| *** erlon has quit IRC | 00:00 | |
| *** tonanhngo has quit IRC | 00:01 | |
| *** krtaylor has quit IRC | 00:02 | |
| *** jrobinson has quit IRC | 00:06 | |
| *** srwilkers has quit IRC | 00:09 | |
| *** rwallner has quit IRC | 00:13 | |
| *** tonanhngo has joined #openstack-kolla | 00:15 | |
| *** tonanhngo has quit IRC | 00:19 | |
| sdake | genpwd doesn't work from a pip install | 00:23 |
|---|---|---|
| sdake | in 4.0.0b3 | 00:23 |
| jascott1 | yikes | 00:24 |
| openstackgerrit | Sayantani Goswami proposed openstack/kolla-ansible: WIP Deployment Guide https://review.openstack.org/427965 | 00:27 |
| *** unicell has quit IRC | 00:34 | |
| hogepodge | sdake: qwang I can moderate. I can also submit if you need me to, but my access to computers is very limited until Monday starting now | 00:35 |
| *** sayantani01 has quit IRC | 00:40 | |
| *** unicell has joined #openstack-kolla | 00:48 | |
| *** jtriley has joined #openstack-kolla | 00:48 | |
| *** yuanying has quit IRC | 00:55 | |
| *** skuda has quit IRC | 00:57 | |
| *** tonanhngo has joined #openstack-kolla | 00:57 | |
| *** tonanhngo has quit IRC | 01:02 | |
| *** openstackgerrit has quit IRC | 01:02 | |
| *** dave-mccowan has quit IRC | 01:05 | |
| *** oomichi has left #openstack-kolla | 01:05 | |
| qwang | hogepodge: no worries I'll submit. I may need your contact information for submission | 01:09 |
| inc0 | kfox1111, around? | 01:10 |
| inc0 | SamYaple, how about you?:P | 01:13 |
| inc0 | around? | 01:13 |
| *** unicell has quit IRC | 01:14 | |
| *** adrian_otto has quit IRC | 01:16 | |
| *** 18WAAG74P has joined #openstack-kolla | 01:20 | |
| *** krtaylor has joined #openstack-kolla | 01:22 | |
| *** inc0 has quit IRC | 01:23 | |
| *** unicell has joined #openstack-kolla | 01:32 | |
| hogepodge | Thanks qwang | 01:38 |
| *** goldyfruit has quit IRC | 01:39 | |
| *** hfu has joined #openstack-kolla | 01:46 | |
| *** tonanhngo has joined #openstack-kolla | 01:49 | |
| *** tonanhngo has quit IRC | 01:50 | |
| *** tonanhngo has joined #openstack-kolla | 01:50 | |
| *** jrobinson has joined #openstack-kolla | 01:59 | |
| SamYaple | inc0 yes | 02:02 |
| *** zhurong has joined #openstack-kolla | 02:08 | |
| *** yuanying has joined #openstack-kolla | 02:08 | |
| *** Jeffrey4l_ has quit IRC | 02:10 | |
| *** unicell has quit IRC | 02:12 | |
| *** Pavo has joined #openstack-kolla | 02:14 | |
| sdake | finally VICtoRY | 02:19 |
| sdake | [root@oscorehost1 kolla-kubernetes]# kubectl get pods -n kolla | 02:19 |
| sdake | NAME READY STATUS RESTARTS AGE | 02:19 |
| sdake | mariadb-0 1/1 Running 0 2m | 02:19 |
| *** zhurong has quit IRC | 02:20 | |
| sdake | bmace_ http://paste.fedoraproject.org/542669/60021081/ -> http://paste.fedoraproject.org/542669/60021081 | 02:22 |
| sdake | 6 minutes to deploy compute kit in kubernetes: http://paste.fedoraproject.org/542672/14860024/ -> http://paste.fedoraproject.org/542672/14860024 | 02:28 |
| bmace_ | sdake: i'm re-working a lot of the dev setup to work start to finish with one command against the halycon vagrant environment. there is a lot of rough code in the gate scripts that hard codes things.. setting up the loop devices on the kubeX nodes for ceph, etc.. but i'll post something when it is ready. maybe friday or very early next week. | 02:36 |
| jascott1 | sdake woot! | 02:36 |
| sdake | bmace_ nice | 02:36 |
| sdake | bmace_ here is something i've got | 02:36 |
| bmace_ | sdake: but i similarly had mariadb / rabbitmq up yesterday | 02:37 |
| bmace_ | sdake: which is nice :) | 02:37 |
| sdake | http://paste.fedoraproject.org/542676/60030801/ -> http://paste.fedoraproject.org/542676/60030801 | 02:38 |
| bmace_ | sdake: might mix in some python scripts into the pure bash we have right now to try to do things like better handling of when services are up in kubernetes.. i know it can be done in bash, but i think it ends up being harder to follow and maintain. | 02:38 |
| sdake | bmace_ i'd recommend just going with the service charts for the moment | 02:38 |
| sdake | bmace_ alot less pokey | 02:38 |
| bmace_ | and the pull_containers thing is just all bad.. | 02:38 |
| bmace_ | sdake: yeah, i was using the ceph service workflow | 02:38 |
| sdake | check out the lsat two pastes | 02:38 |
| bmace_ | sdake: and i agree it is better | 02:39 |
| sdake | one is a fully running compute kit in 6 mins | 02:39 |
| sdake | the other is the script that runs the helm commands to pull the trigger | 02:39 |
| sdake | I have a record of operations I ran to get there | 02:39 |
| sdake | I think its pretty straightforward to turn that into a script | 02:39 |
| sdake | i'd do it now, however, i am ready to the rack | 02:40 |
| sdake | hit the rack that is | 02:40 |
| bmace_ | i find it hard sleeping on a rack. i prefer sacks ;) | 02:42 |
| *** narasimha_SV_ has quit IRC | 02:42 | |
| *** Pavo has quit IRC | 02:43 | |
| sdake | bmace_ here is my bringup history: http://paste.fedoraproject.org/542677/60036591/ -> http://paste.fedoraproject.org/542677/60036591 | 02:47 |
| sdake | bmace_ i think instead of reworking the scripts its alot easier to just start fresh | 02:48 |
| sdake | and not overload the gate scripts with the development environment | 02:48 |
| *** hieulq has joined #openstack-kolla | 02:48 | |
| sdake | given that I've sepnt 10 hours getting the dev scripts operational and managed to get that env setup in under an hour | 02:48 |
| sdake | and the dev scripts are still not operational | 02:48 |
| *** ssurana has quit IRC | 03:02 | |
| *** nihilifer1 has joined #openstack-kolla | 03:03 | |
| *** nihilifer has quit IRC | 03:04 | |
| *** sdake has quit IRC | 03:04 | |
| *** Pavo has joined #openstack-kolla | 03:08 | |
| *** hieulq has quit IRC | 03:09 | |
| *** mgiles has quit IRC | 03:12 | |
| *** Pavo_ has joined #openstack-kolla | 03:12 | |
| *** Pavo_ has quit IRC | 03:28 | |
| *** prameswar has joined #openstack-kolla | 03:36 | |
| *** l4yerffeJ has joined #openstack-kolla | 03:40 | |
| *** Jeffrey4l has joined #openstack-kolla | 03:41 | |
| *** saneax-_-|AFK is now known as saneax | 04:16 | |
| *** openstackgerrit has joined #openstack-kolla | 04:22 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible master: Run nova-manage simple_cell_setup during deploying and upgrading https://review.openstack.org/415472 | 04:22 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible master: Implement nova placement service https://review.openstack.org/427715 | 04:22 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible master: Create a new container when restart a container when necessary https://review.openstack.org/427036 | 04:25 |
| *** mdnadeem has joined #openstack-kolla | 04:29 | |
| *** ssurana has joined #openstack-kolla | 04:40 | |
| *** saneax is now known as saneax-_-|AFK | 04:40 | |
| *** goldyfruit has joined #openstack-kolla | 04:46 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla master: Add openvswitch package for nova-libvirt container https://review.openstack.org/428001 | 04:46 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla stable/mitaka: Fix docker restart policy issue on docker engine 1.13 https://review.openstack.org/424115 | 04:47 |
| *** jrobinson is now known as jrobinson-afk | 04:51 | |
| *** 18WAAG74P has quit IRC | 04:52 | |
| *** tonanhngo has quit IRC | 04:54 | |
| *** zhubingbing has joined #openstack-kolla | 04:55 | |
| *** hfu has quit IRC | 04:59 | |
| *** hfu has joined #openstack-kolla | 05:01 | |
| *** Pavo has quit IRC | 05:01 | |
| *** hfu has quit IRC | 05:01 | |
| *** skramaja has joined #openstack-kolla | 05:01 | |
| *** hfu has joined #openstack-kolla | 05:01 | |
| *** hfu has quit IRC | 05:02 | |
| *** hfu has joined #openstack-kolla | 05:02 | |
| *** hfu has quit IRC | 05:03 | |
| *** Pavo has joined #openstack-kolla | 05:05 | |
| *** jrobinson-afk is now known as jrobinson | 05:16 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible master: Implement nova placement service https://review.openstack.org/427715 | 05:18 |
| *** goldyfruit has quit IRC | 05:26 | |
| *** satyar has joined #openstack-kolla | 05:37 | |
| *** trinaths has joined #openstack-kolla | 05:47 | |
| *** unicell has joined #openstack-kolla | 05:55 | |
| *** tonanhngo has joined #openstack-kolla | 05:56 | |
| *** unicell has quit IRC | 05:57 | |
| *** yuanying has quit IRC | 05:57 | |
| *** unicell has joined #openstack-kolla | 06:01 | |
| *** mdnadeem has quit IRC | 06:06 | |
| *** satyar_ has joined #openstack-kolla | 06:10 | |
| *** yuanying has joined #openstack-kolla | 06:11 | |
| *** gmmaha has quit IRC | 06:11 | |
| *** mdnadeem has joined #openstack-kolla | 06:12 | |
| *** mnaser has quit IRC | 06:12 | |
| *** satyar has quit IRC | 06:13 | |
| *** arslanq has quit IRC | 06:13 | |
| *** prameswar has quit IRC | 06:13 | |
| *** gmmaha has joined #openstack-kolla | 06:13 | |
| *** unicell1 has joined #openstack-kolla | 06:20 | |
| *** unicell has quit IRC | 06:21 | |
| *** japestinho_ has quit IRC | 06:24 | |
| *** japestinho_ has joined #openstack-kolla | 06:24 | |
| *** Jeffrey4l has quit IRC | 06:25 | |
| *** Jeffrey4l has joined #openstack-kolla | 06:26 | |
| *** arslanq has joined #openstack-kolla | 06:26 | |
| *** mnaser has joined #openstack-kolla | 06:37 | |
| *** sudipto_ has joined #openstack-kolla | 06:38 | |
| *** sudipto has joined #openstack-kolla | 06:38 | |
| *** tonanhngo has quit IRC | 06:45 | |
| *** tonanhngo has joined #openstack-kolla | 06:45 | |
| *** clayton has quit IRC | 06:46 | |
| *** tonanhngo has quit IRC | 06:46 | |
| *** jrobinson has quit IRC | 07:00 | |
| *** Pavo has quit IRC | 07:01 | |
| *** saneax-_-|AFK is now known as saneax | 07:05 | |
| *** Pavo has joined #openstack-kolla | 07:06 | |
| *** zhurong has joined #openstack-kolla | 07:13 | |
| *** nihilifer1 is now known as nihilifer | 07:16 | |
| *** bmace_ has quit IRC | 07:20 | |
| *** bmace_ has joined #openstack-kolla | 07:20 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla master: Remove include_header and include_footer in all Dockerfiles https://review.openstack.org/428017 | 07:26 |
| *** prameswar has joined #openstack-kolla | 07:31 | |
| *** unicell1 has quit IRC | 07:36 | |
| *** unicell has joined #openstack-kolla | 07:36 | |
| *** pcaruana has joined #openstack-kolla | 07:39 | |
| *** pcaruana has quit IRC | 07:42 | |
| *** pcaruana has joined #openstack-kolla | 07:43 | |
| *** unicell has quit IRC | 07:49 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible master: [TEST][MASTER][KOLLA-ANSIBLE] https://review.openstack.org/422786 | 07:49 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible master: [wip] enable ceilometer related project in gate https://review.openstack.org/412278 | 07:50 |
| *** tonanhngo has joined #openstack-kolla | 08:02 | |
| *** tonanhngo has quit IRC | 08:04 | |
| *** hfu has joined #openstack-kolla | 08:06 | |
| *** yuanying_ has joined #openstack-kolla | 08:08 | |
| *** yuanying has quit IRC | 08:11 | |
| openstackgerrit | jangpro2 proposed openstack/kolla-ansible master: This feature is supporting for workers config in ironic-api. https://review.openstack.org/428026 | 08:12 |
| *** shardy has joined #openstack-kolla | 08:13 | |
| *** athomas has joined #openstack-kolla | 08:13 | |
| *** zhurong has quit IRC | 08:28 | |
| *** hfu has quit IRC | 08:28 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla master: Using CI mirror host https://review.openstack.org/365721 | 08:36 |
| *** mgoddard_ has joined #openstack-kolla | 08:43 | |
| *** hfu has joined #openstack-kolla | 08:45 | |
| *** egonzalez has joined #openstack-kolla | 08:48 | |
| *** lar_ has joined #openstack-kolla | 08:49 | |
| *** lar__ has joined #openstack-kolla | 08:50 | |
| *** Alexey_Y has quit IRC | 08:52 | |
| *** lar_ has quit IRC | 08:54 | |
| *** matrohon has joined #openstack-kolla | 08:56 | |
| Jeffrey4l | egonzalez, could u review these two patch https://review.openstack.org/415472 https://review.openstack.org/427715 , the latter one depends on the former and unblock the gate. | 09:00 |
| *** Pavo has quit IRC | 09:01 | |
| *** skramaja has quit IRC | 09:01 | |
| * egonzalez looking into PS | 09:02 | |
| *** jmccarthy has quit IRC | 09:04 | |
| *** skuda has joined #openstack-kolla | 09:04 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla master: Using CI mirror host https://review.openstack.org/365721 | 09:04 |
