *** msimonin has joined #openstack-kolla | 14:36 | |
| *** jtriley has joined #openstack-kolla | 14:37 | |
| *** eaguilar_ has joined #openstack-kolla | 14:37 | |
| *** eaguilar has quit IRC | 14:39 | |
| *** absubram has quit IRC | 14:43 | |
| *** salv-orlando has quit IRC | 14:45 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/kolla-kubernetes: Updated from global requirements https://review.openstack.org/416199 | 14:47 |
| *** huikang has joined #openstack-kolla | 14:49 | |
| *** huikang has quit IRC | 14:49 | |
| *** tis has joined #openstack-kolla | 14:50 | |
| *** huikang has joined #openstack-kolla | 14:50 | |
| *** mattmceuen has joined #openstack-kolla | 14:51 | |
| *** goldyfruit has joined #openstack-kolla | 14:55 | |
| *** v1k0d3n has joined #openstack-kolla | 14:59 | |
| *** david-lyle has joined #openstack-kolla | 15:01 | |
| *** Pavo has joined #openstack-kolla | 15:03 | |
| *** Serlex has quit IRC | 15:05 | |
| *** mgiles has joined #openstack-kolla | 15:05 | |
| *** rmart04_ has joined #openstack-kolla | 15:06 | |
| *** rmart04 has quit IRC | 15:06 | |
| *** rmart04_ is now known as rmart04 | 15:06 | |
| *** david-lyle has quit IRC | 15:06 | |
| *** jtriley has quit IRC | 15:10 | |
| erhudy | is there anyone in here with knowledge of the kolla test gate setup? i'm looking at a build failure on a PR of mine and i think it may be a combination of the gate configuration and an existing bug in kolla | 15:11 |
| *** rmart04 has quit IRC | 15:12 | |
| erhudy | it appears that in master if you supply an override for sources.list, it does not get added to the base container anyway because the dockerfile always copies in sources.list.ubuntu or sources.list.debian | 15:12 |
| erhudy | so i fixed that in my branch, but the gate supplies a custom sources.list, and as a consequence the build now tries to proceed with that sources.list and fails with an unauthenticated package error because it's now trying to draw ubuntu main packages from OSIC mirrors instead of canonical's | 15:13 |
| satyar | Hi jeffrey4l | 15:15 |
| satyar | seems nova and cinder changes lot many things | 15:15 |
| satyar | we need to fix those in kolla now | 15:15 |
| Jeffrey4l | hrm. any details? | 15:16 |
| *** rmart04 has joined #openstack-kolla | 15:17 | |
| *** eaguilar_ has quit IRC | 15:17 | |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Move mariadb precheck into its own role https://review.openstack.org/415587 | 15:18 |
| *** jtriley has joined #openstack-kolla | 15:19 | |
| *** sacharya has joined #openstack-kolla | 15:21 | |
| *** eaguilar has joined #openstack-kolla | 15:22 | |
| openstackgerrit | Li Yingjun proposed openstack/kolla-kubernetes: Fix different repo issue for documentation https://review.openstack.org/405054 | 15:23 |
| satyar | When making connections to Ceph-backed volumes via the Libvirt driver, the auth values (rbd_user, rbd_secret_uuid) are now pulled from the backing cinder.conf rather than nova.conf. The nova.conf values are only used if set and the cinder.conf values are not set, but this fallback support is considered accidental and will be removed in the Nova 16.0.0 Pike release. See the Ceph documentation for configuring Cinder for RBD au | 15:24 |
| satyar | before ocata nova would fall back to it's own local rbd auth config for those values if cinder wasn't configured | 15:24 |
| Jeffrey4l | where did u get this? | 15:24 |
| satyar | which points you here docs.ceph.com/docs/master/rbd/rbd-openstack/#configuring-cinder | 15:24 |
| satyar | fom nova irc | 15:25 |
| satyar | i saw you are also there | 15:25 |
| satyar | :) | 15:25 |
| Jeffrey4l | yep. reading. | 15:25 |
| *** sacharya has quit IRC | 15:26 | |
| satyar | :) | 15:26 |
| *** Pavo has quit IRC | 15:26 | |
| kfox1111 | sbezverk: I could have swarn someone tsid they tried it, but yeah, its not currently liking having a empty yaml. | 15:29 |
| *** tovin07 has joined #openstack-kolla | 15:29 | |
| kfox1111 | we could said | 15:29 |
| kfox1111 | one workaround for now would be to load in a 3rd party resource or configmap or something that does nothing. | 15:30 |
| *** sayantan_ has joined #openstack-kolla | 15:30 | |
| Jeffrey4l | satyar, but it should work, right? even though with some log warning. | 15:31 |
| satyar | not sure.. | 15:31 |
| Jeffrey4l | from the code, it says the fall back will be removed in P release. | 15:32 |
| satyar | as jaypipes asked to change rbd_secret_uuid to auth_secret_uuid | 15:32 |
| Jeffrey4l | have u tried this? | 15:33 |
| Jeffrey4l | i can not open docs.ceph.com site right now. | 15:33 |
| srwilkers | hey kfox1111. happy new year | 15:33 |
| sbezverk | kfox1111: yeah I was afraid it would come to a workaround :( | 15:34 |
| sbezverk | kfox1111: but it has to be done one way or another.. | 15:34 |
| kfox1111 | yeah. :/ | 15:36 |
| kfox1111 | sbezverk: would it be more clear if I renamed kolla_val_get_first to be kolla_val_get_raw? | 15:36 |
| SamYaple | o/ | 15:37 |
| *** absubram has joined #openstack-kolla | 15:37 | |
| sbezverk | kfox1111: not sure, from get_first name it is clear what it does, returns first found.. from get_raw I do not get this.. | 15:37 |
| kfox1111 | its only used if you want to ensure the datatype is retrieved unmollested. | 15:37 |
| kfox1111 | otherwise, the _str version is way easier to use. | 15:38 |
| Jeffrey4l | hey SamYaple, about ini parser. what's your approach? copy the code into kolla-ansible? | 15:38 |
| sbezverk | kfox1111: ok so there will be a pair: get_raw and get_str? | 15:39 |
| kfox1111 | yeah. | 15:39 |
| kfox1111 | that would be the alternate. | 15:40 |
| satyar | jeffrey4l yes tried with auth_secret_uuid still got the same error | 15:40 |
| sbezverk | doing the same operation but treating returned data a bit differently | 15:40 |
| SamYaple | Jeffrey4l: youre call. but you should be talking to other people about it. and perhaps make a note about it. referencing requirements.txt is a week defence | 15:40 |
| kfox1111 | sbezverk: yeah. thats really the only difference between the two. | 15:40 |
| Jeffrey4l | satyar, then i think they are different issue. | 15:40 |
| satyar | hmmm | 15:41 |
| sbezverk | kfox1111: in this case it makes sense I think | 15:41 |
| kfox1111 | one returns it as a template snippet (string), the other via dict passback. | 15:41 |
| kfox1111 | k. I'll change it in the next ps. | 15:41 |
| Jeffrey4l | SamYaple, all openstack project is referencing requirements.txt file now. | 15:42 |
| Jeffrey4l | satyar, trying to boot from volume. will tell u the reason later. | 15:42 |
| SamYaple | Jeffrey4l: no i mean for kolla-ansible. oslo_config was not there becuse it is needed in kolla-ansible. its there because its needed in kolla | 15:42 |
| satyar | sure jeffrey4l | 15:42 |
| Jeffrey4l | roger SamYaple | 15:43 |
| SamYaple | Jeffrey4l: and alot of people dont 'pip install kolla-ansible' | 15:43 |
| SamYaple | me included | 15:43 |
| Jeffrey4l | me included too. but i will run `pip install -r requirements.txt ` | 15:43 |
| satyar | we wrote script for that do that does for us :) | 15:44 |
| Jeffrey4l | satyar, sorry? | 15:44 |
| SamYaple | Jeffrey4l: thats not a choice either of us should make is my point. you need to discuss that in a meeting | 15:45 |
| satyar | oops | 15:45 |
| *** sdake has joined #openstack-kolla | 15:45 | |
| satyar | sorry jeffrey4l wrong window | 15:45 |
| kfox1111 | sbezverk: I just changed it to if disbaled make a tiny dummy configmap. seems happy enough with it. | 15:45 |
| Jeffrey4l | roger sam | 15:45 |
| kfox1111 | ugly but should work until helm gets dynamic support. | 15:45 |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Move tacker precheck into its own role https://review.openstack.org/416052 | 15:45 |
| *** salv-orlando has joined #openstack-kolla | 15:46 | |
| sbezverk | kfox1111: cool for now, but if each service when disabled starts creating these configmaps, we will be in trouble.. I think we need to bring it up to helm community.. | 15:46 |
| kfox1111 | +1. | 15:47 |
| kfox1111 | I think we will only need 3. mariadb/rabbit/memcached. | 15:47 |
| kfox1111 | but yeah, lets work with helm upstream to fix it asap. | 15:47 |
| *** inc0 has joined #openstack-kolla | 15:47 | |
| inc0 | good morning | 15:48 |
| sdake | sup folks | 15:48 |
| SamYaple | morning inc0 | 15:48 |
| inc0 | happy new year!:) | 15:48 |
| sdake | snakes on a plane | 15:48 |
| sdake | :) | 15:48 |
| SamYaple | o/ | 15:48 |
| kfox1111 | happy new year! :) | 15:48 |
| sbezverk | kfox1111: sounds good.. I think the workaround you suggested will allows up to move forward.. | 15:48 |
| SamYaple | \o | 15:48 |
| SamYaple | \o/ | 15:48 |
| sbezverk | inc0: morning | 15:48 |
| sdake | inc0 i wasn't sure if you added the ptg brianstorm or I did to the agenda | 15:49 |
| srwilkers | morning sdake inc0 | 15:49 |
| dmsimard | sdake: so I heard you a couple times mention you were moving away from tox to test container builds or something, can you expand on that ? | 15:49 |
| *** TxGirlGeek has joined #openstack-kolla | 15:49 | |
| sdake | ufnortunately I won't be able to make tomorrow's meeting | 15:49 |
| inc0 | sdake, I didn't | 15:49 |
| inc0 | but good thinking | 15:49 |
| *** rmart04 has quit IRC | 15:49 | |
| sdake | dmsimard Jeffrey4l can expnd | 15:49 |
| inc0 | I wanted to do it, but maybe a bit later | 15:49 |
| sdake | dmsimard he is doing the work | 15:49 |
| sdake | inc0 right - typically takes 2-3 sessions ;) | 15:49 |
| erhudy | if somebody could look at the reason behind the ubuntu build failure on https://review.openstack.org/#/c/414639 with me i would appreciate it, per my previous comments about there being a possible problem with those gates | 15:49 |
| dmsimard | Jeffrey4l: so how will containers be built/tested in the gate moving forward ? | 15:49 |
| inc0 | we still have almost 2 months till pth | 15:49 |
| sdake | and I figured after the new year is the time to have to biggest session since poeple haven't been working | 15:49 |
| sdake | :) | 15:50 |
| sdake | dmsimard basically tox -e build is gone - instead you ahve to use environment variables | 15:50 |
| sdake | dmsimard I personally think we want the tox mechanism | 15:50 |
| Jeffrey4l | dmsimard, we still depends on tox, but just move it from project-config to kolla's script. | 15:50 |
| sdake | dmsimard for others to do CI with | 15:50 |
| sdake | Jeffrey4l you removed tox -e build-centos-* right? | 15:50 |
| Jeffrey4l | sdake, no. i don't | 15:51 |
| Jeffrey4l | for kolla project. | 15:51 |
| sdake | Jeffrey4l ok - that was in the original patches | 15:51 |
| *** salv-orlando has quit IRC | 15:51 | |
| dmsimard | Jeffrey4l: okay is there any patches up for review (or a topic) I can follow to look at what's coming ? | 15:51 |
| sdake | maybe there was a followup | 15:51 |
| Jeffrey4l | i removed these from kolla-ansible. | 15:51 |
| sdake | let me look at code | 15:51 |
| *** adrian_otto has joined #openstack-kolla | 15:51 | |
| sdake | oh kolla-ansible | 15:51 |
| Jeffrey4l | yep. | 15:51 |
| sdake | Jeffrey4l i think it makes sense to keep these in kolla deliverable | 15:51 |
| dmsimard | Jeffrey4l: oh so you're not touching kolla, just kolla-ansible ? | 15:51 |
| sdake | agree or disagree? | 15:51 |
| Jeffrey4l | now kolla-ansible will re-user kolla's tox -e build- to build images. | 15:51 |
| Jeffrey4l | re-use* | 15:52 |
| dmsimard | Jeffrey4l: that makes sense | 15:52 |
| sdake | i get why you removed from kolla-ansible - because why test builds in a deployment system? :) | 15:52 |
| dmsimard | so nothing changes in kolla then ? | 15:52 |
| sdake | dmsimard i think i had the wrong deliverable mixed up with the change | 15:52 |
| Jeffrey4l | sdake yep. tox in kolla is useful. | 15:52 |
| sdake | dmsimard which is totally my fault for the miscommunication - apologies for that | 15:52 |
| inc0 | Jeffrey4l, on that note, what is the issue with registry? | 15:52 |
| Jeffrey4l | dmsimard, about build image, yes. | 15:52 |
| inc0 | I'll pick it up if you hit anythin | 15:52 |
| inc0 | g | 15:53 |
| dmsimard | sdake: no worries, I already leveraged tox from kolla, not from kolla-ansible | 15:53 |
| sdake | inc0 which registry issue? | 15:53 |
| Jeffrey4l | inc0, it should work. | 15:53 |
| inc0 | Jeffrey4l, registry in gates? | 15:53 |
| sdake | inc0 i've been on pto since 21st more or less | 15:53 |
| sdake | so i am super out of the loop | 15:53 |
| dmsimard | FWIW I'd like to bring some attention on the required ansible version that was commented in https://review.openstack.org/#/c/411930/ | 15:53 |
| Jeffrey4l | inc0, it is not merged. but it work as expected. | 15:53 |
| *** rmart04 has joined #openstack-kolla | 15:53 | |
| Jeffrey4l | inc0, check the test Patch | 15:54 |
| Jeffrey4l | https://review.openstack.org/413720https://review.openstack.org/413720 | 15:54 |
| dmsimard | My last comment: | 15:54 |
| *** hieulq_ has joined #openstack-kolla | 15:54 | |
| dmsimard | Can we bump that for Ocata ? Ansible 2.0.x is already fairly old, we're not in February yet and Ocata will be supported for a year. | 15:54 |
