openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Basic ceph backend support https://review.openstack.org/373256 | 00:05 |
---|---|---|
*** neilus has joined #openstack-kolla | 00:05 | |
*** yingjun has quit IRC | 00:10 | |
*** neilus has quit IRC | 00:11 | |
*** yingjun has joined #openstack-kolla | 00:13 | |
*** daneyon has joined #openstack-kolla | 00:19 | |
*** inc0 has joined #openstack-kolla | 00:20 | |
inc0 | hey, just jumped in to say thank you guys:) | 00:21 |
*** sdake has quit IRC | 00:22 | |
*** daneyon has quit IRC | 00:23 | |
*** inc0 has quit IRC | 00:26 | |
*** sdake has joined #openstack-kolla | 00:32 | |
*** dave-mccowan has joined #openstack-kolla | 00:34 | |
sdake | hey inc0 ptl election results are in | 00:34 |
*** duonghq has joined #openstack-kolla | 00:46 | |
duonghq | morning, congratulation to inc0 | 00:47 |
*** tonanhngo has joined #openstack-kolla | 00:49 | |
*** tonanhngo has quit IRC | 00:51 | |
sdake | also grats to Jeffrey4l for running ;) | 00:56 |
Jeffrey4l | thank sdake, and congratulation to inc0 | 01:00 |
sbezverk | congratulations inc0!!! | 01:06 |
duonghq | morning sdake, Jeffrey4l, sbezverk | 01:07 |
sdake | sup cat | 01:07 |
sbezverk | it looks more like good night from my side of ocean ;-) | 01:08 |
Jeffrey4l | morning :) | 01:08 |
kfox1111 | congrats inc0! :) | 01:09 |
kfox1111 | sdake: +1. | 01:10 |
sdake | i think inc0 isn't here ;) | 01:10 |
sdake | kind of sucks 100 contributors are not engagedd in kolla enough to vote | 01:12 |
sdake | oh well SEP :) | 01:13 |
*** zhurong has joined #openstack-kolla | 01:14 | |
*** zhubingbing_ has joined #openstack-kolla | 01:16 | |
openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Basic ceph backend support https://review.openstack.org/373256 | 01:17 |
duonghq | due to we are from many tzs, so we need something to "Good..." | 01:17 |
sdake | duonghq huh? | 01:18 |
*** jascott1_ has joined #openstack-kolla | 01:18 | |
sdake | duonghq i like that we have people in every timezone :) | 01:18 |
duonghq | (j/k) from my pov, I "good morning", from your side, you "good night" or "good evening" | 01:18 |
*** zhurong has quit IRC | 01:19 | |
sdake | oh got it ;) | 01:19 |
sdake | struggling to come up with a greatin tht hits all timezones :) | 01:20 |
duonghq | ya | 01:20 |
sdake | besides helo which is prtty boring :) | 01:20 |
*** zhurong has joined #openstack-kolla | 01:20 | |
duonghq | what does Kolla say? | 01:20 |
duonghq | no idea yet | 01:20 |
sdake | kolla mean you mean? | 01:21 |
sdake | it is greek for glue | 01:21 |
sdake | as kolla started out as aa project to glue a bunch ofthings together | 01:21 |
sdake | which itdoes now | 01:22 |
sdake | just different things then originally intended ;-) | 01:22 |
duonghq | ah, I mean the Kolla in our mascot | 01:22 |
sdake | is our mascot live? | 01:22 |
sdake | its a koalla bear | 01:22 |
kfox1111 | heh. thats funny. kubernetes is greek too. :) | 01:22 |
duonghq | sometime I see it alive | 01:22 |
duonghq | wonder what does he/she say for greating other | 01:23 |
sdake | kfox1111 that has become a common pattrn - all the english names are taken | 01:23 |
* kfox1111 chuckles | 01:23 | |
kfox1111 | yeah, I gues that's true. | 01:23 |
duonghq | sdake: As coolsvap comment in https://review.openstack.org/#/c/347056/ : we do not alter specs once they are merged. But no exception for typo fix? | 01:48 |
zhubingbing_ | hello guys | 01:50 |
*** kbaegis has joined #openstack-kolla | 01:54 | |
zhubingbing_ | duonghq | 01:56 |
zhubingbing_ | hello | 01:56 |
duonghq | hi zhubingbing_ | 01:56 |
zhubingbing_ | moring | 01:57 |
duonghq | for the ps, you can see this link: https://docs.docker.com/engine/reference/commandline/volume_ls/#dangling | 01:57 |
*** tonanhngo has joined #openstack-kolla | 01:57 | |
*** tonanhngo has quit IRC | 01:57 | |
zhubingbing_ | sure | 01:58 |
zhubingbing_ | wow | 01:58 |
zhubingbing_ | so good | 01:58 |
duonghq | after cool weather last weekend, it's quite hot now :( | 01:59 |
duonghq | not so bad but still don't like this | 02:00 |
*** kbaegis has quit IRC | 02:04 | |
*** kbaegis has joined #openstack-kolla | 02:05 | |
openstackgerrit | dupengfei proposed openstack/kolla: Support shared path for glance images and nova instances disk https://review.openstack.org/372420 | 02:05 |
*** daneyon has joined #openstack-kolla | 02:07 | |
*** kbaegis has quit IRC | 02:07 | |
*** dupengfei has joined #openstack-kolla | 02:08 | |
dupengfei | Hello core reviewers, Who can help me to merge 'https://review.openstack.org/#/c/372089/', I have rebased many times, thanks. | 02:08 |
sdake | dupengfei looking now | 02:11 |
sdake | dupengfei no exceptons for typo fix | 02:11 |
sdake | duonghq that is | 02:11 |
*** daneyon has quit IRC | 02:12 | |
duonghq | roger | 02:12 |
*** kbaegis has joined #openstack-kolla | 02:14 | |
dupengfei | looking | 02:14 |
sdake | dupengfei reviewed | 02:15 |
sdake | dupengfei regarding rebasing - ya the code has been changingalot | 02:17 |
sdake | so rebasing becomes mandatory unfortunaely in that situation | 02:17 |
dupengfei | OMG, I must do a rebasing again. I have done it many times. I hope this is the last doing. | 02:18 |
openstackgerrit | dupengfei proposed openstack/kolla: Fix keystone,database,memcache for k8s. https://review.openstack.org/372089 | 02:19 |
dupengfei | sdake: Could you review and merge my another bp by the way? 'https://review.openstack.org/#/c/372420/' | 02:20 |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla: integrate gnocchi with ceilometer https://review.openstack.org/372737 | 02:28 |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Integrate gnocchi with ceph https://review.openstack.org/374937 | 02:29 |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Integrate gnocchi with ceph https://review.openstack.org/374937 | 02:29 |
sdake | dupengfei i've got bad news on this second patch | 02:32 |
openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Basic ceph backend support https://review.openstack.org/373256 | 02:32 |
sdake | dupengfei - feature freeze was the 15th | 02:32 |
sdake | no new features for newton | 02:32 |
sdake | further complicating matters is master is also not open for commits | 02:32 |
*** lamt has joined #openstack-kolla | 02:33 | |
sdake | dupengfei so your patch is stuck until we wrap up fixing the bugs for rc2/3.0.0 | 02:33 |
*** sdake has quit IRC | 02:35 | |
dupengfei | ok | 02:42 |
*** yuanying has quit IRC | 02:52 | |
openstackgerrit | Merged openstack/kolla: Change "Virtualhost" to "VirtualHost" in horizon.conf.j2 https://review.openstack.org/375929 | 02:53 |
openstackgerrit | OpenStack Proposal Bot proposed openstack/kolla: Updated from global requirements https://review.openstack.org/375989 | 02:54 |
openstackgerrit | Merged openstack/kolla: Fix heat role creation https://review.openstack.org/375475 | 02:55 |
*** huikang has joined #openstack-kolla | 02:57 | |
*** duonghq_ has joined #openstack-kolla | 03:00 | |
*** daneyon has joined #openstack-kolla | 03:01 | |
*** duonghq has quit IRC | 03:02 | |
*** huikang_ has joined #openstack-kolla | 03:02 | |
*** david-lyle has quit IRC | 03:04 | |
*** huikang has quit IRC | 03:05 | |
*** daneyon has quit IRC | 03:06 | |
*** huikang_ has quit IRC | 03:08 | |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Rename the gnocchi api-paste filename https://review.openstack.org/376092 | 03:12 |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Use keystone-paste.ini template for keystone https://review.openstack.org/373379 | 03:24 |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Fix merge_configs.py file when not provide vars parameter https://review.openstack.org/376098 | 03:24 |
Jeffrey4l | dupengfei, around? | 03:29 |
*** dave-mccowan has quit IRC | 03:35 | |
dupengfei | Jeffrey4l: ?? | 03:35 |
Jeffrey4l | dupengfei, leave a comment in your ps. | 03:35 |
dupengfei | Jeffrey4l: I see. | 03:36 |
kbaegis | Could someone walk me through setting up dashboard extensions in the horizon container? | 03:36 |
dupengfei | Jeffrey4l: i am replying | 03:36 |
kbaegis | I can't seem to figure out what I"m missing | 03:36 |
dupengfei | Jeffrey4l: I agree with you about the glance_registry_host and memcached_servers. But adding new files will increase the confusion of configuration. | 03:36 |
dupengfei | dupengfei: I want to change in this way: to use "memcached" and "glance-registry" inside the *.j2 files, what 's up ? | 03:37 |
dupengfei | Jeffrey4l : I want to change in this way: to use "memcached" and "glance-registry" inside the *.j2 files, what 's up ? | 03:37 |
*** brad[]` has quit IRC | 03:38 | |
dupengfei | Jeffrey4l : how do u think? | 03:38 |
Jeffrey4l | dupengfei, or move all options related to k8s into one place. rather than mix up with ansible deployment | 03:38 |
dupengfei | Jeffrey4l : it is a big work, and will change the kolla whole design | 03:39 |
dupengfei | Jeffrey4l : Right now, the kolla build the configure for k8s, it is very inappositely, but changing is now now。 | 03:40 |
dupengfei | changing is not now | 03:41 |
Jeffrey4l | mix the ansible value and k8s value will be confused for end-user. | 03:42 |
*** tonanhngo has joined #openstack-kolla | 03:42 | |
dupengfei | Jeffrey4l: OK? i will change in this way. | 03:42 |
*** tonanhngo has quit IRC | 03:42 | |
Jeffrey4l | dupengfei, which way? | 03:44 |
Jeffrey4l | I want to change in this way: to use "memcached" and "glance-registry" inside the *.j2 files, what 's up ? // this? | 03:44 |
dupengfei | I want to change in this way: to use "memcached" and "glance-registry" inside the *.j2 files. | 03:44 |
*** eaguilar has quit IRC | 03:44 | |
dupengfei | year | 03:44 |
dupengfei | Jeffrey4l year | 03:44 |
Jeffrey4l | hmm. why not try to introduce a new configuration file hold all the k8s related variables? | 03:44 |
dupengfei | Jeffrey4l: I just fix a small bug, not a whole design, it is a great work. | 03:45 |
Jeffrey4l | dupengfei, before fixing this small bug, i think it is worth to talk about the design part. we should determine which solution we will task and is better. | 03:47 |
dupengfei | Jeffrey4l: Kolla-k8s variables need a new design to make. About all j2 files | 03:47 |
*** yuanying has joined #openstack-kolla | 03:47 | |
Jeffrey4l | dupengfei, when you add such options. it is hard to remove in the future if we think it is bad. | 03:47 |
dupengfei | Jeffrey4l: Big project on the way and we must make kolla-k8s works ok now. | 03:47 |
dupengfei | Jeffrey4l: There are many options more the this 2 are alreay in the code; | 03:48 |
dupengfei | Jeffrey4l: I want to make kolla-k8s works right now, and discuss how to remove them in the future. | 03:49 |
*** yuanying has quit IRC | 03:49 | |
dupengfei | Jeffrey4l: Just the defference as bug and bp; | 03:49 |
kbaegis | grrrr | 03:50 |
kbaegis | Please? | 03:51 |
dupengfei | Jeffrey4l: I have made a big change about k8s before and spent much time, but is refused, see " https://review.openstack.org/#/c/369386/ " | 03:51 |
kbaegis | The horizon container seems immune to plugins | 03:51 |
*** kbaegis has quit IRC | 03:51 | |
*** kbaegis has joined #openstack-kolla | 03:52 | |
*** kbaegis has quit IRC | 03:52 | |
*** kbaegis has joined #openstack-kolla | 03:52 | |
Jeffrey4l | dupengfei, so we need choose the right solution first, rather than fix it first with no matter solution. | 03:53 |
openstackgerrit | dupengfei proposed openstack/kolla: Fix keystone,database,memcache for k8s. https://review.openstack.org/372089 | 03:53 |
Jeffrey4l | dupengfei, rule 1: mix the k8s and ansible variable is very bad. | 03:54 |
Jeffrey4l | we may not separate them in the future. | 03:54 |
*** mdnadeem has joined #openstack-kolla | 03:55 | |
*** yuanying has joined #openstack-kolla | 03:55 | |
*** zhubingbing_ has quit IRC | 03:59 | |
*** schwicht has quit IRC | 04:08 | |
*** schwicht has joined #openstack-kolla | 04:09 | |
*** kbaegis has quit IRC | 04:09 | |
*** kbaegis has joined #openstack-kolla | 04:18 | |
*** schwicht has quit IRC | 04:19 | |
openstackgerrit | OpenStack Proposal Bot proposed openstack/kolla: Updated from global requirements https://review.openstack.org/375989 | 04:21 |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Remove clients from test-requirements already installed by python-openstackclient https://review.openstack.org/375327 | 04:21 |
*** schwicht has joined #openstack-kolla | 04:22 | |
*** bdaca has joined #openstack-kolla | 04:25 | |
bdaca | hello all | 04:25 |
*** kbaegis has quit IRC | 04:26 | |
*** schwicht has quit IRC | 04:29 | |
openstackgerrit | Bartłomiej Daca proposed openstack/kolla: Repair swift reconfiguration https://review.openstack.org/376152 | 04:32 |
*** schwicht has joined #openstack-kolla | 04:32 | |
*** duonghq has joined #openstack-kolla | 04:34 | |
*** duonghq_ has quit IRC | 04:36 | |
*** schwicht has quit IRC | 04:37 | |
*** sdake has joined #openstack-kolla | 04:41 | |
*** sdake has quit IRC | 04:45 | |
*** kbaegis has joined #openstack-kolla | 04:55 | |
*** coolsvap has joined #openstack-kolla | 04:57 | |
*** kbaegis has quit IRC | 05:02 | |
openstackgerrit | Sajauddin proposed openstack/kolla: Changed docker_restart_policy to unless-stopped https://review.openstack.org/375322 | 05:04 |
*** yee3793 has joined #openstack-kolla | 05:11 | |
openstackgerrit | caoyuan proposed openstack/kolla: Update the description when create Glance service endpoints https://review.openstack.org/376006 | 05:11 |