| *** Pavo has joined #openstack-kolla | 09:06 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible master: Create a new container when restart a container when necessary https://review.openstack.org/427036 | 09:08 |
| openstackgerrit | Merged openstack/kolla-ansible master: Fix pull syntax error in cinder https://review.openstack.org/427737 | 09:09 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla master: Using CI mirror host https://review.openstack.org/365721 | 09:11 |
| *** gfidente has joined #openstack-kolla | 09:11 | |
| *** gfidente has quit IRC | 09:11 | |
| *** gfidente has joined #openstack-kolla | 09:11 | |
| *** zioproto has joined #openstack-kolla | 09:14 | |
| *** Jeffrey4l has quit IRC | 09:30 | |
| *** l4yerffeJ has quit IRC | 09:30 | |
| *** l4yerffeJ has joined #openstack-kolla | 09:30 | |
| *** Jeffrey4l has joined #openstack-kolla | 09:31 | |
| openstackgerrit | Merged openstack/kolla stable/newton: Fix the undefined error for vmtp_vm_availability_zone https://review.openstack.org/427762 | 09:34 |
| *** ssurana has quit IRC | 09:37 | |
| *** hrw has quit IRC | 09:37 | |
| *** hrw has joined #openstack-kolla | 09:43 | |
| openstackgerrit | Eduardo Gonzalez proposed openstack/kolla-ansible master: Ignore qemu-img process in cleanup-containers https://review.openstack.org/428061 | 09:50 |
| *** matrohon has quit IRC | 09:51 | |
| *** sambetts_ is now known as sambetts | 09:57 | |
| openstackgerrit | Merged openstack/kolla-ansible master: This feature is supporting for workers config in ironic-api. https://review.openstack.org/428026 | 10:08 |
| *** Serlex has joined #openstack-kolla | 10:10 | |
| *** japestinho has joined #openstack-kolla | 10:11 | |
| *** jascott1 has quit IRC | 10:11 | |
| *** jascott1 has joined #openstack-kolla | 10:12 | |
| *** satyar_ has quit IRC | 10:12 | |
| *** jascott1 has quit IRC | 10:16 | |
| *** japestinho has quit IRC | 10:21 | |
| *** matrohon has joined #openstack-kolla | 10:21 | |
| *** hfu has quit IRC | 10:25 | |
| egonzalez | Jeffrey4l: dunno why gates are complaining about db access for nova_api, in therory should be ussing database.connection values by default | 10:32 |
| *** hfu has joined #openstack-kolla | 10:33 | |
| *** pbourke has quit IRC | 10:33 | |
| *** pbourke has joined #openstack-kolla | 10:34 | |
| *** erlon has joined #openstack-kolla | 10:45 | |
| Jeffrey4l | egonzalez, ubuntu + binary? | 10:55 |
| *** zioproto has quit IRC | 10:56 | |
| egonzalez | Jeffrey4l: oh yeah, forgot what you told me about ubuntu/binary | 10:57 |
| Jeffrey4l | several days ago, nova use "nova_api_cell0" as default database name. | 10:59 |
| Jeffrey4l | egonzalez, end-user may need access placement api to change something. https://review.openstack.org/427715 | 11:00 |
| *** Pavo has quit IRC | 11:01 | |
| egonzalez | Jeffrey4l: thanks | 11:01 |
| Jeffrey4l | egonzalez, https://git.openstack.org/cgit/openstack/openstack-manuals/commit/?id=c3ac94c34fa974e0f6b20647072bae0956714a08 | 11:01 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla master: Using CI mirror host https://review.openstack.org/365721 | 11:04 |
| *** Pavo has joined #openstack-kolla | 11:06 | |
| *** sdake has joined #openstack-kolla | 11:07 | |
| *** l4yerffeJ has quit IRC | 11:09 | |
| *** l4yerffeJ_ has joined #openstack-kolla | 11:09 | |
| *** sdake_ has joined #openstack-kolla | 11:09 | |
| *** sdake_ has quit IRC | 11:11 | |
| *** sdake_ has joined #openstack-kolla | 11:12 | |
| *** sdake has quit IRC | 11:12 | |
| *** matrohon has quit IRC | 11:14 | |
| *** sudipto_ has quit IRC | 11:19 | |
| *** sudipto has quit IRC | 11:19 | |
| *** yuanying_ has quit IRC | 11:20 | |
| *** penfold has joined #openstack-kolla | 11:21 | |
| Jeffrey4l | egonzalez, could u review this again? https://review.openstack.org/415472 | 11:31 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible master: [wip] enable ceilometer related project in gate https://review.openstack.org/412278 | 11:31 |
| egonzalez | Jeffrey4l: i'm testing it | 11:31 |
| Jeffrey4l | cool. thanks. | 11:31 |
| egonzalez | in local centos/source works | 11:32 |
| egonzalez | dont know why sometimes fail in gates and others not | 11:32 |
| Jeffrey4l | i tested centos+source too. | 11:32 |
| Jeffrey4l | this patch do not have placement api. so gate is red. | 11:32 |
| egonzalez | oo right, until the placement api is not merged, compute won't start | 11:33 |
| Jeffrey4l | egonzalez, yes. so the placement patch is almost green ( except ubuntu+binary ) | 11:33 |
| Jeffrey4l | currently, two issue bust the gate. 1. cellv2 is requried and not setup. 2. placement api is requried and not setup. | 11:34 |
| des_consolado | still wrestling with `kolla-kubernetes/tools/secret-generator.py create` today. I've got past the problems with the kolla and kolla config paths not being found because of the perceived root and all that. Now it's moaning that hostvars is undefined. | 11:45 |
| des_consolado | there's a bug for this here, I don't really understand it though https://bugs.launchpad.net/kolla-kubernetes/+bug/1649247 | 11:46 |
| openstack | Launchpad bug 1649247 in kolla-ansible "Secret generate fail with hostvars is undefined for kubernetes orchestration engine" [Critical,In progress] - Assigned to Chung Chih, Hung (lyanchih) | 11:46 |
| des_consolado | oh I think I do I have to do something with the tunnel interface | 11:46 |
| des_consolado | tunnel_interface_address even | 11:47 |
| des_consolado | hm I thought I did this, maybe this means it's not finding the config dir/files properly... | 11:48 |
| *** jtriley has quit IRC | 11:48 | |
| des_consolado | ok so it works when you add tunnel_interface_address: "0.0.0.0" to the config in globals or kolla-kubernetes.yml | 11:51 |
| *** openstackgerrit has quit IRC | 12:03 | |
| *** athomas has quit IRC | 12:04 | |
| *** tonanhngo has joined #openstack-kolla | 12:04 | |
| *** tonanhngo has quit IRC | 12:07 | |
| *** hfu has quit IRC | 12:09 | |
| *** hfu has joined #openstack-kolla | 12:09 | |
| *** hfu has quit IRC | 12:09 | |
| *** hfu has joined #openstack-kolla | 12:10 | |
| *** hfu has quit IRC | 12:10 | |
| *** hfu has joined #openstack-kolla | 12:11 | |
| *** hfu has quit IRC | 12:11 | |
| *** hfu has joined #openstack-kolla | 12:12 | |
| *** hfu has quit IRC | 12:12 | |
| *** hfu has joined #openstack-kolla | 12:12 | |
| *** hfu has quit IRC | 12:12 | |
| *** athomas has joined #openstack-kolla | 12:12 | |
| *** clayton has joined #openstack-kolla | 12:13 | |
| *** dave-mccowan has joined #openstack-kolla | 12:15 | |
| *** rwallner has joined #openstack-kolla | 12:24 | |
| *** rwallner has quit IRC | 12:26 | |
| *** rwallner has joined #openstack-kolla | 12:26 | |
| *** athomas has quit IRC | 12:27 | |
| *** athomas has joined #openstack-kolla | 12:27 | |
| *** eaguilar has joined #openstack-kolla | 12:30 | |
| *** eaguilar has quit IRC | 12:35 | |
| *** eaguilar has joined #openstack-kolla | 12:37 | |
| *** openstackgerrit has joined #openstack-kolla | 12:37 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla master: Using CI mirror host https://review.openstack.org/365721 | 12:37 |
| *** skramaja has joined #openstack-kolla | 12:41 | |
| *** srwilkers has joined #openstack-kolla | 12:49 | |
| srwilkers | morning | 12:49 |
| Jeffrey4l | pbourke, around? | 12:49 |
| Jeffrey4l | morning | 12:49 |
| pbourke | Jeffrey4l: hey | 12:49 |
| Jeffrey4l | could u review this https://review.openstack.org/415472 and its following patch https://review.openstack.org/427715, which unblock the gate. | 12:50 |
| *** satyar has joined #openstack-kolla | 12:51 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible master: Fix typo in cloudkitty register.yml file https://review.openstack.org/428121 | 12:54 |
| *** zioproto has joined #openstack-kolla | 12:58 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible master: Fix gnocchi auth issue https://review.openstack.org/428124 | 13:00 |
| *** jmccarthy has joined #openstack-kolla | 13:01 | |
| *** Pavo has quit IRC | 13:01 | |
| *** zioproto has quit IRC | 13:01 | |
| pbourke | Jeffrey4l: this is strange, are cells mandatory now? | 13:01 |
| Jeffrey4l | mandatory means... | 13:02 |
| pbourke | required | 13:02 |
| Jeffrey4l | aha, pbourke yes. | 13:02 |
| pbourke | I dont see any reference to this in the ocata release notes | 13:02 |
| Jeffrey4l | let me find the link. | 13:02 |
| Jeffrey4l | if not, deploy/upgrade to ocata will raise error. | 13:03 |
| pbourke | I guess ocata is not released yet so no notes as of yet | 13:03 |
| *** eaguilar has quit IRC | 13:04 | |
| *** jmccarthy has quit IRC | 13:05 | |
| *** Pavo has joined #openstack-kolla | 13:05 | |
| Jeffrey4l | pbourke, http://docs.openstack.org/releasenotes/nova/unreleased.html search "Ocata requires that your deployment have created the cell and host mappings in Newton" | 13:06 |
| satyar | :) | 13:06 |
| pbourke | Jeffrey4l: thanks! | 13:06 |
| pbourke | Jeffrey4l: we probably should have referenced this info in the commit message | 13:07 |
| pbourke | actually its not too late, do you mind doing this? | 13:07 |
| Jeffrey4l | hrm. which link should be add? the release note is still in unrelease.html page. | 13:08 |
| Jeffrey4l | it is temporary. | 13:08 |
| pbourke | can you just add the text itself? | 13:08 |
| pbourke | Jeffrey4l: actually dont worry about it, I dont want to remove egonzalez's +2 and slow things down | 13:09 |
| Jeffrey4l | ok if you ask. ( there are several child patch, need rebase them and run the job again. ) | 13:09 |
| pbourke | Jeffrey4l: its ok dont worry about it | 13:09 |
| Jeffrey4l | ok. thanks :) | 13:09 |
| egonzalez | pbourke: i'm here, can add +2 again as soon as it pushed ;) | 13:10 |
| Jeffrey4l | the bug describe this in detail https://bugs.launchpad.net/kolla/+bug/1652580 | 13:10 |
| openstack | Launchpad bug 1652580 in kolla-ansible "nova upgrade failing through kolla-ansible" [Critical,In progress] - Assigned to Jeffrey Zhang (jeffrey4l) | 13:10 |
| Jeffrey4l | As new hosts are added to Nova, the nova-manage cell_v2 discover_hosts command must be run in order to map them into their cell. | 13:11 |
| Jeffrey4l | cell make nova more complicated. when run new nova-compute serivce, need run discover_hosts command again ;( | 13:12 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla master: Using CI mirror host https://review.openstack.org/365721 | 13:19 |
| *** jheroux has joined #openstack-kolla | 13:22 | |
| openstackgerrit | Merged openstack/kolla-ansible master: Run nova-manage simple_cell_setup during deploying and upgrading https://review.openstack.org/415472 | 13:22 |
| *** v1k0d3n has joined #openstack-kolla | 13:24 | |
| *** matrohon has joined #openstack-kolla | 13:26 | |
| *** trinaths has quit IRC | 13:30 | |
| Jeffrey4l | pbourke, placement api fix is blocked by a kolla project patch https://review.openstack.org/427714 , could u review it ;) | 13:32 |
| *** rwallner has quit IRC | 13:39 | |
| *** rwallner has joined #openstack-kolla | 13:42 | |
| *** rwallner has quit IRC | 13:44 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla master: Using CI mirror host https://review.openstack.org/365721 | 13:44 |
| *** rwallner has joined #openstack-kolla | 13:45 | |
| *** lamt has joined #openstack-kolla | 13:47 | |
| *** Bino has joined #openstack-kolla | 13:48 | |
| *** zhurong has joined #openstack-kolla | 13:48 | |
| *** satyar has quit IRC | 13:51 | |
| Bino | Hi, I'm a newbie to kolla. I've already run a basic install with docker-ansible but I would like to experiment Kubernetes Kolla. What is the best option to begin with it, is the doc http://docs.openstack.org/developer/kolla-kubernetes/quickstart.html accurate ? | 13:52 |
| *** eaguilar has joined #openstack-kolla | 13:53 | |
| openstackgerrit | Merged openstack/kolla master: Add openvswitch package for nova-libvirt container https://review.openstack.org/428001 | 13:55 |
| *** eaguilar has quit IRC | 13:56 | |
| *** crushil has quit IRC | 13:56 | |
| *** srwilkers has quit IRC | 13:57 | |
| *** goldyfruit has joined #openstack-kolla | 13:58 | |
| *** srwilkers has joined #openstack-kolla | 13:58 | |
| *** eaguilar has joined #openstack-kolla | 14:01 | |
