Thursday, 2018-01-11

*** jascott1 has joined #openstack-kolla00:00
*** pengdake has quit IRC00:02
*** jascott1 has quit IRC00:03
*** jascott1 has joined #openstack-kolla00:03
*** jascott1 has quit IRC00:07
*** goldyfruit has quit IRC00:11
*** k_mouza has joined #openstack-kolla00:19
*** tovin07_ has joined #openstack-kolla00:44
openstackgerritzhongshengping proposed openstack/kolla-ansible master: Use WSGIApplicationGroup %{GLOBAL} in mod_wsgi configuration  https://review.openstack.org/53243700:50
*** pengdake has joined #openstack-kolla00:52
*** threestrands has quit IRC01:03
*** zhaochao has joined #openstack-kolla01:04
*** livelace-link has quit IRC01:08
*** jtriley has joined #openstack-kolla01:10
*** caowei has joined #openstack-kolla01:11
*** genek_ has quit IRC01:12
*** tovin07_ has quit IRC01:14
*** kiennt26 has joined #openstack-kolla01:17
*** mdnadeem has joined #openstack-kolla01:22
*** kevzha01 has joined #openstack-kolla01:31
*** zhurong has joined #openstack-kolla01:33
*** k_mouza has quit IRC01:43
*** livelace-link has joined #openstack-kolla01:46
*** livelace-link has quit IRC01:48
*** TxGirlGeek has quit IRC01:53
*** skramaja has joined #openstack-kolla02:06
*** tovin07_ has joined #openstack-kolla02:09
*** duonghq has joined #openstack-kolla02:17
*** kevzha01 is now known as kevinz02:31
*** genek has joined #openstack-kolla02:41
*** genek has quit IRC02:46
*** jtriley has quit IRC02:46
*** threestrands has joined #openstack-kolla02:47
*** threestrands has joined #openstack-kolla02:47
*** jtriley has joined #openstack-kolla02:48
*** masber has quit IRC03:02
*** masber has joined #openstack-kolla03:05
*** unicell has quit IRC03:12
*** harlowja_ has quit IRC03:21
*** genek has joined #openstack-kolla03:33
*** genek has quit IRC03:38
openstackgerritDai Dang Van proposed openstack/kolla-ansible master: Support policy.yaml file [part 8]  https://review.openstack.org/53272103:39
*** genek has joined #openstack-kolla03:50
*** genek has quit IRC03:55
*** gkadam has joined #openstack-kolla04:03
*** Angeh has joined #openstack-kolla04:06
*** absubram has quit IRC04:06
*** genek has joined #openstack-kolla04:08
*** zhurong has quit IRC04:12
*** genek has quit IRC04:13
*** daidv has joined #openstack-kolla04:16
*** jtriley has quit IRC04:33
*** unicell has joined #openstack-kolla04:40
*** absubram has joined #openstack-kolla04:44
*** unicell has quit IRC04:45
*** dardelean has joined #openstack-kolla04:49
*** dardelean has quit IRC04:54
*** jtriley has joined #openstack-kolla05:00
*** jtriley has quit IRC05:05
*** jtriley has joined #openstack-kolla05:06
*** janki has joined #openstack-kolla05:07
*** itlinux has quit IRC05:14
*** gkadam has quit IRC05:16
*** unicell has joined #openstack-kolla05:17
*** unicell has quit IRC05:21
*** gema has quit IRC05:26
*** gema has joined #openstack-kolla05:27
*** zhurong has joined #openstack-kolla05:28
*** harlowja has joined #openstack-kolla05:29
*** dardelean has joined #openstack-kolla05:30
*** dardelean has quit IRC05:34
*** genek has joined #openstack-kolla05:40
*** genek has quit IRC05:45
*** coolsvap has joined #openstack-kolla05:47
*** genek has joined #openstack-kolla05:49
*** genek has quit IRC05:53
*** xinliang has quit IRC05:56
*** jbadiapa has quit IRC06:06
*** xinliang has joined #openstack-kolla06:08
*** xinliang has joined #openstack-kolla06:08
*** jtriley has quit IRC06:12
*** Angeh has quit IRC06:15
*** unicell has joined #openstack-kolla06:17
*** feldmann has joined #openstack-kolla06:18
*** unicell has quit IRC06:24
*** feldmann has quit IRC06:32
*** jtriley has joined #openstack-kolla06:35
*** ntpttr_laptop has quit IRC06:36
*** unicell has joined #openstack-kolla06:38
*** jtriley has quit IRC06:43
*** harlowja has quit IRC06:47
*** threestrands has quit IRC07:00
*** threestrands has joined #openstack-kolla07:01
*** threestrands has quit IRC07:01
*** threestrands has joined #openstack-kolla07:01
*** genek has joined #openstack-kolla07:02
*** dardelean has joined #openstack-kolla07:02
*** feldmann has joined #openstack-kolla07:02
*** threestrands has quit IRC07:02
*** threestrands has joined #openstack-kolla07:03
*** threestrands has quit IRC07:03
*** threestrands has joined #openstack-kolla07:03
*** threestrands has quit IRC07:04
*** threestrands has joined #openstack-kolla07:04
*** egonzalez has joined #openstack-kolla07:06
*** absubram has quit IRC07:12
*** cah_link has joined #openstack-kolla07:12
*** jtriley has joined #openstack-kolla07:15
egonzalezcan i have reviews on this tacker fixes? https://review.openstack.org/#/c/531379/ https://review.openstack.org/#/c/531374/07:19
*** pcaruana has joined #openstack-kolla07:21
*** dims_ has quit IRC07:21
*** jtriley has quit IRC07:21
*** dardelean has quit IRC07:21
*** feldmann has quit IRC07:21
*** dardelean has joined #openstack-kolla07:22
*** jtriley has joined #openstack-kolla07:23
*** chmarkus has joined #openstack-kolla07:24
*** dims has joined #openstack-kolla07:25
*** chmarkus has quit IRC07:26
*** feldmann has joined #openstack-kolla07:26
*** chmarkus has joined #openstack-kolla07:26
*** dardelean has quit IRC07:26
*** jtriley has quit IRC07:28
*** feldmann has quit IRC07:31
*** jtriley has joined #openstack-kolla07:34
*** skramaja has quit IRC07:37
*** zhangfei has joined #openstack-kolla07:39
*** jtriley has quit IRC07:40
*** jtriley has joined #openstack-kolla07:45
openstackgerritMerged openstack/kolla-ansible stable/pike: Fix deployment with public TLS enabled  https://review.openstack.org/52115407:46
*** feldmann has joined #openstack-kolla07:47
*** dciabrin has quit IRC07:49
*** jtriley has quit IRC07:50
openstackgerritMerged openstack/kolla-ansible master: Enable vitrage dashboard  https://review.openstack.org/53045507:51
daidvegonzalez, could you review for remain patches of supporting policy YAML file? #link https://review.openstack.org/#/q/topic:bp/support-custom-policy-yaml+(status:open+OR+status:merged)07:53