| dmsimard | The last release of Ansible in 2.0.x was released in April 2016 and there's been a lot of new releases in both 2.1 and 2.2 since then.It's safe to say that 2.0.x is all but unsupported/EOL. | 15:54 |
| *** kristian__ has joined #openstack-kolla | 15:54 | |
| sdake | dmsimard from my understanding x streams are supported | 15:54 |
| sdake | dmsimard let me look at my review | 15:54 |
| *** rmart04 has quit IRC | 15:54 | |
| *** strigazi_ is now known as strigazi_AFK | 15:54 | |
| sdake | rathe that review | 15:54 |
| dmsimard | I've introduced syntax that is >2.1 specific and it looks like we want to pin <2.1 | 15:55 |
| Jeffrey4l | i have to say: ansible made some big change in y stream, which is bad. ;( | 15:55 |
| dmsimard | which I don't particularly agree with | 15:55 |
| dmsimard | Jeffrey4l: yeah Ansible has moved very fast in the past year/year and a half | 15:56 |
| dmsimard | both a blessing and a curse | 15:56 |
| Jeffrey4l | inc0, the libvirt issue is very critical now. | 15:56 |
| inc0 | have you checked if this is qemu issue? | 15:56 |
| dmsimard | but there are very nice improvements, especially in Ansible 2.2 the performance improvements are *very* significant -- it's really fast now. | 15:56 |
| sdake | Jeffrey4l ya they broke their ABI | 15:56 |
| sdake | Jeffrey4l adding block for e.g. :) | 15:56 |
| Jeffrey4l | sdake, yep. i also made some workaround to make kolla_docker work on both ansible 2.0 and ansible 2.1 | 15:57 |
| dmsimard | sdake: not just that, I was personally bitten by other things.. regressions or changes in include behaviors (static includes) | 15:57 |
| sdake | dmsimard we want a wide variety of minimum pin levels since theoretically folks will install these from distro packaging | 15:57 |
| Jeffrey4l | dmsimard, i am agree to bump the ansible dependency. | 15:57 |
| sdake | to 2.1 or 2.2 Jeffrey4l ? | 15:58 |
| *** Serlex has joined #openstack-kolla | 15:58 | |
| Jeffrey4l | inc0, no. i got nothing expect libvirt process segment fault. | 15:58 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: kolla-common value get functions https://review.openstack.org/416069 | 15:58 |
| sdake | vpn no longer works on gogo internet | 15:58 |
| * sdake groans | 15:58 | |
| Jeffrey4l | sdake, hrm. no idea. | 15:58 |
| inc0 | ok it's not qemu issue... | 15:58 |
| inc0 | and it's just rdo package? | 15:59 |
| sdake | what version is in centos? | 15:59 |
| Jeffrey4l | ahh, inc0 got what do u means. | 15:59 |
| sdake | and what version is in ubuntu 16.04? | 15:59 |
| *** rmart04_ has joined #openstack-kolla | 15:59 | |
| *** athomas has quit IRC | 15:59 | |
| Jeffrey4l | sdake, 2.2 | 15:59 |
| inc0 | it still might be qemu | 15:59 |
| sdake | this should ideallly be our minimum pin versions - so people don't install ansible via pip | 15:59 |
| sdake | Jeffrey4l centos is 2.2 - what ver is xenail 16.04? | 15:59 |
| Jeffrey4l | checking~ | 15:59 |
| inc0 | Jeffrey4l, we use qemu or qemu+kvm in this gate? | 15:59 |
| inc0 | I know I had issues on rax vms with kvm | 16:00 |
| Jeffrey4l | we use qemu all the time. inc0 | 16:00 |
| dmsimard | inc0: the important piece is to pass the cpu_mode=None with the qemu hypervisor | 16:00 |
| sdake | inc0 perhaps xen just doesn't work well with qemu | 16:00 |
| dmsimard | because the default provided by nova when using qemu is cpu_mode=host-model | 16:00 |
| inc0 | sdake, xen doesn't affect qemu in any way | 16:00 |
| sdake | i know how its supposed to work | 16:00 |
| sdake | how it is working is a different story :) | 16:00 |
| dmsimard | iirc (although may be confusing with openstack-ansible) no values are provided by nova and you're probably therefore using host-model | 16:01 |
| dmsimard | s/by/to/ | 16:01 |
| *** sayantan_ has quit IRC | 16:01 | |
| Jeffrey4l | dmsimard, this issue only happen on rax-iad, which is xen. and we do not get the same error talked in the ml about cpu_mode. | 16:01 |
| inc0 | dmsimard, but it shouldn't end up in libvirt segfault, rather machine misbehaving | 16:01 |
| sdake | dmsimard our gate jobs use virt_type=qemu | 16:01 |
| sdake | which is probably host-model | 16:01 |
| dmsimard | sdake: yeah but cpu_mode should also be set to none | 16:01 |
| Jeffrey4l | sdake, re ansible version on xenial: 2.0.0.2 -2 ;( | 16:02 |
| *** sayantan_ has joined #openstack-kolla | 16:02 | |
| sdake | dmsimard didn't know that | 16:02 |
| inc0 | Jeffrey4l, and it's only rdo package? | 16:02 |
| Jeffrey4l | dmsimard, i will add cpu_mode in the gate and check what will get. | 16:02 |
| sdake | Jeffrey4l you could try setting cpu_mode to see if that fixes the problem with rax-iad | 16:02 |
| dmsimard | sdake: I would say to at least try that and see if that resolves it | 16:02 |
| Jeffrey4l | sdake, it come from epel repo. | 16:02 |
| Jeffrey4l | dmsimard, yep. | 16:03 |
| *** zhugaoxiao has quit IRC | 16:03 | |
| sdake | Jeffrey4l don't know what to do with that :) | 16:03 |
| sdake | Jeffrey4l if we do make the minimum pin 2.1 (where block is introduced) folks will have to pip install on ubuntu | 16:04 |
| sdake | which is not something i'd necessarily do in a production environment | 16:04 |
| sdake | perhaps inc0 has ideas - he uses ubuntu all the time | 16:04 |
| dmsimard | ¯\_(ツ)_/¯ | 16:04 |
| dmsimard | how are you expecting people to install kolla ? from pip ? :P | 16:04 |
| sdake | dmsimard from distro packaging | 16:04 |
| inc0 | and I used rax public cloud with no issues apart from qemu instead of kvm | 16:04 |
| dmsimard | ubuntu has kolla packaging ? | 16:04 |
| Jeffrey4l | sdake, there is ansible official ppa for ubuntu now, which provide ansible 2.2 | 16:04 |
| Jeffrey4l | https://launchpad.net/~ansible/+archive/ubuntu/ansible | 16:04 |
| inc0 | but it *used to* work for us too | 16:04 |
| sdake | oh misread sentence | 16:04 |
| *** zhugaoxiao has joined #openstack-kolla | 16:04 | |
| sdake | dmsimard i thought you meant ansible | 16:05 |
| sdake | dmsimard atm we require people to pip install kolla because there is no distro packaging | 16:05 |
| Jeffrey4l | inc0, this issue happen since centos 7.3 release ,which upgrade libvirt to 2.0.0. | 16:05 |
| sdake | dmsimard if there was distro packaging, I'd prefer people isntall from that | 16:05 |
| sdake | dmsimard mixing distro packaging with pip in the same workflow ends up with broken ansible for example | 16:05 |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Move tacker precheck into its own role https://review.openstack.org/416052 | 16:06 |
| sdake | dmsimard ie. pip install ansible - yum install ansible | 16:06 |
| sdake | bust | 16:06 |
| sdake | or maybe the order is different | 16:06 |
| dmsimard | yeah | 16:06 |
| sdake | but you get the idea | 16:06 |
| kfox1111 | heading into work. bbiab. | 16:06 |
| sdake | kfox1111 email soon plz :) | 16:06 |
| sbezverk | kfox1111: ttyl | 16:06 |
| sdake | pbourke morning dude | 16:06 |
| pbourke | sdake: o/ morning | 16:07 |
| sdake | pbourke did you see my JJ Abrams photo of my car ;) | 16:07 |
| sdake | well I know you saw it | 16:07 |
| pbourke | sdake: yeah pretty sweet :) | 16:07 |
| inc0 | sdake, show me | 16:07 |
| inc0 | ;) | 16:07 |
| sdake | but did you see the light as subject? :) | 16:07 |
| *** jemcevoy has joined #openstack-kolla | 16:07 | |
| sdake | inc0 its on twitter somewheres | 16:07 |
| sdake | moment | 16:07 |
| Jeffrey4l | qq: dmsimard add `cpu_mode=none` into libvirt group, right? | 16:08 |
| dmsimard | Jeffrey4l: let me double check | 16:08 |
| sdake | inc0 https://twitter.com/steven_dake/status/815964928502743040 | 16:09 |
| sdake | phone photos bit washed out | 16:09 |
| inc0 | nice one | 16:09 |
| sdake | shame I didn't have my wife's fuji camera with me | 16:09 |
| sdake | thanks - i was actually more impressed with the photo then the car :) | 16:09 |
| dmsimard | Jeffrey4l: yes, in libvirt group. | 16:10 |
| Jeffrey4l | roger. thanks. | 16:10 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: TEST MASTER BRANCH https://review.openstack.org/398095 | 16:10 |
| openstackgerrit | Merged openstack/kolla: Not to pull the base images when building. https://review.openstack.org/408940 | 16:11 |
| inc0 | I need to take a shot of my gokart | 16:11 |
| sdake | inc0 lol :) | 16:11 |
| sdake | I took a college photography class | 16:12 |
| sdake | there was one young lady in there - her work was absolutely fantastic | 16:12 |
| inc0 | I have gokart thats allowed to be on road;) | 16:12 |
| sdake | the teacher told me the school was a feeder school and she would end up going to asu for professional photography | 16:12 |
| inc0 | http://www.diseno-art.com/news_content/wp-content/uploads/2014/09/Mazda-MX-5-ND-front-and-rear.jpg | 16:12 |
| sdake | I was there just to learn how to photograph stuff :) | 16:13 |
| sdake | when I had my headshot taken (literally and figuratively!) the dude had all kinds of interesting tricks | 16:13 |
| sdake | inc0 i've seen your car iirc | 16:13 |
| *** awiddersheim has quit IRC | 16:13 | |
| sdake | inc0 i think I stored my jacket in it in austin | 16:14 |
| sdake | or something - maybe a bag | 16:14 |
| inc0 | a yeah | 16:14 |
| sdake | don't recall exactly | 16:14 |
| inc0 | jacket | 16:14 |
| sdake | yup - I do that often jacket or backpack in trunk :) | 16:14 |
| dmsimard | Is there a prequel to http://docs.openstack.org/developer/kolla/image-building.html ? How do I know what to install/setup prior to building images ? ex: installing docker and stuff. | 16:14 |
| inc0 | waiting for spring to get my roof down | 16:14 |
| sdake | inc0 - i'll be indisposed 4th-6th | 16:14 |
| inc0 | thanks for heads up | 16:15 |
| sdake | yup | 16:15 |
| inc0 | dmsimard, quickstart | 16:15 |
| sdake | inc0 also any chance you can get that etherpad updated with notes + video of meeting? | 16:15 |
| sdake | inc0 since its the new year, people may start looking at it | 16:15 |
| dmsimard | inc0: so basically kolla-ansible bootstrap-server ? | 16:15 |
| inc0 | image building is supposed to go after quickstart | 16:15 |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Move grafana precheck into its own role https://review.openstack.org/416055 | 16:15 |
| *** apuimedo|away is now known as apuimedo | 16:16 | |
| inc0 | dmsimard, yeah, and pip install kolla | 16:16 |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Move grafana precheck into its own role https://review.openstack.org/416055 | 16:16 |
| inc0 | sdake, helm guy is silent | 16:16 |
| inc0 | I'll ping him on email | 16:16 |
| sdake | ya they are on pto until the 4th i think I saw | 16:16 |
| inc0 | probably just went for vacation | 16:16 |
| sdake | in their release notes | 16:16 |
| dmsimard | inc0: the reason I ask is because it looks like the kolla gate jobs seem to do so many hacks that are not covered in quickstart/bootstrap-server | 16:16 |
| *** athomas has joined #openstack-kolla | 16:16 | |
| sdake | dmsimard i wouldn't really use the gate jobs as a reference :) | 16:17 |
| *** awiddersheim has joined #openstack-kolla | 16:17 | |
| sdake | dmsimard but i guess your making your own gating jobs so maybe that makes sense | 16:17 |
| inc0 | dmsimard, gates are gates, we need to take into account all the stuff to make it work on all nodepools | 16:17 |
| *** shananigans has joined #openstack-kolla | 16:17 | |
| inc0 | most of these hacks are specific to some nodepool issue | 16:17 |
| sdake | dmsimard all that hack stuff we do in the gate jobs = the work hacks is not strong enough :) | 16:17 |
| sdake | work/word | 16:18 |
| inc0 | I'd suggest starting fresh and adding hacks as needed:) | 16:18 |
| sdake | inc0++ | 16:18 |
| dmsimard | Well, yes and no, I'd expect the quickstart to be used in the gate and not just as a means to test the actual quickstart | 16:18 |
| inc0 | maybe some of these hacks aren't needed any more | 16:18 |
| dmsimard | bootstrap-server isn't tested anywhere right now | 16:18 |
| sdake | dmsimard ya i have wanted that since inception of kolla | 16:18 |
| dmsimard | but let's pretend that's a kolla-ansible specific issue | 16:18 |
| sdake | dmsimard the qsg tested in the gate | 16:18 |
| inc0 | bootstrap servers was designed to provision compute/controller nodes | 16:19 |
| inc0 | not aio/deploy node | 16:19 |
| sdake | dmsimard re bootstrap-server in the gate, I think that is something that could be tackled - if someone knew how to work on the gate and was so inclined :) | 16:19 |
| sdake | inc0 our gate jobs do multinode too :) | 16:19 |
| inc0 | but yeah, I agree we can leverage it | 16:19 |
| sdake | inc0 we actually use ansible in the gate jobs to run against the nodepool inventory that we are passed | 16:19 |
| dmsimard | inc0: a deploy node for you is where the images are built? (making use we use the same vocabulary) | 16:19 |
| inc0 | deploying gates with it would be much clearer | 16:19 |
| sdake | and the ansible we use in our gate jobs is sort of like bootstrap-server | 16:20 |
| inc0 | dmsimard, normally thats the case but there is no specific reason for it | 16:20 |
| inc0 | it can be any node | 16:20 |
| *** lamt has quit IRC | 16:20 | |
| dmsimard | right | 16:20 |