*** SaMnCo_ has joined #openstack-kolla | 05:13 | |
*** sbezverk_ has joined #openstack-kolla | 05:13 | |
*** jidar_ has joined #openstack-kolla | 05:13 | |
*** NobodyCam_ has joined #openstack-kolla | 05:13 | |
*** akwasnie_ has joined #openstack-kolla | 05:13 | |
*** nick-ma_ has joined #openstack-kolla | 05:14 | |
*** elemoine__ has joined #openstack-kolla | 05:14 | |
*** DuncanT_ has joined #openstack-kolla | 05:15 | |
*** serverascode_ has joined #openstack-kolla | 05:15 | |
*** igordcar1 has joined #openstack-kolla | 05:16 | |
*** Guest66676 has joined #openstack-kolla | 05:16 | |
*** senk_ has joined #openstack-kolla | 05:17 | |
*** nihilifer_ has joined #openstack-kolla | 05:18 | |
*** dgonzalez_ has joined #openstack-kolla | 05:19 | |
*** amaged___ has joined #openstack-kolla | 05:19 | |
*** clayton_ has joined #openstack-kolla | 05:20 | |
*** cu5_ has joined #openstack-kolla | 05:20 | |
*** sbezverk has quit IRC | 05:21 | |
*** peterjenkins_ has quit IRC | 05:21 | |
*** manjeets has quit IRC | 05:21 | |
*** igordcard has quit IRC | 05:21 | |
*** elemoine_ has quit IRC | 05:21 | |
*** serverascode has quit IRC | 05:21 | |
*** SaMnCo has quit IRC | 05:21 | |
*** NobodyCam has quit IRC | 05:21 | |
*** ayoung has quit IRC | 05:21 | |
*** clayton has quit IRC | 05:21 | |
*** yee379 has quit IRC | 05:21 | |
*** amaged__ has quit IRC | 05:21 | |
*** akwasnie has quit IRC | 05:21 | |
*** nick-ma has quit IRC | 05:21 | |
*** cu5 has quit IRC | 05:21 | |
*** DuncanT has quit IRC | 05:21 | |
*** Mech422 has quit IRC | 05:21 | |
*** hogepodge has quit IRC | 05:21 | |
*** jidar has quit IRC | 05:21 | |
*** dgonzalez has quit IRC | 05:21 | |
*** timss has quit IRC | 05:21 | |
*** nihilifer has quit IRC | 05:21 | |
*** Guest66666 has quit IRC | 05:21 | |
*** Kdecherf has quit IRC | 05:21 | |
*** logan- has quit IRC | 05:21 | |
*** dgonzalez_ is now known as dgonzalez | 05:21 | |
*** Mech422 has joined #openstack-kolla | 05:21 | |
*** manjeets has joined #openstack-kolla | 05:21 | |
*** clayton_ is now known as clayton | 05:21 | |
*** akwasnie_ is now known as akwasnie | 05:21 | |
*** logan- has joined #openstack-kolla | 05:21 | |
*** mark-casey has joined #openstack-kolla | 05:21 | |
*** SaMnCo_ is now known as SaMnCo | 05:21 | |
*** mark-casey has quit IRC | 05:21 | |
*** elemoine__ is now known as elemoine_ | 05:21 | |
*** NobodyCam_ is now known as NobodyCam | 05:22 | |
*** nick-ma_ is now known as nick-ma | 05:24 | |
*** amaged___ is now known as amaged__ | 05:24 | |
*** ayoung has joined #openstack-kolla | 05:27 | |
*** timss has joined #openstack-kolla | 05:27 | |
*** Kdecherf has joined #openstack-kolla | 05:28 | |
*** peterjenkins_ has joined #openstack-kolla | 05:30 | |
*** serverascode_ is now known as serverascode | 05:31 | |
*** schwicht has joined #openstack-kolla | 05:32 | |
*** DuncanT_ is now known as DuncanT | 05:36 | |
*** bjolo has joined #openstack-kolla | 05:37 | |
bjolo | morning | 05:37 |
*** nihilifer_ is now known as nihilifer | 05:38 | |
*** caowei has joined #openstack-kolla | 05:41 | |
*** neilus has joined #openstack-kolla | 05:43 | |
*** tonanhngo has joined #openstack-kolla | 05:46 | |
*** zhubingbing has joined #openstack-kolla | 05:48 | |
*** tonanhngo has quit IRC | 05:48 | |
*** neilus has quit IRC | 05:49 | |
*** unicell1 has quit IRC | 05:49 | |
*** unicell has joined #openstack-kolla | 05:50 | |
openstackgerrit | Martin Matyas proposed openstack/kolla: Fix Magnum trustee issues https://review.openstack.org/326163 | 05:53 |
*** senk_ has quit IRC | 05:55 | |
*** haplo37_ has quit IRC | 05:56 | |
*** imcsk8_ has joined #openstack-kolla | 05:58 | |
*** haplo37_ has joined #openstack-kolla | 05:58 | |
*** zhubingbing__ has joined #openstack-kolla | 05:59 | |
*** zhubingbing has quit IRC | 06:00 | |
*** imcsk8 has quit IRC | 06:02 | |
openstackgerrit | Martin Matyas proposed openstack/kolla: Fix Magnum trustee issues https://review.openstack.org/326163 | 06:14 |
*** jidar_ is now known as jidar | 06:24 | |
*** egonzalez90 has joined #openstack-kolla | 06:25 | |
*** senk_ has joined #openstack-kolla | 06:32 | |
*** ajpower has quit IRC | 06:43 | |
*** huhaoran has joined #openstack-kolla | 06:44 | |
*** ajpower has joined #openstack-kolla | 06:50 | |
openstackgerrit | zhubingbing proposed openstack/kolla: add check cloudkitty_api_port and cloudkitty api haproxy in port_checks.yml https://review.openstack.org/376222 | 06:54 |
*** tonanhngo has joined #openstack-kolla | 07:01 | |
*** tonanhngo has quit IRC | 07:02 | |
*** jascott1_ has quit IRC | 07:17 | |
*** jascott1_ has joined #openstack-kolla | 07:18 | |
*** cu5_ is now known as cu5 | 07:21 | |
*** msimonin has joined #openstack-kolla | 07:24 | |
*** matrohon has joined #openstack-kolla | 07:25 | |
*** neilus has joined #openstack-kolla | 07:32 | |
openstackgerrit | zhubingbing proposed openstack/kolla: change CollectD to collectd in collectd role https://review.openstack.org/376245 | 07:32 |
openstackgerrit | zhubingbing proposed openstack/kolla: add check cloudkitty prot and api haproxy in port_checks.yml https://review.openstack.org/376222 | 07:35 |
*** neilus has quit IRC | 07:37 | |
openstackgerrit | Merged openstack/kolla: Remove clients from test-requirements already installed by python-openstackclient https://review.openstack.org/375327 | 07:38 |
openstackgerrit | zhubingbing proposed openstack/kolla: add cloudkitty port prechecks https://review.openstack.org/376222 | 07:38 |
*** bdaca has quit IRC | 07:51 | |
*** bdaca has joined #openstack-kolla | 07:56 | |
*** bdaca has quit IRC | 07:57 | |
*** athomas has joined #openstack-kolla | 08:01 | |
*** ankush has joined #openstack-kolla | 08:03 | |
*** phuongnh has joined #openstack-kolla | 08:07 | |
*** bdaca has joined #openstack-kolla | 08:16 | |
*** gfidente has joined #openstack-kolla | 08:18 | |
*** daneyon has joined #openstack-kolla | 08:25 | |
openstackgerrit | zhubingbing proposed openstack/kolla: change CollectD to collectd in collectd role https://review.openstack.org/376245 | 08:30 |
*** daneyon has quit IRC | 08:30 | |
*** rmart04 has joined #openstack-kolla | 08:36 | |
*** Serlex has joined #openstack-kolla | 08:38 | |
*** jascott1_ has quit IRC | 08:42 | |
*** jascott1_ has joined #openstack-kolla | 08:43 | |
*** jascott1_ has quit IRC | 08:48 | |
*** g3ek has quit IRC | 09:10 | |
*** haplo37 has quit IRC | 09:10 | |
*** g3ek has joined #openstack-kolla | 09:10 | |
*** lujinluo has quit IRC | 09:10 | |
jmccarthy | Why does nova upgrade have extra logic for 'Checking if conductor container needs upgrading' ? | 09:11 |
*** haplo37 has joined #openstack-kolla | 09:11 | |
openstackgerrit | Sajauddin proposed openstack/kolla: Changed docker_restart_policy to unless-stopped https://review.openstack.org/375322 | 09:12 |
*** msimonin has quit IRC | 09:18 | |
*** bdaca has quit IRC | 09:28 | |
*** yingjun has quit IRC | 09:31 | |
*** tonanhngo has joined #openstack-kolla | 09:32 | |
*** msimonin has joined #openstack-kolla | 09:33 | |
*** tonanhngo has quit IRC | 09:33 | |
*** zhubingbing__ has quit IRC | 09:33 | |
*** zhubingbing__ has joined #openstack-kolla | 09:34 | |
*** msimonin has quit IRC | 09:34 | |
openstackgerrit | caoyuan proposed openstack/kolla: Fix the url of Cinder endpoints https://review.openstack.org/375169 | 09:38 |
*** flaper87 has joined #openstack-kolla | 09:39 | |
*** flaper87 has quit IRC | 09:39 | |
*** flaper87 has joined #openstack-kolla | 09:39 | |
openstackgerrit | caoyuan proposed openstack/kolla: Fix the url of Cinder endpoints https://review.openstack.org/375169 | 09:40 |
*** tonanhngo has joined #openstack-kolla | 09:53 | |
*** msimonin has joined #openstack-kolla | 09:54 | |
*** msimonin has quit IRC | 09:54 | |
*** tonanhngo has quit IRC | 09:54 | |
*** haplo37_ has quit IRC | 09:55 | |
*** haplo37_ has joined #openstack-kolla | 09:57 | |
*** papacz has joined #openstack-kolla | 09:59 | |
*** zhurong has quit IRC | 10:01 | |
*** sdake has joined #openstack-kolla | 10:06 | |
sdake | yo | 10:06 |
openstackgerrit | zhubingbing proposed openstack/kolla: add upgrade.yml in congress role https://review.openstack.org/376347 | 10:09 |
*** msimonin has joined #openstack-kolla | 10:21 | |
*** huhaoran has quit IRC | 10:26 | |
*** huhaoran has joined #openstack-kolla | 10:26 | |
*** huhaoran has joined #openstack-kolla | 10:27 | |
openstackgerrit | Wei Cao proposed openstack/kolla: Adding the steps of generateing ssh key in tools/init-runonce https://review.openstack.org/376372 | 10:29 |
coolsvap | sdake: hi morning! | 10:34 |
sdake | sup coolsvap | 10:34 |
*** huhaoran has quit IRC | 10:35 | |
coolsvap | good start to the week, my visa approved for spain, all set for barcelona now | 10:36 |
*** berendt has joined #openstack-kolla | 10:41 | |
berendt | hi | 10:41 |
*** rhallisey has joined #openstack-kolla | 10:55 | |
*** haplo37_ has quit IRC | 10:56 | |
*** prithiv has joined #openstack-kolla | 10:56 | |
*** prithiv has quit IRC | 10:56 | |
*** kbaegis has joined #openstack-kolla | 10:56 | |
kbaegis | Good morning everyone | 10:57 |
*** dave-mccowan has joined #openstack-kolla | 10:57 | |
*** haplo37_ has joined #openstack-kolla | 10:59 | |
sdake | good to hear coolsvap | 10:59 |
kbaegis | sdake: I can't seem to extend horizon. Any clues? | 11:00 |
*** duonghq has quit IRC | 11:01 | |
kbaegis | sdake: I'm trying to get the various dashboard-plugins | 11:02 |
*** sdake_ has joined #openstack-kolla | 11:04 | |
*** sdake has quit IRC | 11:04 | |
*** phuongnh has quit IRC | 11:06 | |
*** daneyon has joined #openstack-kolla | 11:08 | |
*** daneyon has quit IRC | 11:13 | |
berendt | when is the change of the PTL? before or after Barcelona? | 11:13 |
*** mkoderer has joined #openstack-kolla | 11:13 | |
kbaegis | Have elections already happened? | 11:15 |
berendt | kbaegis yes, we already have results, the upcoming PTL of kolla is Michal Jastrzebski | 11:15 |
*** huhaoran has joined #openstack-kolla | 11:16 | |
sdake_ | berendt to be more specific, the ptl is elected to serve immediately | 11:17 |
sdake_ | berendt however I still maintain responsibility for rc2 and 3.0.0 | 11:18 |
sdake_ | the tc calls this person a "Release Steward" | 11:18 |
sdake_ | this person being me | 11:18 |
kbaegis | sdake_: Well, excellent work on the 3.0 :) | 11:18 |
openstackgerrit | Wei Cao proposed openstack/kolla: Adding the steps of generateing ssh key in tools/init-runonce https://review.openstack.org/376372 | 11:19 |
sdake_ | kbaegis well that part isn't compelte yet ;) | 11:19 |
sdake_ | that is my problem to solve - along with the core reviewer teams | 11:19 |
kbaegis | sdake_: Sure. It's come strides though | 11:19 |
openstackgerrit | Merged openstack/kolla: add cloudkitty port prechecks https://review.openstack.org/376222 | 11:20 |
kbaegis | Thanks everyone for being so helpful | 11:20 |
sdake_ | for the moment - id expect inc0 and I will share responsibilities as we have a tidy handoff of ptl responsibilities | 11:20 |
sdake_ | as to have that is ;) | 11:20 |
kbaegis | ooc, what is inc0's role? | 11:20 |
sdake_ | hes the ptl dude | 11:21 |
sdake_ | he will be running meanings | 11:21 |
sdake_ | meetings | 11:21 |
sdake_ | interacting with the foundation | 11:21 |
sdake_ | that sort of thing | 11:21 |
kbaegis | I didn't realize his handle was inc0 | 11:21 |
sdake_ | keeping the core review team unblocked | 11:21 |
openstackgerrit | Merged openstack/kolla: fix ironic-inspector setup https://review.openstack.org/374727 | 11:22 |
sdake_ | ptl has alot of responsibilites - they mostly focus around serving the core review team and extended developer family we have | 11:22 |
openstackgerrit | Merged openstack/kolla: install iptables in ironic-inspector container https://review.openstack.org/374970 | 11:22 |
*** rmart04 has quit IRC | 11:23 | |
openstackgerrit | Merged openstack/kolla: Revert "Bind EPMD to api interface address" https://review.openstack.org/375790 | 11:25 |
openstackgerrit | Wei Cao proposed openstack/kolla: Adding the steps of generateing ssh key in tools/init-runonce https://review.openstack.org/376372 | 11:33 |
pbourke__ | kbaegis: what issue are you having with horizon? | 11:43 |
*** sdake_ has quit IRC | 11:43 | |
kbaegis | pbourke__: I never seem to get more than a base installation. I'd like sahara-dashboard, murano-dashboard, etc | 11:45 |
kbaegis | pbourke__: I've been trying to add them, but I can't seem to get them loaded. | 11:45 |
kbaegis | The endpoints are there, the cli works fine | 11:46 |
pbourke__ | kbaegis: have you read http://docs.openstack.org/developer/kolla/image-building.html#plugin-functionality ? | 11:46 |
kbaegis | Yes | 11:46 |
kbaegis | I'll write a proper extension at some point | 11:47 |
pbourke__ | so you have a template overrides file with a block for horizon | 11:47 |
kbaegis | I'm trying it manually first, because I've never actually had to do this | 11:47 |
pbourke__ | and rebuilt horizon? | 11:47 |
kbaegis | pbourke__: I suppose I'm not doing this as part of the container build. I'm trying it on top of the horizon container directly | 11:49 |
kbaegis | pbourke__: Is that a totally invalid approach? | 11:49 |
berendt | kbaegis yes sounds invalid | 11:49 |
pbourke__ | kbaegis: not totally, but its messy | 11:49 |
pbourke__ | kbaegis: much easier to build a new one | 11:50 |
kbaegis | so what I should be doing is RUN git clone https://github.com/openstack/sahara-dashboard ; pip --no-cache-dir install sahara-dashboard, for example? | 11:51 |
openstackgerrit | zhubingbing proposed openstack/kolla: add reconfigure.yml in congress role https://review.openstack.org/376421 | 11:52 |
openstackgerrit | zhubingbing proposed openstack/kolla: add reconfigure.yml in congress role https://review.openstack.org/376421 | 11:53 |
kbaegis | Presumably in the footer? | 11:57 |
*** mliima has joined #openstack-kolla | 11:58 | |
pbourke__ | kbaegis: yes | 11:58 |
*** zhurong has joined #openstack-kolla | 11:59 | |
openstackgerrit | zhubingbing proposed openstack/kolla: change perm 0644 to 0600 in congress role https://review.openstack.org/376425 | 12:00 |
*** haplo37 has quit IRC | 12:01 | |
bjolo | launchpad down | 12:03 |
kbaegis | pbourke__, berendt: thanks guygs | 12:03 |
kbaegis | *guys | 12:04 |
kbaegis | Evidently too early :) | 12:05 |
*** prithiv has joined #openstack-kolla | 12:06 | |
*** prithiv has quit IRC | 12:06 | |
*** haplo37 has joined #openstack-kolla | 12:09 | |
*** zhurong has quit IRC | 12:09 | |
*** zhurong has joined #openstack-kolla | 12:11 | |
bjolo | kbaegis, i have on my list to get the plugins working. Launchpad is down, but i have filed a bug on it | 12:14 |
bjolo | kbaegis, the interesting part is that heat dashboard gets activate if heat is deployed or not. | 12:14 |
*** caowei has quit IRC | 12:14 | |
kbaegis | bjolo: Isn't heat-dashboard part of horizon natively? | 12:15 |
kbaegis | bjolo: I'd check, but I don't remember seeing it | 12:15 |
bjolo | kbaegis, deploy w/o heat -> no heat dashboard. enable_heat yes; deploy -t heat -> heat dashboard | 12:15 |
bjolo | perhaps, but it is interesting how it just appears when heat is deployed | 12:16 |
kbaegis | Trying to get homebrew to work with gpg 8192 bit keys atm | 12:16 |
kbaegis | :) | 12:16 |
bjolo | the horizon container is untouched | 12:16 |
berendt | bjolo heat dashboard is part of horizon core | 12:16 |
kbaegis | bjolo: I think it's supposed to do an endpoint discovery w/ keystone and populate any services it can find | 12:16 |
berendt | bjolo lbaas for neutron, murano and dashboards like this are not part of horizon core | 12:16 |
kbaegis | berendt: That seems to be the direction. I'm expecting heat to be forked as well | 12:17 |