*** unicell1 has joined #openstack-kolla08:00
*** jtriley has joined #openstack-kolla08:00
*** unicell has quit IRC08:00
*** dciabrin has joined #openstack-kolla08:05
*** jtriley has quit IRC08:06
*** jtriley has joined #openstack-kolla08:07
openstackgerritPierre Hanselmann proposed openstack/kolla-ansible master: Designate/Neutron communicaton failure  https://review.openstack.org/53275708:08
*** threestrands has quit IRC08:08
*** dardelean has joined #openstack-kolla08:09
*** jtriley has quit IRC08:12
*** dardelean has quit IRC08:12
*** Zophar has joined #openstack-kolla08:13
*** n0isyn0i1e is now known as n0isyn0ise08:14
*** gkadam has joined #openstack-kolla08:18
*** jtriley has joined #openstack-kolla08:20
*** shardy has joined #openstack-kolla08:22
*** jtriley has quit IRC08:25
openstackgerritNguyen Hung Phuong proposed openstack/kolla master: Remove empty files  https://review.openstack.org/53275908:26
*** knsahm has joined #openstack-kolla08:27
knsahmdoes anybody know a good howto to deploy kolla-kubernetes an on a small kubernetes cluster (master + 3 nodes)08:29
*** jtriley has joined #openstack-kolla08:29
gemaknsahm: we are still working through the scripts to make them work fine, there are issues08:32
*** gfidente has joined #openstack-kolla08:33
gemaknsahm: https://docs.openstack.org/kolla-kubernetes/latest/deployment-guide.html08:33
knsahmgema: thanks a lot08:33
gemaknsahm: you may find help here too: https://github.com/RichWellum/k8s08:34
gemathe deployment guide is being rewritten08:34
*** jtriley has quit IRC08:35
knsahmgema: the guide deploys a all-in-one instance, correct?08:36
*** jtriley has joined #openstack-kolla08:36
gemaknsahm: there should be instructions for both08:36
gemabut I am not sure, I was relaying to you info from my team's internal doc08:36
gemakevinz: ^08:37
gemakevinz: can you help?08:37
openstackgerritZhaokun Fu proposed openstack/kolla-ansible master: FIX inject password error  https://review.openstack.org/53194908:37
*** jbadiapa has joined #openstack-kolla08:40
openstackgerritZhaokun Fu proposed openstack/kolla master: change python-libguestfs to python-guestfs for ubuntu  https://review.openstack.org/53238008:40
*** jtriley has quit IRC08:41
*** pcaruana has quit IRC08:44
*** nathharp has joined #openstack-kolla08:44
*** magicboiz has joined #openstack-kolla08:45
*** magicboiz has quit IRC08:45
kevinzgema: Just finish AIO08:46
kevinzgema: knsahm: But I can help if you need:-)08:47
*** jaosorior has joined #openstack-kolla08:47
*** b_bezak has joined #openstack-kolla08:47
*** Mr_Broken has joined #openstack-kolla08:48
kevinzgema: knsahm: This RichWellum/k8s has been merged to upstream, ref here: https://github.com/openstack/kolla-kubernetes/blob/master/contrib/orchestration/ko.py08:48
kevinzknsahm: This will give you a easy deployment by this script.08:49
knsahmkevinz: ok i will try it08:49
knsahmthanks08:49
*** jtriley has joined #openstack-kolla08:49
kevinzknsahm: yw08:50
kevinzknsahm: ping me if you need help08:50
openstackgerritMerged openstack/kolla-ansible master: Use WSGIApplicationGroup %{GLOBAL} in mod_wsgi configuration  https://review.openstack.org/53243708:52
*** jtriley has quit IRC08:55
*** b_bezak has quit IRC08:57
*** Arminder has joined #openstack-kolla08:58
*** jtriley has joined #openstack-kolla08:58
*** nathharp has quit IRC09:03
*** nathharp has joined #openstack-kolla09:03
*** egonzalez has quit IRC09:03
*** jtriley has quit IRC09:04
*** egonzalez has joined #openstack-kolla09:05
*** jtriley has joined #openstack-kolla09:10
*** k_mouza has joined #openstack-kolla09:13
*** ljjjustin has joined #openstack-kolla09:15
*** jtriley has quit IRC09:15
*** witek has joined #openstack-kolla09:16
*** jtriley has joined #openstack-kolla09:17
*** flaper87 has quit IRC09:17
witekhello09:17
witekcould someone tell me, how often master images are published to Docker Hub?09:18
*** flaper87 has joined #openstack-kolla09:21
*** flaper87 has quit IRC09:21
*** ktibi has joined #openstack-kolla09:22
*** jtriley has quit IRC09:23
*** flaper87 has joined #openstack-kolla09:23
egonzalezwitek, master daily, let me check stable's09:24
egonzalezdaily too09:25
witekdo you use zuul job for that?09:25
openstackgerritDoug Szumski proposed openstack/kolla-ansible master: Local connection to remote Grafana can fail  https://review.openstack.org/53258709:27
egonzalezwitek, yep https://review.openstack.org/#/c/513855/09:30
*** jtriley has joined #openstack-kolla09:31
witekthanks09:31
ktibiHi kolla ;)09:34
ktibiI can see rally container doesn't launch rally db create.09:34
*** Goutham has joined #openstack-kolla09:34
egonzalezktibi, it is https://github.com/openstack/kolla/blob/master/docker/rally/extend_start.sh#L1309:36
*** jtriley has quit IRC09:36
ktibiegonzalez, Yes but when I use rally CLI, I have message => you need to execute rally db create first09:37
GouthamHi All, Facing this issue when performing prechecks using stable pike https://hastebin.com/oyucaxafaw.coffeescript09:38
openstackgerritDoug Szumski proposed openstack/kolla-ansible master: Local connection to remote Grafana can fail  https://review.openstack.org/53258709:39
GouthamDocker version 18.01.0-ce, build 03596f5 in all nodes09:39
egonzalezktibi, need to fix the command https://review.openstack.org/#/c/498697/409:41
egonzalezwas changed to rally db create instead of rally-manage db create09:41
Gouthamegonzalez: any idea on the above error during precheck09:42
Gouthamktibi: ^^09:42
*** jtriley has joined #openstack-kolla09:43
egonzalezGoutham, what python docker you have? min version 2.0.009:46
Goutham pip freeze | grep docker docker==2.7.0 docker-pycreds==0.2.109:47
openstackgerritMerged openstack/kolla-kubernetes master: Changed as per the standard convention set to use "."  https://review.openstack.org/49833209:48
*** jtriley has quit IRC09:49
*** jtriley has joined #openstack-kolla09:49
egonzalezGoutham, The error was: while scanning for the next token\nfound character '\\t' that cannot start any token\n  in \"<unicode string>\", line 2, column 10:\n     Version:\t18.01.0-ce\n             ^"09:49
egonzalez}09:49
egonzalezlooks like doesnt like the \t09:50
Gouthamso how do i fix this ? any idea ?09:51
egonzalezGoutham, ansible version?09:52