| inc0 | well, reason being that people don't re-ssh into other nodes;) | 16:20 |
| sdake | dmsimard typically the deployment host is different node from the other nodes | 16:20 |
| sdake | dmsimard but it can also be on the controller, compute, etc | 16:21 |
| dmsimard | anyway, I sort of had this discussion with SamYaple before the holidays but it didn't seem like we agreed and I didn't fight it as I was going on PTO :p | 16:21 |
| sdake | dmsimard which discussion - what specific request do you have? | 16:21 |
| *** lamt has joined #openstack-kolla | 16:21 | |
| dmsimard | I was asking about how the gate is setup in general, why not use bootstrap-server, why use a mix of bash and ansible, etc. | 16:22 |
| dmsimard | let me try and find it | 16:22 |
| Jeffrey4l | dmsimard, it is historical issue. | 16:23 |
| *** rmart04_ has quit IRC | 16:23 | |
| Jeffrey4l | at the beginning, only bash script. later, we want to move to ansible playbook. | 16:23 |
| Jeffrey4l | and then we added the bootstrap-server feature. | 16:23 |
| dmsimard | http://eavesdrop.openstack.org/irclogs/%23openstack-kolla/%23openstack-kolla.2016-12-21.log.html#t2016-12-21T21:16:38 | 16:23 |
| Jeffrey4l | optimal, we'd better to mvoe to bootstrap-server. | 16:24 |
| dmsimard | Jeffrey4l: then again you're leveraging kolla-ansible inside the kolla gate ? | 16:24 |
| dmsimard | I mean, why not use the bootstrap-server equivalent in kubernetes or now salt | 16:24 |
| Jeffrey4l | dmsimard, yep. | 16:24 |
| dmsimard | (playing devil's advocate) | 16:24 |
| Jeffrey4l | sorry? salt? | 16:25 |
| dmsimard | There's a new kolla-salt thing that's been in the works apparently | 16:25 |
| dmsimard | saw it on openstack-dev | 16:25 |
| Jeffrey4l | dmsimard, yep. it it. | 16:25 |
| Jeffrey4l | it is* | 16:25 |
| Jeffrey4l | afaik, bootstrap-server only exist in kolla-ansible. | 16:26 |
| *** tis has quit IRC | 16:26 | |
| dmsimard | but there can/could/should be something equivalent in other deployment frameworks | 16:26 |
| Jeffrey4l | possible. i think this kinda of feature will be added in other kolla-* project in the future. | 16:27 |
| *** adrian_otto has quit IRC | 16:35 | |
| *** adrian_otto has joined #openstack-kolla | 16:36 | |
| kfox1111 | back | 16:36 |
| kfox1111 | sdake: email? | 16:37 |
| sdake | kfox1111 the one about the api and whats needed | 16:37 |
| kfox1111 | oh. | 16:37 |
| kfox1111 | yeah. I'll try and work on that today. | 16:37 |
| kfox1111 | got about a week of email to catch up on this morning. :/ | 16:38 |
| kfox1111 | and 2 and a half weeks of being out of office. :/ | 16:38 |
| openstackgerrit | Paul Bourke (pbourke) proposed openstack/kolla: Remove source install type for novnc / spice https://review.openstack.org/408599 | 16:40 |
| *** klindgren_ has quit IRC | 16:40 | |
| *** sayantan_ has quit IRC | 16:40 | |
| kfox1111 | sdake: sorry. can you please refresh my memory a bit. what did you want in the email again? is this about what we would like out of helm or what the status of where we're at now, or where we want to be in a week? | 16:42 |
| *** tovin07 has quit IRC | 16:42 | |
| kfox1111 | too many things going on right now to keep things streight. :/ | 16:42 |
| sdake | kfox1111 ok - here is what Id ideally like | 16:43 |
| sdake | kfox1111 I'd like to understand where the API model will be when we release 0.4.0 next wed | 16:44 |
| openstackgerrit | Merged openstack/kolla-ansible: Add Constraints support https://review.openstack.org/413291 | 16:44 |
| sdake | in detail | 16:44 |
| Jeffrey4l | satyar, around? | 16:44 |
| openstackgerrit | Merged openstack/kolla-ansible: Enable horizon deployment in gate https://review.openstack.org/410057 | 16:44 |
| sdake | kfox1111 i think that API model is a crucial thing we have going for us - and I think we want to all understand it in detail | 16:44 |
| sdake | the fact that I've been on pto for 2 weeks + out of the office starting tomorrow isn't helping :) | 16:45 |
| sdake | kfox1111 normally I'd probably just understand this via reviewing, however I mostly unplugged over the holiday shutdown and did other things with family/etc | 16:45 |
| *** adrian_otto has quit IRC | 16:45 | |
| *** klindgren has joined #openstack-kolla | 16:46 | |
| openstackgerrit | Merged openstack/kolla-ansible: Modify the openstack_release in globals.yml https://review.openstack.org/414478 | 16:46 |
| kfox1111 | sdake: problem is, its kind of in the middle of a refactor right now. :/ and will take most of the week to nail down. | 16:47 |
| sdake | kfox1111 ok - well can you just give me your end state in the email | 16:47 |
| sdake | or do you not know it? | 16:47 |
| kfox1111 | the helm set function they gave us for christsmas was awesome, but made a bunch of work. | 16:47 |
| sdake | like what features of helm do we use | 16:47 |
| kfox1111 | k. could give you the desired end state. | 16:47 |
| sdake | as it relates to the API | 16:47 |
| sdake | this is crucial information i'm after | 16:47 |
| sdake | kfox1111 should go to mailing list | 16:48 |
| kfox1111 | k | 16:48 |
| sdake | kfox1111 if you could do today i'd appreciate it | 16:49 |
| kfox1111 | k. I'll try and prioritize it. | 16:49 |
| sdake | as TLDR as possible :) | 16:50 |
| sdake | but if the TLDRness means you won't do it | 16:50 |
| sdake | then trim back on the TLDR :) | 16:50 |
| kfox1111 | nan, the tldr's much easier for me to try and write then a long thing. | 16:50 |
| kfox1111 | nah | 16:50 |
| sdake | kfox1111 i think your irssi is busted again :) | 16:51 |
| kfox1111 | no, it just has too many tabs to see when people ping me sometimes. :/ | 16:51 |
| kfox1111 | especially when users renamethemselves a lot. :/ | 16:51 |
| Jeffrey4l | satyar, got this error when booting from volume. http://paste.openstack.org/show/593799/ will check it in detail tomorrow. | 16:51 |
| sdake | ok peeps gott ajet - airplane landing | 16:53 |
| satyar | Hi jeffrey4l yes | 16:55 |
| satyar | in a meeting now... | 16:55 |
| satyar | sorry for late response | 16:55 |
| Jeffrey4l | satyar np. i am going to sleep. will check this tomorrow ;) | 16:55 |
| *** sdake has quit IRC | 16:57 | |
| *** vhosakot has joined #openstack-kolla | 17:00 | |
| portdirect | any chance a core could give this a look over? https://review.openstack.org/#/c/412205/ | 17:01 |
| portdirect | (kolla core that is) | 17:01 |
| *** david-lyle has joined #openstack-kolla | 17:03 | |
| *** sayantan_ has joined #openstack-kolla | 17:05 | |
| *** inc0 has quit IRC | 17:06 | |
| *** adrian_otto has joined #openstack-kolla | 17:06 | |
| *** matrohon has quit IRC | 17:07 | |
| *** david-lyle has quit IRC | 17:08 | |
| *** chas has quit IRC | 17:09 | |
| *** chas has joined #openstack-kolla | 17:09 | |
| *** chas has quit IRC | 17:14 | |
| *** matrohon has joined #openstack-kolla | 17:15 | |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: kolla-common value get functions https://review.openstack.org/416069 | 17:24 |
| *** mnasiadka has quit IRC | 17:25 | |
| *** l4yerffeJ has joined #openstack-kolla | 17:25 | |
| *** adrian_otto has quit IRC | 17:33 | |
| *** inc0 has joined #openstack-kolla | 17:35 | |
| *** adrian_otto has joined #openstack-kolla | 17:35 | |
| *** msimonin has quit IRC | 17:35 | |
| openstackgerrit | Merged openstack/kolla: Add zun support in README.rst https://review.openstack.org/415628 | 17:36 |
| *** apuimedo is now known as apuimed|away | 17:36 | |
| inc0 | portdirect, out of curiosity, how are we going to deploy k8s images? | 17:36 |
| inc0 | https://review.openstack.org/#/c/412205/ <- talking about this one | 17:37 |
| portdirect | inc0, that's an exercise for the reader :) (serious answer to follow...) | 17:37 |
| portdirect | the images there are drop in compatible with the gcr.io images - but will need retagged to use with k8s deployment tooling | 17:38 |
| portdirect | taking kubeadm as an example - the current method is to pull the image to the hosts required and then retag to match the gcr.io images before running kubeadm | 17:39 |
| portdirect | ^^ horrible - but is the currently reommended way for people beind corb firewalls or in china to use kubeadm | 17:40 |
| inc0 | kinda horrible, yeah | 17:40 |
| portdirect | with other tools eg kargo etc then its a bit simpler as i bleive you can speficy the image name to use - but operaors will still need to retag the version to match the k8s version | 17:40 |
| inc0 | on the other hand, keeping images in sync with gcr.io is not trivial | 17:41 |
| portdirect | and some 3rd party tools eg Project Atomic's commissaire should be able to use them directly (eventually) | 17:42 |
| portdirect | I wanted to get the into the repo so that we could have a source of truth for our gate primarity - and have a set of known working images to use for our purposes | 17:43 |
| inc0 | yeah makes sense | 17:43 |
| *** kristian__ has quit IRC | 17:43 | |
| inc0 | for gates at least - useful | 17:44 |
| inc0 | especially if we crack the registry issue | 17:44 |
| inc0 | which we're pretty close to do | 17:44 |
| portdirect | as for ceph support we are currently relying att-comdev's image - thanks srwilkers/v1k0d3n ! :) | 17:44 |
| inc0 | att-comdev has different ceph images than kolla? | 17:44 |
| portdirect | no - but we need to use thier controller manager image if we want native pvc support | 17:45 |
| portdirect | https://github.com/att-comdev/dockerfiles/blob/master/kube-controller-manager/Dockerfile | 17:46 |
| *** shardy has quit IRC | 17:46 | |
| portdirect | for actual ceph support, both att-comdev and I are using the images from docker-ceph: https://github.com/portdirect/ceph-helm | 17:47 |
| portdirect | ^^ chart for example - theirs is different, though I think about to be updated to use a similar deployment for mons/config | 17:49 |
| portdirect | which reminds me, leseb you about? | 17:49 |
| *** kristian__ has joined #openstack-kolla | 17:55 | |
| *** pcaruana has quit IRC | 17:58 | |
| *** kristian__ has quit IRC | 17:59 | |
| sbezverk | portdirect: Are you planning to use native pvc? | 18:00 |
| sbezverk | I do not remember any discussion here about it, other than "it exists" :-) | 18:00 |
| inc0 | hmm | 18:02 |
| inc0 | so I set up helm stable/mysql | 18:02 |
| inc0 | and deleted stable/mysql | 18:02 |
| inc0 | and now it's stuck pending:( | 18:02 |
| inc0 | ahh here I was, hoping that stuff will work;) | 18:02 |
| v1k0d3n | inc0: did you delete --purge? | 18:02 |
| sbezverk | inc0: mariadb pod is in pending? | 18:02 |
| inc0 | v1k0d3n, I did not, let's see | 18:03 |
| v1k0d3n | that will purge everything, including jobs. | 18:03 |
| sbezverk | v1k0d3n: doubt it will help, as --purge just purges the name from helm | 18:03 |
| inc0 | out of curiosity, is stable/mysql really stable? | 18:03 |
| *** david-lyle has joined #openstack-kolla | 18:03 | |
| inc0 | sbezverk, nope, it purged it all right | 18:04 |
| sbezverk | inc0: you should use mariadb from kolla-kubernetes repo | 18:04 |
| inc0 | sbezverk, my question is | 18:04 |
| sbezverk | it has been proven stable bu many gate runs | 18:04 |
| inc0 | if we have good mariadb | 18:04 |
| v1k0d3n | if it's not sbezverk it may be a bug. it's supposed to. | 18:04 |
| *** eaguilar_ has joined #openstack-kolla | 18:04 | |
| inc0 | in helm | 18:04 |
| inc0 | maybe we should consider using it | 18:04 |
| portdirect | sbezverk: I mush confess I thought that was the intention (native PVC) | 18:04 |
| portdirect | *must | 18:04 |
| *** eaguilar has quit IRC | 18:04 | |
| inc0 | sbezverk, our mariadb is still not that great in terms of HA | 18:04 |
| sbezverk | portdirect: possibly, it is just a surprise for me | 18:05 |
| inc0 | I'd rather have someone to blame for not great HA;) | 18:05 |
| inc0 | narrowing scope of kolla would be good thing tbh | 18:05 |
| sbezverk | inc0: I see your point, I guess we are not there yet | 18:05 |
| inc0 | helm might help | 18:05 |
| v1k0d3n | so inc0... | 18:05 |
| portdirect | sbezverk: though there has not been any objection to date whenever I've mentioned it - should we put it on the ajenda for tomorrow to clarify? | 18:05 |
| inc0 | so v1k0d3n... | 18:06 |
| v1k0d3n | we actually recommend using upstream mariabd if the user wants to do so. | 18:06 |
| v1k0d3n | :) | 18:06 |
| inc0 | well it does make sense ofc | 18:06 |
| sbezverk | portdirec: sure, if it is a must then why we do not create templates anywhere for it?? | 18:06 |
| inc0 | and I'd say we should start thinking in this direction | 18:06 |
| inc0 | most of crap we deal with is with mariadb/rabbimq | 18:06 |
| v1k0d3n | at the end of the day, mariadb is mariadb (as long as it's a galera quorum). in fact separating out client is really nice. | 18:06 |
| inc0 | and if we install stable/mariadb with helm, no reason why we need to support it | 18:06 |
| sbezverk | v1k0d3n: curious how can you do that? or your gate tests upstream? | 18:07 |
| v1k0d3n | most of the issues you're dealing with is mariadb and not PVC related? | 18:07 |
| portdirect | sbezverk: as it only came into being with 1.5 and we were targetting 1.4 when the services that consumed pvc were created | 18:07 |