| *** eaguilar has quit IRC | 14:02 | |
| *** sps has joined #openstack-kolla | 14:03 | |
| *** sps is now known as Guest4444 | 14:04 | |
| *** lrensing has joined #openstack-kolla | 14:04 | |
| *** lamt has quit IRC | 14:04 | |
| *** fguillot has joined #openstack-kolla | 14:06 | |
| *** zhurong has quit IRC | 14:06 | |
| openstackgerrit | Serguei Bezverkhi proposed openstack/kolla-kubernetes master: WIP Adding compute-kit chart https://review.openstack.org/428150 | 14:08 |
| *** eaguilar has joined #openstack-kolla | 14:08 | |
| *** eaguilar has quit IRC | 14:09 | |
| *** eaguilar has joined #openstack-kolla | 14:12 | |
| *** Serlex has quit IRC | 14:13 | |
| *** Serlex has joined #openstack-kolla | 14:13 | |
| *** eaguilar has quit IRC | 14:14 | |
| *** rhallisey has joined #openstack-kolla | 14:14 | |
| openstackgerrit | Serguei Bezverkhi proposed openstack/kolla-kubernetes master: WIP Adding compute-kit chart https://review.openstack.org/428150 | 14:15 |
| *** rwallner has quit IRC | 14:16 | |
| *** zhurong has joined #openstack-kolla | 14:17 | |
| *** crushil has joined #openstack-kolla | 14:19 | |
| *** rwallner has joined #openstack-kolla | 14:19 | |
| *** eaguilar has joined #openstack-kolla | 14:21 | |
| *** eaguilar has quit IRC | 14:22 | |
| *** Jeffrey4l has quit IRC | 14:24 | |
| *** jtriley has joined #openstack-kolla | 14:26 | |
| *** satyar has joined #openstack-kolla | 14:28 | |
| *** eaguilar has joined #openstack-kolla | 14:30 | |
| *** jtriley has quit IRC | 14:31 | |
| *** eaguilar has quit IRC | 14:33 | |
| *** andyjones2001 has joined #openstack-kolla | 14:36 | |
| *** Jeffrey4l has joined #openstack-kolla | 14:36 | |
| *** lamt has joined #openstack-kolla | 14:39 | |
| *** eaguilar has joined #openstack-kolla | 14:40 | |
| *** Jeffrey4l has quit IRC | 14:42 | |
| *** eaguilar has quit IRC | 14:42 | |
| *** crushil has quit IRC | 14:43 | |
| *** crushil has joined #openstack-kolla | 14:44 | |
| *** Jeffrey4l has joined #openstack-kolla | 14:44 | |
| *** m1dev has joined #openstack-kolla | 14:44 | |
| openstackgerrit | Serguei Bezverkhi proposed openstack/kolla-kubernetes master: WIP Adding compute-kit chart https://review.openstack.org/428150 | 14:48 |
| *** krtaylor has quit IRC | 14:48 | |
| *** tonanhngo has joined #openstack-kolla | 14:49 | |
| *** jtriley has joined #openstack-kolla | 14:53 | |
| *** mannamne has joined #openstack-kolla | 14:55 | |
| *** cliles has joined #openstack-kolla | 14:58 | |
| *** m1dev has quit IRC | 14:58 | |
| *** Pavo has quit IRC | 15:01 | |
| *** eaguilar has joined #openstack-kolla | 15:01 | |
| *** sudipto_ has joined #openstack-kolla | 15:03 | |
| *** sudipto has joined #openstack-kolla | 15:03 | |
| *** Pavo has joined #openstack-kolla | 15:05 | |
| *** penfold has quit IRC | 15:08 | |
| *** zhurong has quit IRC | 15:09 | |
| cliles | how are schema changes to the db handled? | 15:13 |
| egonzalez | cliles: depends of the service | 15:16 |
| *** dave-mccowan has quit IRC | 15:18 | |
| cliles | egonzalez: for keystone, I've rebuilt and now getting Unknown column when trying to redeploy | 15:18 |
| *** Bino has quit IRC | 15:18 | |
| egonzalez | rebuild and (upgrade or reconfigure) | 15:19 |
| cliles | I'm wondering if I'm doing it wrong by just redeploying or if I need to be using the upgrade | 15:19 |
| egonzalez | upgrade, otherwise db schema is not sync | 15:19 |
| cliles | ha, yeah | 15:19 |
| egonzalez | using old db schema with new code does not work | 15:19 |
| cliles | ran upgrade and see the migrate log statements | 15:20 |
| cliles | ty | 15:20 |
| *** jmccarthy has joined #openstack-kolla | 15:21 | |
| egonzalez | cliles: are you testing upgrades? | 15:21 |
| jmccarthy | Hmm is this the most up to date image building info ? https://github.com/openstack/kolla/blob/master/doc/image-building.rst | 15:21 |
| *** sp__ has joined #openstack-kolla | 15:21 | |
| cliles | egonzales: I've been rebuilding and redeploying since like aug | 15:21 |
| cliles | seems I've been doing it wrong for a while now.... | 15:21 |
| egonzalez | cliles: help testing upgrades is welcomed :) | 15:22 |
| egonzalez | cliles: have a couple of changes for keystone db migration https://review.openstack.org/#/c/425446/ https://review.openstack.org/#/c/425833/ | 15:23 |
| egonzalez | jmccarthy: think so | 15:25 |
| egonzalez | jmccarthy: found an error? | 15:25 |
| *** sp__ has quit IRC | 15:26 | |
| *** saneax is now known as saneax-_-|AFK | 15:26 | |
| *** eaguilar has quit IRC | 15:26 | |
| *** lamt has quit IRC | 15:26 | |
| jmccarthy | egonzalez: Just with my memory ;) | 15:27 |
| egonzalez | jmccarthy: hehe, nothing changed in the image building for a while iirc. should be fine that guide | 15:28 |
| sdake_ | bmace_ ify our about - have a script that gets kolla-kubernetes running | 15:28 |
| sdake_ | bmace_ its actually a collection of various things | 15:28 |
| sdake_ | bmace_ something to work from | 15:28 |
| *** eaguilar has joined #openstack-kolla | 15:32 | |
| *** adrian_otto has joined #openstack-kolla | 15:33 | |
| openstackgerrit | Serguei Bezverkhi proposed openstack/kolla-kubernetes master: WIP Adding compute-kit chart https://review.openstack.org/428150 | 15:33 |
| Serlex | also interested sdake_ from QA side | 15:33 |
| *** breitz has quit IRC | 15:33 | |
| sdake_ | Serlex could you expand - you mean on the script work? | 15:34 |
| *** dave-mccowan has joined #openstack-kolla | 15:34 | |
| sdake_ | Serlex orspeaking of something else | 15:34 |
| *** breitz has joined #openstack-kolla | 15:34 | |
| Serlex | yes on script work, I probably have no input more testing | 15:34 |
| sdake_ | just got a full from bare metal all the way to openstack on kubernetes in 35 minutes | 15:34 |
| sdake_ | Serlex sorry that sentence didnt' parse - coudl you rephrase | 15:35 |
| des_consolado | what's the deal with tools/setup_kubernetes.sh ? | 15:36 |
| Serlex | Sure. I'm interested in getting kolla-kubernetes running without the pain equivalent to sticking needles into my eyeballs. If you have a script which makes running kolla-kubernetes then i'm very much interested to know more | 15:36 |
| Serlex | if that didn't make sense, then I'm sorry | 15:36 |
| *** lamt has joined #openstack-kolla | 15:37 | |
| des_consolado | having a hard time working out how the positional parameters are being used looking at the script | 15:38 |
| *** mdnadeem has quit IRC | 15:40 | |
| *** sayantani01 has joined #openstack-kolla | 15:43 | |
| *** krtaylor has joined #openstack-kolla | 15:44 | |
| *** rwallner has quit IRC | 15:50 | |
| *** adrian_otto has quit IRC | 15:51 | |
| *** rwallner has joined #openstack-kolla | 15:52 | |
| *** rwallner has quit IRC | 15:55 | |
| *** rwallner has joined #openstack-kolla | 15:55 | |
| *** jtriley has quit IRC | 16:00 | |
| openstackgerrit | Merged openstack/kolla-ansible master: Build deploy guide files and tox environment https://review.openstack.org/426105 | 16:00 |
| kfox1111 | morning | 16:02 |
| sdake_ | portdirect srwilkers ready to go? | 16:02 |
| *** eaguilar has quit IRC | 16:04 | |
| *** jmccarthy has quit IRC | 16:06 | |
| openstackgerrit | Merged openstack/kolla stable/mitaka: Fix the gate failure https://review.openstack.org/427798 | 16:11 |
| openstackgerrit | Merged openstack/kolla stable/mitaka: Fix docker restart policy issue on docker engine 1.13 https://review.openstack.org/424115 | 16:12 |
| *** inc0 has joined #openstack-kolla | 16:14 | |
| inc0 | o/ | 16:14 |
| *** inc0 has quit IRC | 16:14 | |
| *** inc0 has joined #openstack-kolla | 16:15 | |
| kfox1111 | inc0: sorry I missed you yesterday. had meetings go way long. | 16:15 |
| inc0 | np | 16:17 |
| SamYaple | inc0: you ponged my yesterday | 16:17 |
| inc0 | yeah Sam | 16:18 |
| inc0 | soo which image do you use for galera? | 16:18 |
| kfox1111 | yeah. me too. | 16:18 |
| SamYaple | you have to use mariadb:10.1 | 16:18 |
| SamYaple | only 10.1 has galera baked in | 16:18 |
| inc0 | mariadb:10.1 doesn't seem to have wsrep stuff | 16:18 |
| SamYaple | ... its not possible to get it withouth wsrep stuff... | 16:18 |
| SamYaple | its built-in | 16:19 |
| inc0 | kfox1111, wanted to pick up convo where we left off;) | 16:19 |
| inc0 | hold on, let me show you change | 16:19 |
| kfox1111 | inc0: can you wf this please: https://review.openstack.org/#/c/427375/2 | 16:19 |
| kfox1111 | inc0: sure. | 16:19 |
| inc0 | https://review.openstack.org/#/c/427931/ SamYaple | 16:19 |
| inc0 | I'm playing around:) | 16:19 |
| SamYaple | inc0: i saw it | 16:20 |
| sbezverk | kfox1111: morning, could you take a look at the error I am getting, in my local test bed I do not see te same error/ | 16:20 |
| kfox1111 | sbezverk: sure. | 16:20 |
| inc0 | I'll rebuild images with mariadb:10.1 and show you error I'm getting | 16:20 |
| sbezverk | kfox1111: http://logs.openstack.org/50/428150/4/check/gate-kolla-kubernetes-deploy-centos-binary-2-helm-entrypoint-nv/f5a1cdc/console.html#_2017-02-02_15_47_30_225679 | 16:21 |
| *** Guest4444 has quit IRC | 16:21 | |
| inc0 | kfox1111, so I don't think we'll be able to have generic kolla image, but infra images can be much lighter with this approach | 16:22 |
| kfox1111 | sbezverk: thats really weird. | 16:22 |
| *** rwallner has quit IRC | 16:22 | |
| inc0 | also no way to build in dockerhub for now because of jinja | 16:22 |
| kfox1111 | inc0: I just commented on that. I think there's a way. | 16:23 |
| sbezverk | kfox1111: I am thinking that too :( no clue why it is barking | 16:23 |
| inc0 | SamYaple, mysql: unknown variable 'exececute=SHOW STATUS LIKE 'wsrep_ready'' | 16:23 |
| kfox1111 | if we made an option to spit out the rendered docker files into seperate directories, and put it in a zuul merge hook, we could push them in a way docker hub could pull from them. | 16:24 |
| inc0 | yeah, build like that would work | 16:24 |
| kfox1111 | sbezverk:maybe tiller crashed? | 16:24 |
| inc0 | if we have all the scripts curl-able | 16:24 |
| kfox1111 | yeah. | 16:25 |
| kfox1111 | we could make a kolla-docker-hub repo or something that the hub pulls prerenderd stuff from. | 16:25 |
| openstackgerrit | Merged openstack/kolla-ansible master: Fix genconfig for kolla-kubernetes https://review.openstack.org/427375 | 16:25 |
| kfox1111 | so the source is usable by devs to build stuff easily, and for ops to rebuild stuff from scratch, | 16:25 |
| kfox1111 | or the hub git stuff is just a cached partial rendering that the hub can use to build. | 16:25 |
| *** eaguilar has joined #openstack-kolla | 16:26 | |
| kfox1111 | sbezverk: oh. wait. | 16:27 |
| kfox1111 | I think i may know... | 16:27 |
| inc0 | hmm | 16:28 |
| inc0 | I know what I did | 16:28 |
| inc0 | and I'm ashamed | 16:28 |
| SamYaple | inc0: ill have a look at the review | 16:29 |
| inc0 | execucute is really cute | 16:29 |
| inc0 | let me fix this one and try to get it to deployable state, I'm fairly close | 16:29 |
| kfox1111 | sbezverk: ok, here's the deal... | 16:30 |
| inc0 | also these builds in much less than 1min so that's a bonus;) | 16:30 |
| kfox1111 | I think its helm_build_all maybe. it stops running the repo at the end of the build process. | 16:31 |
| SamYaple | inc0: you aren't overriding the ENTRYPOINT? | 16:31 |
| inc0 | I am now | 16:31 |
| inc0 | this isn't up to date | 16:31 |
| inc0 | I'll post working ps later today I hope | 16:31 |
| *** prameswar has quit IRC | 16:32 | |
| SamYaple | kfox1111: inc0: like i said, i have a much better build system in https://github.com/yaodu for cicd. builds from scratch are under a minute _and_ it can be built from patchset references | 16:33 |
| SamYaple | im more than happy to bring it to kolla, its the repo build thing i was refering too | 16:33 |
| SamYaple | _but_ it does mean you would have to deprecate things | 16:33 |
| SamYaple | on the up side, those images build in dockerhub and are ultra tiny | 16:33 |
| sbezverk | kfox1111: I just do not see what is the big difference, why all of a sudden it started failing | 16:34 |
| *** adrian_otto has joined #openstack-kolla | 16:34 | |
| kfox1111 | SamYaple: +1 to tiny builds. and hub builds. would like to explore how an op could build it all themselves though. is that still an option? Is it easy to make chagnes that touch all the containers or do you have to go do a bunch of work? what are the features that will have to be depricated? | 16:35 |