egonzalezi dont have a docker 18-ce by hand to test09:52
Gouthamansible 2.4.0.009:53
*** jtriley has quit IRC09:54
ktibiGoutham, thx ;) do you use rally container on your plateform ?09:56
ktibioups it's for egonzalez09:57
ZopharAny issue with zuul now? Seem a job is hanging currently for more thant 6 hours (https://review.openstack.org/#/c/532721/)09:58
openstackgerritEduardo Gonzalez proposed openstack/kolla-ansible master: Support policy.yaml file [part 8]  https://review.openstack.org/53272109:58
Zophar;) or you too busy with it apparently ;)09:58
egonzalezZophar, yep, zuul is having issues/timeouts/etc09:59
*** jtriley has joined #openstack-kolla10:00
*** kristaps_ has quit IRC10:02
egonzalezktibi, i dont use rally very often, just for test upgrades and things like that10:03
*** pcaruana has joined #openstack-kolla10:04
*** ljjjustin has quit IRC10:05
*** cah_link1 has joined #openstack-kolla10:06
*** jtriley has quit IRC10:06
ktibiegonzalez, because I don't know how use the rally container. rally doesn't have API, so for now, I need to copy OS env in rally container and use docker exec for start bench or unit test :/ not very user friendly, maybe do you have better way ?10:06
*** cah_link has quit IRC10:06
*** cah_link1 is now known as cah_link10:06
egonzalezktibi, install rally python clients and point your config json to your openstack urls10:08
ktibiok, same idea :)10:08
*** jtriley has joined #openstack-kolla10:08
egonzalezktibi, https://github.com/openstack/rally/blob/master/doc/source/quick_start/tutorial/step_1_setting_up_env_and_running_benchmark_from_samples.rst#registering-an-openstack-deployment-in-rally10:10
*** tovin07_ has quit IRC10:11
*** cah_link1 has joined #openstack-kolla10:11
*** cah_link has quit IRC10:11
*** cah_link1 is now known as cah_link10:11
*** jtriley has quit IRC10:14
*** sambetts|afk is now known as sambetts10:14
*** jtriley has joined #openstack-kolla10:14
*** duonghq has quit IRC10:16
*** pbourke has joined #openstack-kolla10:16
*** cah_link has quit IRC10:18
*** jtriley has quit IRC10:20
chmarkuswhile trying to add support for Alpine Linux to the openstack-base image I stumbled upon the issue that the pip package python-qpid-proton (and thus also pyngus) is not installable, because the compilation of proton-c fails. Reason seems to be some missing mutex definition in musl, related: http://qpid.2158936.n2.nabble.com/Compiling-proton-c-for-Linux-Alpine-td7669810.html10:22
chmarkusentry in question: https://github.com/openstack/kolla/blob/c0368764fe6aa4a2f12ac093fff26b0a51ae1262/docker/openstack-base/Dockerfile.j2#L37610:23
*** cah_link has joined #openstack-kolla10:23
hrwchmarkus: how much bigger is base when alpine/glibc is used instead of alpine/musl?10:26
*** jtriley has joined #openstack-kolla10:27
chmarkushrw, with alpine/glibc you mean one of the custom alpine builds including glibc on docker hub?10:28
hrwchmarkus: from quick look at that nabble.com talk it looks like alpine/glibc exists and should not have issues with mutex etc which musl has10:28
hrwchmarkus: if size difference is not so big that it may save you time to get some first images built.10:29
chmarkusif I understand that talk correctly, the glibc alpine docker images include only the runtime part of glibc10:30
chmarkusI'd need the compiler for the pip package10:31
chmarkusif I got that right10:31
*** jtriley has quit IRC10:31
hrwah, ok10:32
*** robbbe has joined #openstack-kolla10:33
*** zhurong has quit IRC10:35
*** mnasiadka has joined #openstack-kolla10:35
*** jtriley has joined #openstack-kolla10:37
chmarkusso essentially we wouldn't be able to support Qpid as a RabbitMQ alternative for Alpine based images at all...10:41
*** jtriley has quit IRC10:42
*** feldmann has quit IRC10:44
*** caowei has quit IRC10:44
*** jtriley has joined #openstack-kolla10:48
*** numans has quit IRC10:54
*** jtriley has quit IRC10:54
*** numans has joined #openstack-kolla10:55
*** cah_link has quit IRC10:57
*** jtriley has joined #openstack-kolla10:57
hrwwith 96 cpu cores and low bandwidth using 8 threads looks like half push, half build11:00
*** jtriley has quit IRC11:02
*** jtriley has joined #openstack-kolla11:03
*** feldmann has joined #openstack-kolla11:04
*** pengdake has quit IRC11:07
*** jtriley has quit IRC11:08
*** jtriley has joined #openstack-kolla11:11
*** k_mouza_ has joined #openstack-kolla11:13
*** k_mouza has quit IRC11:14
*** k_mouza_ has quit IRC11:15
*** jtriley has quit IRC11:16
*** zhangfei has quit IRC11:17
*** jtriley has joined #openstack-kolla11:19
*** Goutham has quit IRC11:19
*** k_mouza has joined #openstack-kolla11:20
*** shardy has quit IRC11:22
*** jtriley has quit IRC11:24
*** shardy has joined #openstack-kolla11:24
*** k_mouza has quit IRC11:24
*** k_mouza has joined #openstack-kolla11:26
*** jtriley has joined #openstack-kolla11:26
*** jtriley has quit IRC11:32
*** jtriley has joined #openstack-kolla11:39
*** jtriley has quit IRC11:45
*** jtriley has joined #openstack-kolla11:49
*** rhallisey has joined #openstack-kolla11:54
*** jtriley has quit IRC11:55
*** jtriley has joined #openstack-kolla11:59
*** janki has quit IRC11:59
*** pcaruana has quit IRC12:02
*** pcaruana|afk| has joined #openstack-kolla12:02
*** jtriley has quit IRC12:04
chmarkuswhat would be the best way to conditionally remove entries from the list of packages in the following template?12:05
chmarkushttps://github.com/openstack/kolla/blob/c0368764fe6aa4a2f12ac093fff26b0a51ae1262/docker/openstack-base/Dockerfile.j2#L28512:05
chmarkusI want to remove 2 incompatible packages from the list if base_distro is alpine12:05
*** jtriley has joined #openstack-kolla12:05
chmarkusis there a way without completely copying the list into an elif statement?12:06
*** feldmann has quit IRC12:08
*** jtriley has quit IRC12:11
*** sean-k-mooney has joined #openstack-kolla12:14
sean-k-mooneyo/12:15
sean-k-mooneyis there anyone familar with deploying Octavia  with kolla-ansibel around? im hoping to deploy it in the next few days but was wondering how to work around https://bugs.launchpad.net/kolla-ansible/+bug/166837712:16