| inc0 | sbezverk, helm install stable/mariadb will work in gates | 18:07 |
| portdirect | sbezverk: so simply just becasue we havent created them yet i think :) | 18:07 |
| awiddersheim | I am having a weird issue with overriding the api_interface for a specific node | 18:08 |
| v1k0d3n | ^ yup | 18:08 |
| inc0 | (now I only need to figure out how to use stable/mariadb with ha and stuff) | 18:08 |
| v1k0d3n | that's all coming though. at least on our side, our jenkins guy has been on nearly a month long vacation. | 18:08 |
| *** david-lyle has quit IRC | 18:08 | |
| v1k0d3n | in fact inc0... | 18:08 |
| awiddersheim | the docs and examples say you can just do "somehost api_interface=eth0" but what I have configured in my globals.yml file for "network_interface" seems to be taking precedence | 18:08 |
| sbezverk | inc0: our mariadb is not just mariadb, it has bunch of things making it eatble by openstack services | 18:08 |
| inc0 | sbezverk, like what? | 18:09 |
| v1k0d3n | just using mysql works perfectly for minikube openstack development :) | 18:09 |
| * srwilkers catches up | 18:09 | |
| v1k0d3n | we started doing it that way...makes things much, much easier! | 18:09 |
| inc0 | v1k0d3n, I'm thinking prod here | 18:09 |
| sbezverk | inc0: svc/pvc/bootstrap | 18:09 |
| portdirect | v1k0d3n: I dont think that upstream mariadb will work with kolla-k8s | 18:09 |
| sbezverk | each is a microservice under mariadb umbrella | 18:09 |
| portdirect | v1k0d3n: though I may be wrong | 18:09 |
| inc0 | sbezverk, boostrap can ofc stay in kolla, pvc, svc should be handled by stable/mariadb maintainers | 18:10 |
| sbezverk | you can pick and choose what needs to be done | 18:10 |
| v1k0d3n | yeah, not sure about kolla-kube. been out of it for a while. | 18:10 |
| v1k0d3n | that's what i've been testing on our side, and works. | 18:10 |
| portdirect | v1k0d3n: obviously a database is a database - but the config will need some twaeking | 18:10 |
| inc0 | yeah I don't think there is any reason why it wouldnt really | 18:10 |
| v1k0d3n | make sense. gotcha. | 18:10 |
| inc0 | I expect helm to allow us to configure it right? | 18:11 |
| inc0 | if it doesn't, helm wouldn't be super useful | 18:11 |
| *** athomas has quit IRC | 18:11 | |
| *** Pavo has joined #openstack-kolla | 18:11 | |
| v1k0d3n | well...maybe. depends. devil is always in the details. | 18:11 |
| sbezverk | inc0: hmm what about integration with API? | 18:11 |
| inc0 | awiddersheim, api_interface defaults to value from network_interface | 18:11 |
| inc0 | sbezverk, what do you mean? db is db | 18:11 |
| inc0 | all openstack services needs is connections string | 18:12 |
| sbezverk | when you said about pvc/pv/svc | 18:12 |
| sbezverk | these microservices gets info from API | 18:12 |
| inc0 | yeah, we need to figure out way to create connection string | 18:12 |
| inc0 | I'm not saying it will work right away | 18:13 |
| inc0 | but I'm saying with a bit of tweaking we can turn kolla-k8s to use helm/mariadb | 18:13 |
| SamYaple | can I get a review on my zookeeper role? its been sitting for 2 weeks https://review.openstack.org/#/c/412423/ | 18:13 |
| inc0 | and stop carying about supporting mariadb, which would be nice | 18:13 |
| inc0 | SamYaple, that's what you get if you propose big hard changes in holiday seaon;) | 18:14 |
| SamYaple | its neither big nor hard! | 18:14 |
| sbezverk | inc0: in future possibly yes, but now stop everything until this tweaking is done. I am not sure if we have this luxury | 18:14 |
| inc0 | sbezverk, oh I'm not saying we should rewrite it now! | 18:14 |
| inc0 | I'm saying, in Pike we can think of rewriting kolla-k8s in a way to support stuff like that | 18:15 |
| inc0 | would make our jobs sooo much easier | 18:15 |
| sbezverk | inc0: ok :) I am still recovering from jumping from jinja to helm and now I thought you want another change in direction | 18:15 |
| inc0 | nah, in ocata we stick to what we have for good or worse | 18:15 |
| sbezverk | inc0: sounds good :) | 18:16 |
| inc0 | for Pike I wanted to take long critical look at it and use all we learned to rearchitect it | 18:16 |
| inc0 | we'll have longer release and lots of knowledge | 18:16 |
| inc0 | I'm just saying that I agree with v1k0d3n and we should limit scope of kolla as much as we can | 18:17 |
| srwilkers | inc0, i share that sentiment | 18:17 |
| inc0 | leave stuff like ceph, mariadb or rabbimq to their communities, we'll end up having better quality | 18:17 |
| v1k0d3n | +1 inc0 | 18:18 |
| inc0 | we kinda have this approach of "batteries included" in kolla-ansible but that's because we didn't have tools that allows us to use different people's expertise to deploy our infra stuff (db and such) | 18:18 |
| inc0 | with helm, we kinda do | 18:18 |
| sbezverk | inc0: as long as tweaking to make these consumable by kolla-kubernetes != amount of work to do it within kolla | 18:18 |
| v1k0d3n | i know a lot of work is done around the API for microservices too...took a lot of work, and is required with Docker...but question...is there a more simple solution with a helm version? | 18:18 |
| v1k0d3n | for instance, is it required? | 18:19 |
| inc0 | sbezverk, trust me, it will be *way* less work than having to support these services in the long run | 18:19 |
| v1k0d3n | just discussing only. i really want to avoid controversy. but the more complex, and more custom, i fear some will shy away (large corps). | 18:19 |
| inc0 | v1k0d3n, again, let's discuss it for pike | 18:20 |
| v1k0d3n | sure, of course. | 18:20 |
| inc0 | as I said on helm meeting | 18:20 |
| v1k0d3n | tabled for now :) | 18:20 |
| SamYaple | yea the "batteries included" was more about making it easier, at least in the beginning | 18:20 |
| sbezverk | inc0 +1 this type of discussions now just unfocusing us | 18:20 |
| inc0 | ideally at the end of the day I'd like kolla-k8s community to support helm native stable charts for openstack | 18:20 |
| inc0 | SamYaple, agree, what I'm saying, not needed in k8s case really | 18:21 |
| inc0 | so we can change this approach I feel | 18:21 |
| SamYaple | i think i would agree with you there | 18:21 |
| *** kristia__ has joined #openstack-kolla | 18:21 | |
| v1k0d3n | sbezverk: it shouldn't unfocus. i can table it. these are good questions for the community to consider and discuss. | 18:21 |
| SamYaple | in kolla-ansible we deploy on an unknown host, k8s is hardly "unknown" in my opinion | 18:21 |
| inc0 | SamYaple, also in k8s we use package manager | 18:21 |
| inc0 | so we can just type helm install stable/mariad | 18:22 |
| inc0 | and that's it | 18:22 |
| v1k0d3n | +1 SamYaple | 18:22 |
| sbezverk | v1k0d3n: it should not but it does unfortunately.. | 18:22 |
| inc0 | sbezverk, it's natural that it does | 18:22 |
| inc0 | that's why I'd suggest starting drafting spec | 18:22 |
| inc0 | for Pike | 18:22 |
| inc0 | not of what we have, but what we want to focus on after 1.0 | 18:23 |
| inc0 | and *not* bloking our work on that spec | 18:23 |
| v1k0d3n | +1 inc0 that keeps things right on track, and plans in the future as we tried to do in Barcelona | 18:23 |
| inc0 | like, work on it in pararell, then in Atlanta we'll just share beer(s) and draft our path for Pike | 18:24 |
| v1k0d3n | baby steps forward is good. | 18:24 |
| inc0 | v1k0d3n, now we know A LOT more than we did in BCN | 18:24 |
| inc0 | now we can make educated plans | 18:24 |
| inc0 | so bottom line, we don't change our direction in Ocata | 18:24 |
| inc0 | but I'd suggest we come to BCN with clean minds and start fresh | 18:25 |
| portdirect | to this end I've been reaching out as much as possible to our up-streams - as for ceph esp - it seemed to make much more sense to help them get to where we need than to build a "propitiatory" implementation, as this way we can capitalise on their expertise and offload ownership to people who *really* care about it working for as many people and use cases as | 18:25 |
| portdirect | possible. It would be great if somone could do the same for mariadb - to get it offering what we need ootb. | 18:25 |
| inc0 | I'd also be ok telling that to people that Ocata kolla-k8s is ready to deploy, but it might change a lot for pike | 18:25 |
| v1k0d3n | agree with you there. the whole exercize has been a learning experience for everyone involved. | 18:25 |
| inc0 | we'll provide upgrade path, but don't expect code backward compatibility | 18:25 |
| *** kristia__ has quit IRC | 18:25 | |
| inc0 | I'm fairly sure that however we want to change our chart structure, upgrade O->P will be possible | 18:26 |
| inc0 | portdirect, also, nothing stops us from helping them | 18:26 |
| inc0 | just, share the community, don't silo-ize it | 18:27 |
| portdirect | +2 exactly - though time, and expertese, mean I cant really take that on for mariadb | 18:27 |
| sbezverk | inc0: btw we have already some service level packages in a very native to helm form, mariadb is one of them | 18:27 |
| portdirect | though alanmeadows work there (mariadb) is pretty awesome, from what I've seen | 18:28 |
| inc0 | sbezverk, so my suggestion would be to review helm native mariadb and improve it using our learnings | 18:29 |
| inc0 | of if there is no service, and we have it | 18:29 |
| inc0 | for example idk...memcache (probably there is but for sake of argument) | 18:29 |
| inc0 | we can propose it ourselves | 18:29 |
| inc0 | and support it as official helm chart for memcache rather than part of kolla | 18:30 |
| v1k0d3n | portdirect: since you know the kolla parts...what are the main differences between kolla and what we have? | 18:30 |
| inc0 | also, again, I'd love for us to provide helm native charts for openstack | 18:30 |
| v1k0d3n | i know the differences between upstream and ours, and honestly...i agree with inc0, having that upstream will be great. | 18:30 |
| portdirect | v1k0d3n: by you, you mean att-comdev? | 18:31 |
| srwilkers | portdirect, yeah | 18:31 |
| v1k0d3n | yes | 18:31 |
| portdirect | in that case the charts you have, taking mariadb as an example - are closer to being consumable by the charts repo | 18:31 |
| portdirect | as they are not so split into subcharts for each component | 18:32 |
| v1k0d3n | ah, yes. that's a big one. | 18:32 |
| portdirect | (which has pros and cons) | 18:32 |
| inc0 | portdirect, imho this is issue we should resolve with helm | 18:32 |
| v1k0d3n | using configmaps, seeds, single service, etc? right? | 18:32 |
| inc0 | v1k0d3n, or figure out helm way to do it with helm community | 18:33 |
| portdirect | though once ther service layer is done I expect our charts to be much mor similar in terms of consumption by end users | 18:33 |
| *** hieulq_ has quit IRC | 18:33 | |
| inc0 | code in helm feature if needed | 18:33 |
| *** hieulq_ has joined #openstack-kolla | 18:34 | |
| portdirect | but the unserlying struture will be a bit different - I'm not yet sure if we have decided how the config will be generated yet in kolla | 18:34 |
| *** hieulq_ has quit IRC | 18:34 | |
| sbezverk | portdirect: after API is finilized config gen would be big thing to decide | 18:34 |
| inc0 | yeah, we might end up forking ansible configs | 18:35 |
| *** prameswar has quit IRC | 18:35 | |
| portdirect | but kolla-k8s's mariadb chart will consume a subchart for the pod, svc, pvc and pv - whihc is very different from most helm charts | 18:35 |
| portdirect | ^^ again with pros (controll) and cons (lack of uniformity with other charts) | 18:36 |
| inc0 | lack of uniformity is an issue in my book tbh | 18:36 |
| sbezverk | portdirect: subcharting is very native for helm, so what we have done by splitting mariadb into a bunch of subcharts/microservices is not a problem imho | 18:36 |
| *** sayanta__ has joined #openstack-kolla | 18:37 | |
| inc0 | sbezverk, it is if we'd like to have chart-per-repo whch is requirement for having charts in kubernetes repo | 18:37 |
| sbezverk | I consider it as a more efficient and flexible approach than shoving all into the same chart | 18:37 |
| inc0 | and again, I think we should aim to do it | 18:37 |
| inc0 | sbezverk, I was thinking that chart per service seems intuitive | 18:38 |
| inc0 | nova - a chart | 18:38 |
| inc0 | helm install nova | 18:38 |
| inc0 | as opposed to helm install openstack | 18:38 |
| *** adrian_otto has quit IRC | 18:38 | |
| *** sayantan_ has quit IRC | 18:38 | |
| *** adrian_otto has joined #openstack-kolla | 18:39 | |
| sbezverk | inc0 what if you need something special for one of nova components? | 18:39 |
| inc0 | and keep some "signle source of truth" to connect services with each other | 18:39 |
| inc0 | sbezverk, if helm lacks configurability, that's something we should address in helm | 18:39 |
| portdirect | inc0 that's exactly what the service layer should deliver - helm install openstack is madness in my option (and I think we are all in agreement there) | 18:39 |
| sbezverk | if you have everything in the same chart then it becomes less intuitive to deal with cases like that | 18:39 |
| inc0 | (I'm talking about end goal, nto what we have now) | 18:39 |