berendt | when you install murano dashboard inside the horizon container and do not provide murano endpoints the murano dashboard is not visible | 12:17 |
bjolo | will all dashboards to endpoint discovery or just heat? | 12:17 |
bjolo | s/to/do/ | 12:18 |
berendt | i think it is working for all dashboards | 12:18 |
bjolo | ok | 12:19 |
bjolo | will try | 12:20 |
*** zhubingbing__ has quit IRC | 12:23 | |
openstackgerrit | OpenStack Proposal Bot proposed openstack/kolla: Updated from global requirements https://review.openstack.org/375989 | 12:23 |
*** zhubingbing__ has joined #openstack-kolla | 12:24 | |
*** zhubingbing__ has quit IRC | 12:24 | |
*** tonanhngo has joined #openstack-kolla | 12:25 | |
*** lamt has quit IRC | 12:27 | |
*** tonanhngo has quit IRC | 12:27 | |
*** prithiv has joined #openstack-kolla | 12:27 | |
*** schwicht has quit IRC | 12:28 | |
*** kbaegis has quit IRC | 12:41 | |
*** kbaegis has joined #openstack-kolla | 12:42 | |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Rename the gnocchi api-paste filename https://review.openstack.org/376092 | 12:51 |
*** dmsimard2 is now known as dmsimard | 12:51 | |
*** david-lyle has joined #openstack-kolla | 12:56 | |
*** schwicht has joined #openstack-kolla | 12:57 | |
*** jtriley has joined #openstack-kolla | 12:57 | |
*** sdake has joined #openstack-kolla | 12:57 | |
*** b_bezak has joined #openstack-kolla | 12:58 | |
*** sdake_ has joined #openstack-kolla | 13:03 | |
*** sdake has quit IRC | 13:06 | |
*** dwalsh has joined #openstack-kolla | 13:09 | |
jmccarthy | Anyone run into an error when upgrading where the rabbitmq cluster complains: '{error,{corrupt_or_missing_cluster_files,{error,enoent},{error,enoent}}}' ? | 13:12 |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Make the service user only have read permission for config https://review.openstack.org/376465 | 13:12 |
*** caowei has joined #openstack-kolla | 13:13 | |
kbaegis | Can someone remind me how to ONLY redeploy horizon? | 13:15 |
kbaegis | —tag? | 13:15 |
*** ankush has left #openstack-kolla | 13:19 | |
*** LamT_ has joined #openstack-kolla | 13:23 | |
berendt | kbaegis i think o | 13:23 |
Daviey | Hey, can someone look at https://review.openstack.org/#/c/372161/ please? | 13:24 |
Daviey | Ceilometer is currently completely fubared on Newton | 13:25 |
*** lamt has joined #openstack-kolla | 13:26 | |
*** schwicht has quit IRC | 13:29 | |
sdake_ | jmccarthy haven't seen that one | 13:29 |
bjolo | kbaegis, look in site.yml | 13:32 |
bjolo | kbaegis, should be -t horizon | 13:32 |
*** skramaja_ has joined #openstack-kolla | 13:34 | |
openstackgerrit | Merged openstack/kolla: Fix the failure of ceilometer-api container https://review.openstack.org/372161 | 13:35 |
kbaegis | the plugin didn't appear to work | 13:35 |
*** skramaja has quit IRC | 13:36 | |
*** tonanhngo has joined #openstack-kolla | 13:39 | |
kbaegis | My fault. | 13:39 |
kbaegis | :) | 13:40 |
*** kbaegis has quit IRC | 13:40 | |
*** tonanhngo has quit IRC | 13:40 | |
*** skramaja has joined #openstack-kolla | 13:45 | |
*** skramaja_ has quit IRC | 13:46 | |
*** fguillot has joined #openstack-kolla | 13:46 | |
*** brad[] has joined #openstack-kolla | 13:47 | |
*** sdake has joined #openstack-kolla | 13:47 | |
*** eaguilar has joined #openstack-kolla | 13:49 | |
*** schwicht has joined #openstack-kolla | 13:50 | |
*** sdake_ has quit IRC | 13:50 | |
*** sdake_ has joined #openstack-kolla | 13:51 | |
*** cinerama` is now known as cinerama | 13:52 | |
*** sdake has quit IRC | 13:54 | |
*** rmart04 has joined #openstack-kolla | 13:57 | |
sdake_ | https://image-store.slidesharecdn.com/b5eceaaa-15af-44dd-8c06-0995ea2e8647-original.jpeg | 13:57 |
sdake_ | most proud of: | 13:57 |
sdake_ | Try to take a sinking ship and make it float | 13:58 |
sdake_ | groan | 13:58 |
sdake_ | ceo's using the word try | 13:58 |
sdake_ | emberrassing | 13:58 |
*** MarMat has joined #openstack-kolla | 14:00 | |
*** zhurong has quit IRC | 14:01 | |
*** tonanhngo has joined #openstack-kolla | 14:02 | |
pbourke__ | nice design though | 14:03 |
openstackgerrit | Andrew Widdersheim proposed openstack/kolla: Remove redundant common play and add haproxy hosts https://review.openstack.org/376524 | 14:03 |
*** huhaoran has quit IRC | 14:04 | |
sdake_ | pbourke__ you got that ;) | 14:04 |
sdake_ | all flash no substance | 14:04 |
sdake_ | the shit about the light really rubs me the wrong way | 14:04 |
sdake_ | its a leader's job to make sure other don't stand in their sshadows | 14:04 |
*** awiddersheim has joined #openstack-kolla | 14:05 | |
sdake_ | anyway shes a ceo and i'm not so I guess she knows something I dont ;) | 14:05 |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Add a note for how to get prior release kolla doc https://review.openstack.org/376525 | 14:05 |
*** inc0 has joined #openstack-kolla | 14:06 | |
*** b_bezak has quit IRC | 14:06 | |
*** b_bezak has joined #openstack-kolla | 14:07 | |
inc0 | uff good morning | 14:07 |
egonzalez90 | morning inc0 | 14:09 |
*** imcsk8_ is now known as imcsk8 | 14:09 | |
sdake_ | pbourke__ fwiw this is her tool shee used to create her resume if your interested :) https://enhancv.com/ | 14:09 |
*** haplo37 has quit IRC | 14:09 | |
*** lrensing has joined #openstack-kolla | 14:09 | |
pbourke__ | sdake_: yeah was just looking at it there | 14:10 |
sdake_ | sup inc0 | 14:10 |
pbourke__ | sdake_: all well and good till everyones CVs start to look like that :) | 14:10 |
sdake_ | haha | 14:10 |
sdake_ | ya no originality ;) | 14:10 |
pbourke__ | inc0: congrats on the election | 14:10 |
sdake_ | did you find their price sheet yet | 14:10 |
inc0 | thanks:) | 14:10 |
sdake_ | inc0 ya you missed numerous praising today - check archive ;) | 14:10 |
inc0 | lol, looking | 14:11 |
inc0 | never miss any praising;) | 14:11 |
sbezverk_ | inc0: qq when you use jinja2 macro and keep them in a separate file, do you know how ansible loads this extra template to use during the main template rendering? | 14:11 |
*** b_bezak has quit IRC | 14:11 | |
*** g3ek has quit IRC | 14:11 | |
pbourke__ | sbezverk_: we only use macros as part of the build process afaik | 14:12 |
sdake_ | Show who you really are | 14:12 |
sdake_ | Enhancv helps you express your true personality and show the real person behind the resume | 14:12 |
inc0 | yeah, I was drunk by the time praises came;) | 14:12 |
sdake_ | I thought that was what the resume was for ;) | 14:12 |
pbourke__ | inc0: lol | 14:12 |
sdake_ | rather interiew | 14:12 |
*** g3ek has joined #openstack-kolla | 14:12 | |
inc0 | who is writing a CV? | 14:13 |
*** haplo37 has joined #openstack-kolla | 14:13 | |
sdake_ | "It's freee" | 14:13 |
berendt | inc0 i do not know, i never wrote one | 14:13 |
sdake_ | inc0 https://image-store.slidesharecdn.com/b5eceaaa-15af-44dd-8c06-0995ea2e8647-original.jpeg | 14:13 |
sbezverk_ | inc0: it is not for kolla, it is for kolla-kube in order to be able to use small chunks/modules and then to use them in the main service template with {% import... | 14:13 |
*** absubram has joined #openstack-kolla | 14:14 | |
inc0 | sbezverk_, macros we wrote so far aren't used by ansible | 14:14 |
inc0 | it's for building | 14:14 |
inc0 | but it should be fairly easy to write one for ansible too | 14:14 |
sbezverk_ | inc0: here is example you have your main template as Dockerfile.j2 and inside you have import macro which is located in a separate file, could you point me to the place in python code where you load these two pieces for combined rendering? | 14:16 |
*** lrensing has quit IRC | 14:16 | |
inc0 | hold on | 14:17 |
inc0 | https://github.com/openstack/kolla/blob/master/docker/cinder/cinder-api/Dockerfile.j2#L6 | 14:17 |
*** neilus has joined #openstack-kolla | 14:18 | |
sbezverk_ | inc0: right but can you point me to python code which deals with it | 14:19 |
inc0 | that's it | 14:19 |
inc0 | it will import macros from j2 root dir | 14:19 |
inc0 | which is docker/ in our case | 14:19 |
inc0 | it's jinja2 thing, nothing I wrote | 14:20 |
sbezverk_ | inc0: hm, in kube case it cannot find template with macro. I thoought there was some magic done in python to preload macro template.. | 14:20 |
*** britthouser has joined #openstack-kolla | 14:21 | |
britthouser | morning! | 14:21 |
*** neilus has quit IRC | 14:22 | |
Jeffrey4l | berendt, could u review this? https://review.openstack.org/372737 | 14:23 |
*** zhubingbing_ has joined #openstack-kolla | 14:23 | |
*** wirehead_1 has quit IRC | 14:23 | |
inc0 | sbezverk_, it's jinja2 thing, it should just work | 14:24 |
berendt | Jeffrey4l tomorrow, i do not have time for tests today | 14:24 |
Jeffrey4l | ok. thanks. | 14:24 |
Jeffrey4l | inc0, sdake_ this PS please https://review.openstack.org/372737 | 14:24 |
*** wirehead_ has joined #openstack-kolla | 14:25 | |
*** Guest92615 is now known as zeus | 14:26 | |
*** zeus has quit IRC | 14:26 | |
*** zeus has joined #openstack-kolla | 14:26 | |
*** lrensing has joined #openstack-kolla | 14:26 | |
zhubingbing_ | hi | 14:29 |
zhubingbing_ | http://logs.openstack.org/89/375989/4/gate/gate-kolla-pep8-ubuntu-xenial/925a4e8/console.html#_2016-09-26_14_16_27_425524 | 14:29 |
zhubingbing_ | what's problems? | 14:29 |
zhubingbing_ | who can help me | 14:29 |
Jeffrey4l | zhubingbing_, seems a CI issue. how about try `recheck`? | 14:30 |
sbezverk_ | inc0: here is what I see, it seems "just work" is not the case :-( http://paste.openstack.org/show/582960/ | 14:32 |
*** zhubingbing_ has quit IRC | 14:32 | |
*** prithiv has quit IRC | 14:33 | |
*** britthouser has quit IRC | 14:33 | |
sdake_ | Jeffrey4l trying to teach people how to debug ci gate problems | 14:34 |
sdake_ | Jeffrey4l rther then mindlessly rechecking | 14:34 |
*** zhubingbing_ has joined #openstack-kolla | 14:34 | |
Jeffrey4l | in zhubingbing_ 's log we got | 14:34 |
sdake_ | Jeffrey4l in this cae your correct tho, looking at ogs its a ci issue needss a recheck :) | 14:34 |
Jeffrey4l | /usr/zuul-env/bin/zuul-cloner -m /tmp/tmp.bobcohUrh1 --cache-dir /opt/git --branch git://git.openstack.org openstack/requirements | 14:34 |
Jeffrey4l | this is what ci does. kolla can do nothing for this. | 14:35 |
sdake_ | damn see what melissa just did there | 14:35 |
Jeffrey4l | yes :) | 14:35 |
sdake_ | she made me use the word try | 14:35 |
sdake_ | try is an acknowledgement of failure before even getting going | 14:37 |
sdake_ | taking a look = acknowledgement that failure is an option that could occur | 14:37 |
sdake_ | thats what i get for reading inked in articles ;) | 14:37 |
Jeffrey4l | what's melissa? | 14:38 |
sdake_ | https://image-store.slidesharecdn.com/b5eceaaa-15af-44dd-8c06-0995ea2e8647-original.jpeg | 14:38 |
sdake_ | she was the ceo of yahoo | 14:39 |
sdake_ | or still may be i guess | 14:39 |
Jeffrey4l | got. | 14:39 |
*** senk_ has quit IRC | 14:39 | |
Jeffrey4l | seems a bad ceo ever ;) | 14:39 |
sdake_ | she didn't add any value - thats for sure | 14:40 |
sdake_ | i dont now yahoo however | 14:41 |
sdake_ | so hard for me to say | 14:41 |
sdake_ | did i ever tell you the time my manager at intel said "we have 7 mllion for capex budget, we need to spend it in the next 2 weeks" | 14:42 |
sdake_ | deer in headlights looking back ;) | 14:43 |
wznoinsk | sdake_: sweet :-) | 14:44 |
sdake_ | i scored a display out of the deal | 14:44 |
sdake_ | that is all I could think of | 14:44 |
sdake_ | we had 20 palettes of equipment going to recycle per week | 14:44 |
Jeffrey4l | like what State-owned enterprise does in china :) | 14:44 |
sdake_ | an entire wall in c6 near the c6 lab on first floor | 14:45 |
*** b_bezak has joined #openstack-kolla | 14:45 | |
wirehead_ | My yahoo buddies were trying to convince me to come back when Marissa joined. | 14:45 |
sdake_ | wirehead_ good move dude | 14:45 |
wirehead_ | They told me it was totally different, and then told me about schooling a programmer that writing Java code with 50 threads that all were waiting on a lock was a bad idea. | 14:47 |
wirehead_ | That was different. | 14:47 |
wirehead_ | Just not better. :) | 14:47 |
sdake_ | | o | | 14:47 |
*** egonzalez90 has quit IRC | 14:49 | |
sdake_ | 7 million in capex could fund a fantastic startup | 14:50 |
*** dmsimard has quit IRC | 14:52 | |
*** eaguilar has quit IRC | 14:52 | |
inc0 | sbezverk_, you might need to get into code to find out what is real root dir for templates there | 14:52 |
*** b_bezak has quit IRC | 14:53 | |
*** zhubingbing_ has quit IRC | 14:54 | |
inc0 | to retry particular zuul job you can use reproduce.sh | 14:55 |
inc0 | but you need images from infra | 14:55 |
Jeffrey4l | fyi: gate ci bust due openstack-infra, the fix is merge | 14:57 |
*** kbaegis has joined #openstack-kolla | 14:57 | |
*** hogepodge has joined #openstack-kolla | 14:58 | |
*** jtriley has quit IRC | 14:59 | |
*** jtriley has joined #openstack-kolla | 14:59 | |
*** sdake_ has quit IRC | 14:59 | |
*** Marcellin__ has joined #openstack-kolla | 15:01 | |
*** kbaegis has quit IRC | 15:02 | |
*** kbaegis has joined #openstack-kolla | 15:02 | |
*** huhaoran has joined #openstack-kolla | 15:04 | |
openstackgerrit | Christian Berendt proposed openstack/kolla: Use keystone-paste.ini template for keystone https://review.openstack.org/373379 | 15:04 |
*** zhubingbing_ has joined #openstack-kolla | 15:05 | |
*** zhubingbing_ has quit IRC | 15:07 | |
*** huhaoran has quit IRC | 15:09 | |
*** zhubingbing_ has joined #openstack-kolla | 15:09 | |
wirehead_ | oooh. Breakfast disaster averted. Only put salsa in my omlette, forgot to put hot sauce on it as well. | 15:12 |
*** bdaca has joined #openstack-kolla | 15:13 | |
*** lrensing has quit IRC | 15:13 | |
*** zhubingbing_ has quit IRC | 15:13 | |
*** britthouser has joined #openstack-kolla | 15:14 | |
*** sdake has joined #openstack-kolla | 15:15 | |
britthouser | morening sdake! | 15:17 |
sdake | sup britthouser | 15:17 |
sdake | i've actually been awake forawhile :) | 15:17 |
britthouser | In SFO this week. | 15:17 |
britthouser | so in your timezone. :smiley: | 15:17 |
sdake | must be easier then your normal timezone ;) | 15:18 |
britthouser | Well, body is still in EST | 15:18 |
*** dmsimard has joined #openstack-kolla | 15:18 | |
*** prithiv has joined #openstack-kolla | 15:22 | |
*** prithiv has quit IRC | 15:23 | |
*** openstackstatus has joined #openstack-kolla | 15:27 | |
*** ChanServ sets mode: +v openstackstatus | 15:27 | |
*** caowei has quit IRC | 15:29 | |
*** kbaegis has quit IRC | 15:36 | |
*** SamYaple has joined #openstack-kolla | 15:36 | |
*** harlowja_at_home has joined #openstack-kolla | 15:36 | |
inc0 | hey SamYaple :) nice to see you around! | 15:36 |