openstackLaunchpad bug 1668377 in kolla-ansible "kolla-ansible certificates not generating certs for Octavia" [Wishlist,Triaged]12:16
*** jtriley has joined #openstack-kolla12:20
openstackgerritChason Chan proposed openstack/kolla-ansible master: Quickstart rewrite  https://review.openstack.org/52620012:21
egonzalezchmarkus, {% set openstack_base_pip_packages_remove = [1, 2] %}12:22
hrwchmarkus: and NOTE(chmarkus): those two packages got removed because ...12:22
*** jtriley has quit IRC12:24
chmarkusegonzalez, thanks - now I'm curious, how is "openstack_base_pip_packages_remove" evaluated by kolla? I can't find that string anywhere else by grep'ing.12:25
*** jtriley has joined #openstack-kolla12:25
*** llua has quit IRC12:27
*** llua has joined #openstack-kolla12:27
*** egonzalez has quit IRC12:29
*** jtriley has quit IRC12:30
*** jtriley has joined #openstack-kolla12:31
*** egonzalez has joined #openstack-kolla12:34
*** Mr_Broken has quit IRC12:34
*** feldmann has joined #openstack-kolla12:35
*** jtriley has quit IRC12:36
hrwchmarkus: we did not had to use it probably12:36
*** zshi has quit IRC12:39
*** jtriley has joined #openstack-kolla12:40
chmarkusI mean there has to be a function that takes the list "X_remove" and subtracts the elements from the list named "X" defined somewhere12:40
hrwchmarkus: check jinja2 parser code12:41
*** zshi has joined #openstack-kolla12:41
egonzalezchmarkus, https://github.com/openstack/kolla/blob/master/kolla/template/filters.py#L1912:41
chmarkusoh I see12:42
chmarkusmany thanks12:42
chmarkusit was the "customizable()" call in the filter chain for the install_pip macro then12:42
chmarkusnow it makes sense :)12:42
egonzalezyep12:43
*** jtriley has quit IRC12:45
*** absubram has joined #openstack-kolla12:48
*** pcaruana|afk| has quit IRC12:49
*** jtriley has joined #openstack-kolla12:50
*** krtaylor_ has quit IRC12:50
*** absubram_ has joined #openstack-kolla12:52
*** absubram has quit IRC12:53
*** absubram_ is now known as absubram12:53
*** jtriley has quit IRC12:56
*** pcaruana|afk| has joined #openstack-kolla13:02
*** jtriley has joined #openstack-kolla13:02
*** ljjjustin has joined #openstack-kolla13:03
*** jtriley has quit IRC13:07
*** lvdombrkr has joined #openstack-kolla13:09
*** lvdombrkr has quit IRC13:09
*** jaosorior has quit IRC13:09
*** Mr_Broken has joined #openstack-kolla13:11
*** janki has joined #openstack-kolla13:14
*** lvdombrkr has joined #openstack-kolla13:14
*** jtriley has joined #openstack-kolla13:16
*** jtriley has quit IRC13:21
*** chas has joined #openstack-kolla13:21
*** pengdake has joined #openstack-kolla13:21
*** jtriley has joined #openstack-kolla13:24
*** absubram has quit IRC13:25
*** jtriley has quit IRC13:29
*** jtriley has joined #openstack-kolla13:32
*** lvdombrkr has quit IRC13:34
*** jtriley has quit IRC13:38
ktibiWith cinder and ceph enabled, I have a fail with tempest and snapshot volume. I think we need to change to True rbd_flatten_volume_from_snapshot here https://github.com/openstack/kolla-ansible/blob/master/ansible/roles/cinder/templates/cinder.conf.j2#L11413:39
*** mdnadeem has quit IRC13:39
ktibiwith false, we can't remove a volume created from a snapshot => https://bugs.launchpad.net/cinder/+bug/148589713:40
openstackLaunchpad bug 1485897 in Cinder "fail to delete a snapshot after creating a volume from it with RBD driver" [Undecided,Won't fix]13:40
ktibiI found this fix but not merged https://review.openstack.org/#/c/281550/13:40
*** jtriley has joined #openstack-kolla13:46
*** masber has quit IRC13:52
sean-k-mooneyqq if you anre not useing cinder drivers that require it is there any reason to enable the multipathd container13:54
openstackgerritChristian Berendt proposed openstack/kolla-ansible master: Make custom kibana configuration files possible  https://review.openstack.org/53282513:55
*** jtriley has quit IRC13:56
openstackgerritChristian Berendt proposed openstack/kolla-ansible master: Make custom kibana configuration files possible  https://review.openstack.org/53282513:56
openstackgerritChristian Berendt proposed openstack/kolla-ansible master: Make custom elasticsearch configuration files possible  https://review.openstack.org/53282613:58
*** jtriley has joined #openstack-kolla13:59
*** shardy has quit IRC14:02
*** jtriley has quit IRC14:07
*** jtriley has joined #openstack-kolla14:08
*** dave-mccowan has joined #openstack-kolla14:10
*** jtriley has quit IRC14:13
*** dave-mccowan has quit IRC14:14
*** yangyapeng has quit IRC14:15
*** dave-mccowan has joined #openstack-kolla14:16
*** ljjjustin has quit IRC14:29
*** wojdec1 has joined #openstack-kolla14:32
*** wojdec has quit IRC14:32
*** zhaochao has quit IRC14:40
*** lvdombrkr has joined #openstack-kolla14:41
*** mchlumsky has joined #openstack-kolla14:42
*** mchlumsky has quit IRC14:44
*** mchlumsky has joined #openstack-kolla14:45
*** mchlumsky has quit IRC14:45
*** jtriley has joined #openstack-kolla14:46
*** goldyfruit has joined #openstack-kolla14:52
*** mchlumsky has joined #openstack-kolla14:55
*** Rodrigo_BR has joined #openstack-kolla14:56
*** goldyfruit has quit IRC14:59
*** yangyapeng has joined #openstack-kolla15:00
Rodrigo_BRHello guys, this will be my first deploy using kolla I deployed another environment using other tools, I would like deploy a multi node environment using 3 controllers on vms and 2 computes nodes on baremetal.15:00
Rodrigo_BRbut the quick start https://docs.openstack.org/kolla-ansible/latest/user/quickstart.html is a litle confuse.15:01
Rodrigo_BRI need have a separate node for deployment ?15:02
*** wojdec has joined #openstack-kolla15:05
*** shardy has joined #openstack-kolla15:08
*** wojdec1 has quit IRC15:08
pbourkeRodrigo_BR: you don't need to no15:08
hrwmachine used for deployment may be used as one of target nodes iirc15:13
sean-k-mooneykolla is written to use an optional dedicated deployment node for cases where a it is required for you network topology or b you just want to use a dedicated deployment node though i generally prefer to use one of my controler nodes as the deployment node15:15
Rodrigo_BRThanks, and about the local registry, Can I use any host or a separate node?15:15
