Friday, 2016-03-04

« Thursday, 2016-03-03 Index Saturday, 2016-03-05 »
sbezverksdake another question for swift volumes, I do not see it mounts files systems..00:01
*** ayoung has quit IRC00:07
*** sdake has quit IRC00:09
*** ssurana has quit IRC00:13
*** diogogmt has quit IRC00:28
openstackgerritNikolai Korablin proposed openstack/kolla: Reconfigure for Memcached  https://review.openstack.org/28819600:28
dave-mccowansdake is your cloud accessible from the interwebs?00:29
dave-mccowanit would be fun to turn on tls and run https://www.ssllabs.com/ssltest/ on it.00:29
*** daneyon_ has quit IRC00:31
*** vhosakot has joined #kolla00:35
*** vhosakot has quit IRC00:35
*** vhosakot has joined #kolla00:36
*** ayoung has joined #kolla01:05
*** phuongnh has joined #kolla01:14
*** achanda has joined #kolla01:14
*** mbound has quit IRC01:16
*** mbound has joined #kolla01:17
*** iceyao has joined #kolla01:21
*** iceyao_ has joined #kolla01:25
*** sdake has joined #kolla01:27
*** iceyao has quit IRC01:29
*** diogogmt has joined #kolla01:35
*** sdake has quit IRC01:35
*** jtriley has joined #kolla01:35
*** vhosakot has quit IRC01:40
*** vhosakot has joined #kolla01:41
*** vhosakot has quit IRC01:48
*** tfukushima has joined #kolla01:50
*** dmsimard has quit IRC01:50
*** dmsimard has joined #kolla01:51
*** vhosakot has joined #kolla01:51
*** yuanying has quit IRC01:58
*** yuanying has joined #kolla01:59
*** dwalsh has joined #kolla01:59
*** iceyao has joined #kolla02:02
*** iceyao_ has quit IRC02:05
*** iceyao_ has joined #kolla02:09
*** unicell has joined #kolla02:09
*** mbound has quit IRC02:12
*** iceyao has quit IRC02:12
openstackgerritIldar proposed openstack/kolla: Fixing typo in the documetation  https://review.openstack.org/28822302:14
SamYapledave-mccowan: mine is02:16
SamYapleill test your patch if youd like02:16
SamYapleoh yay its all merged02:16
dave-mccowanthanks SamYaple. whenever you get a chance.  unless you have a trusted cert the grade won't be that great anyways. :-)02:17
SamYaplei have a trusted cert :)02:19
SamYapledo you know about letsencrypt?02:19
SamYapleits all free02:19
SamYapleall of my things are signed nowadays02:20
*** stvnoyes has quit IRC02:24
*** stvnoyes has joined #kolla02:25
dave-mccowancool.  it'll be interesting what it says.  we should be able to get it to an A.  i don't think it's there yet though, maybe a B+.02:31
SamYaplewell ill tell you in a few minutes its deploying anew02:32
*** sdake has joined #kolla02:39
*** iceyao has joined #kolla02:43
*** iceyao_ has quit IRC02:44
*** vhosakot has quit IRC02:44
SamYapledave-mccowan: that patch does not work for me....02:45
SamYaplespits out invalid config and doesnt copy the right files02:45
dave-mccowanat deploy?02:46
SamYapleyea02:46
*** unicell has quit IRC02:47
dave-mccowanit's working for me.  how do we figure out what's different?02:47
*** tfukushima has quit IRC02:47
SamYaplekolla_external_fqdn_cacert is an unused var too02:47
dave-mccowanyea, cacert isn't used during deploy... it's really the client that needs it.02:48
SamYaplecan you post your globals.yml? also do a fresh deploy with master (not your patchset)02:48
SamYapleright but its not referenced _anywhere_ in the playbooks02:48
SamYapleits a dead variable02:48
dave-mccowando i need to do anything to make sure it's fresh deploy, and not using an old generated file?02:49
SamYaplei would recommend a fresh fresh deploy yes02:50
SamYapleyou (shouldn't) dont need to rebuild your images02:50
*** sdake has quit IRC02:51
dave-mccowanis that just cleanup-containers and deploy?  or is there something else to wipe /etc/kolla/*?02:51
*** sdake has joined #kolla02:52
SamYaplei would recommend removing all containers and volumes and cleaning out /etc/kolla/*02:52
*** sdake has quit IRC02:52
*** tfukushima has joined #kolla02:55
openstackgerritSam Yaple proposed openstack/kolla: Fix haproxy  https://review.openstack.org/28822502:57
SamYapledave-mccowan: please review that patch02:57
SamYaplewith those changes its working for me02:57