| sbezverk | portdirect: for some cases even madness might be useful ;-) like roadshows | 18:40 |
| *** sdake has joined #openstack-kolla | 18:41 | |
| sbezverk | portdirect: but I agree that service level probably will be used mostly | 18:41 |
| inc0 | sbezverk, service level makes much more sense | 18:42 |
| *** david-lyle has joined #openstack-kolla | 18:42 | |
| inc0 | we just need to figure out "single source of truth" | 18:42 |
| *** salv-orlando has joined #openstack-kolla | 18:42 | |
| inc0 | and get helm to develop featureset for us to customize installations the way you described | 18:42 |
| inc0 | I think between me, v1k0d3n's team and other volunteers we could even find people to do it if we define what we need exactly | 18:43 |
| inc0 | and Pike is longer, gives us more flexibilty | 18:44 |
| sdake | hey folks | 18:44 |
| sbezverk | inc0: service level is might be the most common but it does not mean microservice level is not usefull. | 18:44 |
| inc0 | sbezverk, we should do service level with microservice mechanism | 18:45 |
| sbezverk | inc0 +1 | 18:45 |
| inc0 | although, in kolla-ansible we don't really have that much of microservice management right | 18:45 |
| inc0 | ? | 18:45 |
| inc0 | like, nova will always deploy nova-api | 18:45 |
| inc0 | it might not deploy nova-spiceproxy, but there is only so much configurability that makes sens | 18:46 |
| inc0 | e | 18:46 |
| inc0 | well ok I take that back, you can config it all with inventory modification | 18:46 |
| inc0 | but regardless, helm should support that. | 18:46 |
| sdake | inc0 catching up on the logs - i don't see a strong rationale to rearchitect kolla-kubernetes | 18:48 |
| *** haplo37 has quit IRC | 18:48 | |
| sdake | inc0 that would result in second system effect which is to be avoided at all costs | 18:48 |
| sdake | inc0 instead incrementally changing it to suit is better and doesn't result in a monstrosity :) | 18:49 |
| inc0 | sdake, having kolla charts as official charts would need rearchitect | 18:49 |
| inc0 | and there is only so much you can do incrementally | 18:49 |
| sdake | i disagree with both statements :) | 18:50 |
| sdake | heat has been completely incrementally changed atleast 20 times | 18:50 |
| sdake | kolla-ansible 8 times atleast | 18:50 |
| sdake | and the kolla deliverable has been through 5+ refactors that make it look much different | 18:50 |
| sdake | anything can be refactored via incremental change to fit the shoe we want it to fit in | 18:50 |
| inc0 | but changing compose->ansible was not incrementsl | 18:51 |
| SamYaple | inc0: it was, you just werent around | 18:51 |
| *** salv-orlando has quit IRC | 18:51 | |
| *** vhosakot has quit IRC | 18:51 | |
| SamYaple | inc0: i wrote a docker compose module in ansible originally | 18:51 |
| SamYaple | then i refactored it out | 18:51 |
| inc0 | SamYaple, I was around. | 18:51 |
| *** salv-orlando has joined #openstack-kolla | 18:51 | |
| inc0 | but what I'm saying | 18:51 |
| inc0 | it's revolution | 18:52 |
| SamYaple | inc0: https://github.com/ansible/ansible-modules-extras/pull/354 | 18:52 |
| sdake | start fresh is terrible idea | 18:52 |
| SamYaple | but ^ that was incremental? | 18:52 |
| SamYaple | im not sure what your point is | 18:52 |
| sdake | the community will have state loaded -if you start fresh that will dump state completely | 18:52 |
| SamYaple | compose->ansible was _acutally_ incremetnal | 18:52 |
| sdake | SamYaple agreed | 18:52 |
| inc0 | sdake, fresh as in "we learned a lot, now let's get to the drawing board" | 18:53 |
| v1k0d3n | well, this is all very unfortunate. | 18:53 |
| SamYaple | inc0: im not disagreeing with you, i just dont know your point | 18:53 |
| sdake | we are already at the ddrawing board | 18:53 |
| inc0 | however we'll push code | 18:53 |
| inc0 | sdake, no, we're not, we're sprinting with development | 18:53 |
| v1k0d3n | gotta run guys. you know where to find me. got other things to do. later. | 18:53 |
| inc0 | SamYaple, my point is, at some point you need to drop old approach and start new approach | 18:53 |
| sdake | right - dev + drawing board happens at same time in my mind :) | 18:53 |
| sdake | inc0 here is what I get from reading the log | 18:54 |
| inc0 | and that's what I'd like to discuss for Pike | 18:54 |
| sdake | you intend to delete kolla-kubernetes repo and start over | 18:54 |
| inc0 | sdake, no | 18:54 |
| inc0 | god no | 18:54 |
| sdake | thats what I get from reading the log | 18:54 |
| inc0 | I'm saying we have ocata the way we have today | 18:54 |
| sdake | whereas incremental change (like the dockerfile refactor you led earlier this year) is the way to go | 18:54 |
| inc0 | with microservice arch we have | 18:54 |
| sdake | we are doing services as well as a meta package | 18:55 |
| inc0 | and we will discuss, given learnings we aquired, how to make it better without need of backwards compatibility | 18:55 |
| inc0 | but with need of upgrades | 18:55 |
| sdake | kolla-kubernetes is on a release-independent schedule | 18:55 |
| inc0 | but we have milestones | 18:55 |
| sdake | if we release a 1.0.0 it needs to be upgradeable | 18:55 |
| inc0 | ofc | 18:55 |
| sdake | right we have milestones - which lead into april :) | 18:55 |
| *** vhosakot has joined #openstack-kolla | 18:55 | |
| sdake | our end date is april not jan 26 | 18:55 |
| sdake | summit is may | 18:56 |
| sdake | we have plenty o time | 18:56 |
| inc0 | where did you get april? we submitted for Ocata | 18:56 |
| inc0 | Ocata is Jan | 18:56 |
| inc0 | we can get to feb | 18:56 |
| sdake | openstack summit is in may | 18:56 |
| inc0 | but design summit is in Atlanta | 18:56 |
| sdake | we dont need to release on the schedule | 18:56 |
| inc0 | and Ocata is in Feb | 18:56 |
| inc0 | but we want to try | 18:56 |
| sdake | we pick the date we release 1.0.0 on release-independent | 18:57 |
| inc0 | when did we pick it? | 18:57 |
| sdake | we are not beholden to the release schedule according to Doug on the ml | 18:57 |
| inc0 | because I'm fairly sure we were talking about ocata | 18:57 |
| inc0 | ofc we're not | 18:57 |
| *** haplo37 has joined #openstack-kolla | 18:57 | |
| sdake | the core team while you were schlepping your stuff across country :) | 18:57 |
| inc0 | ok...can I have irc logs please? | 18:57 |
| sdake | my point is this - march 31 or thereabouts, we need to have kolla-kubernetes 1.0.0 with ready to wrap a bow on it | 18:57 |
| inc0 | because I'm pretty sure it wasnt on ML | 18:57 |
| v1k0d3n | can we hit the pause button for a moment guys? | 18:58 |
| sdake | and 2.0.0 changed to release:trailing_cycle | 18:58 |
| sdake | so our 2.0.0 release is actually synced with Pike | 18:58 |
| inc0 | guys, this is policy decision | 18:58 |
| inc0 | which we agreed we solve on ML | 18:58 |
| sdake | I talked to doug on the ml about it | 18:58 |
| inc0 | with voting | 18:58 |
| inc0 | don't overrule your own rules please | 18:58 |
| sdake | i agree changing the release cycle is a policy decision | 18:58 |
| sdake | however we are currently release_independt | 18:58 |
| sdake | that doesn't make any sense for 2.0.0 | 18:59 |
| *** v1k0d3n has left #openstack-kolla | 18:59 | |
| sdake | we are unable to change the release cycle model at present | 18:59 |
| inc0 | sdake, timeframe when we release 1.0 is a poliy decision | 18:59 |
| inc0 | afk for 30min | 18:59 |
| sdake | when its ready is the answer | 18:59 |
| inc0 | brb | 18:59 |
| *** inc0 has quit IRC | 18:59 | |
| *** salv-orlando has quit IRC | 18:59 | |
| sdake | portdirect i was pretty sure we agreed to release 1.0.0 beginning of april - is my memory fuzzy? | 19:00 |
| sdake | sbezverk kfox1111 ?? | 19:01 |
| *** salv-orlando has joined #openstack-kolla | 19:01 | |
| sbezverk | sdake: +1 | 19:01 |
| sdake | +1 my memory is fuzzy? :) | 19:01 |
| sbezverk | both ;-) | 19:01 |
| sbezverk | sdake: seriously I think april target is correct | 19:02 |
| sdake | well i want to make sure we have a majority in agreement | 19:02 |
| *** sdake has quit IRC | 19:04 | |
| * portdirect reading scollback | 19:09 | |
| portdirect | just got back home 2 mins | 19:10 |
| *** chas has joined #openstack-kolla | 19:10 | |
| Pavo | evening gents | 19:12 |
| *** kristian__ has joined #openstack-kolla | 19:15 | |
| *** chas has quit IRC | 19:15 | |
| *** dave-mccowan has quit IRC | 19:16 | |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: kolla-common value get functions https://review.openstack.org/416069 | 19:17 |
| kfox1111 | back | 19:17 |
| kfox1111 | wow... | 19:18 |
| * kfox1111 starts reading back | 19:18 | |
| sbezverk | kfox1111Ь looks good now, will ack when the gate is green | 19:19 |
| sbezverk | kfox1111: you missed another storm in the glass of tea ;) | 19:19 |
| *** kristian__ has quit IRC | 19:20 | |
| *** MasterOfBugs has joined #openstack-kolla | 19:20 | |
| *** pramodrj07 has joined #openstack-kolla | 19:20 | |
| *** pramodrj07 has quit IRC | 19:23 | |
| *** MasterOfBugs has quit IRC | 19:23 | |
| openstackgerrit | Merged openstack/kolla: Add Kubernetes Images for Kolla-Kubernetes https://review.openstack.org/412205 | 19:24 |
| srwilkers | you say storm, i say the beauty of a community expressing differing opinions | 19:25 |
| *** MasterOfBugs has joined #openstack-kolla | 19:27 | |
| *** inc0 has joined #openstack-kolla | 19:28 | |
| *** v1k0d3n has joined #openstack-kolla | 19:29 | |
| inc0 | back | 19:29 |
| inc0 | had quick meeting, sorry | 19:29 |
| srwilkers | wb inc0 | 19:29 |
| sbezverk | srwilkers: I guess it depends on a person ;-) | 19:30 |
| *** kristian__ has joined #openstack-kolla | 19:31 | |
| *** salv-orlando has quit IRC | 19:35 | |
| *** serverascode has quit IRC | 19:35 | |
| *** alanmeadows has quit IRC | 19:35 | |
| *** NachoDuck has quit IRC | 19:35 | |
| *** kong_ has quit IRC | 19:35 | |
| *** JoseMello has quit IRC | 19:35 | |
| *** peterjenkins_ has quit IRC | 19:35 | |
| *** igordcard has quit IRC | 19:35 | |
| *** markmcclain has quit IRC | 19:35 | |
| *** gmmaha has quit IRC | 19:35 | |
| *** hyakuhei has quit IRC | 19:35 | |
| *** irtermite has quit IRC | 19:35 | |
| *** igordcard has joined #openstack-kolla | 19:35 | |
| *** irtermite has joined #openstack-kolla | 19:36 | |
| *** sdake has joined #openstack-kolla | 19:36 | |
| sdake | yo | 19:36 |
| sdake | did portdirect or kfox1111 respond to my query? | 19:36 |
| *** gmmaha has joined #openstack-kolla | 19:36 | |
| *** markmcclain has joined #openstack-kolla | 19:36 | |
| sdake | srwilkers ? | 19:36 |
| *** gmmaha has quit IRC | 19:36 | |
| *** gmmaha has joined #openstack-kolla | 19:36 | |
| sdake | (you weren't on earlier when I asked) | 19:36 |
| *** hyakuhei has joined #openstack-kolla | 19:36 | |
| *** JoseMello has joined #openstack-kolla | 19:38 | |
| *** harlowja has joined #openstack-kolla | 19:38 | |
| *** gmmaha has quit IRC | 19:38 | |
| *** rmart04 has joined #openstack-kolla | 19:39 | |
| *** kristian__ has quit IRC | 19:40 | |
| *** kristia__ has joined #openstack-kolla | 19:40 | |
| *** gmmaha has joined #openstack-kolla | 19:40 | |
| openstackgerrit | Merged openstack/kolla-ansible: Update the place of Watcher HAProxy by alphabet https://review.openstack.org/416194 | 19:41 |
| *** kong_ has joined #openstack-kolla | 19:41 | |
| *** peterjenkins_ has joined #openstack-kolla | 19:41 | |
| *** alanmeadows has joined #openstack-kolla | 19:44 | |
| *** yee37913 has quit IRC | 19:48 | |
| *** yee379 has joined #openstack-kolla | 19:49 | |
| *** rmart04 has quit IRC | 19:50 | |
| *** serverascode has joined #openstack-kolla | 19:51 | |
| *** salv-orlando has joined #openstack-kolla | 19:53 | |
| *** NachoDuck has joined #openstack-kolla | 19:54 | |
| *** sdake has quit IRC | 19:54 | |
| *** msimonin has joined #openstack-kolla | 19:59 | |
| *** dave-mccowan has joined #openstack-kolla | 20:02 | |
| *** TxGirlGeek has quit IRC | 20:03 | |
| *** openstackgerrit has quit IRC | 20:03 | |
| *** haplo37_ has quit IRC | 20:09 | |
| *** openstackgerrit has joined #openstack-kolla | 20:09 | |
| openstackgerrit | Merged openstack/kolla: Enable zookeeper for CentOS/OracleLinux https://review.openstack.org/409965 | 20:09 |
| *** adrian_otto has quit IRC | 20:11 | |
| *** haplo37_ has joined #openstack-kolla | 20:12 | |
| kfox1111 | alot to read. havnet caught up yet. | 20:14 |
| kfox1111 | one thing for upstream mariadb vs kolla's. | 20:14 |
| kfox1111 | it totally depends on whats tested. | 20:15 |
| kfox1111 | and by whom. | 20:15 |
| kfox1111 | rdo for example, tests/builds their stuff against their own mariadb, not what upstream mariadb. | 20:15 |
| kfox1111 | ideally switching to upstream mariadb would "just work". but the devil's always in the details. | 20:15 |
| kfox1111 | I'm guessing it will be an operator level decision. some will want upstream, some will want distro packages to always match. | 20:16 |
| sbezverk | kfox1111: do you have any recollection on kolla-kubernetes target date for 1.0 release? | 20:16 |
| kfox1111 | I'm not aware of any date set. | 20:16 |
| SamYaple | sbezverk: hard deadline of april 1st | 20:16 |