*** goldyfruit has joined #openstack-kolla15:16
sean-k-mooneyRodrigo_BR: same fro registry, the only requirement is that all nodes can reach it over the network15:16
Rodrigo_BRsean-k-mooney: thanks15:17
hrwRodrigo_BR: you will use official images from hub.docker or build own ones?15:17
Rodrigo_BRhrw: I will try usage the official images15:19
sean-k-mooneyRodrigo_BR: oh in that case you can skip the registry if all nodes have internet acess to pull them form docker hub. im guessing thats what hrw was going to suggest15:20
hrwyep15:20
hrwbuilding own images in such case would be best with having local registry and then build-with-push to local and then deploy from local15:21
hrwwe built all kolla images for debian/source/aarch64 combo in 1.5h today on our new CI builder15:21
hrwand then waited for an hour to get them pushed to hub.docker ;D15:22
Rodrigo_BRhrw: sean-k-mooney: Ok I understood !15:23
sean-k-mooneyhrw do you use apt-cacheng to cache the packages? i used to get a lot os speed ups from a local apt and pip cache in the past15:23
hrwsean-k-mooney: proxy is next step. we just started using that beast15:23
*** chmarkus has quit IRC15:23
*** chmarkus has joined #openstack-kolla15:24
sean-k-mooneyhrw: i have found that alot of the time spend building is either due to slow disks or slow network. that said 1.5 hours is not terible15:24
hrwsean-k-mooney: the good part is that all changes we wanted/needed are merged into kolla and kolla-ansible so we can concentrate on testing now15:26
hrwsean-k-mooney: I plan to do one more build next week just to have second set for upgrade tests15:27
hrwall-in-one setup runs on 29 images15:27
*** chmarkus has left #openstack-kolla15:29
Rodrigo_BRWhat is the stable version ocata or pike ?15:29
hrwboth are stable15:29
hrwpike is latest stable15:29
Rodrigo_BRhrw: I will try with pike, thanks15:31
*** M4he has joined #openstack-kolla15:33
sean-k-mooneyRodrigo_BR: queens will be avilable in about 6 weeks so if your testing time permits you can also upgrade  to queens before going to production15:33
*** M4he has left #openstack-kolla15:33
*** yangyapeng has quit IRC15:33
*** janki has quit IRC15:35
Rodrigo_BRGood news ! Yes I can wait and test the upgrade in my LAB15:35
*** shardy has quit IRC15:35
*** pengdake has quit IRC15:35
*** itlinux has joined #openstack-kolla15:37
*** yangyapeng has joined #openstack-kolla15:38
*** lvdombrkr has quit IRC15:39
*** chas has quit IRC15:39
ktibi6 weeks for queens ?15:42
*** chmarkus has joined #openstack-kolla15:43
ktibiwith luminous for ceph ? :D I saw last talk between inc0 and gema.15:43
ktibidate for first RC of kolla ?15:45
*** mahe has joined #openstack-kolla15:46
Zopharhttps://bugs.launchpad.net/kolla-ansible/+bug/174045515:46
openstackLaunchpad bug 1740455 in kolla-ansible "Kolla Designate Pike "Could not find 1514518466 for openstack.example. on enough nameservers."" [Undecided,New]15:46
ZopharGuys.... the worker is sending NOTIFY to dns backend (Bind)... but doesn't seems to work... if i disable notify on worker, mdns will do the notify on dns backend... and it works15:47
*** itlinux has quit IRC15:47
ZopharSeems more a workaround that a real solution... any clue why bind doesn't accept the notify from the worker?15:48
Zopharnamed.conf?15:48
*** mnasiadka has quit IRC15:49
*** mnasiadka has joined #openstack-kolla15:50
sean-k-mooneypbourke: egonzalez inc0 qq do we gate on external ceph. im going to try and deploy with it shortly just wondering if we test it regularly15:51
*** mahe has left #openstack-kolla15:51
pbourkesean-k-mooney: no it's untested unfortuknately15:51
pbourke*unfortunately15:51
sean-k-mooneyok i tried it last weekend at home with no luck but i was on a weird config so we will see how it goes with centos15:52
*** jmcevoy has joined #openstack-kolla15:52
sean-k-mooneypbourke: are we still planning to move to external ceph only in the future?15:52
ktibiZophar I use pike and designate and I can create record which are created in my bind.15:52
pbourkesean-k-mooney: honestly I'm not sure15:53
pbourkeI guess it depends if any external solution is viable15:53
pbourkefrom what I hear ceph-ansible right now is not15:54
*** mnasiadka has quit IRC15:55
sean-k-mooneypbourke: i actully deployed with ceph ansible for this deployment but not the dockerised mode15:55
pbourkeyeah I'm not saying its not good, just unsure how well it will work as a drop in replacement15:55
sean-k-mooneyi would be happier with it as one if the docker version was robust but i think its still a little young that said we have bugs in our native support too so you can pick your poision i guess15:57
pbourkeyeah16:00
*** itlinux has joined #openstack-kolla16:08
*** jbadiapa has quit IRC16:08
gemasean-k-mooney: we have tested ceph queens16:12
gemasean-k-mooney: smoke test after manual installation worked, we don't support bluestore config16:12
gemasean-k-mooney: not sure if that helps, but there you go :)16:13
gemapbourke: ^16:13
pbourkegood to know gema16:13
sean-k-mooneygema: external ceph or ceph deployed by kolla16:13
gemasean-k-mooney: ceph deployed by kolla-ansible16:13
gemaluminous16:13
gemawe have also tested with pike containers and they also work16:14
sean-k-mooneygema: ah as an fyi there are still 2 bugs with that wich we should really fix.16:14
gemasean-k-mooney: do you have ids?16:14
gemaI can put someone on those16:14
*** lvdombrkr has joined #openstack-kolla16:15
sean-k-mooneyi dont know if the bugs are still open or if i created them ... ill try to open them later if not already there.16:15
sean-k-mooney1 we dont set the cache size when you enable ceph caching so you have to do that as a manual step after or you data will never leave the cache and be written tothe backing teir16:16
gemasean-k-mooney: sounds good, keep me posted16:16
gemaunless you ahve the bandwith16:16
*** egonzalez has quit IRC16:16
gemathen we can review16:16
sean-k-mooney2 we use /dev/sdX names for the external journal paths in the contianers instead of the uuids so external journals can fail if those names change on a reboot16:16