inc0 | also everyone, please vote for sessions if you haven't http://civs.cs.cornell.edu/cgi-bin/vote.pl?id=E_8368e1e74f8a0049&akey=91bbcf4baeff0a2f | 15:36 |
*** mdnadeem_ has joined #openstack-kolla | 15:36 | |
berendt | inc0 put it into the room topic | 15:37 |
inc0 | let's see if I can | 15:37 |
*** ccesario has joined #openstack-kolla | 15:37 | |
inc0 | nope:( | 15:37 |
berendt | you should be able to do that | 15:38 |
*** kbaegis has joined #openstack-kolla | 15:38 | |
inc0 | /topic not working | 15:39 |
*** mkoderer has quit IRC | 15:39 | |
inc0 | I'll ask infra folks | 15:39 |
*** athomas has quit IRC | 15:39 | |
rhallisey | inc0, yo | 15:39 |
inc0 | hey Ryan:) | 15:40 |
rhallisey | congrats :) | 15:40 |
berendt | inc0 i think you have to +o first? | 15:40 |
inc0 | thanks! yeah I'll need +o | 15:40 |
*** mdnadeem has quit IRC | 15:40 | |
berendt | inc0 /msg chanserv op #openstack-kolla should give you +o | 15:42 |
-openstackstatus- NOTICE: Earlier job failures for "zuul-cloner: error: too few arguments" should now be solved, and can safely be rechecked | 15:42 | |
MarMat | Jeffrey4l ping | 15:43 |
*** ChanServ sets mode: +o inc0 | 15:43 | |
inc0 | oh, cool | 15:43 |
inc0 | thanks:) | 15:43 |
*** inc0 changes topic to "Please vote for summit sessions! http://civs.cs.cornell.edu/cgi-bin/vote.pl?id=E_8368e1e74f8a0049&akey=91bbcf4baeff0a2f" | 15:44 | |
Jeffrey4l | MarMat, pong | 15:44 |
berendt | inc0 now you kicked the rest of the topic.. | 15:44 |
*** inc0 changes topic to "Please vote for summit sessions! http://civs.cs.cornell.edu/cgi-bin/vote.pl?id=E_8368e1e74f8a0049&akey=91bbcf4baeff0a2f; RC2 DEADLINE (HARD) Oct 10th-Oct 13th; Kolla IRC meetngs on Wednesdays @ 16:00 UTC - see agenda @ https://goo.gl/OXB0DL - IRC channel is *LOGGED* @ http://goo.gl/3mzZ7b (old logs from #kolla http://goo.gl/VKpPzA)" | 15:44 | |
inc0 | yeah I've noticed | 15:44 |
*** mdnadeem has joined #openstack-kolla | 15:45 | |
*** jascott1_ has joined #openstack-kolla | 15:45 | |
MarMat | Jeffrey4l Hello, did you see latest comments in the magnum trustee patchset? Should repeated deployment be supported? | 15:45 |
*** ChanServ sets mode: -o inc0 | 15:45 | |
*** jascott1_ has quit IRC | 15:46 | |
Jeffrey4l | sorry, no. i am checking. | 15:46 |
dmsimard | ohai SamYaple o/ | 15:46 |
inc0 | :( | 15:46 |
*** lrensing has joined #openstack-kolla | 15:46 | |
sdake | MarMat repeated deployment? | 15:46 |
dmsimard | inc0: grats on PTL, let me know if you need anything RDO related | 15:46 |
inc0 | dmsimard, thanks! also you let me know if you need anything RDO related;) | 15:46 |
MarMat | Jeffrey4l yes, like execution of deploy task multiple times | 15:47 |
-openstackstatus- NOTICE: Launchpad SSO is offline, preventing login to https://review.openstack.org/, https://wiki.openstack.org/ and many other sites; no ETA has been provided by the LP admin team | 15:47 | |
*** ChanServ changes topic to "Launchpad SSO is offline, preventing login to https://review.openstack.org/, https://wiki.openstack.org/ and many other sites; no ETA has been provided by the LP admin team" | 15:47 | |
*** mdnadeem_ has quit IRC | 15:47 | |
sdake | MarMat thats the theory - that shuold work | 15:47 |
dmsimard | inc0: I don't have the bandwidth to keep up with Kolla but I can help with questions or trunk/packaging/bugs in regards to RDO | 15:47 |
dmsimard | inc0: are RDO jobs voting yet? :P | 15:48 |
inc0 | dmsimard, sure, I appreciate it | 15:48 |
berendt | why does the infra team change our topic..? | 15:48 |
*** unicell1 has joined #openstack-kolla | 15:48 | |
*** mark-casey1 has joined #openstack-kolla | 15:48 | |
sdake | berendt they will change it back | 15:48 |
sdake | its a global notification system for problems | 15:48 |
inc0 | dmsimard, I don't think they will ever will - RHEL has some license issues which prevents us from putting it into gate afaik | 15:48 |
dmsimard | berendt: it's an escalated notificatoin | 15:48 |
inc0 | sdake, is this accurate? ^ | 15:48 |
dmsimard | inc0: Well, RDO on CentOS, yes ? | 15:48 |
sdake | berendt its mostly to avoid the probelm of 1k people showing up in #openstack-infra and saing "wtf' | 15:49 |
*** unicell has quit IRC | 15:49 | |
dmsimard | inc0: not RDO on RHEL or OSP. | 15:49 |
Jeffrey4l | MarMat, Got. we need all task idempotent. so the task should be able run repeated. | 15:49 |
inc0 | ahh, rdo as repos? | 15:49 |
berendt | ah ok makes sense :) | 15:49 |
inc0 | well, they'll be voting as soon as we have any voting gates;) | 15:49 |
inc0 | which I want to be the case next release | 15:49 |
Jeffrey4l | so make sure create the domain when not exist, or return the exist one. MarMat | 15:49 |
dmsimard | inc0: ack. | 15:50 |
sdake | inc0 good lluck with that one inc0 - i've been cracking on that nut going on 2 years... | 15:50 |
inc0 | (issue is, we already had problems with UCA that would prevent us from merging anything ) | 15:50 |
inc0 | couple weeks ago | 15:50 |
Jeffrey4l | rather then trying to create new one on every running. | 15:50 |
dmsimard | inc0: puppet-openstack has those too | 15:50 |
inc0 | so it's non-trivial to say the least | 15:50 |
dmsimard | inc0: RDO provides new repositories on every commit so updates are incremental and it's easy to bisect if there's anything | 15:50 |
mark-casey1 | hey inc0 congrats! | 15:50 |
sdake | training problem | 15:50 |
inc0 | thanks Mark:) | 15:50 |
dmsimard | inc0: but UCA updates in a batch at every milestone | 15:50 |
sdake | nobody knows how to use the gate interfaces | 15:50 |
*** shardy has joined #openstack-kolla | 15:51 | |
inc0 | yeah, and people makes mistakes | 15:51 |
sdake | exceept a select few | 15:51 |
sdake | pbourke__ Jeffrey4l know how to write gate code | 15:51 |
sdake | SamYaple sdake know | 15:51 |
sdake | need *everyone* to know | 15:51 |
inc0 | one I refer to happened Friday afternoon so we had good few days of gates being red | 15:51 |
openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Basic ceph backend support https://review.openstack.org/373256 | 15:55 |
MarMat | Jeffrey4l ok, then it seems that then inserting stuff and reading inserted IDs need to be split. Based on patchset comments, can you suggest how to do that? Keep everything in bash script as it is now, or use kolla_toolbox for inserting structures and read in bash script? Also, should the bash script be moved to docker/magnum/magnum-base ? | 15:55 |
Jeffrey4l | MarMat, np. i will leave a comment in PS. | 15:56 |
MarMat | Jeffrey4l cool, thanks | 15:56 |
*** berendt has quit IRC | 16:01 | |
*** kbaegis1 has joined #openstack-kolla | 16:02 | |
*** eaguilar has joined #openstack-kolla | 16:04 | |
*** kbaegis has quit IRC | 16:06 | |
*** msimonin has quit IRC | 16:06 | |
*** mdnadeem_ has joined #openstack-kolla | 16:06 | |
Jeffrey4l | MarMat, done please check the comments https://review.openstack.org/326163 | 16:09 |
*** ChanServ changes topic to "Please vote for summit sessions! http://civs.cs.cornell.edu/cgi-bin/vote.pl?id=E_8368e1e74f8a0049&akey=91bbcf4baeff0a2f; RC2 DEADLINE (HARD) Oct 10th-Oct 13th; Kolla IRC meetngs on Wednesdays @ 16:00 UTC - see agenda @ https://goo.gl/OXB0DL - IRC channel is *LOGGED* @ http://goo.gl/3mzZ7b (old logs from #kolla http://goo.gl/VKpPzA)" | 16:09 | |
-openstackstatus- NOTICE: Launchpad SSO logins are confirmed working correctly again | 16:09 | |
*** mdnadeem has quit IRC | 16:10 | |
*** pbourke__ is now known as pbourke | 16:11 | |
*** awiddersheim has quit IRC | 16:13 | |
*** david-lyle has quit IRC | 16:13 | |
*** david-lyle has joined #openstack-kolla | 16:14 | |
*** mdnadeem has joined #openstack-kolla | 16:15 | |
openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Adds sidecar logging container to common-deployment pod https://review.openstack.org/375682 | 16:16 |
sdake | inc0 sweet i get to vote twice! | 16:17 |
inc0 | sdake, you cheater | 16:17 |
sdake | well i didn't say i did vote wice | 16:17 |
britthouser | vote early and often sake :smiley: | 16:17 |
sdake | i was pointing out the oting was gamable | 16:17 |
inc0 | lol | 16:17 |
inc0 | yeah, I didn't put too many restrictions, I trust in good heart of our community | 16:18 |
sdake | one voice one vote - thats how openstack works | 16:18 |
inc0 | and monitor results to see if proportions are plausable:P | 16:18 |
*** mdnadeem_ has quit IRC | 16:18 | |
inc0 | but I would need a list of voters to limit it | 16:18 |
sdake | inc0 personally i'd take that vote as advisory | 16:18 |
inc0 | I will | 16:19 |
sdake | ultimately its up to the ptl to schedule summit | 16:19 |
inc0 | it's for next meeting really | 16:19 |
inc0 | I want to get this list up for discussion before we'll make final call | 16:19 |
sdake | sounds good | 16:19 |
wznoinsk | congratz Michal | 16:19 |
sdake | show-me-how - I know where you learned that one inc0 ;) | 16:20 |
sdake | ok so on to pressing matters | 16:21 |
sdake | which is rc2 | 16:21 |
sdake | anyone in need of direction | 16:21 |
MarMat | Jeffrey4l thanks, any other comments on latest PS from you? My testing loops are quite time-expensive, so it would be good to have complete review from you. | 16:21 |
sdake | i see our bug triage process was already defined by openstack, so no need to redefine that - > Jeffrey4l found the correct link :) | 16:21 |
*** senk_ has joined #openstack-kolla | 16:21 | |
inc0 | thanks Waldek:) | 16:22 |
sdake | marmat destroy, build magnum conainers, deploy is time expensive in which way? | 16:22 |
Jeffrey4l | MarMat, add another nit. others are fine. | 16:23 |
*** mdnadeem_ has joined #openstack-kolla | 16:23 | |
gmmaha | inc0: Congrats on the election.. :) | 16:24 |
inc0 | thanks:) | 16:24 |
sdake | i see we have an army of intcers :) | 16:24 |
*** kbaegis1 has quit IRC | 16:24 | |
*** lamt has quit IRC | 16:25 | |
*** kbaegis has joined #openstack-kolla | 16:25 | |
*** mdnadeem has quit IRC | 16:26 | |
MarMat | sdake deployment takes 10 mins for me, usually I have to repeat several times until code is good enough. | 16:27 |
sdake | marmat roger - so deplooy is at 10 minutes, your test cycle is at what? | 16:27 |
*** lamt has joined #openstack-kolla | 16:27 | |
MarMat | sdake + magnum tests like creating clusters etc. | 16:27 |
sdake | i see - so magnum is kind of beasty here | 16:28 |
sdake | since it does alot of extra work | 16:28 |
*** matrohon has quit IRC | 16:28 | |
sdake | i think your stuck with a problem that most people don't suffer from | 16:28 |
sdake | when developing for kolla | 16:28 |
*** kbaegis has quit IRC | 16:28 | |
sdake | the solution for these prolems is just what you asked for - a complete and thourough review | 16:28 |
sdake | these problems/this problem | 16:29 |
MarMat | sdake there are few other magnum-related issues which I don't want to take into this patchset, like the etcd proxy stuff | 16:29 |
inc0 | MarMat, splitting patches into multiple makes review easier | 16:30 |
*** rmart04 has quit IRC | 16:30 | |
*** kbaegis has joined #openstack-kolla | 16:30 | |
inc0 | when you have huge patchsets, it's harder to review, hence lesser change to get thorough review | 16:30 |
openstackgerrit | Merged openstack/kolla-kubernetes: More deployments merged https://review.openstack.org/375794 | 16:31 |
*** huhaoran has joined #openstack-kolla | 16:31 | |
*** jheroux has joined #openstack-kolla | 16:31 | |
sdake | the more red in gerrit, the less likely a patch to be reviewed ;-) | 16:32 |
sdake | even if its critical | 16:32 |
sdake | wish we could fix that one | 16:33 |
*** senk_ has quit IRC | 16:33 | |
sdake | wouold be nice if gerrit had a severity for reviews related to bugs associated with it | 16:33 |
MarMat | MarMat I'm just trying to bring already posted PS to a mergeable state :-) | 16:34 |
MarMat | inc0 ^^^ | 16:34 |
*** eaguilar has quit IRC | 16:35 | |
*** huhaoran has quit IRC | 16:35 | |
*** harlowja_at_home has quit IRC | 16:39 | |
*** unicell1 has quit IRC | 16:40 | |
*** mdnadeem has joined #openstack-kolla | 16:42 | |
*** shardy has quit IRC | 16:45 | |
*** irtermit- is now known as irtermite | 16:45 | |
*** mdnadeem_ has quit IRC | 16:45 | |
openstackgerrit | Bartłomiej Daca proposed openstack/kolla: Add external rabbitmq https://review.openstack.org/374525 | 16:48 |
*** mdnadeem_ has joined #openstack-kolla | 16:50 | |
sdake | folks - there is a tool called reviewday: http://status.openstack.org/reviews/ | 16:50 |
sdake | i'm not quite sure how it works, but i'm going to dig into it | 16:50 |
sdake | the short of it is, it takes launchpad priority and review activity into account | 16:50 |
sdake | rather then placing the last review at the top of the gerrit queue | 16:51 |
sdake | you see kolla isn't in the list of projects | 16:51 |
sdake | anyone interested in using this tool? | 16:51 |
*** lrensing has quit IRC | 16:53 | |
*** mdnadeem has quit IRC | 16:53 | |
*** lamt has quit IRC | 16:54 | |
bdaca | MarMat do you have an idea how to get these domain IDs you need for the configuration? (context: magnum) | 16:55 |
MarMat | bdaca that's what I'm trying to figure out, probably os_keystone_domain module can be used. | 16:58 |
*** senk_ has joined #openstack-kolla | 17:02 | |
bdaca | can we use modules that are in ansible 2.1? | 17:03 |
openstackgerrit | Merged openstack/kolla-kubernetes: Basic ceph backend support https://review.openstack.org/373256 | 17:04 |
inc0 | bdaca, we can bump ansible requirements to 2.1 from ocata | 17:04 |
inc0 | so not until we branch stable/newton | 17:04 |
inc0 | after that, I think that's ok | 17:04 |
inc0 | but we need to make sure that we fix docs and bootstrap-servers play | 17:04 |
bdaca | thx inc0 | 17:05 |
inc0 | also I would like to table discussion of re-using ansible docker module | 17:06 |
inc0 | I know people are reluctant, me too | 17:06 |
inc0 | but if current ansible docker module will handle features we currently have | 17:07 |
inc0 | we're safe. | 17:07 |
*** kbaegis has quit IRC | 17:08 | |
*** kbaegis has joined #openstack-kolla | 17:08 | |
sdake | inc0 wait until after newton plz :) | 17:09 |
inc0 | oh, obviously | 17:09 |
MarMat | Jeffrey4l we are trying still to squeeze magnum fix into newton, and from this point of view we cannot use os_keystone_domain_module which is ansible >= 2.1, right? | 17:10 |
inc0 | I'm nto convinced myself, I'm not saying let's do this, but I think we should review this issue | 17:10 |
bdaca | MarMat so if your changes in magnum are scheduled for newton, you can't use os_keystone_domain :( | 17:10 |