SamYaplethough i dont see what would have changed to make it work02:57
SamYaplewell its deploying now dave-mccowan so ill let you know02:58
dave-mccowani use cacert in this patch out for review https://review.openstack.org/28816502:58
dave-mccowando you have the original error message you got?02:59
SamYapledave-mccowan: that won't work on remote nodes that dont have that cacert03:00
SamYapleim not sure we want to generate things like that03:00
dave-mccowanit's more a developer tool03:01
SamYaplepost-deploy isnt just for devs though, its for everyone03:02
SamYaplethe other openrc files we create can be tranfered around03:02
SamYapleyea dave-mccowan im not sure we want to include the CACERT stuff like that03:03
SamYaplelike anywhere03:03
SamYapleI did get an A rating though03:04
dave-mccowanhaving the certificate generator kind of got us in that business.  but, if we don't have a generator then no devs will use TLS.03:04
dave-mccowanwoot!03:04
SamYapleim not sure i care about that though. the playbooks are for production03:05
dave-mccowanhow about against the horizon endoint?03:05
SamYaplewhat about that?03:05
dave-mccowandid you run the scanner against 443 or 5000?03:06
SamYapleits not like you inject the cacert into webbrowsers those will still complain03:06
SamYaple44303:06
SamYaplehorizon03:06
*** vhosakot has joined #kolla03:08
*** dwalsh has quit IRC03:09
*** vhosakot has joined #kolla03:09
*** yuanying has quit IRC03:16
*** coolsvap has joined #kolla03:21
*** sha has quit IRC03:21
*** rhallisey has quit IRC03:27
*** dims has quit IRC03:40
*** dims has joined #kolla03:43
*** dims has quit IRC03:44
openstackgerritHui Kang proposed openstack/kolla: Set mongodb log file to shared volume by heka  https://review.openstack.org/28823803:52
*** vhosakot has quit IRC03:53
openstackgerritMD NADEEM proposed openstack/kolla: WIP: Reconfigure for Murano  https://review.openstack.org/28823903:53
*** vhosakot has joined #kolla03:55
openstackgerritSam Yaple proposed openstack/kolla: Fix horizon with ssl  https://review.openstack.org/28822503:55
*** yuanying has joined #kolla04:08
*** yuanying has quit IRC04:12
*** Jeffrey4l has joined #kolla04:22
openstackgerritMD NADEEM proposed openstack/kolla: WIP:Reconfigure for ceph  https://review.openstack.org/28824504:23
*** openstack has joined #kolla04:29
*** yuanying has joined #kolla04:30
*** coolsvap has quit IRC04:30
*** coolsvap has joined #kolla04:30
*** openstack has quit IRC04:30
*** openstack has joined #kolla04:32
*** tfukushima has quit IRC04:34
*** openstack has quit IRC04:34
*** openstack has joined #kolla04:36
*** openstack has quit IRC04:36
*** openstack has joined #kolla04:37
*** openstack has quit IRC04:39
*** openstack has joined #kolla14:07
vhosakotsdake: I have tested both nova reconfig  (https://review.openstack.org/#/c/288146/)  and ceph reconfig  (https://review.openstack.org/#/c/288252/)14:07
sdakenice checking14:07
vhosakotcool14:07
sdakevhosakot if your jobs ever fail, please type in "recheck" on a single line in the reply field14:07
sdakethis will cause the gate job to run again14:07
sdakewe dont approve patches unless the  gate is compeltely green14:08
vhosakotah, ok.. I will recheck both then... they failed due to un-related gate errors...14:08
*** macsz has quit IRC14:08
sdakethis looks problematic http://logs.openstack.org/52/288252/1/check/gate-kolla-dsvm-deploy-centos-binary/1fbedca/console.html#_2016-03-04_06_02_59_36214:09
sdakei already recheck tehm14:09
vhosakotcool14:09
sdakeJeffrey4l around?14:10
Jeffrey4lsdake, yep14:10
sdakeJeffrey4l re reconfigure, I'd like to approve changes even if unteted if they are syntatically correct14:10
sdake we can sort out bugs of reconfigure during the rcs14:11
*** mandre has joined #kolla14:11
sdakehey mandre14:11
sdakebut I do want them green in the gate14:11
Jeffrey4lsdake, roger. agree.14:12
Jeffrey4lasking the owner make sure we really testd it locally. :p14:13
sdakeJeffrey4l if people say no, thatso k, that jusst means it oculd be a trouble spot when debugging reconfigure14:13
sdaketoday I am tagging14:13