gemasean-k-mooney: one problem we are fixing is that we don't allow to install ceph on a partition instead of an entire drive and that is bad for testing in our env16:17
sean-k-mooneygema: well these have been around since ceph support was added so sinc ei have not fixed them yet im proably going to keep putting them off....16:17
gemasean-k-mooney: we are going to upgrade our cluster to the containers in queens, so we need to make those work16:18
ktibigema why bluestore no works ?16:18
gemaktibi: it cannot be configured, I am told, with kolla-ansible16:18
sean-k-mooneygema: ya the first time i used kolla to deploy ceph i tried to use a partition on my os drive and nuked the whole thing... that was interesting...16:18
gemasean-k-mooney: yep16:18
*** feldmann has quit IRC16:19
gemasean-k-mooney: we have moonshots for testing, and those only have one SSD per cartridge16:19
gemaso our current ceph deployment is not very testable16:19
sean-k-mooneyktibi: the partition layout is different for bluestore and you have to skip steps like creating a file system on the main osd partion16:19
sean-k-mooneygema: for testing loopback devices work well16:20
sean-k-mooneyi think thats what we use in the gate16:20
gemasean-k-mooney: ok, will tell my team16:20
sean-k-mooneygema: they might find this helpful https://github.com/openstack/kolla-ansible/blob/master/tests/setup_ceph_disks.sh16:20
gemasean-k-mooney: thanks!16:21
sean-k-mooneyby the way ye added dpdk support last release correct? do ye test that in your ci?16:22
gemasean-k-mooney: our CI is two guys for now16:22
sean-k-mooneyah :)16:23
gemasean-k-mooney: we are working throught the system to add to tmonjalo's patchwork automation16:23
gemasean-k-mooney: we found out he has stuff automated16:23
gemasean-k-mooney: we are also talking to the unh guys16:23
gemaabout the interop automation lab16:23
gemasome of our members are going to put HW there, I think16:24
gemasean-k-mooney: so still on the making16:24
gemasean-k-mooney: we can catch up in dublin, I am guessing you'll be there?16:24
sean-k-mooneyone of my own personal goals (never have time to work on it) is to extend kolla-ansible to deploy a workign copy of zuul + nodepool so you can easily setup a ci16:24
gemathat'd be nifty16:25
sean-k-mooneygema: ya its looking like i will. i havent booked a ticket/hotel yet but i will likely do that soon16:25
*** lvdombrkr has quit IRC16:25
gemasean-k-mooney: sounds good16:25
ktibiwhen do you think can I test upgrade P=>Q ? I have a big dev plateform with ceph cluster (80 OSDs)16:26
ktibimaybe good to test on a RC :)16:26
sean-k-mooneyktibi: if you feel like testing an rc on production? feel free16:26
ktibidev plateform*16:26
sean-k-mooneythat more sane hehe sorry missread16:27
ktibiplateform for test my deploy :)16:27
*** ManoX has joined #openstack-kolla16:28
ktibiI had make a lot of benchmark on CEPH, very16:29
ktibiexcited to test ceph upgrade :D16:29
kolla-slack<aslancimbom100> are you using nvme ktibi?16:30
*** absubram has joined #openstack-kolla16:31
*** shardy has joined #openstack-kolla16:32
*** knsahm has quit IRC16:32
sean-k-mooneyinc0: i could have sworn kolla had upgrade gates at one point. was i imaginging it? or was that a future goal16:32
*** mnasiadka has joined #openstack-kolla16:33
pbourkesean-k-mooney: future goal!16:33
sean-k-mooneyktibi: it look like all the luminous patches have merged https://review.openstack.org/#/q/topic:bp/ceph-luminous so you shoudl be able to test ceph upgrades16:33
ktibiaslancimbom100 no full SSD16:34
ktibiwith sata16:35
*** absubram has quit IRC16:36
sean-k-mooneypro anti hack tip. if you only have one ssd in the server don thave it hold a cache partiton, 2 journals and the os unless you like really shortening life fo your ssd's16:36
*** wojdec has quit IRC16:39
ktibisean-k-mooney, do you test plugin with ceph mgr ? like metrics in influx ?16:41
*** absubram has joined #openstack-kolla16:41
ktibimaybe we can enable module if influx is deploy by kolla :)16:42
sean-k-mooneyktibi: i have not looked into the fucntionality provided by the ceph mgr yet but i was hoping to add ceph monitoring via collectd in the future.16:42
sean-k-mooney*to my cloud not sure if ill have time to automate it via kolla16:43
kolla-slack<aslancimbom100> ceph-mgr sounds/looks like a good addition...bluestore also available?16:45
ktibiaslancimbom100, no http://eavesdrop.openstack.org/irclogs/%23openstack-kolla/latest.log.html#t2018-01-11T16:19:4116:46
sean-k-mooneythe lack of bluestore is why im currently deploying with ceph ansible and trying the external ceph support16:47
kolla-slack<aslancimbom100> I see ok thanks16:47
*** jtriley has quit IRC16:51
*** jtriley has joined #openstack-kolla16:52
*** pcaruana|afk| has quit IRC16:57
*** k_mouza has quit IRC17:14
*** wojdec has joined #openstack-kolla17:17
*** ntpttr_laptop has joined #openstack-kolla17:32
*** feldmann has joined #openstack-kolla17:35
kfox1111morning.17:36
*** coolsvap has quit IRC17:36
*** gkadam has quit IRC17:37
*** mnasiadka has quit IRC17:41
*** nathharp has quit IRC17:42
*** absubram has quit IRC17:46
-openstackstatus- NOTICE: Due to an unexpected issue with zuulv3.o.o, we were not able to preserve running jobs for a restart. As a result, you'll need to recheck your previous patchsets17:47
*** mnasiadka has joined #openstack-kolla17:50
*** mnasiadka has quit IRC17:54
*** sambetts is now known as sambetts|afk17:54
*** jistr is now known as jistr|afk17:55
inc0good morning18:00
*** martinst has joined #openstack-kolla18:03
*** jbadiapa has joined #openstack-kolla18:04
*** mnasiadka has joined #openstack-kolla18:05
*** wojdec has quit IRC18:06
*** mnasiadka has quit IRC18:10
*** ntpttr_laptop has quit IRC18:11
*** unicell1 has quit IRC18:13
*** robbbe has quit IRC18:17
*** wojdec has joined #openstack-kolla18:18
*** mnasiadka has joined #openstack-kolla18:20
*** wojdec1 has joined #openstack-kolla18:20
*** wojdec has quit IRC18:22
*** mnasiadka has quit IRC18:24
sean-k-mooneyinc0: 0/ how is life states side18:32
inc0cold rain today18:32
inc0probably my least favorite weather18:33