mark-casey1 | inc0: something that needs to go into whatever module is used is option to wait a few seconds for container to exit cleanly. it hasn't been killer but it's something we need for example for when db containers are restarted during upgrade | 17:10 |
mark-casey1 | on my list to add at some point so if we switch I thinkk the ansible one would need that | 17:11 |
*** LamT_ has quit IRC | 17:11 | |
MarMat | bdaca good point, thanks for bringing this. | 17:12 |
openstackgerrit | Eduardo Gonzalez proposed openstack/kolla: Add newline in grafana template https://review.openstack.org/376837 | 17:12 |
inc0 | mark-casey1, agree, in fact I'm not sure how to handle this | 17:12 |
inc0 | I mean, I know of timeout | 17:13 |
inc0 | but things like cinder-backup can run for hours | 17:13 |
bdaca | MarMat no problem, just want to help. There is os_user module, I'll check if this can be used instead | 17:13 |
inc0 | and we can't run our upgrades for hours | 17:13 |
*** msimonin has joined #openstack-kolla | 17:13 | |
*** gfhellma has joined #openstack-kolla | 17:13 | |
*** kbaegis has quit IRC | 17:14 | |
*** kbaegis has joined #openstack-kolla | 17:14 | |
mark-casey1 | inc0: is there a common method for that outside of kolla? I assume you mean "I'm upgrading... lets back up the disks" so that seems like other types of deploys would ahve that problem too | 17:15 |
*** unicell has joined #openstack-kolla | 17:16 | |
mark-casey1 | i'd say make it a task prior to the actual upgrade but I guess then there are things to be lost between that backup and running upgrade | 17:17 |
inc0 | mark-casey1, nope | 17:17 |
inc0 | I don't even know how something like that would work | 17:17 |
inc0 | there has to be long-running process copying things | 17:18 |
mark-casey1 | inc0: https://blueprints.launchpad.net/cinder/+spec/incremental-backup "—incr option specified for incremental backups. To keep the API implementation simple, we only support incremental from the last full backup." | 17:19 |
mark-casey1 | kolla-ansible backup-the-things; kolla-ansible upgrade-with-incremental-rebackup? | 17:20 |
*** eaguilar has joined #openstack-kolla | 17:20 | |
inc0 | well I don't want to make these types of calls, that's operators role | 17:20 |
inc0 | on the other hand, it might be good to document it | 17:21 |
sdake | Jeffrey4l - say http://docs.openstack.org/developer/kolla/bug-triage.html doesnt render properlly | 17:21 |
sdake | Jeffrey4l can you taclke that | 17:22 |
sdake | or anone else for that matter ;) | 17:22 |
*** Serlex has quit IRC | 17:23 | |
*** kbaegis has quit IRC | 17:25 | |
openstackgerrit | Serguei Bezverkhi proposed openstack/kolla: WIP Adds Fluentd service to Kolla https://review.openstack.org/375615 | 17:26 |
*** rmart04 has joined #openstack-kolla | 17:28 | |
*** harlowja has joined #openstack-kolla | 17:30 | |
*** kbaegis has joined #openstack-kolla | 17:34 | |
*** eaguilar_ has joined #openstack-kolla | 17:36 | |
*** eaguilar has quit IRC | 17:37 | |
openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Adds sidecar logging container to common-deployment pod https://review.openstack.org/375682 | 17:38 |
*** tonanhngo has quit IRC | 17:40 | |
kfox1111 | sdake: ceph support is in. :) | 17:42 |
sdake | kfox1111 rockin | 17:42 |
sdake | so where is ceph running precissely kfox1111 ? | 17:42 |
kfox1111 | I've got it running in k8s for minikube, but its really not intended for production. just for testing. | 17:43 |
kfox1111 | ceph requires a lot of orcestration and doens't fit well right now in k8s I think. | 17:43 |
*** salv-orlando has joined #openstack-kolla | 17:43 | |
kfox1111 | the mons probably would fit well. | 17:43 |
kfox1111 | the osd's are tricky though. as they may require differnet numbers of containers per host. | 17:44 |
kfox1111 | so it doesnt' fit very well with daemonsets. | 17:44 |
*** eaguilar_ has quit IRC | 17:44 | |
kfox1111 | you could just launch a whole pile of pods, but might be hard to maintain. | 17:44 |
*** eaguilar has joined #openstack-kolla | 17:44 | |
kfox1111 | I'm kind of leaning, for a 1.x release to just do ceph with kolla-ansible or ceph-deploy. both are good options. | 17:45 |
kfox1111 | most ops I think will want an external ceph anyway. | 17:46 |
kfox1111 | it feels more safe to me. | 17:46 |
kfox1111 | can't accidently run a command and nuke all your important data. | 17:46 |
kfox1111 | but, I'm able to launch a minikube, follow the instructions in doc/source/minikube.rst, and go from nothing at all to vm's with cinder volumes attached. :) | 17:49 |
kfox1111 | that means the templates should work for an external ceph too. :) | 17:50 |
inc0 | kfox1111, they do in kolla-ansible | 17:51 |
*** britthouser has quit IRC | 17:53 | |
*** britthouser has joined #openstack-kolla | 17:54 | |
sdake | kfox1111 i'd encourage 1.0.0 to be tagged when its ready to production | 17:54 |
sdake | if its ready for playtime, 0.y.z is better | 17:54 |
*** britthouser1 has joined #openstack-kolla | 17:55 | |
*** britthouser has quit IRC | 17:55 | |
*** bmace has joined #openstack-kolla | 17:57 | |
*** g3ek has quit IRC | 17:57 | |
*** haplo37 has quit IRC | 17:57 | |
*** rmart04 has quit IRC | 17:59 | |
*** rmart04 has joined #openstack-kolla | 18:00 | |
*** gfhellma1 has joined #openstack-kolla | 18:02 | |
*** gfhellma has quit IRC | 18:03 | |
*** tonanhngo has joined #openstack-kolla | 18:03 | |
*** tonanhngo has quit IRC | 18:04 | |
*** britthouser1 has quit IRC | 18:04 | |
*** g3ek has joined #openstack-kolla | 18:06 | |
*** haplo37 has joined #openstack-kolla | 18:07 | |
*** dwalsh has quit IRC | 18:07 | |
*** rmart04 has quit IRC | 18:09 | |
bdaca | MarMat ping | 18:13 |
MarMat | bdaca pong | 18:14 |
bdaca | MarMat: os_user should be ok: http://paste.openstack.org/show/582986/ only tricky part is with setting api version in environment | 18:15 |
bdaca | hope it helps a little | 18:16 |
MarMat | cool, thanks, i'm also playing with it | 18:16 |
*** awiddersheim has joined #openstack-kolla | 18:17 | |
*** daneyon has joined #openstack-kolla | 18:21 | |
*** gfhellma1 has quit IRC | 18:22 | |
*** dwalsh has joined #openstack-kolla | 18:23 | |
*** vhosakot has joined #openstack-kolla | 18:24 | |
*** daneyon has quit IRC | 18:26 | |
*** gfhellma has joined #openstack-kolla | 18:29 | |
*** thumpba has joined #openstack-kolla | 18:30 | |
kfox1111 | sdake: yeah, I'm trying to get it into production asap. :) | 18:33 |
kfox1111 | I think there's two different potential ways of tagging it 1.0 though. | 18:34 |
kfox1111 | in the 4 layer cake: config, workflow, deployment, software | 18:34 |
*** eaguilar has quit IRC | 18:34 | |
kfox1111 | it could be production ready with a manual workflow. which is what I'm going for with my systems. | 18:34 |
kfox1111 | but if we want to wait to tag it 1.0 until there is a batteries included worlflow piece, that could take a lot longer. | 18:35 |
*** pbourke has quit IRC | 18:38 | |
*** pbourke has joined #openstack-kolla | 18:39 | |
*** Marx314 has joined #openstack-kolla | 18:41 | |
*** Marx314 has quit IRC | 18:41 | |
*** lrensing has joined #openstack-kolla | 18:43 | |
*** Marx314 has joined #openstack-kolla | 18:43 | |
*** eaguilar has joined #openstack-kolla | 18:44 | |
*** britthouser has joined #openstack-kolla | 18:45 | |
*** tonanhngo has joined #openstack-kolla | 18:46 | |
inc0 | kfox1111, 1.0 will be important | 18:47 |
*** tonanhngo_ has joined #openstack-kolla | 18:47 | |
inc0 | it's ok to take longer | 18:48 |
inc0 | you only get one first impression | 18:48 |
*** tonanhngo_ has quit IRC | 18:48 | |
kfox1111 | I'm on the fence on that one. its the balance of first impresion to first with a "production system". | 18:48 |
*** tonanhngo_ has joined #openstack-kolla | 18:49 | |
kfox1111 | if one of the other openstack on k8s gets production systems first, it might attract more devs. :/ | 18:49 |
kfox1111 | sometimes the best technology doesn't win, but those that either make it there first, or gain mindshare first. | 18:50 |
*** tonanhngo has quit IRC | 18:50 | |
inc0 | well, 1.0 means prod | 18:50 |
inc0 | I don't want to release beta out there | 18:51 |
*** vhosakot has quit IRC | 18:51 | |
inc0 | but agree, we need workflow automation before | 18:51 |
kfox1111 | yes. I agree. but does prod mean needing an easy to use worflow? | 18:51 |
inc0 | should be doable in ocata tho | 18:51 |
inc0 | I think we can figure some workflow out | 18:51 |
kfox1111 | cause to me, it doesn't nessisarily. | 18:51 |
kfox1111 | yeah. | 18:51 |
inc0 | well, you know this system in and out | 18:52 |
inc0 | if your first impression means lots of pain during initial deployment | 18:52 |
inc0 | that's bad first impression | 18:52 |
kfox1111 | true. true. but often documentation can take the place of workflow. | 18:52 |
* kfox1111 nods. | 18:52 | |
kfox1111 | there's a lot that can turn you off during initial deployment though. | 18:53 |
kfox1111 | to me, automation I don't understand is much worse then manual deployment steps I can understand. | 18:53 |
sdake | kfox1111 i've seenthe initial deployment | 18:53 |
sdake | 9aftr the deploy of kubernetes) | 18:53 |
sdake | its pretty rough :( | 18:53 |
inc0 | kfox1111, what happend to ansible playbook - driven initial deployment? | 18:53 |
kfox1111 | because I can understand the manual steps in case I need to debug. | 18:53 |
inc0 | it should be fairly easy to do at least to some extend | 18:53 |
inc0 | and ops would have easy time digging through workflow as ansible is well known | 18:54 |
inc0 | if we keep it simple, which we should | 18:54 |
kfox1111 | inc0: thats coming too. but its a time/dev issue. we need a dev with free time to implement workflow. | 18:54 |
kfox1111 | inc0: disagree. | 18:54 |
inc0 | elaborate plz | 18:54 |
kfox1111 | it took me 2 days to figure out how ceph deployment worked in kolla. | 18:54 |
inc0 | ceph deployment in kolla is not simple | 18:55 |
inc0 | this will be easier | 18:55 |
* kfox1111 nods. | 18:55 | |
inc0 | this will be glorified kubectl wrapper | 18:55 |
inc0 | with some wait conditions | 18:55 |
kfox1111 | yeah. should be pretty simple. | 18:55 |
sdake | kfox1111 bit of cherry-picking examples there :) | 18:55 |
*** britthouser2 has joined #openstack-kolla | 18:55 | |
inc0 | I'm reluctant to even running ceph on k8s;) | 18:55 |
sdake | ceph is a beast | 18:55 |
kfox1111 | sdake: true true. was just on the tip of my brain as I had just done it. | 18:56 |
inc0 | if it makes even remote sense | 18:56 |
sdake | oh makes sense then kfox1111 | 18:56 |
*** senk_ has quit IRC | 18:56 | |
kfox1111 | inc0: +1. I think k8s needs some more features first. | 18:56 |
inc0 | thats true, but what I'm saying is....this is not an k8s'y app | 18:56 |
inc0 | you don't want to migrate osd out of disk it uses | 18:56 |
*** vhosakot has joined #openstack-kolla | 18:57 | |
kfox1111 | not nocking ceph deployment with ansible either. its quite clever how its done. it was just pretty twisty to understand for someone not very familior with ansible. | 18:57 |
inc0 | yeah, most of complexity comes from keyring generation | 18:57 |
inc0 | which will be issue with k8s too | 18:57 |
kfox1111 | part of it is in the implied orderings of running things too. it took me a while to follow what was running where, when. | 18:57 |
inc0 | regardless, ceph is different matter, I'd be really reluctant to put it to k8s at all costs | 18:58 |
kfox1111 | +1 | 18:58 |
inc0 | some things shouldn't run on k8s | 18:58 |
*** britthouser has quit IRC | 18:58 | |
inc0 | imho | 18:58 |
kfox1111 | so, yeah. I agree, workflow might not be too hard on top of k8s. | 18:58 |
inc0 | not yet anyway, we'll see where k8s goes | 18:58 |
inc0 | that'd be my plan | 18:58 |
inc0 | just wrap around kubectl | 18:59 |
kfox1111 | basically we need to ansibleize this: https://github.com/openstack/kolla-kubernetes/blob/master/doc/source/minikube.rst | 18:59 |
inc0 | maybe write simple ansible module for k8s api if it doesn't exist yet | 18:59 |
kfox1111 | there is one, but its blocked for some reason. :/ | 18:59 |
kfox1111 | ryan and I were talking about forking it locally for a while. | 19:00 |
*** msimonin1 has joined #openstack-kolla | 19:00 | |
inc0 | forking is a no-no in ansible | 19:00 |
inc0 | GPL v3 | 19:00 |
inc0 | write our own from scratch | 19:00 |
inc0 | would be way to go | 19:00 |
kfox1111 | k. :/ | 19:00 |
kfox1111 | yeah. | 19:01 |
inc0 | shouldn't be super hard too | 19:01 |
*** dwalsh has quit IRC | 19:01 | |
kfox1111 | yeah. we realy just need to wrap the status command. | 19:01 |
inc0 | we did it with kolla-docker which is arguably much harder case | 19:01 |
*** vhosakot has quit IRC | 19:01 | |
sdake | atleast kolla-docker had docker-py binding | 19:01 |
inc0 | still, ack people need for it | 19:01 |
sdake | there is no bidning for kubernetes | 19:01 |
inc0 | it's coding and it takes time | 19:01 |
inc0 | sdake, k8s has neat REST api | 19:02 |
*** tonanhngo has joined #openstack-kolla | 19:02 | |
sdake | neat != language binding | 19:02 |
kfox1111 | just need to wrap something to exec 'kubectl get <x> -o yaml --namespace=<y>' | 19:02 |
inc0 | so you hardly need any real binding | 19:02 |
sdake | been down that road | 19:02 |
sdake | kfox1111 that is an easier way to go ;) | 19:02 |
kfox1111 | will work until ansible gets native. :) | 19:02 |
inc0 | we can start with command module around kubectl | 19:02 |
inc0 | kfox1111, if you'll help me work through multinode deploy of kolla-k8s | 19:03 |
sdake | inc0 ctually the language bindings are autogenerated | 19:03 |
inc0 | I can write playbook | 19:03 |
sdake | inc0 with some kind of garbage converter | 19:03 |
kfox1111 | inc0: would be happy to. | 19:03 |
*** msimonin has quit IRC | 19:03 | |
inc0 | let's get on hangout later this week | 19:03 |
inc0 | I'll deploy couple of VMs | 19:03 |
kfox1111 | if we can get it working in the minikube env, that should be the last bit needed to start gate testing too. | 19:03 |
inc0 | deploy k8s on them | 19:03 |
inc0 | and have ceph around to access | 19:04 |
*** msimonin has joined #openstack-kolla | 19:04 | |
kfox1111 | a hangout sounds good. | 19:04 |
inc0 | well I'd love to have it at least similar to prod | 19:04 |
kfox1111 | sounds like a plan. | 19:04 |
inc0 | kk, which day works for you best? | 19:04 |
*** msimonin1 has quit IRC | 19:04 | |
inc0 | Wed just after meeting would work? | 19:05 |
kfox1111 | hows thurday work for you? | 19:05 |
inc0 | sure, what time? | 19:05 |
*** tonanhngo_ has quit IRC | 19:05 | |
kfox1111 | 2:00pm pacific? | 19:06 |