| kfox1111 | sbezverk: on your host, maybe you had the helm repo server running the whole time? | 16:35 |
| *** jtriley has joined #openstack-kolla | 16:35 | |
| *** mannamne has quit IRC | 16:35 | |
| sbezverk | kfox1111: yes I do have it running all the time | 16:35 |
| inc0 | so overrides is something I wouldn't be too fond of leaving out | 16:35 |
| kfox1111 | sbezverk: yeah. thats one thing I think may be different between your system and the gate. | 16:36 |
| inc0 | I'd rather do what kfox1111 suggested - make images tiny but keep jinja2 | 16:36 |
| kfox1111 | if you did the helm package build in helm_build_all.sh I think it might work? | 16:36 |
| inc0 | also not having openstack-base will make overall build much longer | 16:36 |
| sbezverk | kfox1111: I usually do 4 separate steps in my local cluster | 16:36 |
| SamYaple | kfox1111: right now, since you all opened up templating to use configs to the Nth degree, thats a feature that will need to go through the deprecation process | 16:37 |
| kfox1111 | I'm ok having a much longer build, as its not done frequiently, and ideally would be ci'ed, so people don't notice it so much. | 16:37 |
| SamYaple | inc0: youre wrong. i build keystone in under a minute. dockerhub does it in 6 minutes | 16:37 |
| SamYaple | kfox1111: builds _should_ be done frequently for CI/CD stuff | 16:38 |
| inc0 | SamYaple, but we have much more images | 16:38 |
| kfox1111 | SamYaple: yeah. agreed. but its not something a user will be sitting and waiting on, was my point. | 16:38 |
| inc0 | and these won't reuse much | 16:38 |
| kfox1111 | it just "happens" | 16:38 |
| inc0 | and yes, you can setup apt-cache and stuff | 16:38 |
| SamYaple | inc0: no, all openstack services will build in about a minute | 16:38 |
| inc0 | each will build under a miute | 16:39 |
| SamYaple | yep | 16:39 |
| SamYaple | unless it has _alot_ of packages (nova-compute qemu stuff) | 16:39 |
| inc0 | all of them combined will build in much more than if they share common base | 16:39 |
| portdirect | o/ | 16:39 |
| SamYaple | inc0: no. youre forgetting parallel | 16:39 |
| kfox1111 | hey portdirect | 16:39 |
| SamYaple | running in parallel, it will be _faster_ than kolla. build hardware depending | 16:39 |
| SamYaple | plus, now it can be distributed to a build farm | 16:40 |
| portdirect | On call so only reading tiny context | 16:40 |
| portdirect | I can build all yaodu images in under 12 mins | 16:40 |
| SamYaple | and thats in serial :P | 16:41 |
| SamYaple | kolla serial is well over an hour (albiet, far more images) | 16:41 |
| portdirect | Correct | 16:41 |
| SamYaple | inc0: given the nature of the unlinked images, you could build _all_ services at the same time | 16:41 |
| inc0 | or you could have common base image with python deps | 16:41 |
| inc0 | but regardless | 16:41 |
| SamYaple | which means you can't CICD well | 16:42 |
| inc0 | this is just testing | 16:42 |
| SamYaple | that was the whole things | 16:42 |
| inc0 | SamYaple, you could if you build them with squash | 16:42 |
| SamYaple | that doesnt help.... | 16:42 |
| SamYaple | if you want to build keystone, you have to take 15minutes out of your day in kolla | 16:42 |
| inc0 | anyway, again, matter of testing and research | 16:42 |
| SamYaple | i mean, testing shows kolla is pretty awful at CICD | 16:43 |
| SamYaple | and having to build _ALL_ images at the same time is bad | 16:43 |
| SamYaple | but i dont know how to take kolla to the next level with all the deprecation thats needed | 16:43 |
| kfox1111 | +1 for more ci/cd support in kolla. | 16:43 |
| inc0 | yeah, we'd effectively deprecate base_distro | 16:44 |
| inc0 | and I'm pretty ok with that tbh | 16:44 |
| SamYaple | you'd need to get rid of _all_ layers | 16:44 |
| SamYaple | layers are whats killing CICD | 16:44 |
| kfox1111 | why have to depricate distro support? or am I misunderstanding? | 16:44 |
| inc0 | Sam wants to get rid of jinja2 | 16:44 |
| portdirect | kfox1111: we dont have to do that - but the layering nature of kolla kills it for ci/cd pipeline | 16:45 |
| inc0 | so all images are dockerhub-buildable | 16:45 |
| SamYaple | more accurately, i want to get rid of external tooling | 16:45 |
| kfox1111 | portdirect: why? is it due to the lack of dep suport in ci/cd systems? or somethign else? | 16:45 |
| inc0 | and without jinja2, distro support becomes pretty hard | 16:45 |
| SamYaple | kfox1111: you can't publish an image that was built in CICD | 16:45 |
| portdirect | ^^ | 16:45 |
| kfox1111 | SamYaple: why not? | 16:45 |
| SamYaple | kfox1111: because it has its own unique base and openstack-base layer | 16:46 |
| SamYaple | so if you build keystone, then glance on a different node, they _DONT_ share layers | 16:46 |
| kfox1111 | so? you can docker push that? | 16:46 |
| portdirect | the tracability becomes very hard - as dow the need for extrnal tooling to build | 16:46 |
| * portdirect keyboard mashing as usual | 16:46 | |
| SamYaple | ok. now kolla has images that are over 800MB. no sharing | 16:46 |
| inc0 | kfox1111, if you build base in 2 different ci runs | 16:46 |
| kfox1111 | ok, but then why not bild the shared image first, then the rest in parallel? | 16:46 |
| inc0 | and one breaks another | 16:46 |
| *** rwallner has joined #openstack-kolla | 16:46 | |
| SamYaple | kfox1111: but only keystone has changed, anytime keystone changes you have to rebuild and REPUSH all images | 16:47 |
| inc0 | they can't share anything if you want auto-uploaded per-image cicd | 16:47 |
| kfox1111 | no, not saying everythign needs to be in the base image. | 16:47 |
| kfox1111 | but like, python maybe. | 16:47 |
| SamYaple | inc0: well thats not true. they can | 16:47 |
| SamYaple | kfox1111: really, im telling you the overhead is tiny | 16:47 |
| inc0 | not in 100% reliable way | 16:47 |
| SamYaple | the yaodu layer for keystone is under 40MB | 16:47 |
| SamYaple | inc0: you really can | 16:47 |
| SamYaple | inc0: dockerhub has triggers | 16:48 |
| kfox1111 | SamYaple: if you have 10 containers on a controller, how much does that blow up though? | 16:48 |
| SamYaple | kfox1111: ~40MB per service (or far far less than kolla) | 16:48 |
| inc0 | SamYaple, on debian tho | 16:48 |
| SamYaple | kfox1111: overall, it uses significanlty less space than kolla | 16:48 |
| kfox1111 | is taht due to broken layering in kolla though? | 16:48 |
| SamYaple | inc0: there is ubuntu and centos as well | 16:48 |
| inc0 | ubuntu and centos are 200+MB | 16:48 |
| SamYaple | inc0: youre wrong | 16:48 |
| kfox1111 | ie, could it be even less if more sharing happened? | 16:48 |
| SamYaple | inc0: images sizes are here https://github.com/yaodu/docker-keystone | 16:49 |
| srwilkers | morning o/ | 16:49 |
| portdirect | hey o/ | 16:49 |
| bmace_ | sdake_: around? you can email or pastebin or use whatever mechanism to get me that script you were mentioning? i'll give it a look. | 16:49 |
| SamYaple | kfox1111: the space could potentially be less, yes. | 16:49 |
| inc0 | ubuntu is 129MB | 16:49 |
| portdirect | bmace_: on call with sdake now | 16:49 |
| inc0 | according to docker | 16:49 |
| SamYaple | kfox1111: however, the space savings of that for breaking ci/cd is no good | 16:49 |
| inc0 | pure ubuntu 16.04 | 16:49 |
| kfox1111 | SamYaple: still not sure why it breaks ci? | 16:50 |
| inc0 | centos 191 | 16:50 |
| SamYaple | inc0: compressed vs uncompressed. if you want to compare those sizes Kolla == 1GB for keystone. yaodu images == 300MB | 16:50 |
| kfox1111 | SamYaple: is the ci your using not suport dependencies? | 16:50 |
| inc0 | SamYaple, but kolla has *a lot more* images today | 16:50 |
| SamYaple | kfox1111: you can't build one service independant from another. so if keystone has a change, to use that keystone change you must rebuild _ALL_ images | 16:50 |
| SamYaple | inc0: no, 1GB PER SERVICE | 16:50 |
| SamYaple | inc0: keystone is 900MB build on my system | 16:50 |
| inc0 | with a lot of sharing tho | 16:50 |
| kfox1111 | only if keystone is in the base image? | 16:51 |
| SamYaple | inc0: not in cicd :) | 16:51 |
| inc0 | point taken | 16:51 |
| kfox1111 | why would a common layer have to change if keystone changes. thats whats confusing me. | 16:51 |
| SamYaple | kfox1111: because the common layer has things like openstack-requirements | 16:51 |
| inc0 | but regardless, if we do it in prod | 16:51 |
| inc0 | it will be bigger space eater | 16:51 |
| kfox1111 | SamYaple: ah. well, what if it diddnt? | 16:51 |
| SamYaple | kfox1111: then you wouldnt need a common layer | 16:52 |
| kfox1111 | SamYaple: but you still have a bunch of stuff that really is shared? like python and a bunch of common python stuff? | 16:52 |
| inc0 | anyway, regardless, without jinja2 we can't have multi-distro imho | 16:52 |
| SamYaple | inc0: are you saying yaodu images will be more space than kolla in total? | 16:52 |
| openstackgerrit | Merged openstack/kolla master: Fix mistral _50_mistral.py location https://review.openstack.org/417301 | 16:52 |
| SamYaple | because thats blatantly wrong too | 16:52 |
| openstackgerrit | Merged openstack/kolla stable/newton: Fix typo heat.conf https://review.openstack.org/427764 | 16:52 |
| kfox1111 | I still think that can be managed in a seperate step. | 16:52 |
| *** rwallner has quit IRC | 16:52 | |
| inc0 | SamYaple, idk, I never tested it, I'm saying deployed infra might | 16:53 |
| pbourke | inc0: we could do the sed stuff. but unfortunately its too late to change I think | 16:53 |
| kfox1111 | ie, do a jinja step to make manging all the options easy and do less code, then render the Docker files and use those for ci? | 16:53 |
| SamYaple | inc0: it really one. the entire yaodu image with _ubuntu_ base is still less than the single keystone layer in kolla | 16:53 |
| SamYaple | kfox1111: thats doable for sure. some other projects have done simliar things | 16:53 |
| inc0 | pbourke, not it this context I think, you can't sed image you're building | 16:53 |
| pbourke | inc0: you should take a look at what SamYaple is doing in yaodu. Im not saying we can use it but its very interesting | 16:54 |
| SamYaple | kfox1111: generating the dockerfile on commit is something official docker iamges do, albeit with a bash script. and not normally entire sections of code (just versions) | 16:54 |
| inc0 | well, ok, I grant you that, it will share base distro image which is biggest part of it | 16:54 |
| SamYaple | inc0: no you misunderstand | 16:54 |
| *** jtriley has quit IRC | 16:54 | |
| inc0 | pbourke, I did and this is what I'm working on now:P https://review.openstack.org/#/c/427931/ | 16:54 |
| SamYaple | keystone kolla layer is greater than yaodu keystone in total | 16:54 |
| inc0 | ok | 16:55 |
| inc0 | I wonder why tho | 16:55 |
| SamYaple | look, my point is, these images are what kolla _NEEDS_ to get to. but frankly, i don't see a path forward | 16:55 |
| SamYaple | inc0: single layer, removed all cruft. no build tools | 16:55 |
| SamYaple | lots of reasons | 16:55 |
| inc0 | not without deprecating whole bunch of stuff, which will have to happen at some point | 16:55 |
| *** rwallner has joined #openstack-kolla | 16:56 | |
| SamYaple | inc0: hey man, you want to put it on the roadmap, ill work with you | 16:56 |
| *** rwallner has quit IRC | 16:56 | |
| SamYaple | i would love to see kolla switch to this image model | 16:56 |
| inc0 | we will have session on PTG | 16:56 |
| *** rwallner has joined #openstack-kolla | 16:56 | |
| SamYaple | it can be built automatically in dockerhub _and_ its much much faster | 16:56 |
| inc0 | I recognize that we need to rething our images | 16:56 |
| srwilkers | inc0, no ones saying it needs to happen overnight. but i agree with SamYaple that it should be on the roadmap in the future | 16:56 |
| kfox1111 | what must be depricated? | 16:57 |
| inc0 | template overrides comes to mind for sure | 16:57 |
| kfox1111 | +1 to talking about it more at the ptg. | 16:57 |
| SamYaple | kfox1111: so right now, you can right a single macro and change a whole much of stuff in the base image | 16:57 |
| srwilkers | kfox1111, agree. +1 to PTG | 16:57 |
| SamYaple | well, if no common base is shared.... you can't do that | 16:57 |