inc0it could only get worse if cold rain would be freezing rain18:33
sean-k-mooneywe had fog up to like an hour ago when i think it just froze instead of clearing18:33
sean-k-mooneyim currently preparing to deploy the docker hub stable pike images with external ceph which could be interesting18:35
*** mnasiadka has joined #openstack-kolla18:35
inc0yeah, totally18:35
sean-k-mooneyim personaly hoping its boaring and just works :P18:35
inc0gates are using hub images all the time now18:35
sean-k-mooneyhow has that effect speed/reliablity?18:36
inc0not really18:36
inc0mechanism for publishing is similar to that of tarballs18:36
inc0but I have change in pipeline that's adding testing before publishing18:36
inc0so that should have positive impact18:36
sean-k-mooneyoh ya i forgot about that.18:36
inc0lol18:37
inc0I just wanted to exec to my standing container18:37
inc0with sleep 99999 and was baffled when I couldnt18:38
inc0I guess 99999 seconds passed..18:38
sean-k-mooneylol ya maybe try sleep infinity next time18:38
sean-k-mooneyregarding the gates we dont run tempest yet correct18:39
sean-k-mooneyis that what your patch is doing or something else?18:39
*** mnasiadka has quit IRC18:40
inc0no, my patch just adds testing to pipeline18:40
inc0no tempest or upgrades yet18:40
inc0:(18:40
inc0we have all the tools we'd need tho18:40
inc0with stable images up there we can totally do upgrade gates18:40
sean-k-mooneywell we have the tempest container already so it would not take too much to add tempest18:40
inc0yeah18:41
sean-k-mooneywe have the role too actully18:41
inc0yup18:41
sean-k-mooneythe only annoyance with the role is you have to do two deploy to use it18:41
inc0hmm..I wonder why18:42
inc0I mean if you put tempest role at the very end, it really have whole cloud there18:42
sean-k-mooneyactully you could do it in one. you just need to know the flavor ids, image ids and network ides you will use ahead of time18:43
sean-k-mooneythe role uses that info to create a tempest config18:43
*** wojdec1 has quit IRC18:47
inc0you can create these in tempest role18:47
inc0I'd be ok with that18:47
sean-k-mooneyinc0: yes you could but it does not do it currently if you did it would work the way devstack does18:47
*** mnasiadka has joined #openstack-kolla18:50
*** shardy has quit IRC19:01
*** gkadam has joined #openstack-kolla19:17
*** bmace has quit IRC19:19
*** bmace has joined #openstack-kolla19:19
*** feldmann has quit IRC19:23
openstackgerritChristian Berendt proposed openstack/kolla-ansible master: kibana: allow use of server.basePath  https://review.openstack.org/53294619:25
*** gfidente is now known as gfidente|afk19:26
*** itlinux has quit IRC19:26
*** feldmann has joined #openstack-kolla19:38
*** harlowja has joined #openstack-kolla19:43
*** wojdec has joined #openstack-kolla19:52
*** feldmann has quit IRC19:54
*** itlinux has joined #openstack-kolla19:54
*** itlinux has quit IRC19:55
*** jistr|afk is now known as jistr19:57
jmcevoyHow is the Spectre/Meltdown bug patched with with containers?  We are running version kolla-ansible 3.0.3 with Newton.  We will be patching the hypervisors and controllers and the cinder images.  Do the kolla containers need to be patched since updates to kvm and qemu are part of the patch?20:00
*** ManoX has quit IRC20:05
*** wojdec has quit IRC20:05
*** gema has quit IRC20:06
kfox1111jmcevoy: I believe you need to create updated containers with new qemu version with the fixes in it applied.20:07
*** jtriley has quit IRC20:09
*** feldmann has joined #openstack-kolla20:10
*** wojdec has joined #openstack-kolla20:11
*** gkadam has quit IRC20:12
*** jtriley has joined #openstack-kolla20:12
*** mnasiadka has quit IRC20:15
*** mchlumsky has quit IRC20:17
*** gema has joined #openstack-kolla20:33
*** Rodrigo_BR has quit IRC20:36
*** gfidente|afk has quit IRC20:47
jmcevoyThanks kfox111120:48
*** mm6021 has joined #openstack-kolla20:53
sean-k-mooneyyeah my new team dev cluster is alive :) time to go home before it explodes and i spend all evening fixing it20:54
jmcevoykfox1111 is there an easy way to pass kolla-build the version of the updated qemu rpm for centos?  I have been building from binaries rpms.20:55
*** nathharp has joined #openstack-kolla20:55
sean-k-mooneyjmcevoy: you sould have to use a template override20:55
sean-k-mooneykfox1111: for  Spectre/Meltdown you should really just need to update the kernel i think it should not need a container rebuild20:56
sean-k-mooneykfox1111: we use kvm from the host kernel so that would be pathced when the host is patched. qemu i dont think is effected by either iteslf.20:58
jmcevoyseal-k-mooney OK I did that before to install a private CA certificate.  And what do I do to push out the change... Have not gotten a chance to run an upgrade yet...  Or do I just run deploy again?20:59
sean-k-mooneyjmcevoy: you would build the images with a new tag version then upgrade to that tag version21:00
sean-k-mooneybut for meltdown and specter i dont think you need a contaienr rebuild just a host kernel update21:00
*** krtaylor has joined #openstack-kolla21:02
jmcevoysean-k-mooney:  Where does /usr/libexec/qemu-kvm run from a container or the host syste,?21:02
sean-k-mooneymeltdown is mitigated by segmenting kernel and user spaces page so qemu does not need to be updated to facilite that and since kvm is from the host kernel it should be fine also21:02
sean-k-mooneyjmcevoy: the container21:02
*** david-lyle has quit IRC21:03
*** krtaylor has quit IRC21:03
*** david-lyle has joined #openstack-kolla21:04
*** krtaylor has joined #openstack-kolla21:04
sean-k-mooneyjmcevoy: i could be wrong but i would guess the qemu changes are updates to the cpu feature set to expose the PCID feautre for all cpu models that support it. that prevents the performance hit intoduces by the kernel adress spaces segmentation and the tlb flushes that introduces.21:05
sean-k-mooneyto be extra safe it would be worth rebuilding the images if for nothing else then to pick up any other security patches that are out standing, but not 100% sure it would be required to patch meltdown on its own21:06