inc0 | ok, I'm blocking this slot | 19:06 |
kfox1111 | cool. | 19:06 |
inc0 | thanks m8 | 19:06 |
kfox1111 | anytime. :) | 19:06 |
kfox1111 | thanks for helping with the ansible stuff. | 19:07 |
kfox1111 | sdake: any chance you could start the conversation with infra to get some bare metal nodes going for testing? | 19:08 |
kfox1111 | I think we're close to being about to test this stuff in an automated way. :) | 19:09 |
*** eaguilar has quit IRC | 19:09 | |
sdake | kfox1111 i dont think that is possible | 19:11 |
sdake | kfox1111 tripleo does that via third party ci | 19:11 |
sdake | and a bunch of others do as well | 19:12 |
*** eaguilar has joined #openstack-kolla | 19:12 | |
sdake | tripleo's third party ci is somehow integrated into oepsntack ci | 19:12 |
sdake | notsure how | 19:12 |
kfox1111 | hmm... k. | 19:12 |
sdake | neeed to make it run in a vm imo :) | 19:12 |
kfox1111 | so, then I guess we would have to do a single vm, without minikube, and just deploy k8s manually. | 19:12 |
sdake | path of least resistance and all that | 19:12 |
kfox1111 | so, basically just the minikube instructions with manual k8s deployment added on. | 19:13 |
*** huhaoran has joined #openstack-kolla | 19:13 | |
kfox1111 | should work still. | 19:13 |
kfox1111 | is taht something you can ask for? | 19:13 |
sdake | which gate jobs? | 19:13 |
kfox1111 | yeah. | 19:13 |
sdake | we can add gate obs | 19:13 |
sdake | thats self-service | 19:14 |
kfox1111 | ones that just request a bare vm and do stuff on it? | 19:14 |
sdake | yup thats what we got | 19:14 |
kfox1111 | ah. cool. | 19:14 |
sdake | usually we run a script that runs tox | 19:14 |
kfox1111 | if you can get one submitted that calls a script, I can fill it in with the rest? | 19:15 |
*** dwalsh has joined #openstack-kolla | 19:17 | |
sbezverk_ | kfox1111: for kube installation in VM I would suggest to do really manual install | 19:21 |
*** matrohon has joined #openstack-kolla | 19:21 | |
sbezverk_ | kfox1111: it works in 100% of cases since you control everything | 19:22 |
kfox1111 | sbezverk_: yeah. thats what we'll have to do it soulds like. | 19:22 |
sbezverk_ | I just copy kube binaries and then create services files in systemd | 19:22 |
sbezverk_ | that is it. | 19:22 |
*** haplo37 has quit IRC | 19:23 | |
kfox1111 | yeah. I usually just install kubelet and use manifests for the rest. | 19:24 |
kfox1111 | either way, pretty easy. | 19:24 |
kfox1111 | just not as easy as: minikube start --vm-driver=kvm --memory=$((8*1024)) :) | 19:24 |
*** g3ek has quit IRC | 19:25 | |
kfox1111 | as that buys you kubelet, the api bits, dns, and the dashboard all in one go. :) | 19:25 |
kfox1111 | btw, if you havent played with the 1.4 dashboard, its much nicer. :) | 19:26 |
*** haplo37 has joined #openstack-kolla | 19:26 | |
*** g3ek has joined #openstack-kolla | 19:26 | |
*** lrensing has quit IRC | 19:28 | |
*** sdake has quit IRC | 19:33 | |
*** britthouser2 has quit IRC | 19:34 | |
*** britthouser has joined #openstack-kolla | 19:34 | |
*** tonanhngo_ has joined #openstack-kolla | 19:34 | |
*** tonanhngo has quit IRC | 19:37 | |
kfox1111 | sbezverk_: you there? | 19:38 |
kfox1111 | services/openvswitch/openvswitch-vswitchd-daemonset.yml.j2 has: | 19:39 |
kfox1111 | ovs-vsctl --no-wait --db=unix:/var/run/openvswitch/db.sock add-br {{ neutron_bridge_name }} ; | 19:39 |
kfox1111 | but why the --no-wait? seems to cause a race condition. | 19:39 |
sbezverk_ | kfox1111: I have never seen it and no-wait only tels that there is no vswitchd running for executing this command | 19:41 |
*** kbaegis has quit IRC | 19:41 | |
kfox1111 | the setting of the ip afterwards seems unrelyable though, as vswitch doesn't always apply it right away. | 19:42 |
sbezverk_ | kfox1111: I was advised to use this form by ovs people | 19:42 |
*** salv-orl_ has joined #openstack-kolla | 19:42 | |
openstackgerrit | Merged openstack/kolla: Use keystone-paste.ini template for keystone https://review.openstack.org/373379 | 19:43 |
*** kbaegis has joined #openstack-kolla | 19:43 | |
kfox1111 | hmm.... | 19:43 |
sbezverk_ | kfox1111: if you do not use no-wait it will hang indefinetly waiting for vswitchd running | 19:44 |
kfox1111 | but the socket is from the db, not the vswitch? | 19:44 |
sbezverk_ | also it is recommended to add this before vswitchd running | 19:44 |
*** bjolo has quit IRC | 19:44 | |
*** haplo37_ has quit IRC | 19:45 | |
sbezverk_ | ovs-vsctl uses vsciwtd to execute commands if no-wait is not specified | 19:45 |
kfox1111 | k. I guess I'll have to totally rethink how I do this then. | 19:45 |
*** salv-orlando has quit IRC | 19:45 | |
sbezverk_ | kfox1111: can you describe in which scenario you hit the issue? | 19:46 |
kfox1111 | if it requres ovs to be up before it actually builds the br-ex, the ip command will never run successfully. | 19:46 |
*** sdake has joined #openstack-kolla | 19:46 | |
sbezverk_ | I will like to try to see if I can repro, because so far it has been very reliable | 19:46 |
kfox1111 | sbezverk_: https://github.com/openstack/kolla-kubernetes/blob/master/services/openvswitch/openvswitch-vswitchd-daemonset.yml.j2#L29-L32 | 19:46 |
kfox1111 | I'm guessing you never used that feature. | 19:47 |
SamYaple | kfox1111: --no-wait allows you to updated the database without verifying the update took place. you never talk ot vswitchd | 19:47 |
SamYaple | you alwaays talk to the database and vswitchd reads from the db | 19:47 |
kfox1111 | I think probably the next best thinkg I can think of is to move that chunk out into its own job. | 19:47 |
kfox1111 | then it can just run until the br-ex shows up. | 19:47 |
SamYaple | IIRC i belelieve that was implemetned because we ran commands in the db container before teh vswitchd container was up | 19:47 |
*** haplo37_ has joined #openstack-kolla | 19:48 | |
sbezverk_ | kfox1111: I understand what you try to do, but I think setting up ip in daemonset is not right approach | 19:49 |
kfox1111 | yeah. I think your right. it will be hard to make work with ovs working the way it is. | 19:49 |
kfox1111 | I'll try it with a job. I think that will fix it. | 19:49 |
openstackgerrit | Merged openstack/kolla-kubernetes: Adds sidecar logging container to common-deployment pod https://review.openstack.org/375682 | 19:50 |
sbezverk_ | kfox1111: I hope so, cause I spent two weeks trying make ovs stable | 19:50 |
sbezverk_ | and cut all fat whihc was making it not very stable | 19:51 |
*** vhosakot has joined #openstack-kolla | 19:52 | |
kfox1111 | yeah. | 19:52 |
kfox1111 | I still need to test ovs in the host use case as wel. | 19:52 |
kfox1111 | as that's actualy the more common use case for me. | 19:52 |
kfox1111 | just tryign to get a gate going for now though, to prevent regressions. | 19:53 |
sbezverk_ | kfox1111: right that part is not covered. but I do not think this split kind of deployment is populat at least in kolla community.. I might be wrong but I have not heard other folks using OVS from the host and runnig kolla on top of that | 19:54 |
*** vhosakot_ has joined #openstack-kolla | 19:56 | |
*** vhosakot has quit IRC | 19:56 | |
kfox1111 | I'm probably your first 40gig deployment. ;) | 19:56 |
kfox1111 | If we're successful, I won't be your last though. :) | 19:56 |
kfox1111 | chicken and the egg though. if kolla doesn't support it, then it wont' be poplular in the community. | 19:57 |
*** vhosakot_ has quit IRC | 19:58 | |
*** britthouser has quit IRC | 20:00 | |
*** britthouser has joined #openstack-kolla | 20:00 | |
*** vhosakot has joined #openstack-kolla | 20:01 | |
sbezverk_ | kfox1111: 40gig is past my friend try 100Gig ;-) and 400gige bundles | 20:01 |
sbezverk_ | kfox1111: planning to add side car logging to converted deployments, any objections? | 20:02 |
bdaca | can anyone explain me how should I backport changes to stable/* branches when cherry-pick generates conflicts please? | 20:06 |
kfox1111 | sbezverk_: yeah,t hat just means big orgs will go to 100g and 40g will start hitting mb's soon. :) | 20:06 |
kfox1111 | sbezverk_: I thought that just merged? | 20:07 |
sbezverk_ | kfox1111: it has, but example glance does not have check for fluentd in the template | 20:08 |
*** britthouser has quit IRC | 20:08 | |
kfox1111 | ah. glance is the exception, as I'm going to merge that into the deployments too. | 20:09 |
*** britthouser has joined #openstack-kolla | 20:09 | |
kfox1111 | but was waiting for your patch to go in, before making a macro version of it. | 20:09 |
kfox1111 | as it has a far amount of conditional stuff. | 20:09 |
kfox1111 | needed the engine fix that was in your ps to go in first. | 20:09 |
sbezverk_ | ok I might have missed it, I will re-run resource-template on update workspace | 20:09 |
sbezverk_ | I has all been just merged | 20:10 |
kfox1111 | k | 20:10 |
kfox1111 | yeah. :) | 20:10 |
kfox1111 | trunks looking very good now. :) | 20:10 |
kfox1111 | I'm compliing a ps with a few misc fixes in it. | 20:10 |
kfox1111 | should smooth out fresh deployments. | 20:11 |
*** msimonin1 has joined #openstack-kolla | 20:12 | |
*** inc0_ has joined #openstack-kolla | 20:12 | |
*** msimonin has quit IRC | 20:12 | |
*** inc0 has quit IRC | 20:13 | |
*** haplo37 has quit IRC | 20:15 | |
*** g3ek has quit IRC | 20:18 | |
*** bdaca has quit IRC | 20:18 | |
*** g3ek has joined #openstack-kolla | 20:20 | |
*** huhaoran has quit IRC | 20:21 | |
*** Kimmo_ has quit IRC | 20:24 | |
*** haplo37 has joined #openstack-kolla | 20:24 | |
openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: WIP: Misc fixes https://review.openstack.org/377082 | 20:26 |
*** jmccarthy has quit IRC | 20:29 | |
*** jmccarthy has joined #openstack-kolla | 20:29 | |
*** jtriley has quit IRC | 20:34 | |
*** Jeffrey4l has quit IRC | 20:35 | |
*** Jeffrey4l has joined #openstack-kolla | 20:36 | |
*** tonanhngo_ has quit IRC | 20:36 | |
*** tonanhngo has joined #openstack-kolla | 20:37 | |
*** msimonin1 has left #openstack-kolla | 20:42 | |
sdake | kfox1111 around? | 20:51 |
sdake | kfox1111 re gate - if you want that, infra requires a functional tox environment first to do the job | 20:52 |
sdake | and a sscript to run | 20:53 |
sdake | it doesn't have to be fancy | 20:53 |
*** absubram has quit IRC | 20:54 | |
*** inc0_ has quit IRC | 20:54 | |
*** inc0 has joined #openstack-kolla | 20:55 | |
kfox1111 | sdake: yeah. whats up? | 21:00 |
sdake | see scrollback | 21:00 |
kfox1111 | tox works now, the pep8/py27 stuff. something more then that? | 21:00 |
sdake | ya you need something that builds the images - no? | 21:01 |
kfox1111 | I've been using the hub build images for that. | 21:01 |
sdake | those work for you? | 21:01 |
kfox1111 | yeah. | 21:01 |
sdake | good enuf | 21:01 |
*** neilus has joined #openstack-kolla | 21:01 | |
sdake | you need to deploy k8s right? | 21:01 |
kfox1111 | right. | 21:01 |
sdake | i guess that doesnt have to be in tox | 21:02 |
kfox1111 | that code would be pretty specific to the gate though, so not suited to a tox env? | 21:02 |
sdake | could be in a script | 21:02 |
kfox1111 | yeah. | 21:02 |
*** eaguilar has quit IRC | 21:02 | |
sdake | let me show yoou something real qauck | 21:02 |
sdake | quick | 21:02 |
kfox1111 | k | 21:02 |
sdake | https://review.openstack.org/#/c/326163/ | 21:03 |
sdake | see all those gate jobs on the right? | 21:03 |
sdake | notice a pattern deploy build distro | 21:03 |
kfox1111 | yup. | 21:03 |
sdake | those are tox jobs | 21:03 |
kfox1111 | the dsvm ones or the non ones? | 21:04 |
sdake | if you dont need to build you dont need a tox job for that | 21:04 |
sdake | the ones with dsvm | 21:04 |
sdake | need atleast one tox job | 21:05 |
sdake | the setup_gate.sh script we run sets up the gate | 21:05 |
sdake | i.e. installs minikube | 21:05 |
sdake | or deploys k8s | 21:05 |
sdake | or does whatever is needed to get the env into an operational state | 21:05 |
*** neilus has quit IRC | 21:06 | |
kfox1111 | so, do we want to copy setup_gate from kolla/tools/ into kolla-kubernetes/tools, strip it to bare bones, then get that in the gate? | 21:06 |
sdake | yup but infra is not a fan of jobs that fail immediately | 21:07 |
sdake | kfox1111 i can train you after summit or help you after summit on how ot dod the job | 21:08 |
sdake | or even during summit ;) | 21:08 |
sdake | i'm a bit busy atm to write a gate from scratch :( | 21:08 |
kfox1111 | k. well, if we removed the kolla specific bits and just made it a no-op? | 21:08 |
kfox1111 | k. | 21:08 |
sdake | i think that would be ok | 21:08 |
sdake | but that wuldn't produce any useful result and just suck up vm resources | 21:09 |
kfox1111 | yeah. but would give me a place then to start adding real checks. | 21:09 |
sdake | infra wants the gate jobs doing something useful | 21:09 |
kfox1111 | I've got a pile I could add in short order. | 21:09 |
sdake | spin up spin down are expensive | 21:09 |
kfox1111 | k. so, I'll work on getting a no-op, then a simple but useful check. | 21:10 |
sdake | note they will actuallly look at the gate jobs | 21:10 |
*** vhosakot has quit IRC | 21:10 | |
sdake | and may not approve the patch if there is no work being done | 21:10 |
kfox1111 | then if we can get that in the gate, then I can iterate on adding more checks. | 21:10 |
*** haplo37 has quit IRC | 21:11 | |
kfox1111 | I think even just getting it doing a kolla genconfig with the k8s options enabled would be a useful bit of work. | 21:11 |
sdake | what would the useful check be? | 21:11 |
kfox1111 | as we've seen regressions. | 21:11 |
sdake | cool that sounds like the beginnings of a deploy job to me | 21:11 |
kfox1111 | yeah. | 21:11 |
sdake | if you handle the k8s side i can handle the infra gate changes | 21:11 |
kfox1111 | sounds good. I've recently deployed a k8s so its all still pretty fresh. | 21:12 |
*** britthouser has quit IRC | 21:12 | |
*** britthouser has joined #openstack-kolla | 21:12 | |
rhallisey | kfox1111, I already set up the gate from infra | 21:13 |
rhallisey | kfox1111, you need to add the script on kolla's end to setup | 21:14 |
sdake | rhallisey we are talking about a deploy job here for centos | 21:14 |
rhallisey | yup | 21:14 |
rhallisey | let me find the name of the script that gets called.. | 21:14 |
kfox1111 | oh really? cool. where does the scrip need to be? | 21:14 |
kfox1111 | k. | 21:14 |
sdake | rhallisey hopefullly its the same syntax as the kolla gate job | 21:15 |
*** msimonin has joined #openstack-kolla | 21:15 | |
* sdake wtb phone locator | 21:15 | |
rhallisey | pretty sure it's the same name | 21:16 |