| sbezverk | SamYaple: works for me ;-) I was just trying to see if all folks developing for kolla kubernetes are on the same page. | 20:19 |
| kfox1111 | I think its something we shuld talk about at the ptg. | 20:20 |
| kfox1111 | come up with a list of requirements we think 1.0 should cover. | 20:20 |
| kfox1111 | that will inform when we think it might be possible to have it done by. | 20:20 |
| sbezverk | kfox1111: sure it is just I saw Feb mentioned by somebody and it kind of raised an alarm in my head as being not realistic.. | 20:21 |
| kfox1111 | for finishing 1.0? | 20:22 |
| kfox1111 | heh. | 20:22 |
| kfox1111 | like i said, depends on what the requirements are. should totally be able to launch an openstack by then. | 20:23 |
| kfox1111 | "easily", not so much. | 20:23 |
| sbezverk | kfox1111: yeah 1.0 should cover openstack deployment preferably not using copy and paste command from the notepad ;-) | 20:27 |
| * kfox1111 nods | 20:27 | |
| kfox1111 | inc0: chart per service is very corse graned. | 20:30 |
| kfox1111 | and doesn't fit helm very well. | 20:30 |
| kfox1111 | take nova for example. | 20:30 |
| kfox1111 | seems streight foward at first. | 20:30 |
| kfox1111 | nova-api, nova-conductor, nova-schedular, nova-compute. | 20:30 |
| kfox1111 | sept it starts getting complicated when you realize you might want N different nova-compute instances. | 20:31 |
| *** satyar has quit IRC | 20:31 | |
| kfox1111 | one for your cheep nodes, one for those with gpgpu's, etc. | 20:31 |
| kfox1111 | for me, I want the flexibilty to upgrade just little bits at a time. | 20:32 |
| kfox1111 | one of the hugest benifits of containerization. | 20:32 |
| kfox1111 | and very important for openstack upgrades. | 20:32 |
| kfox1111 | yes, you can take the big hammer aproach and just delete 'nova' launch some upgrades, and then recreate 'nova'. | 20:33 |
| kfox1111 | that hardcodes the unideal world of non live upgrades. | 20:33 |
| *** kristia__ has quit IRC | 20:34 | |
| *** kristian__ has joined #openstack-kolla | 20:34 | |
| kfox1111 | on the other end of the spectrum, you should be able to have a ci/cd system detect a new kolla helm package, test it, and push just that one update out to your live system with zero downtime. | 20:35 |
| kfox1111 | should be possible today with our microserivice archetecture for those openstack components that support it. | 20:35 |
| *** TxGirlGeek has joined #openstack-kolla | 20:37 | |
| SamYaple | wow. wall of kfox1111 messages | 20:38 |
| sbezverk | SamYaple: but it makes perfect sense :-) | 20:39 |
| SamYaple | yea kfox1111 i agree | 20:39 |
| SamYaple | kfox1111: just to clarify though, nova isnt affecting the instances when it upgrades. in fact it shouldnt affect anything | 20:39 |
| SamYaple | other than api availability | 20:39 |
| kfox1111 | SamYaple: yeah. but api availability is an uptime requirement in some places too. | 20:40 |
| kfox1111 | just saying. a lot of effort is being put into live upgrades in a lot of projects. | 20:40 |
| kfox1111 | if we go and try to hard code assumptions that they don't happen, it will be an even harder thing to solve. | 20:41 |
| kfox1111 | I've heard most of compute kit's close at this point to being able to do full major rolling upgrades. | 20:43 |
| SamYaple | kfox1111: oh i agree. i think you may have mis understood me. even _today_ keystone has 100% zero-downtime upgrades | 20:43 |
| kfox1111 | ah. nice. :) | 20:43 |
| SamYaple | kfox1111: i wouldnt count on it until Q | 20:43 |
| kfox1111 | yeah. I think nova<->neutron might be the trickiest bit. | 20:44 |
| kfox1111 | even just rolling upgrade support for minor upgrades is a huge thing though. | 20:45 |
| inc0 | kfox1111, what I'm saying is, we should be able to support complex modifications and upgrades on chart level | 20:45 |
| inc0 | what I would like at the end of the day is to potentially be able to split each service to different repo | 20:46 |
| inc0 | so you can do helm install stable/openstack-nova | 20:46 |
| SamYaple | kfox1111: neutron has 2 more bits to make it full rolling upgrade | 20:46 |
| SamYaple | kfox1111: i think we'll see that in P releaes | 20:46 |
| kfox1111 | inc0: with a service level chart, helm would have to do it, which it can't I think. | 20:47 |
| inc0 | I hear what you're saying and I agree, but I think our approach with chart nova-bootstap will quickly become messy | 20:47 |
| kfox1111 | inc0: with microservices, you have full flexibility there. | 20:47 |
| * srwilkers reads | 20:47 | |
| inc0 | kfox1111, yeah, but we might want to draft change that needs to be made to enable it | 20:47 |
| inc0 | it is valid use case | 20:48 |
| srwilkers | i agree with kfox1111 -- i wasnt aware of a 1.0 date set either | 20:48 |
| kfox1111 | inc0: we can assemble service packages from microservices though. | 20:48 |
| sbezverk | kfox1111: +1 | 20:48 |
| kfox1111 | so helm install stable/openstack-nova works for those with only one nova-compute need. | 20:48 |
| kfox1111 | or that kind of thing. | 20:48 |
| inc0 | as for dates, I was under impression that we're aiming for ocata, which is this month, potentially nex | 20:48 |
| inc0 | t | 20:49 |
| inc0 | so we can use PTG to plan for future | 20:49 |
| kfox1111 | like Isaid, it depends on definition of what 1.0 is. if its, you can spawn an openstack, 0.4.0 will be able to. | 20:49 |
| sbezverk | inc0: we are still debating API and then big thing is config management, how do you think it could be possible? | 20:49 |
| inc0 | kfox1111, what I'm saying is, we should plan for changes in helm too | 20:50 |
| kfox1111 | so if its still stable by end of ocata, we can retag it 1.0. | 20:50 |
| kfox1111 | if its heavy orchestration, upgrades, etc, maybe not. | 20:50 |
| kfox1111 | inc0: ah. yeah, we should work with helm to see if they can get the features needed in. | 20:50 |
| kfox1111 | inc0: but relying on it happening is a seperate issue. | 20:50 |
| inc0 | kfox1111, I was thinking 1.0 == just deploy working openstack | 20:50 |
| inc0 | that won't hurt you | 20:50 |
| kfox1111 | that may be possible in the timeframe. | 20:51 |
| inc0 | yeah, no argument there, we do what we have to do | 20:51 |
| kfox1111 | we still have service packages, and configmap stuff to figure out. | 20:51 |
| inc0 | but nothing will happen unless we specify stuff we need | 20:51 |
| kfox1111 | and dep handling. | 20:51 |
| kfox1111 | and operators, of those will be needed.... | 20:51 |
| inc0 | so here's thing | 20:51 |
| inc0 | I'd code the thing myself if we agree on what we want | 20:52 |
| inc0 | and by we I mean kolla, att and maybe even SAP | 20:52 |
| inc0 | and ofc helm | 20:52 |
| kfox1111 | and... thats what the ptg kind of thing is for. :/ | 20:52 |
| inc0 | I'll be frank, I really don't want to have 3 different openstacks on k8s using helm | 20:52 |
| kfox1111 | its been realy hard to get everyone on the same page without being in person. :/ | 20:52 |
| inc0 | yeah, and reason I even mentioned 1.0 date | 20:53 |
| kfox1111 | +1 to that. | 20:53 |
| kfox1111 | better to cooperate. | 20:53 |
| inc0 | is because I would like us all to use PTG to draft this stuff | 20:53 |
| inc0 | but not block our current path until we all draft it and agree on it | 20:53 |
| kfox1111 | yeah | 20:54 |
| kfox1111 | meeting next. bbiab. | 20:54 |
| *** absubram has quit IRC | 20:56 | |
| *** eaguilar_ has quit IRC | 20:57 | |
| *** athomas has joined #openstack-kolla | 20:59 | |
| *** adrian_otto has joined #openstack-kolla | 21:02 | |
| *** JoseMello has quit IRC | 21:05 | |
| *** chas has joined #openstack-kolla | 21:11 | |
| *** matrohon has quit IRC | 21:11 | |
| *** gfidente has quit IRC | 21:12 | |
| *** jheroux has quit IRC | 21:14 | |
| *** jheroux has joined #openstack-kolla | 21:15 | |
| *** chas has quit IRC | 21:15 | |
| *** matrohon has joined #openstack-kolla | 21:24 | |
| *** sdake has joined #openstack-kolla | 21:25 | |
| *** Serlex has quit IRC | 21:29 | |
| *** athomas has quit IRC | 21:30 | |
| *** l4yerffeJ_ has joined #openstack-kolla | 21:34 | |
| *** l4yerffeJ has quit IRC | 21:34 | |
| *** dave-mccowan has quit IRC | 21:35 | |
| *** l4yerffeJ has joined #openstack-kolla | 21:35 | |
| *** l4yerffeJ__ has quit IRC | 21:35 | |
| *** Jeffrey4l has quit IRC | 21:35 | |
| *** Jeffrey4l has joined #openstack-kolla | 21:36 | |
| *** srwilkers has quit IRC | 21:41 | |
| jemcevoy | inc0 Happy New Year | 21:42 |
| inc0 | thank you good sir, happy new year to you | 21:48 |
| *** sdake has quit IRC | 21:51 | |
| *** chas has joined #openstack-kolla | 21:51 | |
| *** msimonin1 has joined #openstack-kolla | 21:53 | |
| *** sdake has joined #openstack-kolla | 21:53 | |
| sdake | hey peeps | 21:53 |
| sdake | hey peeps | 21:53 |
| *** chas has quit IRC | 21:56 | |
| *** msimonin has quit IRC | 21:56 | |
| *** matrohon has quit IRC | 22:01 | |
| jemcevoy | I had to shutdown my 4 node cluster... Is there a procedure for restarting it? and when to run the kolla-ansible mariadb_recovery -i ~/inventory/multinode | 22:06 |
| jemcevoy | inc0: Do I just ipmitool power on all 4 nodes? | 22:07 |
| jemcevoy | I tried before and this was the result: http://paste.openstack.org/show/593796/ | 22:08 |
| jemcevoy | so I resorted to a cobbler kickstart | 22:08 |
| *** jrobinson has joined #openstack-kolla | 22:11 | |
| *** jtriley has quit IRC | 22:11 | |
| inc0 | mariadb_recovery is for full failure | 22:12 |
| inc0 | if you gracefully shut down mariadb, it should be ok | 22:12 |
| *** chas has joined #openstack-kolla | 22:12 | |
| inc0 | but SamYaple can help here - I'm not good at this one | 22:12 |
| *** huikang has quit IRC | 22:13 | |
| *** matrohon has joined #openstack-kolla | 22:15 | |
| *** chas has quit IRC | 22:16 | |
| kfox1111 | back. | 22:16 |
| kfox1111 | https://review.openstack.org/#/c/416069/ is ready for a second. | 22:17 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Visual feedback https://review.openstack.org/416077 | 22:20 |
| sbezverk | kfox1111: how do you want to move forward with microservice conversion? | 22:21 |
| bmace | this may end up being a silly question. i'm just getting up to speed on the k8s codebase, but is there any rhyme or reason for the file name differences in helm/microservice/<service>/templates/*.. seems like a mix between dashes and underscores in the filenames. | 22:22 |
| kfox1111 | sbezverk: I was thinking there were a couple of aproaches. | 22:22 |
| *** fguillot has quit IRC | 22:22 | |
| kfox1111 | one would be to do all the templates that shared a single common_X template in a ps. | 22:23 |
| kfox1111 | less merge conflicts that way, but potentially less opertunity to generalize some code. | 22:23 |
| kfox1111 | the other way would be per variable kind of thing, so like "image_pull_policy" for example, could be made a function, looked up in vals and updated everywhere. and then do the next thing. | 22:24 |
| kfox1111 | probably much more merge conflicts that way though. | 22:24 |
| kfox1111 | so, maybe the first option might work better? | 22:25 |
| inc0 | kfox1111, ad https://review.openstack.org/#/c/416069/15 do we store these values somewhere outside of yaml? | 22:26 |
| inc0 | so here's my thingking | 22:26 |
| inc0 | in ansible we have multiple layers of config, each will override latter | 22:26 |
| kfox1111 | inc0: all values come from values, or the users specified values. | 22:26 |
| inc0 | at the end of the day, we have single flat list of all things overriden | 22:26 |
| inc0 | here, can we store it in k8s | 22:27 |
| inc0 | ? | 22:27 |
| inc0 | like a configmap? | 22:27 |
| kfox1111 | inc0: similar to the way ansible does it, but can be overridden at a per package level too. | 22:27 |
| kfox1111 | so more flexible. | 22:27 |
| kfox1111 | there's two different things too... | 22:27 |
| kfox1111 | configmap data, which this isn't about. | 22:27 |
| kfox1111 | or the config of the k8s objects themselves. | 22:27 |
| kfox1111 | which this is about. | 22:27 |
| *** salv-orlando has quit IRC | 22:28 | |
| kfox1111 | so like, a config option to attach an extra logging container to each pod. | 22:28 |
| inc0 | ok, well in ansible case, they both kinda overlap, here it might not | 22:28 |
| kfox1111 | what image name to use for the container, etc. | 22:28 |
| kfox1111 | yeah. different use case for now. | 22:28 |
| inc0 | yeah, it's resource declaration phase | 22:29 |
| kfox1111 | right. | 22:29 |
| inc0 | so here is why I'm thinking about it | 22:29 |
| inc0 | if we'd have signle source of truth available somewhere (configmap probably), we could use it to both configure deployment, pass values between charts and create config files | 22:30 |
| *** sdake has quit IRC | 22:30 | |
| sbezverk | kfox1111: not sure if first approach is scalabale | 22:30 |
| kfox1111 | sbezverk: yeah. depends on how fast we can crank through it too. | 22:30 |
| sbezverk | unless you insist to do everything yourself ;-) | 22:31 |
| kfox1111 | I won't insist. :) but I am going to try and spend as much time on it as I can for the next few days. | 22:31 |
| kfox1111 | really wana get it finished. ;) | 22:32 |