*** wojdec has quit IRC21:07
kfox1111sean-k-mooney: I patched all my machines with a newer qemu-kvm-ev package too21:09
kfox1111as they mention specifically, fixing spectre issues.21:09
kfox1111I'm guesssing it is for a vm to vm attack, not the meltdown vm to kernel attack.21:09
*** Mr_Broken has quit IRC21:10
*** masber has joined #openstack-kolla21:10
sean-k-mooneykfox1111: ah perhaps, jmcevoy an upgrade is your safet bet but without the kernel host update you will still be vulnerable. i think redhat release an ansible playbook to patch RHEL/centos systems21:16
sean-k-mooneyjmcevoy: if the qemu pathed rpms are availabel in the centos main repo(they should be) you should just be able to do a normal image build and it will pick them up. i dont think we pin the version specifically in kolla21:18
kfox1111+1.21:24
*** nathharp has quit IRC21:26
*** nathharp has joined #openstack-kolla21:27
*** nathharp has quit IRC21:31
*** nathharp has joined #openstack-kolla21:32
*** numans has quit IRC21:36
*** numans has joined #openstack-kolla21:36
*** unicell has joined #openstack-kolla21:37
*** threestrands has joined #openstack-kolla21:58
*** threestrands has quit IRC21:58
*** threestrands has joined #openstack-kolla21:58
*** threestrands has quit IRC21:59
*** threestrands has joined #openstack-kolla21:59
*** threestrands has quit IRC21:59
*** threestrands has joined #openstack-kolla21:59
*** threestrands has quit IRC22:00
*** threestrands has joined #openstack-kolla22:01
*** jbrooks has quit IRC22:01
*** jascott1 has joined #openstack-kolla22:02
*** threestrands has quit IRC22:02
*** threestrands has joined #openstack-kolla22:02
*** jbrooks has joined #openstack-kolla22:02
*** bjolo has quit IRC22:09
*** ktibi has quit IRC22:12
*** jtriley has quit IRC22:12
*** feldmann has quit IRC22:16
*** dave-mccowan has quit IRC22:16
*** dave-mccowan has joined #openstack-kolla22:17
*** nathharp has quit IRC22:17
*** goldyfruit has quit IRC22:22
*** dave-mccowan has quit IRC22:25
*** itlinux has joined #openstack-kolla22:30
*** jtriley has joined #openstack-kolla22:41
*** k_mouza has joined #openstack-kolla22:44
*** bjolo has joined #openstack-kolla22:44
*** jbadiapa has quit IRC22:47
*** ruhe has quit IRC22:51
*** ruhe has joined #openstack-kolla22:52
*** Jeffrey4l has quit IRC22:53
*** k_mouza has quit IRC22:54
*** itlinux has quit IRC22:55
*** Jeffrey4l has joined #openstack-kolla22:56
*** itlinux has joined #openstack-kolla22:57
*** goldyfruit has joined #openstack-kolla23:03
*** jascott1 has quit IRC23:06
*** jascott1 has joined #openstack-kolla23:07
*** jascott1 has quit IRC23:09
*** jascott1 has joined #openstack-kolla23:09
*** jascott1 has quit IRC23:11
*** jascott1 has joined #openstack-kolla23:11
*** itlinux has quit IRC23:12
*** jascott1 has quit IRC23:12
*** jascott1 has joined #openstack-kolla23:13
kfox1111SamYaple: have you played with clear containers yet?23:13
*** jbadiapa has joined #openstack-kolla23:14
kfox1111was wondering if cri might be a good/better fit for vms. you could reuse cni and maybe csi easier then?23:14
*** jascott1 has quit IRC23:17
*** jtriley has quit IRC23:19
SamYaplekfox1111: i have not, but its on thelist for sure23:21
kfox1111I got clear containers and cri working together.23:23
kfox1111crio23:23
kfox1111crio sets up the image dir, and then calls runc or cc to do the launch.23:23
kfox1111can be selected by pod annotation.23:24
kfox1111would recommend looking at wha tthey did for the qemu thing you are doing.23:24
kfox1111it would make things a little more runtime specific, but might give you much more direct access to things.23:25
*** jascott1 has joined #openstack-kolla23:29
*** jbadiapa has quit IRC23:32
*** jtriley has joined #openstack-kolla23:36
klindgren_howdy - We (GoDaddy) are wanting to make some changes to kolla-ansible to add in some better controls around deploying containers.  Such as doing things like. If its an API node, start the new container, test the new container, if tests pass, manage harpoxy for that server to place the service in maintenace mode (letting connections drain off).  making the new container active in haproxy.  Then stopping the old container.23:46
klindgren_then on failure rolling back whatever amount of work, to make the original container running/active if something doesn't work with the new container.  Doing something similar for RPC containers....23:48
SamYapleklindgren_: good plan, but that does pretty radically change the way containers are laid down right now (static naming)23:48
SamYaplewhat about just draining haproxy connections, and recreating the container?23:48
SamYaplethat would be far more easily doable23:48
klindgren_Also having vm's go through a _new -> tests -> active.  _active -> _ld.  Which gives the ability to also do a "rollback"23:49
kfox1111kind of more what k8s provides.23:50
SamYaplebut this is for kolla-ansible, no?23:50
klindgren_it would be... we have our own container_deployment which does this and its working now.  But we want to switch to kolla-ansible23:51
klindgren_so we want to port this concept/ability into kolla-ansible23:51
SamYapleklindgren_: for reference, is that public or a known one?23:51
SamYapleor internal only23:52
kfox1111yes, ansible could be made to do it, but would take a fair amount of code. where as k8s was built for that kind of use case.23:52
klindgren_internal - but nothing "godaddy" specific about it - so I can move it externally23:52
SamYaplei was just curious what you were coming from, or if even ansible23:53
*** k_mouza has joined #openstack-kolla23:54
*** k_mouza has quit IRC23:56
*** k_mouza has joined #openstack-kolla23:56
klindgren_its 100% ansible23:56
klindgren_https://etherpad.openstack.org/p/kolla-ansible_better_container_deployments - is the rough flow chart23:56
klindgren_give me a few and I will drop the role and an example playbook in a public git repo.23:59
SamYapleklindgren_: so spinning up a *new* rpc server is no bueno23:59
klindgren_(might actually be tommorrow23:59
SamYaplethatll actually make a new one in the database and the like23:59

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!