| kfox1111 | we were going to talk about ci/cd at the ptg nayway. this would be a good thing to talk about too, as it affects ci/cd | 16:57 |
| kfox1111 | SamYaple: I think you might still? | 16:58 |
| openstackgerrit | Merged openstack/kolla master: Add nova-placement-api container https://review.openstack.org/427714 | 16:58 |
| kfox1111 | you just have to make sure you can aplpy the base template change to all the containers? | 16:58 |
| kfox1111 | rather then the base container? | 16:58 |
| kfox1111 | ie, if you did the jinja2 -> docker file thing, | 16:58 |
| SamYaple | kfox1111: true, but that could kill some peoples cicd (doing the same step for each image) | 16:58 |
| kfox1111 | you could just include the base override macro in all the individual templates? | 16:58 |
| inc0 | yeah, we could keep lots of stuff because of jinja | 16:58 |
| SamYaple | im just saying, its a major change that might be hard to fix now that jinja2 is even more embrassed | 16:58 |
| kfox1111 | but wouldn't need a base container naymore. | 16:59 |
| SamYaple | true enough. and thats the way to go | 16:59 |
| inc0 | SamYaple, I don't think removing all the external tooling is feasable for general use case | 16:59 |
| SamYaple | along with not carrying rabbitmq and mariadb anymoe | 16:59 |
| inc0 | like, RH people won't want to deploy debian based containers | 16:59 |
| SamYaple | inc0: i mean, it is | 16:59 |
| kfox1111 | yeah. I like the fact I can kubectl exec into a container, and know its layout better, as its redhat based. | 17:00 |
| SamYaple | inc0: but they do! mariadb:10.1 | 17:00 |
| kfox1111 | yeah, I understand ubuntu, but it still takes longer to learn how to admin more then one distro. | 17:00 |
| *** athomas has quit IRC | 17:00 | |
| qwang | Please feel free to comment or revise. https://etherpad.openstack.org/p/openstack-on-helm | 17:01 |
| *** Pavo has quit IRC | 17:01 | |
| inc0 | SamYaple, what I'm saying, we can tell them "don't worry about it" | 17:01 |
| qwang | alanmeadows: hogepodge ^^^^^ | 17:01 |
| kfox1111 | SamYaple: all in all, I think the idea is a good one. I think we can maybe refine the idea so that we don't loose so many features. I don't think they are multually exclusive ideas. | 17:01 |
| inc0 | we can break this down to - thin up images and remove external tooling | 17:02 |
| *** rwallner has quit IRC | 17:02 | |
| inc0 | these are 2 distinct steps that can be handled separately | 17:02 |
| kfox1111 | which "external tooling" are you talking about? the jinja2 stuff? | 17:02 |
| inc0 | build.py | 17:02 |
| SamYaple | kfox1111: yea. build.py | 17:02 |
| kfox1111 | ah. yeah. | 17:03 |
| kfox1111 | I think that can be split up simlarly to what were doing in kolla-kubernetes. | 17:03 |
| kfox1111 | have a prebuild step and a build step. | 17:03 |
| SamYaple | kfox1111: i dont think there is one or the other going on, but kolla is _REALLY_ far away from anything that resembles a docker project at the moment | 17:03 |
| inc0 | I mean without layers | 17:03 |
| *** jtriley has joined #openstack-kolla | 17:03 | |
| kfox1111 | prebuild just renders the templates to something that can then be used natively. | 17:03 |
| kfox1111 | like docker hub. | 17:03 |
| kfox1111 | or ci | 17:03 |
| inc0 | we can use build.py to render self-contained dockerfiles | 17:03 |
| kfox1111 | build is a script for those that don't want to do that. | 17:03 |
| openstackgerrit | Eduardo Gonzalez proposed openstack/kolla-ansible master: Integrate designate with neutron https://review.openstack.org/424679 | 17:03 |
| kfox1111 | SamYaple: yeah, and I think in some ways, kolla makes it much easer not to have a security incident. :/ | 17:04 |
| *** japestinho_ has quit IRC | 17:04 | |
| kfox1111 | caue you can build a whole, patched set of containers at once. :/ | 17:04 |
| kfox1111 | docker hub doesn't give you any mechanism to keep packages up to date in your containers. :/ | 17:04 |
| SamYaple | kfox1111: i think rebuilding and redeploying daily (without downtime or lost packets) is a better way | 17:04 |
| kfox1111 | SamYaple: I agree. | 17:05 |
| SamYaple | kfox1111: it realyl does. you can trigger a rebuild if ubuntu:latest changes | 17:05 |
| inc0 | SamYaple, yes...for 1% of people th | 17:05 |
| inc0 | 99% will deploy when needed | 17:05 |
| kfox1111 | SamYaple: no, there are still use cases where that fails. | 17:05 |
| kfox1111 | docker autobuild watches for git changes. | 17:05 |
| *** Pavo has joined #openstack-kolla | 17:05 | |
| SamYaple | kfox1111: this is different | 17:06 |
| kfox1111 | but if your container depends on a rpm package, | 17:06 |
| kfox1111 | it can't notice that. | 17:06 |
| *** sdake has joined #openstack-kolla | 17:06 | |
| kfox1111 | ie, if the next openssl volnerability gets found/fixed, | 17:06 |
| inc0 | kfox1111, but thats cron-able | 17:06 |
| kfox1111 | there isn't an easy way for containers to notice, and rebuild. | 17:06 |
| inc0 | redeploy dailt even if nothing changed in git | 17:06 |
| sdake | portdirect can you finish the submission - my intenet imploded | 17:06 |
| des_consolado | guys could you tell me what you are using to provision hosts with kubernetes before deploying kolla-kubernetes? are you using this setup_kubernetes.sh at all? | 17:06 |
| pbourke | kfox1111: I think most will want to build with jenkins | 17:07 |
| kfox1111 | inc0: I think openstack's far away from a seamless redeploy. :/ | 17:07 |
| *** mattmceuen has joined #openstack-kolla | 17:07 | |
| SamYaple | kfox1111: ubuntu:latest puts out an update and it triggers a rebuild of keystone. but wait. you can always manually rebuild too with a click of a button | 17:07 |
| inc0 | des_consolado, this is only for gates | 17:07 |
| inc0 | I personally use project called kargo | 17:07 |
| SamYaple | kfox1111: thats all around better than kolla which is all manual | 17:07 |
| kfox1111 | pbourke: +1 | 17:07 |
| kfox1111 | pbourke: or some other ci system. I'm all for that. | 17:07 |
| *** sdake_ has quit IRC | 17:07 | |
| kfox1111 | SamYaple: yeah, but when you have 70 containers, clicking on each is a PAIN. :/ | 17:07 |
| kfox1111 | it needs to be ci friendly. :/ | 17:08 |
| SamYaple | kfox1111: .... these all accept POSTs for rebuilts | 17:08 |
| des_consolado | inc0: you mean setup_kubernetes.sh is for gates? | 17:08 |
| SamYaple | CI friendly rarely if ever means "manual" | 17:08 |
| inc0 | correct des_consolado | 17:08 |
| kfox1111 | SamYaple: we really need: | 17:08 |
| *** egonzalez has quit IRC | 17:08 | |
| kfox1111 | a way a ci job can check to see if a container needs updates ('yum upgrade makes changes?) | 17:08 |
| kfox1111 | then triggers a rebuild only if something would change. | 17:09 |
| SamYaple | i disagree. things change so quickly you can just assume something changed | 17:09 |
| kfox1111 | then once the containers are built, it cann trigger a helm upgrade of packages that refverence that container in koloa-kubernetes. | 17:09 |
| SamYaple | openstack-requirements changed? rebuild | 17:09 |
| kfox1111 | SamYaple: rarely. | 17:09 |
| SamYaple | new day has passed? rebuild | 17:09 |
| kfox1111 | SamYaple: ah. I see where your coming from. | 17:09 |
| kfox1111 | SamYaple: you are assuming someone will want to run openstack from trunk. | 17:09 |
| SamYaple | kfox1111: im really not | 17:10 |
| kfox1111 | so changes are so frequent that you just follow trunk. | 17:10 |
| SamYaple | there are updates almost everyday _somewhere_ | 17:10 |
| SamYaple | im talking about stable | 17:10 |
| kfox1111 | rarely that often. | 17:10 |
| kfox1111 | redhat is monthlyish. | 17:10 |
| SamYaple | i forget. we are also talking binary deploy vs source deploy, right? | 17:11 |
| kfox1111 | yeah. I use binary deploys based on rdo. | 17:11 |
| SamYaple | because openstack-requiremetns changes often enough on stable | 17:11 |
| SamYaple | yea i guess thats going ot be adifferent | 17:11 |
| SamYaple | honestly, i havent even attempted to look into binary stuff in those yaodu images | 17:11 |
| SamYaple | as it is now, i wouldnt even let them merge | 17:12 |
| SamYaple | using the same structure in kolla would be different | 17:12 |
| kfox1111 | so, thats one thing we will have to resolve. | 17:12 |
| pbourke | i was thinking yaodu images could be offered as a drop in alternative in kolla. The main work really is in ansible/k8s to bind in the start scripts at runtime rather than expecting them to be baked in | 17:12 |
| pbourke | i almost got this working with keystone | 17:13 |
| kfox1111 | uh, | 17:13 |
| kfox1111 | that is kind of counter to docker philosophy. | 17:13 |
| *** tonanhngo has quit IRC | 17:13 | |
| inc0 | pbourke, we need to host openstack service images in kolla | 17:13 |
| kfox1111 | the container should contain all the software needed. | 17:13 |
| kfox1111 | poking software into the container at runtime's really ugly. :/ | 17:13 |
| inc0 | we are meant to be source of services | 17:13 |
| pbourke | SamYaple: ^ ? | 17:13 |
| inc0 | I'm ok with not hosting mariadb and stuff like that | 17:13 |
| kfox1111 | I'd rather inherit from the containers and poke stuff in then poke them in at runtime. | 17:14 |
| inc0 | but openstack - that's us. | 17:14 |
| SamYaple | kfox1111: disagree. the official docker images often say "just bind in an override entrypoint" | 17:14 |
| kfox1111 | like, copy this binary in and run it? | 17:14 |
| SamYaple | yea | 17:14 |
| kfox1111 | thats really ugly. :/ | 17:14 |
| kfox1111 | how do you know the two sofware is compatible? | 17:15 |
| SamYaple | kfox1111: some projects have a goal of agnostic images. so you can bring your own, bring your redhat licensed private image build, it doesnt matter | 17:15 |
| kfox1111 | docker's claim to faime, is the ability to build someting once, and deploy the same tested thing everywhere. | 17:15 |
| SamYaple | the only way to make that work is by treating it like a binary, and you bringing the service script | 17:15 |
| kfox1111 | that elminates that. :/ | 17:15 |
| SamYaple | it really doesnt | 17:15 |
| pbourke | the image is still the binary. the script is just starting it | 17:16 |
| SamYaple | that 'deploy the same tested thing' doesnt include configs | 17:16 |
| kfox1111 | so, I really want atomic images. they never change, they are tested before deployed, and garenteed to be the same running on the production system as the test system. | 17:16 |
| SamYaple | and these could be considered configs and versioned as such (ie, kolla-ansible 4.0.0) | 17:16 |
| kfox1111 | a x86_64 binary is not a config file though. | 17:16 |
| kfox1111 | thats stretching the analogy too far I think. | 17:17 |
| SamYaple | its pretty darn close with all the versioning involved | 17:17 |
| kfox1111 | I'd much rateher just inherit from the parent container and add the binary into a new build. | 17:17 |
| kfox1111 | then you can tag it/ pull it everywhere. | 17:17 |
| SamYaple | kfox1111: also the config file i hope | 17:17 |
| SamYaple | using only ENV vars to add anything new | 17:18 |
| kfox1111 | env doesn't scale. :/ | 17:18 |
| SamYaple | because _THATS_ the docker model | 17:18 |
| SamYaple | ENV vars are, however, controlable | 17:18 |
| kfox1111 | dockers phylosophy on evn is the same the config management tools took. | 17:18 |
| SamYaple | so you can control the configuration thats running | 17:18 |
| kfox1111 | "we provide an abstraction that you must write a ton of code for to map the abstraction to the underlying config system of the program" | 17:19 |
| kfox1111 | and its rare its fully configurable. :/ | 17:19 |
| SamYaple | i know... this is my point. the docker model is flawed for complex applications | 17:19 |
| kfox1111 | theres a huge amount of chef/puppet code to try and do that, and they all are incomplete. :/ | 17:19 |
| SamYaple | you can't pick and choose here | 17:19 |
| kfox1111 | agreed. | 17:19 |
| kfox1111 | some things they got right, some things they got wrong. | 17:20 |
| kfox1111 | the build/deploy seperation they got right I think. | 17:20 |
| SamYaple | anyway. official docker has recommended 'bind entrypoint script in at run time' in the past. its not nearly as frowned on as it was intiially | 17:20 |
| kfox1111 | the config via envs, they got wrong. config file passing in by bind mount works though. | 17:20 |