| SamYaple | kfox1111: its easy if you stop caring and start coding! | 22:32 |
| *** salv-orlando has joined #openstack-kolla | 22:32 | |
| sbezverk | kfox1111: ok | 22:32 |
| kfox1111 | SamYaple: easier yet if meetings stop. ;) | 22:32 |
| *** salv-orlando has quit IRC | 22:32 | |
| *** salv-orlando has joined #openstack-kolla | 22:33 | |
| SamYaple | kfox1111: i think my previous statement still applies! | 22:33 |
| sbezverk | inc0: could you expand a bit your idea with config map? | 22:33 |
| kfox1111 | configmaps arn't readable by helm. | 22:33 |
| inc0 | ok, so I'm still thinking on having separate services from each other | 22:33 |
| openstackgerrit | Merged openstack/kolla-kubernetes: kolla-common value get functions https://review.openstack.org/416069 | 22:34 |
| kfox1111 | I think you get to what your thinking though, with: | 22:34 |
| inc0 | kfox1111, helm has no way to access configmaps? | 22:34 |
| kfox1111 | * microservice packages with the new api, | 22:34 |
| kfox1111 | * configmap packages | 22:34 |
| kfox1111 | * service packages | 22:34 |
| kfox1111 | user provided config file, ala values.yaml | 22:35 |
| sbezverk | kfox1111: what is configmap pakcages? | 22:35 |
| kfox1111 | they simply helm install service-package.tgz --values mycloud.yaml and off they go. | 22:35 |
| inc0 | so here is what I was thinking | 22:35 |
| kfox1111 | sbezverk: one possible idea. was tryhing to avoid talking about config stuff till after 0.4.0.... | 22:35 |
| inc0 | you create your config files | 22:36 |
| sbezverk | kfox1111: got it, placeholder for opensatck gen config | 22:36 |
| inc0 | then you deploy mariadb | 22:36 |
| inc0 | helm install mariadb | 22:36 |
| inc0 | mariadb installed will create couple of configmaps/secrets | 22:36 |
| inc0 | how to access it and so on | 22:36 |
| inc0 | helm install nova | 22:37 |
| kfox1111 | how do you reuse existing configmaps/secrets? | 22:37 |
| inc0 | it will run bootsrap pod which will access configmaps created by amriadb, bootstrap nova and create config string and put it into configmap | 22:37 |
| kfox1111 | what is "it"? | 22:38 |
| inc0 | nova db username, nova db password | 22:38 |
| kfox1111 | ie, what is orchestrating. | 22:38 |
| inc0 | you mean so it wont re-bootsrap later? | 22:39 |
| kfox1111 | helm/kubernetes have no ordering garantees. | 22:39 |
| inc0 | but init container solves that | 22:39 |
| SamYaple | not for upgrades | 22:39 |
| SamYaple | (if i partially understand things) | 22:40 |
| kfox1111 | an init container is not for bootstrapping a system, but bootstrapping a pod. | 22:40 |
| inc0 | yeah, but system is sum of pods as far as I understand | 22:40 |
| sbezverk | kfox1111: I think they want to use init contianer to bootstrap service and including dependecies | 22:40 |
| kfox1111 | assertion still holds. | 22:40 |
| inc0 | don't start nova-scheduler until nova-bootstrap is done | 22:40 |
| kfox1111 | you bootstrap a system once, not n times. | 22:41 |
| kfox1111 | on most services, you do n of them. | 22:41 |
| inc0 | and nova-bootsrap will wait till mariadb-configmaps it needs exists | 22:41 |
| kfox1111 | so.... theres a different way to do it, which we have some prototypes of. | 22:41 |
| inc0 | it's a matter of locks | 22:41 |
| kfox1111 | inc0: https://review.openstack.org/#/c/412239/ | 22:41 |
| inc0 | init container is nothing more than a locking mechanism | 22:41 |
| inc0 | kfox1111, I'll need some time to understand this blob of code | 22:42 |
| *** jheroux has quit IRC | 22:42 | |
| kfox1111 | inc0: k. please do. | 22:43 |
| inc0 | what I'm esentially saying is deployments (including upgrades if we're careful) can be done by lock management | 22:43 |
| kfox1111 | but, tldr; | 22:43 |
| inc0 | and configmaps can act as locks | 22:43 |
| kfox1111 | helm service package has deps for all the microservices: | 22:43 |
| kfox1111 | https://review.openstack.org/#/c/412239/21/helm/service/neutron/requirements.yaml | 22:43 |
| kfox1111 | each microservice has a kubernetes-entrypoint init-contiainer and blocks until its ready: | 22:44 |
| kfox1111 | https://review.openstack.org/#/c/412239/21/helm/all_values.yaml | 22:44 |
| kfox1111 | you helm install the service package, | 22:44 |
| kfox1111 | helm installs all the microservices, | 22:44 |
| kfox1111 | and they start up in their own dep order. | 22:44 |
| openstackgerrit | Mark Giles proposed openstack/kolla-kubernetes: Fix helm charts that are broken behind a proxy https://review.openstack.org/416366 | 22:44 |
| inc0 | yeah, that's essentially what I was saying | 22:45 |
| kfox1111 | and unfortunately, you can't see the green jenkins test, as the recent merges reverted the tests.... :) | 22:45 |
| inc0 | only instead of locking on keystone running, lock on existance of certain configmaps | 22:45 |
| openstackgerrit | Marcus Williams proposed openstack/kolla-ansible: Work in Progress: Add ODL to kolla-ansible https://review.openstack.org/416367 | 22:46 |
| inc0 | only bootstrap jobs will care about other services | 22:46 |
| inc0 | or even not, because they will lock on configmaps these other services make | 22:46 |
| kfox1111 | inc0: pods already block if configmaps don't exist. | 22:46 |
| kfox1111 | inc0: kubernetes-entrypoint provides blocking on jobs finishing successfully and services existing. | 22:47 |
| inc0 | yeah, but if we overhaul using configmaps for all the cross-service communication | 22:47 |
| inc0 | we can have services completely outside of openstack | 22:47 |
| inc0 | k8s* | 22:47 |
| openstackgerrit | Marcus Williams proposed openstack/kolla: Work in Progress: Draft add ODL to kolla https://review.openstack.org/416369 | 22:47 |
| inc0 | like totally external mariadb | 22:47 |
| kfox1111 | you can now too. | 22:47 |
| kfox1111 | svc's an point directly to external things, | 22:47 |
| kfox1111 | and so can the configmap data itself. | 22:47 |
| kfox1111 | works today. :) | 22:48 |
| inc0 | I'm just strugling with idea why we have microservices directory tbh | 22:48 |
| kfox1111 | because if you just had nova, | 22:48 |
| kfox1111 | it wouldn't work when you have gpgpu nodes, and regular ones, | 22:48 |
| kfox1111 | just want to upgrade just nova-api without touching anything else, | 22:49 |
| kfox1111 | orchestrate with ansible/chef or heat, | 22:49 |
| kfox1111 | etc. | 22:49 |
| inc0 | I don't understand why it won't work? | 22:49 |
| kfox1111 | why one super package wont work? | 22:49 |
| inc0 | one superpackage per service | 22:49 |
| kfox1111 | with no microservices? | 22:49 |
| inc0 | helm install nova -set nova.enable_spiceproxy=false | 22:49 |
| inc0 | microservices as resources rather than charts | 22:50 |
| kfox1111 | helm's packages are all or nothing. upgrade them all, or not. delete them all, or not. | 22:50 |
| kfox1111 | 1 package -> 1 release. | 22:50 |
| kfox1111 | a release gets manipulated all at once. | 22:50 |
| inc0 | ok, but helm guys mentioned people worked around it with templating right? | 22:50 |
| inc0 | it's ugly, but maybe we could amke it easier in helm | 22:51 |
| kfox1111 | maybe. but I doubht that will be a thing for at least 6 months. | 22:51 |
| inc0 | let me guess, this is whole att-comdev vs kolla-k8s difference discussion right? | 22:52 |
| kfox1111 | what your asking is? | 22:52 |
| *** adrian_otto has quit IRC | 22:52 | |
| sbezverk | inc0: other than you do not like to have microservice in its own directory, what is the technical reason? | 22:53 |
| sbezverk | I mean like/do not like it not enough really | 22:53 |
| inc0 | so I'm trying to understand what would need to be done for helm to accept kolla's charts to official helm repo | 22:54 |
| kfox1111 | your suggesting the aproach taken by att-comdev. make it inflexible but easy to deploy for now. then, maybe hopefully some day they will make helm make it easier to do more fine grained things. | 22:54 |
| kfox1111 | I'm worried about day 2 operations. how do I upgrade it. how do I kick just one service. etc. | 22:54 |
| inc0 | I want configurable and easy to install | 22:54 |
| kfox1111 | inc0: so, theres a lot of things to discuss down that road. | 22:54 |
| inc0 | totally | 22:55 |
| inc0 | I'm just trying to do just this - discuss;) | 22:55 |
| inc0 | how to find middle ground - easy and robust | 22:55 |
| inc0 | that's the goal | 22:55 |
| kfox1111 | inc0: configurable and easy to install's basically every deployment tool. what does kolla-kubernetes bring to the table there? | 22:55 |
| sbezverk | inc0: are you sure that their charts will be accepted as is into official helm chart repo? | 22:55 |
| inc0 | well, it's deployment tool which uses specific set of underlying tech | 22:56 |
| sbezverk | they do funny things as well | 22:56 |
| kfox1111 | I do want to get us there, but most distro's focus on making it easy to install on day one, then cause pain on day 2. and make you cry when you go and upgrade it. | 22:56 |
| inc0 | sbezverk, I'm not saying we should do what they're doing | 22:56 |
| kfox1111 | then the distro gets stuck for years trying to figure out how to do upgrades. | 22:56 |
| inc0 | but if our chart has issues in being accepted to helm, then we're doign something helm community doesn't like | 22:56 |
| inc0 | and that's bad | 22:56 |
| kfox1111 | I've been trying for the other aproach. focus on day 2 life. and upgrades. | 22:56 |
| kfox1111 | and day one falls out from there. | 22:57 |
| inc0 | kfox1111, upgrades, reconfigure and everything are a must | 22:57 |
| sbezverk | inc0: nobody from helm community ever commented about our charts | 22:57 |
| inc0 | I mean if kolla-k8s can't be both, why even use it? | 22:57 |
| inc0 | kolla ansible is both easy and day-2 robust | 22:57 |
| kfox1111 | sbezverk: +1 | 22:57 |
| inc0 | sbezverk, ok, let's ask them to comment then | 22:58 |
| kfox1111 | so, more issues with upstreaming: | 22:58 |
| inc0 | guys, I'm not against you here | 22:58 |
| kfox1111 | governence: how does that work with openstack? | 22:58 |
| SamYaple | openstack-ansible pushes to ansible-galaxy | 22:58 |
| *** sdake has joined #openstack-kolla | 22:58 | |
| kfox1111 | their repo doesn't handle building multiple packages well at all. needs a lot of work. | 22:58 |
| kfox1111 | inc0: please don't feel like we're piling up on you, but we've been down this discussion, a LOT... :/ | 22:59 |
| sbezverk | inc0: so far we heard negative comments ONLY from one guy | 22:59 |
| inc0 | sbezverk, so I've heard from multiple | 22:59 |
| SamYaple | yea screw that guy | 22:59 |
| inc0 | that's why I'm worried | 22:59 |
| sbezverk | but it makes sense everybody want to sell their shoes ;-) | 23:00 |
| inc0 | one guy is particulary vocal about them, but it's not just him:) | 23:00 |
| sbezverk | inc0: subcharts approach is perfectly native for helm | 23:01 |
| * portdirect back at keyboard - sees loads of scrollback | 23:01 | |
| sbezverk | so I do not see any architectural issues | 23:02 |
| SamYaple | portdirect: i made some comments too! | 23:02 |
| portdirect | SamYaple: ;-) | 23:02 |
| sbezverk | helm is relatively new project, mostly used to deploy simple things | 23:02 |
| sbezverk | now more complex req required more complex solutions | 23:03 |
| sdake | inc0 kfox1111 portdirect sbezverk srwilker- re schedule - here is the ocata schedule: https://releases.openstack.org/ocata/schedule.html | 23:03 |
| sdake | that schedule there points out the PTG is feb 20-24th and trailing cycle deadline is March 8th | 23:04 |
| sdake | one thing to keep in mind is kolla-kubernetes is on a release independent schedule 1.0.0 != necessarily ocata | 23:04 |
| sdake | or aligned with the ocata schedule | 23:04 |
| sdake | what we want out of 1.0.0 is something that is upgradeable in the field with 2.0.0 in 6 months | 23:05 |
| sdake | when pike releases | 23:05 |
| sdake | the minimum viable 1.0.0 includes an upgrade path - which means the implementation needs to be fairly complete | 23:06 |
| sdake | and of course subject to refactoring | 23:06 |
| sdake | it doesn't have to upgrade at 1.0.0 | 23:06 |
| sdake | it has to be upgradeable from 1.0.0 | 23:06 |
| portdirect | inc0: (and I'm just answering things as I see them, not voicing an option on right/wrong etc) configmaps can be used as a very effective locking mechanism - if they don;t exist the pod doesn't launch - i used to use secrets for this very purpose. | 23:06 |
| sdake | I further think helm installing a million microservices works well for big operations but poorly for small/medium shops (100-1k employees) | 23:07 |
| *** matrohon has quit IRC | 23:09 | |
| sdake | inc0 i find it difficult to believe you ahve heard there is issue with kolla-kubernetes dev since you have been on pto for 5 weeks | 23:09 |
| sdake | inc0 things are going very well from my POV | 23:09 |
| sdake | in 1 month we have implemented ALL microservices | 23:09 |
| openstackgerrit | Mark Giles proposed openstack/kolla-kubernetes: Fix helm charts that are broken behind a proxy https://review.openstack.org/416366 | 23:09 |
| *** kristian__ has quit IRC | 23:10 | |