rhallisey | tools/setup_aio.sh | 21:16 |
rhallisey | or something | 21:16 |
kfox1111 | ../kolla/tools/deploy_aio.sh ? | 21:17 |
rhallisey | ya that's right | 21:19 |
kfox1111 | so is the job just passing if the file doesn't exist? | 21:19 |
kfox1111 | hmmm... I gues I'm not seeing it in the list. | 21:20 |
rhallisey | to run the job add to a review | 21:20 |
*** haplo37 has joined #openstack-kolla | 21:20 | |
kfox1111 | https://review.openstack.org/#/c/375682/ for example. | 21:20 |
rhallisey | 'check experimental' | 21:20 |
*** dwalsh has quit IRC | 21:20 | |
kfox1111 | oh... interesting... | 21:20 |
rhallisey | I made it experimental | 21:20 |
kfox1111 | sec | 21:20 |
rhallisey | just post a test patch | 21:21 |
sdake | https://github.com/openstack-infra/project-config/blob/master/jenkins/jobs/kolla-kubernetes.yaml | 21:21 |
sdake | looks goodd | 21:21 |
rhallisey | sdake, ah thank you | 21:21 |
rhallisey | setup_gate.sh | 21:21 |
kfox1111 | so how does that get triggered? | 21:22 |
rhallisey | kfox1111, check experimental | 21:22 |
rhallisey | I'll show you | 21:22 |
openstackgerrit | Ryan Hallisey proposed openstack/kolla-kubernetes: GATE TESTING https://review.openstack.org/377101 | 21:23 |
rhallisey | kfox1111, k check that out | 21:24 |
*** eaguilar has joined #openstack-kolla | 21:24 | |
sdake | https://github.com/openstack-infra/project-config/blob/master/zuul/layout.yaml#L10412 | 21:24 |
sdake | kfox1111 ^^ | 21:24 |
kfox1111 | yeah. I've got the flag on one of mine though. was wondering how it was hooked up in the gate to hvae 'check experimental' work. | 21:24 |
sdake | run check experimental | 21:24 |
kfox1111 | ah. thats what I was wondering. thanks sdake. | 21:25 |
*** jheroux has quit IRC | 21:25 | |
*** matrohon has quit IRC | 21:26 | |
*** ayoung has quit IRC | 21:26 | |
kfox1111 | the gate's really backed up right now. :/ | 21:27 |
*** matrohon has joined #openstack-kolla | 21:27 | |
rhallisey | is it `run check experimental`? | 21:27 |
rhallisey | though it was just `check experimental` | 21:27 |
kfox1111 | its just check experimental. they are showing up in the zuul dashboard. | 21:28 |
rhallisey | col | 21:28 |
rhallisey | cool | 21:28 |
kfox1111 | I think I see how it all works together though. so I should be able to whip something up. | 21:28 |
kfox1111 | I wont be able to test too much without some rudimentery workflow bits, | 21:29 |
kfox1111 | but I can get a bit of groundwork layed. | 21:29 |
kfox1111 | then we can fill in the workflow bits when they land. | 21:29 |
openstackgerrit | Serguei Bezverkhi proposed openstack/kolla-kubernetes: WIP Adding logging to kolla-kubernetes services https://review.openstack.org/377104 | 21:30 |
kfox1111 | like I said though, even just doing a few genconfig's and testing the templates render would go along way. | 21:30 |
*** salv-orl_ has quit IRC | 21:30 | |
sbezverk_ | kfox1111, rhallisey: please check this PS to see if you are ok with tkaen approach.. | 21:30 |
kfox1111 | sec | 21:31 |
kfox1111 | generally. | 21:32 |
*** salv-orlando has joined #openstack-kolla | 21:32 | |
kfox1111 | do you think there will be more then one generic format? | 21:32 |
kfox1111 | otherwise, just using resource_name might be easier then defining a new var that = resource_name. | 21:33 |
kfox1111 | well... | 21:33 |
*** sdake has quit IRC | 21:33 | |
kfox1111 | if we ever have multiple rabbits/mariadb's, then resource_name may not be that clean cut. | 21:34 |
sbezverk_ | kfox1111: not sure as example there wil be particularities in neutron | 21:34 |
kfox1111 | ok, I'm fine with the way it is. | 21:34 |
*** lamt has joined #openstack-kolla | 21:35 | |
sbezverk_ | kfox1111: cool, thanks | 21:35 |
*** portdirect has joined #openstack-kolla | 21:35 | |
kfox1111 | one other thought... | 21:36 |
kfox1111 | shoudl we rename logging-configmap to fluentd-configmap, and rename common-fluentd to common-logging? | 21:36 |
kfox1111 | the logging-configmap is rather specific to fluentd. | 21:37 |
kfox1111 | while we could make common-logging have a switch statement in it and support other loggers too, with the same set of macros. | 21:37 |
*** Kimmo_ has joined #openstack-kolla | 21:37 | |
*** neilus has joined #openstack-kolla | 21:38 | |
kfox1111 | then, if we add heka,filebeat, or logstash support, it just goes into common-logging and we don't have to touch any of the other templates. | 21:38 |
portdirect | Hi, I've put OpenStack in Kubernetes, which makes quite extensive use of Kuryr, and OVN, replacing Kubernetes default networking, Cinder for persistant storage, and federated keystone for auth. I've put the code up here: https://github.com/portdirect/harbor and plan to have AMI's and an ISO to download in the next few days - if anyone has any feedback I'd really appreciate it. I've also a demo running at | 21:39 |
portdirect | https://port.direct if you wanna check out a running install | 21:39 |
MarMat | Jeffrey4l ping | 21:42 |
rhallisey | neat | 21:42 |
kfox1111 | portdirect: thats a lot of code to drop. how long have you been working on that? | 21:43 |
portdirect | too long :) about 5 months to get it to the point its at - when i started I hardly knew anything :) | 21:44 |
portdirect | I'd be very keen to get invovled in kolla if possible - you are starting to get going on kube now? | 21:45 |
rhallisey | portdirect, yes | 21:45 |
rhallisey | https://github.com/openstack/kolla-kubernetes | 21:45 |
*** eaguilar has quit IRC | 21:45 | |
kfox1111 | the docker containers based on alpine are interesting. very light weight. :) | 21:46 |
kfox1111 | kolla could really use some of those. :) | 21:46 |
rhallisey | portdirect, more contributors welcome :) | 21:46 |
portdirect | yup - only cinder seems to really need a glibc becasue of deps | 21:47 |
kfox1111 | some great stuff here. we'd love it if you could help us with kolla-kubernetes. | 21:48 |
rhallisey | ya that's fantastic work | 21:49 |
kfox1111 | we havent gotten kuryr working, nor keystone federation. thats awesome. | 21:49 |
portdirect | Ok - I had to build my own adapter based on some work stal midonet did | 21:49 |
portdirect | replaces kube-proxy with neutron lbaas | 21:49 |
kfox1111 | looks like you have some nice networking bits too. | 21:49 |
kfox1111 | cool. | 21:50 |
portdirect | pure ovn - only have the l3 agent to replace with openflow | 21:50 |
portdirect | whats the area I could help most with? | 21:51 |
rhallisey | portdirect, looks like you have more services | 21:51 |
kfox1111 | still looking, | 21:51 |
rhallisey | portdirect, what's the genearl deploy procedure? | 21:52 |
kfox1111 | off hand I'd say the ovs stuff is really cool. | 21:52 |
kfox1111 | never gotten to play with ovs yet. | 21:52 |
rhallisey | looks like you deloy a container that orchestrats the other bits? | 21:52 |
portdirect | yup - i've called that marina | 21:52 |
portdirect | it just bootstrapps the templates and cenerates keytabs/certs for services for FreeIPA | 21:53 |
kfox1111 | oh, so you got it all integreated with freeipa too? | 21:54 |
kfox1111 | thats awesome. :) | 21:54 |
portdirect | there is a few bugs but I have an ami I could talk you though - It runs on a custom CentOS atomic host (newer kernel - OVN support etc) | 21:54 |
rhallisey | nice :) | 21:54 |
portdirect | you literally just fire it up and run systemctl start kublet :) | 21:55 |
portdirect | it then takes over - and poopulates this with random values https://github.com/portdirect/harbor/blob/latest/config/harbor-auth.conf | 21:55 |
kfox1111 | we're not quite as easy yet, but do have instructions from gonig from nothing, to getting stuf working in a minikube. | 21:55 |
kfox1111 | no custom software needed. | 21:55 |
kfox1111 | so should deploy on any k8s. | 21:56 |
rhallisey | portdirect, one think we've been thinking about in kolla-k8s end is the bootstrap/deplpoy/upgrade mechanism | 21:56 |
rhallisey | and the robustness of kubernetes when it comes to orchestrating/ugrading openstack services properly | 21:56 |
*** kbaegis has quit IRC | 21:57 | |
kfox1111 | and the debugging aspect. | 21:57 |
kfox1111 | if there is a lot of automation, and no hooks to get in and help debug, then operators are going to suffer. | 21:57 |
portdirect | getting rid of kube-proxy did wonders in that department - though I'm sure I could have worked round it - neutron with the standard agents and it fought loads over who had the last say on iptables :) | 21:58 |
*** daneyon has joined #openstack-kolla | 21:58 | |
kfox1111 | so we've spent a lot of time on getting things fine grained enough it should be easy to debug. | 21:58 |
kfox1111 | we're planning on working on the workflow bits soon. | 21:58 |
rhallisey | portdirect, neat :) | 21:58 |
rhallisey | portdirect, curious on your thoughts about how things should be orchestrated/upgraded | 21:59 |
kfox1111 | +1. upgrades are very hard to do. been doing them for many years now. never fun. | 22:01 |
rhallisey | I mean there may not be a right answer to the question | 22:02 |
rhallisey | just curious | 22:02 |
*** daneyon has quit IRC | 22:03 | |
portdirect | tricky - i think we have taken the same approach so far from what I can see. I run a job to do db management, and then just iterate the rc and do a rolling upgrade. It's worked for me... but... | 22:03 |
portdirect | by keeping the db's in cinder (well for everything apart from keystone, and cinder itself) its easy to take a snapshot of them and I suppose you could then roll back if you needed. | 22:04 |
rhallisey | how about the ordering though | 22:04 |
rhallisey | that was one thing that concerned me | 22:05 |
kfox1111 | yeah. similar aproach. we've got 0 downtime rolling minor upgrades workign with some of the services now. | 22:05 |
kfox1111 | for the major upgrades though, we're thinkign we need to do it outside of kubernetes, as it really wanst designed for long workflows. go through the tweaks to tthe k8s managed resources a step/job at a time. | 22:06 |
rhallisey | there are some lifecycle hooks that seem to be coming that could help | 22:07 |
kfox1111 | yeah. | 22:07 |
kfox1111 | we've been using advanced k8s objects too. like deployemnts and petsets. | 22:07 |
kfox1111 | they fit a little better for some of the services then regular rc's do. | 22:08 |
rhallisey | portdirect, the question we asked in the community is: if openstack deployment and upgrade is a workflow, can kubernetes function as a workflow enigne? | 22:08 |
*** salv-orlando has quit IRC | 22:08 | |
*** msimonin has quit IRC | 22:08 | |
portdirect | Yes, I think so for what we need to do | 22:09 |
kfox1111 | our conclusion was no. in my experience with upgrades, there often are steps like: | 22:09 |
kfox1111 | * redeply computes with config tagged with version -1 so they will continue to work | 22:10 |
kfox1111 | * redeploy controllers with new version tagged with version -1 | 22:10 |
kfox1111 | * upgrade the database | 22:10 |
kfox1111 | * redeploy computes taged current | 22:10 |
kfox1111 | redeployo controllers tagged current | 22:10 |
kfox1111 | etc. | 22:10 |
*** inc0 has quit IRC | 22:10 | |
portdirect | got it - my deployments are much simpler than yours :) | 22:11 |
openstackgerrit | Serguei Bezverkhi proposed openstack/kolla-kubernetes: WIP Adding logging to kolla-kubernetes services https://review.openstack.org/377104 | 22:11 |
kfox1111 | not sure how to do those steps inside k8s itself. | 22:11 |
rhallisey | portdirect, if you want add you scripts to deploy from containers I"m all for it btw | 22:11 |
kfox1111 | yeah. we're trying to arange things so that its possible to run an upgrade like that. | 22:12 |
rhallisey | kubectl start or whatever it was would be awesome | 22:12 |
portdirect | its the first two that present difficulty- are you using config files on the host? | 22:12 |
kfox1111 | portdirect: so, whats your goal for your project? | 22:12 |
portdirect | job :) | 22:13 |
kfox1111 | portdirect: I get that. :) | 22:13 |
kfox1111 | well, your releaseing source... so is it more for consulting with a software stack? | 22:13 |
portdirect | used to be in renweable engineering, but spend so much time doing CFD that I realised my time was better here. | 22:14 |
kfox1111 | ah. | 22:14 |
portdirect | thats the idea at the moment, though I'm very keen to get involved in anything I can :) | 22:14 |
*** ayoung has joined #openstack-kolla | 22:15 | |
portdirect | how are you configuring the containers atm? | 22:15 |
kfox1111 | so what about contributing to kolla-kubernetes? we are trying to organzie a generic platform for deploying openstack on k8s. | 22:15 |
kfox1111 | being generic doesn't nessisarily mean easy though. | 22:15 |
rhallisey | portdirect, generate config from kolla and have the cli place them in configmaps | 22:15 |
kfox1111 | so having sub distro's that make decisions for you and make it easy to deploy seems like a good reseller like opertunity. | 22:15 |
kfox1111 | so, I see there being 4 things to getting thigns working: | 22:16 |
sbezverk_ | portdirect: qq if you replace kube-proxy with neutron lbaas, curious how dynamic dns functionaity works in thisw case? I was under impression that kube-proxy was the key in this. | 22:16 |
*** huikang has joined #openstack-kolla | 22:16 | |
kfox1111 | config management, workflow, deployment, software. | 22:16 |
kfox1111 | kolla-kubernetes focus right now is on deployment. generating the k8s templates and handing them over to k8s to deploy. | 22:17 |
kfox1111 | software is containers provided by kolla. | 22:17 |
portdirect | kfox - I'm starting work on putting this all into murano - which would provide exatly what your talking about :) | 22:17 |
kfox1111 | and config management is currently kolla-genconfig, but could be others. I hand roll config files sometimes. | 22:17 |
*** neilus has quit IRC | 22:18 | |
kfox1111 | the workflow bit is the least developed at the moment. the plan is to wok through a set of ansible playbooks to make it easy to bootstrap things in the right order, and perform upgrades. | 22:18 |
portdirect | skydns is the critical element for DNS - kube-proxy just uses iptables to point at virtual IPs | 22:18 |
kfox1111 | but other workflow managers would work too. | 22:18 |
sbezverk_ | portdirect: we do use skydns, but would lbaas be able completely replace what kube-proxy does? I throught it is not just iptable but also service registration and nodeport functionality get assisted by kube-proxy | 22:20 |
portdirect | nodeport is done via docker itself, registration is performed by kube-controller-manager | 22:21 |