| SamYaple | that dependso n who you ask | 17:20 |
| SamYaple | the ENVs make config ultra controllable | 17:20 |
| SamYaple | which some people want | 17:20 |
| SamYaple | it fits some models well | 17:20 |
| kfox1111 | yeah. it could work, if its a very simple thing. but I don't think generally it holds. | 17:21 |
| SamYaple | ive got to head out. good conversation as normal | 17:21 |
| kfox1111 | yeah. when you don't have very much configurabiliyt, the mapping code isn't hard to write. | 17:21 |
| des_consolado | are any of you guys using kolla-kubernetes in production? I guess not right? ;p | 17:21 |
| alanmeadows | qwang: what is social summary? | 17:21 |
| SamYaple | i land firmly in bind-in-startup-scripts though | 17:21 |
| openstackgerrit | Eduardo Gonzalez proposed openstack/kolla-ansible master: Integrate designate with neutron https://review.openstack.org/424679 | 17:21 |
| kfox1111 | des_consolado: I'm working on deploying it. | 17:21 |
| kfox1111 | SamYaple: and I disagree. and thats ok. :) | 17:21 |
| SamYaple | :) agree | 17:21 |
| kfox1111 | the startup scripts are inimately tied to the continer. | 17:22 |
| kfox1111 | so belong in the container. | 17:22 |
| kfox1111 | so long as both use cases can be handled, I'm good. :) | 17:22 |
| qwang | alanmeadows: I think it's about the purpose of the talk to the community | 17:22 |
| * portdirect reading scollback, in boring meeting | 17:22 | |
| kfox1111 | I think my use case can stil be handled by just inheriting from your containers. | 17:22 |
| alanmeadows | qwang: so expand on abstract essentially | 17:22 |
| qwang | alanmeadows: right | 17:23 |
| openstackgerrit | Eduardo Gonzalez proposed openstack/kolla-ansible master: Integrate designate with neutron https://review.openstack.org/424679 | 17:24 |
| kfox1111 | SamYaple: so, I think kolla compatible images could be built from yaodu images by just inheriting from each one and adding the kolla_start stuff on top? | 17:24 |
| pbourke | do we end up with the layering problem again then? | 17:25 |
| srwilkers | i say we shelve a lot of this discussion for the PTG. definitely something that needs quite a bit of attention overall | 17:26 |
| kfox1111 | pbourke: would be a 1:1 mapping, so would be easier to solve. | 17:27 |
| kfox1111 | srwilkers: +1. | 17:27 |
| pbourke | kfox1111: agree | 17:27 |
| *** sdake_ has joined #openstack-kolla | 17:27 | |
| pbourke | would like to resolve whos right on the config stuff though | 17:27 |
| portdirect | kfox1111: that would work, though like SamYaple I'm firmly in favour of binding startup script at container lauch, this allows the orchestration platform to take full ownership of the infra | 17:27 |
| pbourke | s/config/start script/g | 17:27 |
| kfox1111 | portdirect: I'm worried about incompatabilites. :/ | 17:28 |
| kfox1111 | it would work well with shell. | 17:28 |
| portdirect | and config as well for that matter - the only think i want to rely on being present in the container is the binaies that are required to run the service itself | 17:28 |
| kfox1111 | it would be maybe ok with go statically linked binaries. | 17:28 |
| *** sdake has quit IRC | 17:28 | |
| inc0 | SamYaple, do you know if anyhthing changed in bootstrap method between galera we had and one from image? | 17:28 |
| *** jascott1 has joined #openstack-kolla | 17:28 | |
| kfox1111 | the rest, I'm really nervious the container will lack a pip package, or a c library, or whatever. | 17:28 |
| inc0 | wsrep_ready never gets ON | 17:28 |
| portdirect | kfox1111: if scoped correctly thats no problem | 17:29 |
| kfox1111 | it would be way easier if kolla_start was a go binary. :/ | 17:30 |
| portdirect | I'm working toward using containers that don't contain a shell at all, the only assumption being that the python stdlib will be present | 17:30 |
| portdirect | kfox1111: i hope that was a joke :) | 17:30 |
| kfox1111 | portdirect: only partially. | 17:30 |
| kfox1111 | doing what your suggesting with putting all of kolla_start in a configmap and passing the whole virtualenv in I think woudl be safest, but ugly ugly. :/ | 17:31 |
| kfox1111 | if it was a go binary, it would at least only be one file. :/ | 17:31 |
| jascott1 | kfox1111 i see the koolaid is seeping in | 17:31 |
| * jascott1 twiddles fingers and says "excellent" | 17:32 | |
| kfox1111 | jascott1: heh. not nessisarily. | 17:32 |
| jascott1 | :D | 17:32 |
| kfox1111 | I still think software belongs in the container. it makes this problem so much easier. | 17:32 |
| qwang | hey jascott1 | 17:32 |
| kfox1111 | it makes the ci/cd issue slightly harder. | 17:32 |
| jascott1 | hi qwang! | 17:32 |
| kfox1111 | but that shoudl be tractable. | 17:32 |
| qwang | jascott1: Please check it out https://etherpad.openstack.org/p/openstack-on-helm | 17:33 |
| jascott1 | qwang good start! hmm we might actually need to land some helm patches before the talk ;) | 17:34 |
| qwang | jascott1: we do need some for the talk, not the abstract | 17:35 |
| des_consolado | also what does kolla-kubernetes have to say about coreos? I guess the underlying os just isn't really that important? | 17:38 |
| *** matrohon has quit IRC | 17:38 | |
| des_consolado | I get the impression the kubernetes community prefers coreos for baremetal deployments... | 17:38 |
| sbezverk | kfox1111: any suggestions to try? | 17:41 |
| jascott1 | des_consolado Id love to explore the coreos option but doesnt seem like we are there yet | 17:44 |
| kfox1111 | sbezverk: did you try doing the package build in the build_all script? | 17:44 |
| inc0 | des_consolado, in all reality, you should be able to run it on coreos | 17:44 |
| inc0 | we don't really care what is distro on your servers | 17:45 |
| kfox1111 | des_consolado: coreos would like you to believe that. :) | 17:45 |
| sbezverk | kfox1111: do you mean on the local cluster? | 17:45 |
| kfox1111 | sbezverk: in the gate. but should work there too. | 17:46 |
| sbezverk | kfox1111: it looks like at the gate it is already uses build_all | 17:47 |
| sbezverk | tools/setup_gate.sh:tools/helm_build_all.sh ~/.helm/repository/kolla | 17:47 |
| kfox1111 | sbezverk: yeah, I'm saying, I think the problem why it wasn't building for you in the gate, was the package isn't being built in the build_all script? | 17:47 |
| *** satyar_ has joined #openstack-kolla | 17:48 | |
| sbezverk | kfox1111: ok let me check then.. | 17:48 |
| *** Serlex has quit IRC | 17:49 | |
| *** satyar has quit IRC | 17:49 | |
| *** rhallisey has quit IRC | 17:50 | |
| inc0 | SamYaple, can I take a look at your boostrap script for galera? | 17:52 |
| *** rhallisey has joined #openstack-kolla | 17:52 | |
| *** sudipto_ has quit IRC | 17:53 | |
| *** sudipto has quit IRC | 17:53 | |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes master: Unpin kolla-ansible https://review.openstack.org/427365 | 17:53 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes master: Pull container tallballs into the gate https://review.openstack.org/426598 | 17:59 |
| *** tonanhngo has joined #openstack-kolla | 18:03 | |
| *** mliima_ has quit IRC | 18:05 | |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes master: Check to see if Horizon is working in the gate. https://review.openstack.org/426025 | 18:06 |
| *** rwallner has joined #openstack-kolla | 18:08 | |
| *** rhallisey has quit IRC | 18:21 | |
| *** unicell has joined #openstack-kolla | 18:26 | |
| *** unicell has joined #openstack-kolla | 18:26 | |
| *** hieulq has joined #openstack-kolla | 18:29 | |
| cliles | anyone else having issues with the cinder handlers from master? ansible is barfing on the conditional check for me... | 18:31 |
| openstackgerrit | Eduardo Gonzalez proposed openstack/kolla stable/newton: This feature is supporting for workers config in ironic-api. https://review.openstack.org/428342 | 18:33 |
| *** srwilkers has quit IRC | 18:33 | |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes master: Support rolling upgrades of horizon https://review.openstack.org/333996 | 18:35 |
| openstackgerrit | Eduardo Gonzalez proposed openstack/kolla stable/mitaka: This feature is supporting for workers config in ironic-api. https://review.openstack.org/428343 | 18:35 |
| *** skuda has quit IRC | 18:36 | |
| *** gfidente is now known as gfidente|afk | 18:38 | |
| sdake_ | bmace_ serlex here are my notes: http://paste.openstack.org/raw/597410/ -> setup.sh -> http://paste.openstack.org/raw/597412/ create-configmap.sh: http://paste.openstack.org/show/597413/ and /etc/kolla/globals.yaml -> http://paste.openstack.org/show/597288/ | 18:41 |
| *** salv-orlando has joined #openstack-kolla | 18:41 | |
| sdake_ | bmace_ i'll turn that into a proper review that is extracted from the gate jobs today | 18:41 |
| *** shardy has quit IRC | 18:45 | |
| bmace_ | sdake_: i think using something like the python kube library we might be able to do some nicer / automated waiting for things to get up and running to make it push button end to end. at least there is where i want to be with the halcyon-vagrant env setup. run the container and done. | 18:52 |
| kfox1111 | +1. | 18:53 |
| sdake_ | bmace_ i am working on distilling the basic instructions into something humans can understand first;) | 18:53 |
| sdake_ | bmace_ if you wnt to run with those directions wfm :) | 18:53 |
| kfox1111 | hoping we can figure out how to get to the point where its all helm though. | 18:53 |
| kfox1111 | so no python env needed to deploy with the packages. | 18:53 |
| kfox1111 | that will take a while to get to though. | 18:54 |
| sdake_ | kfox1111 ya that sounds like a good end goal - in the meanitme we need steps to get there, and the first step is to understand how what we have works :) | 18:54 |
| kfox1111 | +1 | 18:54 |
| sdake_ | the current dev env script in the queue is integrated into the gating, which makes that extremely difficult | 18:55 |
| bmace_ | sdake_: we have some directions written up in our internal wiki that i was able to use to get 100% up, with some small tweaks here and there. the most obnoxious parts were certainly things like the pull_containers and other wait logic in the scripts that don't work at all.. and ceph setup. i'm pulling vagrant into the dev setup image so it can make sure ceph is set up on kube2 "automatically" | 18:56 |
| kfox1111 | sdake_: we have to figure out how to do that better, yeah. | 18:56 |
| kfox1111 | the problem is, our docs bitrotted when it wasn't gated. | 18:57 |
| *** portdirect is now known as portdirect_away | 18:57 | |
| kfox1111 | so we will have broken devenvs if we cant gate it, or at least gate most of it. | 18:57 |
| sdake_ | kfox1111 the cause of the bitrot is in inaccurate analsys | 18:57 |
| bmace_ | sdake_: I think most of the gating scripts, like the ceph service workflow scripts, are mostly ok. pull_containers imo should get yanked and the wait poll needs to be fixed to be more reliable. | 18:57 |
| sdake_ | the reason the docs bitrotted is because nobody amintined them | 18:57 |
| kfox1111 | sdake_: if we knew what broke, when it broke in the gate, we would fix it right away. | 18:57 |
| bmace_ | sdake_: and a lot of hard code values need to be made into variables that can be passed in | 18:57 |
| kfox1111 | its much harder to do it in a panic later once you realize its all broken. :/ | 18:57 |
| sdake_ | the docs were never maintained | 18:58 |
| sdake_ | that is being fixed now | 18:58 |
| kfox1111 | that too. | 18:58 |
| sdake_ | it really has little to do with the gate | 18:58 |
| kfox1111 | but the docs contained a large amount of "paste this code in" | 18:58 |
| kfox1111 | and that code wasn't tested. | 18:58 |
| sdake_ | right- which is fine - if the docs are tested | 18:58 |
| kfox1111 | so when we broke that docs, then we didn't know it. | 18:58 |
| sdake_ | the docs were not manually tested which is mandatory | 18:58 |
| kfox1111 | the problem was there wasn't enough time to test every ps going in, and manually test the docs each time. :/ | 19:00 |
| kfox1111 | thats why I came up with the kolla-kubernetes gate in the first place. I couldn't keep up with manually testing every ps coming in. | 19:00 |
| kfox1111 | now I can review ps's much more quickly. | 19:01 |
| *** Pavo has quit IRC | 19:01 | |