| SamYaple | i mean im not involved in k8s at all and I have seen the issues even in the past two weeks..... | 23:10 |
| *** kristian__ has joined #openstack-kolla | 23:10 | |
| sdake | SamYaple which issues in particular? you mean startup issues? | 23:12 |
| *** srwilkers has joined #openstack-kolla | 23:12 | |
| sdake | srwilkers schedule here: https://releases.openstack.org/ocata/schedule.html | 23:12 |
| sdake | march 8th is when kolla-ansible releases | 23:13 |
| inc0 | sdake, well, I was on pto but I wasn't totally away | 23:13 |
| *** openstack has joined #openstack-kolla | 23:16 | |
| inc0 | you said that's no longer the case and said it as a matter of fact, well, I'd like to see when it was discussed | 23:17 |
| inc0 | because *that is not what we agreed on* | 23:17 |
| sdake | inc0 new decisions can be made - i was not at most of the summit sessions | 23:18 |
| inc0 | well, I wasn't during new decision making, does it make it less or more valid decision | 23:19 |
| inc0 | ? | 23:19 |
| inc0 | if you weren't on session, that doesn't mean decision we've done there is invalid | 23:20 |
| sdake | inc0 i htink we should have a proper vote of the kolla-kubernetes team as it stands now to determine the 1.0.0 target | 23:20 |
| sdake | since the core reviewers are responsible ultimately for the quality of the work and future success of kolla-kubernetes | 23:20 |
| inc0 | so instead of vote on ML, let's make this decision on PTG | 23:22 |
| inc0 | as most of core team will be there | 23:22 |
| sbezverk | inc0: i will not but happy to express my opinion now or later via ml | 23:23 |
| inc0 | shame sbezverk :( | 23:24 |
| sdake | inc0 ok well 1.0.0 can't be released before PTG if that is the case - meaning at minimum 1.0.0 is early March | 23:24 |
| sdake | inc0 imo all votes should be on mailing list | 23:24 |
| sdake | inc0 votes at ptgs or summits don't include the core review team | 23:24 |
| sdake | er all of em anyway | 23:24 |
| sdake | i have to say feature freeze by feb 26 is too agressive - pretty sure eveyrone has been operating under a model where we would release when it was ready - because that is what I told them based upon advice from Doug | 23:25 |
| sdake | Doug posted the mail on the mailing list | 23:25 |
| inc0 | ok, so here's my take | 23:26 |
| inc0 | we're not release bound | 23:26 |
| inc0 | but I want to *aim* for Ocata | 23:26 |
| inc0 | not follow it to the letter | 23:27 |
| inc0 | and Ocata is Feb 20 | 23:27 |
| *** lamt has quit IRC | 23:27 | |
| *** rhallisey has quit IRC | 23:28 | |
| *** rhallisey has joined #openstack-kolla | 23:28 | |
| sdake | actually march 6-10 for kolla-ansible and kolla deliverables | 23:29 |
| sdake | ptg is feb20 | 23:29 |
| *** saneax-_-|AFK is now known as saneax | 23:29 | |
| sdake | the release traling happens after the PTG | 23:29 |
| openstackgerrit | Joshua Harlow proposed openstack/kolla: Allow for image additions archives (as well as plugins) https://review.openstack.org/411534 | 23:30 |
| portdirect | inc0: If possible it would be great to have a vote immediately following ptg, or even during it if notes are on etherpad | 23:30 |
| openstackgerrit | Marcus Williams proposed openstack/kolla: Work in Progress: Add OpenDaylight Container to kolla https://review.openstack.org/416369 | 23:30 |
| inc0 | oh btw portdirect sbezverk and others who might not be on ptg | 23:31 |
| inc0 | we got green light to hold teleconference | 23:31 |
| inc0 | so we will | 23:31 |
| inc0 | you can all call into sessions | 23:31 |
| sbezverk | inc0: !!!!! Great !!!!!! | 23:31 |
| portdirect | sweet! | 23:31 |
| srwilkers | thats fantastic | 23:32 |
| portdirect | :D no need to put on trousers is always a win | 23:32 |
| portdirect | (though I think others in my office may not be that happy about that...) | 23:32 |
| *** vhosakot has quit IRC | 23:32 | |
| sdake | portdirect that depends on a whole lotta variables | 23:33 |
| kfox1111 | awesome. :) | 23:33 |
| portdirect | sdake: I think it's my turn to say 'I dont wanna know' :) | 23:33 |
| sdake | portdirect you got it ;) | 23:34 |
| kfox1111 | inc0: to be fair, I wasn't at the summit either, | 23:34 |
| kfox1111 | nore probably were the majority of the kolla-kubernetes cores now? | 23:35 |
| kfox1111 | so what was "agreed" to there, is a guideline. we're doing our best, but not bound to it. | 23:35 |
| sbezverk | kfox1111: I was there but overwhelming amount of info, especially new, made it impossible for me to realize it | 23:36 |
| sbezverk | after spending few month now, it is much easier to navigate in this info and take right decision.. | 23:37 |
| kfox1111 | yeah. makes sense. | 23:37 |
| kfox1111 | its a jurney the community makes together. | 23:37 |
| kfox1111 | and as we're release independent at the moment, | 23:37 |
| kfox1111 | my vote would be to reach 1.0 when its ready, not at some arbitrary date. | 23:38 |
| kfox1111 | then after that, we can go to date based. | 23:38 |
| kfox1111 | that does require setting a definition of "ready" though. | 23:38 |
| portdirect | kfox1111: I think the majority of current cores (at least sdake, sbezverk, srwilkers, inc0, rhallisey and me were) but many of use were not at the time | 23:38 |
| portdirect | kfox1111: +1 to release when ready | 23:39 |
| portdirect | kfox1111: (stayong out of ready discussion for now if pos - it's 11:30pm here..) | 23:39 |
| openstackgerrit | Mark Giles proposed openstack/kolla-kubernetes: Fix helm charts that are broken behind a proxy https://review.openstack.org/416366 | 23:39 |
| kfox1111 | portdirect: ah. didn't think you made it. glad you did. :) | 23:40 |
| kfox1111 | portdirect: yeah. I'm staying out of the ready discussion too. that I think is a much bigger conversation. :) | 23:40 |
| portdirect | yeah - my side of the pond :) | 23:40 |
| srwilkers | yeah. last minute approval to go -- glad i was able to | 23:40 |
| kfox1111 | inc0: oh.... btw.... | 23:40 |
| kfox1111 | inc0: the patch Mark Giles just posted? | 23:40 |
| kfox1111 | it is needed only because we put in a workaround to some broken containers on the hub. | 23:41 |
| mgiles | kfox1111 yes | 23:41 |
| kfox1111 | can we get those updated please so the workaroundd can be dropped? | 23:41 |
| mgiles | although we may need something else similar in other cases | 23:41 |
| kfox1111 | then the patch to fix the workaround won't be needed. | 23:41 |
| *** msimonin1 has quit IRC | 23:41 | |
| *** msimonin has joined #openstack-kolla | 23:41 | |
| *** msimonin has quit IRC | 23:41 | |
| *** msimonin has joined #openstack-kolla | 23:41 | |
| mgiles | but if we can get the containers updated, I'm happy to abandon the ps | 23:42 |
| kfox1111 | mgiles: the docker ideal is that the containers should be fully baked. they should never need to reach out to download stuff. | 23:42 |
| *** msimonin has quit IRC | 23:42 | |
| kfox1111 | otherwise, they aren't tested. :/ | 23:42 |
| mgiles | kfox1111 I hope that's true :-) | 23:42 |
| SamYaple | kfox1111: hmm im not sure i agree with that "the docker ideal" | 23:42 |
| SamYaple | or wait. we arent talking about configs are we? | 23:43 |
| kfox1111 | mgiles: should be able to test/enforce that. :) if we disable internet access in the gate. | 23:43 |
| portdirect | kfox1111: I could keep a rolling set of containers from master up on the docker hub? | 23:43 |
| kfox1111 | SamYaple: not talking about configs. just programs. | 23:43 |
| kfox1111 | portdirect: would love to see that. hard to do at the moment. :/ | 23:43 |
| portdirect | kfox1111: how come? | 23:44 |
| kfox1111 | portdirect: alternately, maybe a tarball with them all in there for easy loading. | 23:44 |
| kfox1111 | portdirect: short anser is, they take forever to upload, as they upload the same underlayers over and over due to the way the hub is designed. :/ | 23:44 |
| portdirect | (I have more bandwidth than normal here...) | 23:44 |
| openstackgerrit | Marcus Williams proposed openstack/kolla-ansible: Work in Progress: Add OpenDaylight to kolla-ansible https://review.openstack.org/416367 | 23:44 |
| kfox1111 | I think the last number I heard was something like 10 hours to upload the set? | 23:45 |
| portdirect | also the newer docker client regognises shared layers - takes about 2-3 hours for me to upload | 23:45 |
| portdirect | I have 1.2gbs average upload | 23:46 |
| *** rhallisey has quit IRC | 23:46 | |
| kfox1111 | for the whole set? | 23:46 |
| portdirect | yes | 23:46 |
| kfox1111 | nice. :) | 23:46 |
| portdirect | easy for me to set up a cron job and push to a repo fo use to use for devel | 23:47 |
| kfox1111 | nightlies wouldn't be too bad then. I wonder if we could replicate that in zuul. | 23:47 |
| sdake | lord knows I've tried kfox1111 :) | 23:49 |
| sdake | the gate for kolla-kubernetes should be buildilng images fresh rather then pulling from hub imo :) | 23:49 |
| *** TxGirlGeek has quit IRC | 23:50 | |
| kfox1111 | I think it shoudl do both. but a cached version, not the hub. | 23:50 |
| sbezverk | mgiles: could you quick explain what will use proxy setting when user gets created? | 23:50 |
| kfox1111 | so we can ensure the current release is deployable. | 23:50 |
| kfox1111 | sbezverk: its for the pip upgrade. :/ | 23:50 |
| mgiles | sbezverk The container does a pip install | 23:50 |
| kfox1111 | sbezverk: if we can get the hub images upgraded, we can drop the pip install. | 23:50 |
| mgiles | and then I can abandon the ps :) | 23:51 |
| sbezverk | mgiles: got it | 23:51 |
| portdirect | mgiles: that would be great - i really dont like it | 23:51 |
| mgiles | but without one or the other, I'm pretty well blocked | 23:51 |
| srwilkers | id like to see this resolved elsewhere. not a fan of seeing proxy stuff thrown in the helm charts | 23:51 |
| sdake | mgiles which image | 23:51 |
| portdirect | mgiles: but like locking you out of development even more... | 23:51 |
| kfox1111 | inc0: what do you think about getting the hub image fixed? it has a faulty openstack client. | 23:51 |
| mgiles | portdirect Have to agree. I didn't love it either | 23:51 |
| portdirect | *DONT - whoops! | 23:52 |
| portdirect | you're cool :) | 23:52 |
| mgiles | portdirect Wow - wondered what I had done to you :) | 23:52 |
| sdake | ok - early knock off day for me - not enough sleep | 23:52 |
| sdake | and hungree | 23:52 |
| portdirect | mgiles: nothing yet - but i think i just gave you a motive :) | 23:52 |
| sdake | i'll probably be back later | 23:52 |
| openstackgerrit | Marcus Williams proposed openstack/kolla-ansible: [WIP] Add OpenDaylight to kolla-ansible https://review.openstack.org/416367 | 23:53 |
| openstackgerrit | Marcus Williams proposed openstack/kolla: [WIP] Add OpenDaylight Container to kolla https://review.openstack.org/416369 | 23:53 |
| portdirect | kfox1111: who has push access to docker hub? | 23:54 |
| kfox1111 | not sure who all does. sbezverk, sdake, and I do for sure. Ithink inc0 too. | 23:54 |
| kfox1111 | I'm relucant to change any existing container though, as I don't know the rules. | 23:55 |
| sdake | kfox1111 typically someone builds new images when a new tag is created | 23:55 |
| portdirect | roger - who does Jeffrey4l ? | 23:55 |
| sdake | kfox1111 other then that, they are poorly maintained | 23:55 |
| sdake | portdirect coolsvap was maintaining them | 23:55 |
| sdake | in a very limited way | 23:56 |
| kfox1111 | what about when the tag doesn't need updating, but just a rebuild is needed? | 23:56 |
| kfox1111 | can we just do that? | 23:56 |
| kfox1111 | or alternately, when will 3.0.2 show up? | 23:56 |
| portdirect | what do you think the best way to get this fixed - it was a bit of an anyoance before - but now its blocking devs its getting a bit ugly | 23:56 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/kolla-kubernetes: Updated from global requirements https://review.openstack.org/416199 | 23:57 |
| kfox1111 | portdirect: yeah. :/ | 23:57 |
| sdake | kfox1111 3.0.2 - no idea when that will show up - the last tag of the z stream for kolla was Nov 17 | 23:57 |
| kfox1111 | can we cut one at any time? | 23:58 |
| sdake | we are due for another zstream shortly | 23:58 |
| sbezverk | we can easily substitute existing image with the same tag but new content, it is a bit dangerous but maybe less dangerous that puyshing proxy setting everywhere | 23:58 |
| sdake | kfox1111 depends when the release liason wants to pull the trigger on the release (who is Jeffrey4l ) | 23:58 |
| sdake | but then someone has to build them and push them to the hub | 23:58 |
| sdake | i spoke to Jeffrey4l prior to break about releasing new images | 23:58 |
| sdake | he said its in progress | 23:58 |
| kfox1111 | in progress? cool. | 23:59 |
| sdake | i think something with the gates is blocking | 23:59 |
| kfox1111 | so, within the next day or two? | 23:59 |
| sdake | kfox1111 no idea | 23:59 |
| sdake | kfox1111 given the pace of z streams i find it unlikely to happen on that time scale | 23:59 |
| kfox1111 | :/ | 23:59 |
| kfox1111 | hmm.... | 23:59 |
| sdake | this is a problem for inc0 to sort out unfortunately | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!