kfox1111 | so, what I'm thinking is, maybe you add some of your stuff to kolla-kubernetes, as we're kind of an inclusive lego box. then you can work on making some choises for the users, selecting your favorite network config defaults for example, etc, and selling that out of the box experience? That way, everyone wins. the toolbox gets more tools, and you don't have to maintain such a huge code base. | 22:21 |
rhallisey | portdirect, agree with kevin. I think it would be awesome for you to take the pieces and merge them in. Also , you can add the kubernetes container that will run the bits if you like. | 22:22 |
*** MarMat has quit IRC | 22:23 | |
kfox1111 | yeah. I've already started down the same path: https://github.com/kfox1111/kolla-kubernetes-toolbox | 22:23 |
portdirect | sounds great - I'll need some hand holding with gerrit (I've never used it before) but I'd be very keen | 22:23 |
kfox1111 | running the bits out of a continer makes it soo much easier. :) | 22:23 |
rhallisey | portdirect, sure! | 22:23 |
kfox1111 | awesome. yeah, I'd be glad to help with gerrit. :) | 22:24 |
rhallisey | portdirect, I'd recommend starting with https://github.com/openstack/kolla-kubernetes/blob/master/doc/source/minikube.rst | 22:24 |
rhallisey | to get a feel for what is there and how it works | 22:24 |
kfox1111 | yeah. thats our most up to date documentation at the moment. | 22:25 |
kfox1111 | like I said, the workflow piece isn't there yet, so there's a lot of copy/paste steps in it for now. | 22:25 |
rhallisey | oh speaking of which I need to rebase | 22:25 |
kfox1111 | but we plan on automating that soon. so any ideas you have there woudl be good. :) | 22:25 |
portdirect | yeah I was looking at that - how about i add heat to get a feel for things ? | 22:26 |
kfox1111 | portdirect: cool. yeah. :) | 22:26 |
kfox1111 | see the common-deployment.yml.j2 template. heat-api woudl best map to that. | 22:26 |
kfox1111 | and heat-engine to a petset. | 22:26 |
rhallisey | portdirect, ya that would be great | 22:27 |
kfox1111 | cinder api would probably be a good example to copy from. | 22:27 |
portdirect | and then get going on the real work, fyi you can checkout here https://86.24.220.168:9090/system (username harbor, pass acoman) if you want to see where I am at the moment - Iits just a test rig so you can't break anything :) | 22:27 |
kfox1111 | we do rpc engines as petsets as they register their dns entries into the database. | 22:27 |
*** schwicht has quit IRC | 22:28 | |
portdirect | thats a great idea - I've been cleaning the dead egines every so often... | 22:28 |
* kfox1111 nods | 22:28 | |
*** matrohon has quit IRC | 22:29 | |
*** neilus has joined #openstack-kolla | 22:30 | |
kfox1111 | see services/mariadb/mariadb-pod.yml.j2 for a petset example. | 22:30 |
*** lamt has quit IRC | 22:31 | |
*** huikang has quit IRC | 22:32 | |
*** fguillot has quit IRC | 22:32 | |
*** huikang has joined #openstack-kolla | 22:33 | |
openstackgerrit | Ryan Hallisey proposed openstack/kolla-kubernetes: Documentation rework https://review.openstack.org/375866 | 22:35 |
*** neilus has quit IRC | 22:35 | |
sbezverk_ | portdirect: very interested in Kuryr, currently I am using flanneld in my multinode cluster, can you point me in your repo where you deal with Kuryr | 22:35 |
*** portdirect has quit IRC | 22:36 | |
sbezverk_ | I would like to replace flanneld | 22:36 |
*** portdirect_ has joined #openstack-kolla | 22:36 | |
*** portdirect_ is now known as portdirect | 22:36 | |
kfox1111 | I'd like to support flanneld and something more advanced. | 22:36 |
portdirect | sry - powercut, might have missed the last few messages | 22:37 |
kfox1111 | flanneld is good for an undercloud like concept. but not as good as more advanced things. | 22:37 |
*** huikang has quit IRC | 22:37 | |
sbezverk_ | kfox1111: it looks like people these days go more for l3 solutions | 22:38 |
sbezverk_ | portdirect: I was asking if you could point me to your Kuryr bits in your repo | 22:38 |
kfox1111 | yeah. | 22:38 |
portdirect | ok - thats the biggest mess :) | 22:38 |
kfox1111 | the thing that makes flannel good for undercloud stuff though is how simple it is. the only thing you need for a db is etcd. | 22:39 |
kfox1111 | while doing more advanced sdn requires a lot more stuff. then its hard to get those bits started. | 22:39 |
portdirect | https://github.com/portdirect/harbor/tree/latest/docker/openstack/openstack-kuryr/openstack-kuryr-libnetwork | 22:39 |
kfox1111 | like, basing on neturon is awesome, but then how do you start/manage neutron? | 22:39 |
sbezverk_ | kfox1111: agree it works like a charm in my test bed, but somebody mentioned that it is kind of fading out | 22:39 |
sbezverk_ | portdirect: thank you | 22:40 |
kfox1111 | you have to manage keystone/neutron/mariadb for it. if you want to make it resielient, you need multiple nodes, then you need something to manage it. :/ | 22:40 |
*** thumpba has quit IRC | 22:40 | |
portdirect | though the k8s stuff is in here https://github.com/portdirect/kuryr/tree/k8s/kuryr/raven | 22:40 |
*** thumpba has joined #openstack-kolla | 22:40 | |
kfox1111 | sbezverk_: yeah. its is very simple. which is kind of why I like it. | 22:40 |
kfox1111 | it would be very difficult to add features to it. | 22:41 |
kfox1111 | but for the: bootstrap off of this vm, then deploy bare metal and extend your seed to the bare metal case, it would work well. | 22:41 |
*** thumpba has quit IRC | 22:42 | |
kfox1111 | you could then use that physical seed system to deploy all your other nodes with ironic+neutron+k8s. :) | 22:42 |
sbezverk_ | kfox1111: I do not think it will be suitable for multitenancy, but l3 based solution could you some policies for this | 22:42 |
portdirect | its possible to bootstrap using neutron - and then pass over to the 'real' neutron api | 22:42 |
kfox1111 | yeah. you'd have a 3 node flannel+k8s undercloud in that case, and then all the other nodes would use neutron. :) | 22:42 |
kfox1111 | so the undercloud woldn't need multitenancy. | 22:43 |
kfox1111 | portdirect: I'm thinking of a world where you use ironic + neutron with direct access to the switch for port reconfigures. | 22:44 |
kfox1111 | you might have a minimal 3 node k8s cluster for managing that neutron and that ironic stack, | 22:45 |
kfox1111 | and then use that cluster to build a huge k8s cluster (or multiple) | 22:45 |
portdirect | flannel definialty makes things easier - i used to use it but now run one pod for each service in the host's network namespace which also works though may not be acceptible? | 22:45 |
kfox1111 | we're running a few pods that way, ones that realy need to, | 22:46 |
kfox1111 | but the rest we're trying to avoid that and useing k8s's services/deployments. | 22:46 |
kfox1111 | that lets us do rolling upgrades and such. | 22:46 |
portdirect | this is exacltly where i was going - I've not a copy of magnum that I'm confortable sharing at the moment - but I will soon :) | 22:46 |
kfox1111 | awesome. :) | 22:46 |
sbezverk_ | kfox1111: is it not what ACI solution does? | 22:47 |
kfox1111 | yeah. having an undercloud deploy k8s using magnum on ironic bare metal would be amazing. :) | 22:47 |
kfox1111 | sbezverk_: I think so. but neutron just sits in front to let you be vendor agnositc. | 22:48 |
kfox1111 | I think arista was the first to suport ironic neutron drivers. | 22:48 |
*** MarMat has joined #openstack-kolla | 22:51 | |
sbezverk_ | kfox1111: not sure who was first or second but cisco definetly supports it and it is on my to do list to bring up aci solution, I want to see how neutron configures my nexus switch :-) | 22:52 |
*** haplo37 has quit IRC | 22:52 | |
kfox1111 | sbezverk_: +1. :) thats been on my to explore list for a while too. :) | 22:52 |
kfox1111 | we've got some cisco hardware too and want to experiment with bare metal. | 22:53 |
kfox1111 | last I looked the driver was experimental, but maybe its finished now. | 22:53 |
kfox1111 | we totally have a use case for a bare metal cloud though. | 22:53 |
kfox1111 | mostly its been network security holding us back. so if thats fixed, awesome. :) | 22:54 |
kfox1111 | is the experimental queue lowest priority? | 22:54 |
kfox1111 | its been stuck for 1hr 33min so far. :/ | 22:55 |
sbezverk_ | kfox1111: it takes lots of time to experiment ;-) | 22:55 |
kfox1111 | well, it does when your cisco admin isn't well versed in neutron. or cloud sdn in general. :/ | 22:56 |
*** eaguilar has joined #openstack-kolla | 22:56 | |
kfox1111 | plus all of our switches are in production. well, at least until recently. | 22:56 |
kfox1111 | so we gota dig one up. I think we have a spare now we can use. | 22:57 |
kfox1111 | portdirect: I noticed there's a way to get k8s to deploy k8s nodes too. been itching to try that. :) | 22:57 |
kfox1111 | part of the autoscaling stuff. | 22:58 |
kfox1111 | I wonder if you could just get nova+ironic going, then point k8s as it and tell it to scale up. :) | 22:58 |
kfox1111 | rhallisey: looks good to me. | 22:59 |
kfox1111 | we can do the networking bits in a followup. these docs are still better then the previous. | 22:59 |
*** haplo37 has joined #openstack-kolla | 23:01 | |
kfox1111 | my grand vision is to get an ironic based undercloud, and kolla-kubernetes based over cloud running on that hardware, then mix the over and undercloud into the same region. | 23:01 |
kfox1111 | so in one region, users can launch vms and bare metal. | 23:01 |
kfox1111 | and as an admin, I can take nodes from the ironic undercloud and just make them overcloud vm nodes. | 23:01 |
portdirect | +1 | 23:02 |
kfox1111 | and share the k8s layer with users when there is enough multitenancy security there. | 23:02 |
portdirect | yup - This would be almost possible with clear conainers or the rkt atm | 23:03 |
rhallisey | cool | 23:03 |
* kfox1111 nods | 23:04 | |
kfox1111 | portdirect: I've got a patch in to try and get k8s authenticating keystone tokens too. :) | 23:06 |
portdirect | I also wondered about makeing a nova k8s driver - running the infrastruture container via nova-docker. Would offer some hugae advantages in terms of talking to the rest of the Opensatck Ecosystem but I think that the addittional overhead and incread startup time would be a killer? | 23:06 |
kfox1111 | yeah. nova's a pretty inflexable project at this point. I've tried getting code in, and its really hard. :/ | 23:07 |
kfox1111 | I'd like to see a nova compatable api written on top of launching k8s containers with kvm running in them prototyped.... :) | 23:07 |
kfox1111 | I think it would be a pretty small amount of code. :) | 23:08 |
kfox1111 | and would be way easier to manage. would just be a small thin api wrapper around k8s's api. stateless. :) | 23:09 |
openstackgerrit | Merged openstack/kolla-kubernetes: Documentation rework https://review.openstack.org/375866 | 23:11 |
openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: WIP: Misc fixes https://review.openstack.org/377082 | 23:15 |
*** mdnadeem has joined #openstack-kolla | 23:15 | |
kfox1111 | wow... almost 2 hours now for the experimental gate. :/ | 23:16 |
*** schwicht has joined #openstack-kolla | 23:17 | |
kfox1111 | yay. k8s 1.4 :) | 23:18 |
*** mdnadeem_ has quit IRC | 23:18 | |
kfox1111 | the changelog is very interesting. :) | 23:23 |
*** mdnadeem_ has joined #openstack-kolla | 23:25 | |
*** Pavo has quit IRC | 23:25 | |
kfox1111 | hmm.... I gota play more with taints/tolerations... very interesting. | 23:26 |
*** mdnadeem has quit IRC | 23:28 | |
*** lamt has joined #openstack-kolla | 23:28 | |
portdirect | thats a great idea, it would only take half an hour to convert nova-docker to nova-kubectl but it would be horrible... | 23:29 |
kfox1111 | yeah. I think nova's api would be easier to map to k8s. host aggregates = nodeSelectors, etc. then just have a container that pulls an image from glance and runs it under libvirt directly. start/stop can map to a kubectl exec virsh start/stop/ etc. | 23:31 |
kfox1111 | no schedular needed. k8s will take care of it already. etc. | 23:32 |
*** Pavo has joined #openstack-kolla | 23:33 | |
openstackgerrit | Serguei Bezverkhi proposed openstack/kolla-kubernetes: WIP Adding logging to kolla-kubernetes services https://review.openstack.org/377104 | 23:33 |
kfox1111 | sbezverk_: what do you think of the idea of sed -i 's/fluentd/logging/' of the pods in that PS? | 23:35 |
*** neilus has joined #openstack-kolla | 23:35 | |
kfox1111 | then if we ever want to add non fluentd loggers we just need to edit common-logging.yml.j2? | 23:36 |
kfox1111 | with a big conditional that changes the logger type, redefining the macro's? | 23:36 |
*** salv-orlando has joined #openstack-kolla | 23:39 | |
*** neilus has quit IRC | 23:40 | |
kfox1111 | yay. scheduledjob made it! :) | 23:41 |
kfox1111 | hmm... helm is starting to mature too... nice. | 23:43 |
kfox1111 | heh. but their mariadb chart still a toy. :/ | 23:44 |
sbezverk_ | kfox1111: I kind of liked it that is explicit and clear to show what exactly it enables | 23:44 |
*** lujinluo has joined #openstack-kolla | 23:44 | |
*** sdake has joined #openstack-kolla | 23:44 | |
*** salv-orlando has quit IRC | 23:45 | |
sbezverk_ | also I am not sure if enable_logging will fly on kolla side | 23:45 |
kfox1111 | we could call it kolla_kubernetes_enable_logging: flannel? | 23:45 |
sbezverk_ | it could be confusing for people and by mistake might turn it on | 23:45 |
sdake | hey peeps | 23:46 |
kfox1111 | I'm just thinkgin, every time we want to add another logger type, we'd have to add 7 lines to every file if we don't put it in a central logger template. | 23:46 |
kfox1111 | seems very messy. | 23:47 |
kfox1111 | sdake: hey | 23:47 |
sbezverk_ | ok I will change it but not for something long | 23:48 |
sdake | say kfox1111 did you get my msg from earlier | 23:48 |
*** zhurong has joined #openstack-kolla | 23:48 | |
sbezverk_ | how about enable_kube_logging | 23:48 |
kfox1111 | sdake: no. missed it. let me look for it. | 23:48 |
kfox1111 | sbezverk_: that sounds good to me. | 23:49 |
sbezverk_ | also for macro name kube_logger | 23:50 |
*** tonanhngo has quit IRC | 23:50 | |
sbezverk_ | {{ kube_logger.kube_logger_container() | 23:50 |
*** haplo37_ has quit IRC | 23:51 | |
*** haplo37_ has joined #openstack-kolla | 23:53 | |
*** tonanhngo_ has joined #openstack-kolla | 23:55 | |
*** tonanhngo has joined #openstack-kolla | 23:57 | |
*** tonanhng_ has joined #openstack-kolla | 23:58 | |
*** tonanhngo_ has quit IRC | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!