| kfox1111 | manually testing each ps against the devenv docs won't scale well. :/ | 19:01 |
| kfox1111 | I'm afraid there wont be enough reviewer time to ensure the devenv doesn't regress without some kind of automation. | 19:03 |
| inc0 | kfox1111, it will on it's own | 19:04 |
| inc0 | when people will use devenv | 19:04 |
| kfox1111 | it just pushes the problem onto the new devs I think. :/ | 19:05 |
| kfox1111 | then they think "oh, this isn't mature, maybe I should go look at a different project" | 19:05 |
| kfox1111 | :/ | 19:05 |
| *** Pavo has joined #openstack-kolla | 19:06 | |
| sdake_ | kfox1111 both you and inc0 are correct, however, that is the nature of documentation | 19:06 |
| kfox1111 | I guess its the struggle, do you release something but broken, or wait to release until its ready. :/ | 19:06 |
| kfox1111 | yeah. | 19:06 |
| kfox1111 | still, as much of the docs that can be put into something testable, instead of human text, means its gate testable and less likely to be broken when they go and try it. | 19:07 |
| kfox1111 | as well as being easier for the devs. | 19:07 |
| kfox1111 | we just have to find the right balance there. | 19:07 |
| *** inc0 has quit IRC | 19:10 | |
| *** inc0 has joined #openstack-kolla | 19:14 | |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes master: WIP: test rebooting in the gate https://review.openstack.org/413193 | 19:16 |
| *** karlamrhein has quit IRC | 19:17 | |
| *** mgoddard_ has quit IRC | 19:17 | |
| *** ipsecguy has quit IRC | 19:18 | |
| *** ipsecguy has joined #openstack-kolla | 19:19 | |
| *** rhallisey has joined #openstack-kolla | 19:19 | |
| *** karlamrhein has joined #openstack-kolla | 19:22 | |
| openstackgerrit | Sayantani Goswami proposed openstack/kolla-ansible master: Update links in documents https://review.openstack.org/428392 | 19:35 |
| kfox1111 | oh.. that sucks.... | 19:37 |
| kfox1111 | lost a datacenter... | 19:37 |
| kfox1111 | there goes the rest of my day. | 19:37 |
| openstackgerrit | Merged openstack/kolla stable/mitaka: Fix typo heat.conf https://review.openstack.org/427768 | 19:47 |
| *** rhallisey_ has joined #openstack-kolla | 19:48 | |
| *** rhallisey_ has quit IRC | 19:48 | |
| *** rhallisey_ has joined #openstack-kolla | 19:48 | |
| sdake_ | inc0 yo dude | 19:49 |
| sdake_ | entrypoint is busted | 19:49 |
| sdake_ | inc0 can you get it fixed plz | 19:49 |
| openstackgerrit | Sayantani Goswami proposed openstack/kolla-ansible master: WIP Deployment Guide https://review.openstack.org/427965 | 19:51 |
| *** rhallisey has quit IRC | 19:52 | |
| *** matrohon has joined #openstack-kolla | 19:52 | |
| openstackgerrit | Merged openstack/kolla stable/newton: Add neutron vpnaas code into neutron-server container https://review.openstack.org/419977 | 19:52 |
| *** skuda has joined #openstack-kolla | 19:56 | |
| *** gfidente|afk is now known as gfidente | 19:58 | |
| *** rhallisey_ is now known as rhallisey | 19:58 | |
| *** jascott1 has quit IRC | 20:00 | |
| *** jascott1 has joined #openstack-kolla | 20:01 | |
| *** jascott1 has quit IRC | 20:01 | |
| *** jascott1 has joined #openstack-kolla | 20:02 | |
| harlowja | is there a non-draft logo for kolla up anywhere? | 20:03 |
| harlowja | (want one for a meetup presentation i'm doing, ha | 20:03 |
| harlowja | but all i see on a simple ML search is the draft one | 20:04 |
| inc0 | yeah, I think community is ready for this awesomeness | 20:04 |
| inc0 | hold on | 20:04 |
| harlowja | thx | 20:04 |
| inc0 | https://drive.google.com/file/d/0B9SxQBJsT7y2Nnk0QjVnSTJ5Skg0NUxpcllvXzI2SXJYZHRr/view?usp=sharing | 20:05 |
| harlowja | niceee | 20:05 |
| harlowja | thx | 20:05 |
| inc0 | it should have caption "you wot m8?" | 20:05 |
| *** rwallner has quit IRC | 20:05 | |
| harlowja | ha | 20:06 |
| harlowja | https://www.meetup.com/openstack/events/236906794/ btw inc0 | 20:09 |
| harlowja | if u ant to advertise it at all, ha | 20:09 |
| harlowja | its pretty much openstack + kolla + containers + where we are at | 20:09 |
| inc0 | I'd love to see it myself:) | 20:09 |
| harlowja | not a ton of alloted time, but i'll go over the top level stuff | 20:09 |
| *** rwallner has joined #openstack-kolla | 20:09 | |
| *** salv-orlando has quit IRC | 20:09 | |
| inc0 | I'd really like to see where you guys are at;) | 20:09 |
| harlowja | ha | 20:10 |
| harlowja | somewhere | 20:11 |
| harlowja | we are building out our ansible stuff and how it integrates with jenkins (jenkins will kick off the ansible 'subroutines') | 20:11 |
| harlowja | but jenkins will (to start) be the thing that does the overall orchestration | 20:11 |
| harlowja | with it invoking ansible playbooks as needed | 20:11 |
| harlowja | that's the current plan (so far) | 20:12 |
| inc0 | you might be interested in our conversation about rethinking images | 20:12 |
| harlowja | klindgren ^ | 20:12 |
| harlowja | mdorman ^ | 20:12 |
| harlowja | lol | 20:12 |
| harlowja | we might, ha | 20:12 |
| inc0 | part of it is to make them more cicd friendly | 20:12 |
| harlowja | ?? | 20:12 |
| harlowja | do tell | 20:12 |
| harlowja | lol | 20:12 |
| inc0 | https://review.openstack.org/#/c/427931/ also stuff like this | 20:13 |
| *** adrian_otto has quit IRC | 20:13 | |
| harlowja | klindgren u want to look at that sometime, u prob have some thoughts | 20:14 |
| harlowja | personally mariadb from a perspective of a container for us isn't so useful | 20:15 |
| *** eaguilar has quit IRC | 20:15 | |
| harlowja | cause we have a managed DB service that is managing mariadb and such | 20:15 |
| inc0 | well it's more about "remove non openstack stuff from kolla" | 20:15 |
| inc0 | as the images go | 20:15 |
| inc0 | let mariadb community handle these | 20:16 |
| inc0 | anyway, one session in ptg will be meant for that stuff | 20:17 |
| harlowja | ya, trying to see if klindgren can show up to that | 20:17 |
| harlowja | though i'm pretty sure the eventbrite has been all taken up | 20:18 |
| harlowja | (i won't be there, skiing with family) | 20:18 |
| harlowja | pretty sure klindgren should just show up and inc0 will let him in | 20:18 |
| harlowja | lol | 20:18 |
| *** crushil has quit IRC | 20:26 | |
| *** crushil has joined #openstack-kolla | 20:26 | |
| harlowja | inc0 sometime though if u want me to talk about wtf godaddy is doing i can | 20:35 |
| *** salv-orlando has joined #openstack-kolla | 20:35 | |
| harlowja | just let me know where to show up | 20:35 |
| harlowja | at irc or whatever | 20:35 |
| harlowja | i can go into the current things | 20:35 |
| inc0 | cool, I'll keep that in mind | 20:35 |
| harlowja | *in more detail | 20:35 |
| harlowja | ya, i don't mind | 20:35 |
| inc0 | we will have bronwfield deployment session too and I'd love to hear your story tbh | 20:35 |
| *** rhallisey has quit IRC | 20:37 | |
| *** rhallisey has joined #openstack-kolla | 20:37 | |
| *** rhallisey has quit IRC | 20:39 | |
| *** rhallisey has joined #openstack-kolla | 20:39 | |
| *** rhallisey has quit IRC | 20:39 | |
| *** rhallisey has joined #openstack-kolla | 20:40 | |
| openstackgerrit | Sayantani Goswami proposed openstack/kolla-ansible master: WIP Deployment Guide https://review.openstack.org/427965 | 20:42 |
| *** rhallisey has quit IRC | 20:44 | |
| harlowja | all at the PTG or at the summit? | 20:47 |
| harlowja | doesn't seem like anyone from GD folks will be at the PTG due to existing vacations/plans ... :( | 20:47 |
| inc0 | wherever I just want to know how you guys did this and I'm sure you'd have ton of feedback | 20:47 |
| inc0 | shame:( | 20:47 |
| *** hieulq has quit IRC | 20:48 | |
| harlowja | ya, maybe i can spend a few hours on zoom | 20:48 |
| harlowja | with whoever is interested | 20:48 |
| harlowja | show the goods | 20:48 |
| harlowja | lol | 20:48 |
| harlowja | (and/or what we got so far) | 20:48 |
| inc0 | kk I'll get sth arranged | 20:48 |
| inc0 | also if you want to provide any feedback or want us to discuss any pain you've experienced, that'd be good place | 20:49 |
| harlowja | if i send a carrier piegon to the PTG | 20:49 |
| harlowja | that'd be cool right? | 20:49 |
| harlowja | or multiple of them | 20:49 |
| *** jascott1 has quit IRC | 20:49 | |
| inc0 | pings aren't great with IPoAC | 20:49 |
| inc0 | IP over Avian Carriers | 20:50 |
| harlowja | ha | 20:50 |
| *** jascott1 has joined #openstack-kolla | 20:50 | |
| *** jascott1 has quit IRC | 20:50 | |
| *** rhallisey has joined #openstack-kolla | 20:58 | |
| sbezverk | inc0: we really need to get some help for entrypoint, can you help? | 20:58 |
| inc0 | whats up with it? | 20:59 |
| sbezverk | inc0: it is a bit broken | 20:59 |
| inc0 | elaborate please.. | 20:59 |
| sbezverk | we filed several issues but got 0 response | 20:59 |
| sbezverk | https://github.com/stackanetes/kubernetes-entrypoint/issues/19 | 21:00 |
| *** harlowja has quit IRC | 21:00 | |
| inc0 | I'll ask around | 21:00 |
| sbezverk | inc0: what happened that some dependencies are not enforced | 21:00 |
| *** jascott1 has joined #openstack-kolla | 21:01 | |
| *** Pavo has quit IRC | 21:01 | |
| sbezverk | inc0: thank you! | 21:01 |
| inc0 | however we might have to fix it ourselves | 21:01 |
| *** fguillot has quit IRC | 21:01 | |
| *** rwallner has quit IRC | 21:02 | |
| sbezverk | inc0: that is an option too ;) | 21:02 |
| sbezverk | but let's explore first easier way | 21:02 |
| inc0 | well, we can try to debug issue ourselves and push pull request | 21:03 |
| *** jascott1 has quit IRC | 21:05 | |
| *** rwallner has joined #openstack-kolla | 21:05 | |
| *** Pavo has joined #openstack-kolla | 21:05 | |
| *** salv-orlando has quit IRC | 21:05 | |
| *** schwicht has joined #openstack-kolla | 21:06 | |
| *** sharon_so has joined #openstack-kolla | 21:08 | |
| *** rwallner has quit IRC | 21:09 | |
| *** dave-mccowan has quit IRC | 21:10 | |
| *** Pavo has quit IRC | 21:17 | |
| *** mannamne has joined #openstack-kolla | 21:21 | |
| *** matrohon has quit IRC | 21:23 | |
| cliles | so, how the heck are you suppose to setup different interfaces between your compute and controllers? | 21:23 |
| sbezverk | inc0: ok I will add bunch of debug prints into the code see if I can spot any issues | 21:25 |
| inc0 | cliles, as in different name of interfaces? | 21:25 |
| inc0 | you can do it in inventory | 21:25 |
| inc0 | like [compute] /n my_compute_host network_interface=eth666 | 21:26 |
| cliles | yeah, I did that but I have some failuers where a host with non-standard interfaces is trying to use eth0 | 21:26 |
| inc0 | would you mind pasting a log? | 21:26 |
| cliles | sure | 21:26 |
| inc0 | and inventory? | 21:26 |
| cliles | can do | 21:26 |
| cliles | 1 sec | 21:26 |
| cliles | http://paste.openstack.org/show/n1KojB97ZWbBbSEWiKKR/ | 21:28 |
| cliles | in this specific case the compute host is still using the tunnel_interface as defined in globals.yml | 21:29 |
| inc0 | so this is one ansible quirk | 21:29 |
| inc0 | I don't like it but that's the reality | 21:29 |
| inc0 | try specyfying ifaces in inventory for vortex too | 21:30 |
| inc0 | and remove these from globals | 21:30 |
| v1k0d3n | sdake_: hi | 21:30 |
| inc0 | because globals.yml generally override whatever | 21:30 |
| inc0 | including stuff in inventory | 21:30 |
| v1k0d3n | did you guys find out anything on where halcyon is failing you? | 21:30 |
| cliles | inc0: ah, ok | 21:31 |
| v1k0d3n | inc0: you know anyone else using it? | 21:32 |
| v1k0d3n | i didn't really think anyone was using it here or needed it really. | 21:32 |
| v1k0d3n | wasn't sure. | 21:32 |
| inc0 | halcyon? | 21:32 |
| inc0 | I dont know...I dont use it personally | 21:32 |
| *** l4yerffeJ has joined #openstack-kolla | 21:34 | |
| *** Jeffrey4l has quit IRC | 21:35 | |
| *** l4yerffeJ_ has quit IRC | 21:35 | |
| *** sayantani01 has quit IRC | 21:36 | |
| *** sayantani01 has joined #openstack-kolla | 21:37 | |
| *** Jeffrey4l has joined #openstack-kolla | 21:38 | |
| *** gfidente has quit IRC | 21:38 | |
| *** adrian_otto has joined #openstack-kolla | 21:39 | |
| *** l4yerffeJ has quit IRC | 21:45 | |
| *** l4yerffeJ_ has joined #openstack-kolla | 21:45 | |
| *** dave-mccowan has joined #openstack-kolla | 21:47 | |
| *** sambetts is now known as sambetts|afk | 21:49 | |
| *** jtriley has quit IRC | 21:55 | |
| *** salv-orlando has joined #openstack-kolla | 21:57 | |
| *** Pavo has joined #openstack-kolla | 21:59 | |
| *** Pavo has quit IRC | 22:01 | |
| *** lrensing has quit IRC | 22:05 | |
| *** mannamne has quit IRC | 22:06 | |
| *** satyar_ has quit IRC | 22:06 | |
| *** jascott1 has joined #openstack-kolla | 22:11 | |
| *** sayantan_ has joined #openstack-kolla | 22:12 | |
| *** sayantani01 has quit IRC | 22:12 | |
| *** Pavo has joined #openstack-kolla | 22:13 | |
| *** Pavo has quit IRC | 22:13 | |
| *** breitz has quit IRC | 22:14 | |
| *** salv-orl_ has joined #openstack-kolla | 22:15 | |
| jascott1 | kfox1111 ping | 22:15 |
| *** Pavo has joined #openstack-kolla | 22:16 | |
| *** salv-orlando has quit IRC | 22:18 | |
| *** srwilkers has joined #openstack-kolla | 22:19 | |
| srwilkers | afternoon | 22:19 |
| srwilkers | o/ | 22:19 |
| *** crushil has quit IRC | 22:19 | |
| *** rwallner has joined #openstack-kolla | 22:22 | |
| *** Pavo_ has joined #openstack-kolla | 22:23 | |
| *** jheroux has quit IRC | 22:25 | |
| *** harlowja has joined #openstack-kolla | 22:25 | |
| jascott1 | o/ | 22:26 |
| *** breitz has joined #openstack-kolla | 22:26 | |
| *** rwallner has quit IRC | 22:26 | |
| *** mgoddard_ has joined #openstack-kolla | 22:27 | |
| *** harlowja has quit IRC | 22:32 | |
| *** harlowja has joined #openstack-kolla | 22:33 | |
| *** sdake_ has quit IRC | 22:34 | |
| *** skuda has quit IRC | 22:35 | |
| *** sdake has joined #openstack-kolla | 22:37 | |
| *** Pavo_ has quit IRC | 22:38 | |
| *** mgoddard_ has quit IRC | 22:40 | |
| *** erlon has quit IRC | 22:40 | |
| *** Pavo has quit IRC | 22:46 | |
| *** portdirect_away is now known as portdirect | 22:49 | |
| -openstackstatus- NOTICE: Restarting gerrit due to performance problems | 22:51 | |
| portdirect | sbezverk: had a peek at that issue you filed at k8s-entrypoint | 22:52 |
| portdirect | this is pretty weird: http://logs.openstack.org/97/424697/3/check/gate-kolla-kubernetes-deploy-centos-binary-2-helm-entrypoint-nv/f687966/logs/pods/kolla-cinder-scheduler-0-dependencies.txt.gz | 22:53 |
| *** salv-orl_ has quit IRC | 22:53 | |
| portdirect | looks like at some point an endpoint got added to that service :( | 22:53 |
| portdirect | which confuses me, as I'm not sure how thats possible | 22:54 |
| *** MasterOfBugs has joined #openstack-kolla | 22:54 | |
| *** adrian_otto has quit IRC | 22:57 | |
| *** Pavo has joined #openstack-kolla | 22:59 | |
| *** Pavo has quit IRC | 22:59 | |
| *** Pavo has joined #openstack-kolla | 23:00 | |
| *** yuanying has joined #openstack-kolla | 23:08 | |
| *** Pavo_ has joined #openstack-kolla | 23:09 | |
| *** schwicht has quit IRC | 23:11 | |
| *** jascott1 has quit IRC | 23:14 | |
| *** jrobinson_ has joined #openstack-kolla | 23:18 | |
| *** jrobinson_ is now known as jrobinson | 23:19 | |
| sdake | inc0 portdirect - entrypoint seems to not function as advertised - could someone who can fix i t please fix it ? :) | 23:26 |
| *** Pavo has quit IRC | 23:26 | |
| *** Pavo_ has quit IRC | 23:26 | |
| sdake | inc0 portdirect a whole slew of services go into a crashloopbackoff without acknowledging the dependencies defined | 23:27 |
| *** Pavo has joined #openstack-kolla | 23:27 | |
| *** Pavo has quit IRC | 23:27 | |
| kfox1111 | sdake: you trying from service package? | 23:27 |
| sdake | kfox1111 yup | 23:27 |
| kfox1111 | hmm | 23:27 |
| *** Pavo has joined #openstack-kolla | 23:27 | |
| kfox1111 | that should work. | 23:27 |
| sdake | if i launch them all at once, nova api fails to migrate the db - which is troubling | 23:28 |
| *** dave-mccowan has quit IRC | 23:28 | |
| sdake | i have to launch everything minus nova+libvirt, then nova+libvirt seprarately | 23:28 |
| sdake | which is another issue | 23:28 |
| *** Pavo has quit IRC | 23:28 | |
| sdake | which I'd file however i've been up since 4am and its bedtime | 23:28 |
| kfox1111 | what does the logs in the dependency container for the nova-api-migradte-db job look like? | 23:28 |
| kfox1111 | k | 23:28 |
| *** tonanhngo has quit IRC | 23:29 | |
| *** rhallisey has quit IRC | 23:29 | |
| *** Pavo has joined #openstack-kolla | 23:29 | |
| portdirect | sdake: you get halcyon working for you? or do we still need to fix something up? | 23:30 |
| sdake | portdirect i have not got back to that - hopefully this weekend - although I believe bmace_ has it operational | 23:31 |
| sdake | portdirect I really believe it was the mispelling in the docs you identified, however, haven't had time to verify that | 23:31 |
| sdake | portdirect for the moment i wouldn't panic about it until I come back with the requested ddata | 23:31 |
| *** mattmceuen has quit IRC | 23:31 | |
| inc0 | sdake, can we push patchset with bugfix? | 23:31 |
| portdirect | sweet - be good to get that issue closed out if so | 23:32 |
| sdake | portdirect or alternately i'd come back and say it was an error in our review process (which happens) | 23:32 |
| kfox1111 | sdake: was wondiering something, as a stopgap until we discuss configmaps at the ptg. | 23:32 |
| kfox1111 | is the halcyon static? | 23:32 |
| kfox1111 | the ip's, the ethernet interfaces, etc? | 23:33 |
| portdirect | lol - thats up to you sdake, though i think its a bit unfair to bring other projects into bugs in our docs :) | 23:33 |
| portdirect | kfox1111: yea | 23:33 |
| sdake | portdirect i don't follow - the bugs are in kolla docs | 23:33 |
| kfox1111 | could we make a kolla_config for halcyon, generate a bunch of configmaps, | 23:34 |
| sdake | portdirect or alternately in the code | 23:34 |
| *** tonanhngo has joined #openstack-kolla | 23:34 | |
| kfox1111 | and just dump them to a devenv-configmap helm package? | 23:34 |
| sdake | portdirect if you don't want me to report its a bug in the kolla docs, I'll just say it was an invalid bug? | 23:34 |
| kfox1111 | then we might be able to leave out all the python bits. | 23:34 |
| sdake | portdirect been through the ringer today - let me know how you want to proceed and i'll proceed that way - however, I haven't the energy to actually validate the problem today | 23:34 |
| portdirect | ah sorry think I got wrong end of stick, was asking if i could close out this: https://github.com/att-comdev/halcyon-vagrant-kubernetes/issues/50 | 23:35 |
| bmace_ | portdirect: i got most of my env operational on halcyon. i'm trying to improve some of the containerization / scripting around the bring up now | 23:35 |
| *** schwicht has joined #openstack-kolla | 23:35 | |
| sdake | portdirect if bmace_ has validated it works for him, that is good enough for me | 23:35 |
| kfox1111 | bmace_: great news! :) | 23:36 |
| sdake | portdirect as you cn see in the last report- my wife got back tue and was in jetlag utnil about today | 23:36 |
| sdake | portdirect our household is still recovering :) | 23:36 |
| portdirect | lol - all good :) just trying to clean the table :) | 23:36 |
| sdake | me too portdirect me too :) | 23:37 |
| sdake | at the moment family is a big chunk of my table | 23:37 |
| *** Pavo has quit IRC | 23:37 | |
| sdake | sadly she is in europe again monday and tuesday | 23:39 |
| * sdake groans | 23:39 | |
| *** tonanhngo has quit IRC | 23:39 | |
| sdake | i'm out for a bit - possibly the count :) | 23:39 |
| *** oomichi has joined #openstack-kolla | 23:41 | |
| kfox1111 | bmace_: portdirect: what do you think of that configmap idea? | 23:43 |
| *** jascott1 has joined #openstack-kolla | 23:43 | |
| kfox1111 | jascott1: ping. :) | 23:43 |
| *** Pavo has joined #openstack-kolla | 23:44 | |
| *** sdake has quit IRC | 23:44 | |
| portdirect | kfox1111: predefined configmaps for halcyon? | 23:44 |
| kfox1111 | yeah. | 23:44 |
| bmace_ | kfox1111: do we really need them? not like it is that hard to generate the configmaps | 23:44 |
| kfox1111 | bmace_: the idea being, in theory you might be able to deploy the whole thing from just helm packages then, not needing a python env at all. | 23:45 |
| bmace_ | assuming things aren't borked, like fluentd not working right if you don't have network interfaces that match your target env. | 23:45 |
| *** srwilkers has quit IRC | 23:45 | |
| *** Pavo has quit IRC | 23:45 | |
| jascott1 | kfox1111 pong | 23:45 |
| kfox1111 | we need to figure out longer term how to do configmaps at the ptg. | 23:45 |
| kfox1111 | jascott1: you pinged earlier? | 23:45 |
| jascott1 | sorry i got a call from da boss | 23:45 |
| kfox1111 | no worries. | 23:46 |
| kfox1111 | I"m in the middle of a datacenter outage. | 23:46 |
| bmace_ | kfox1111: since all the halcyon values are static i don't think having pre-generated charts for it would be a bad thing particularly.. but as new services get added, etc.. they would need to be kept up to date. | 23:46 |
| kfox1111 | but kind of in wait mode. | 23:46 |
| portdirect | kfox1111: i'm a bit out of touch :( but if we can generate configmaps without *too* many hoops I'd kinda prefer that | 23:46 |
| kfox1111 | bmace_: yeah. it would just be temporary until we figure out the prefered configmap generation aproach. | 23:47 |
| kfox1111 | just to make the dev env easier to get off the ground. | 23:47 |
| bmace_ | kfox1111: although, some values are slightly different since there is an option for ubuntu and centos | 23:47 |
| kfox1111 | its possible then we wouldn't need a dev env container at all at that point? | 23:47 |
| portdirect | kfox1111: Idealy, we should provide a setup that would allow someone to start working on a better solution, rather than supply a set of configs with no supporting framwork | 23:47 |
| kfox1111 | bmace_: those arn't in the configmaps, so could be done in a values.yaml | 23:47 |
| bmace_ | kfox1111: i know that myself and mark have had a lot of success with centos.. i know someone else here that tried to get stuff on ubuntu and had some problems. | 23:48 |
| bmace_ | kfox1111: kk | 23:48 |
| *** salv-orlando has joined #openstack-kolla | 23:48 | |
| kfox1111 | portdirect: yeah. we should discuss the real solution at the ptg. | 23:48 |
| kfox1111 | portdirect: not entirly shure what that would look like. so would be good to get a bunch of eyes/ears/brains on it and work it out together. | 23:48 |
| portdirect | +1, so i suppose whatever is easiest to implement wins for now? | 23:49 |
| v1k0d3n | kfox1111: the networks are variables in halcyon. but if you need something in halcyon for users, just contribute it. | 23:49 |
| v1k0d3n | portdirect originally added a script in there for kolla-k8s anyway, so it would be more usable for your folks. welcome to drop in or replace. | 23:50 |
| *** Pavo has joined #openstack-kolla | 23:50 | |
| v1k0d3n | although i'm still reading through backlog of comments. | 23:50 |
| kfox1111 | v1k0d3n: ok. cool. :) | 23:50 |
| kfox1111 | ok. just got notification that I've got some access to power again... gotta go for a bit to boot some things... | 23:51 |
| kfox1111 | bbiab | 23:51 |
| v1k0d3n | cool | 23:51 |
| *** salv-orlando has quit IRC | 23:52 | |
| *** Pavo has quit IRC | 23:54 | |
| *** schwicht has quit IRC | 23:56 | |
| *** Pavo has joined #openstack-kolla | 23:56 | |
| kfox1111 | actually, back for a little bit longer. | 23:56 |
| sbezverk | kfox1111: about entrypoint, man when all services are pushed without waiting, bunch of pods keep crashing, they do stabilize in some time | 23:57 |
| sbezverk | but with entrypoint there should be NO crashing | 23:57 |
| jascott1 | sbezverk how long is some time? :) | 23:57 |
| sbezverk | deploy all services took I think less than 10 minutes | 23:58 |
| kfox1111 | sbezverk: are we missing a few deps? | 23:58 |
| *** schwicht has joined #openstack-kolla | 23:58 | |
| sbezverk | nope, because it is not consistent, I am 99 positive that the way entrypoint waits for service is wrong | 23:58 |
| kfox1111 | like, maybe we're depending on keystone being up, but not having keystone endpoints created? | 23:58 |
| jascott1 | in what way? | 23:58 |
| *** schwicht has quit IRC | 23:59 | |
| sbezverk | they either do not or incorrectly detect that the service has associated backend | 23:59 |
| kfox1111 | but keystone without endpoints may happen occationally? | 23:59 |
| *** Pavo has quit IRC | 23:59 | |
| sbezverk | example I saw nova scheduler must wait for nova api to be up is stays in init state | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!