sdakeso lets ram and jam :)14:13
Jeffrey4lok. np14:13
*** akwasnie has quit IRC14:14
openstackgerritMichal Rostecki proposed openstack/kolla-mesos: Autodetect number of instances  https://review.openstack.org/28464314:14
openstackgerritMichal Rostecki proposed openstack/kolla-mesos: Support dependencies inside one host  https://review.openstack.org/28473714:14
openstackgerritMichal Rostecki proposed openstack/kolla-mesos: Dependency between nova-libvirt and nova-compute inside host  https://review.openstack.org/28481514:14
openstackgerritMichal Rostecki proposed openstack/kolla-mesos: Use constraints only on multinode deployment  https://review.openstack.org/28597014:14
openstackgerritMichal Rostecki proposed openstack/kolla-mesos: Allow to override resources in globals.yml  https://review.openstack.org/28483614:14
*** macsz has joined #kolla14:16
*** macsz has left #kolla14:17
*** vhosakot has quit IRC14:17
*** jmccarthy has quit IRC14:20
*** dwalsh has quit IRC14:21
*** jmccarthy has joined #kolla14:21
sdakeJeffrey4l this could use an ack https://review.openstack.org/#/c/288239/214:21
sdakerhallisey plz remove workflow -1 this patch looks good to go https://review.openstack.org/#/c/287760/314:23
sdakewe are looking at these  for syntactical correctness, not if they have been or haven't been tested14:23
sdakeI dont want to block for 1 week waiting for someone to setup an ironic environment14:23
sdakeJeffrey4l can clean up any messes made :)14:24
sdakehopefully :)14:24
*** mandre has quit IRC14:24
*** macsz1 has joined #kolla14:25
openstackgerritMerged openstack/kolla: Reconfigure mistral service  https://review.openstack.org/28828114:25
*** vhosakot has joined #kolla14:26
openstackgerritOpenStack Proposal Bot proposed openstack/kolla: Updated from global requirements  https://review.openstack.org/28835214:28
*** akwasnie has joined #kolla14:28
openstackgerritMerged openstack/kolla: Reconfigure cinder service  https://review.openstack.org/28605014:28
openstackgerritMerged openstack/kolla: Fixing typo in the documentation  https://review.openstack.org/28822314:29
*** jtriley has joined #kolla14:30
openstackgerritEric Lemoine proposed openstack/kolla: Use use_forwarded_for in manila config  https://review.openstack.org/28847814:30
sdakethis could use a review https://review.openstack.org/#/c/288239/214:30
openstackgerritMerged openstack/kolla: Reconfigure Ironic  https://review.openstack.org/28776014:31
*** dwalsh has joined #kolla14:32
openstackgerritMichal Rostecki proposed openstack/kolla: Playbook for magnum upgrade  https://review.openstack.org/27579814:33
*** macsz1 has quit IRC14:33
SamYaplemorning14:33
sdakethis is the second time I've seen this gate failure: http://logs.openstack.org/85/277085/3/check/gate-kolla-dsvm-deploy-centos-binary/d3da970/console.html#_2016-03-04_13_50_14_86514:34
sdakemorning SamYaple14:34
pbourkehey sam14:34
SamYaplesdake: at least its now returning the error14:35
openstackgerritEric Lemoine proposed openstack/kolla: Use use_forwarded_for in manila config  https://review.openstack.org/28847814:35
SamYaplesdake: slow server. this can be fixed with an early check, ill work it in14:35
sdakeSamYaple thanks :)14:35
sdakeSamYaple btw I figured out why the centos gate is timign out14:35
sdakewe lack pypi mirrors for centos14:36
sdakewhereas pypi mirrors are being used for ubuntu14:36
SamYaplesdake: thats not true at all14:36
sdakepablinger is working on centos mirrors14:36
SamYaplewe use pypi for both14:36
sdakeclarkb just said it was so14:36
sdakei know the code use spypi14:36
SamYaplebut we use them...14:36
SamYapleoh nvm14:36
sdakebut infra lacks a centos pypi mirror14:36
openstackgerritMerged openstack/kolla: Add reconfigure for heka  https://review.openstack.org/28774014:36
*** openstackgerrit_ has quit IRC14:36
openstackgerritOpenStack Proposal Bot proposed openstack/kolla: Updated from global requirements  https://review.openstack.org/28835214:37
SamYaplepypi doesnt change per distro though14:37
*** dave-mccowan has joined #kolla14:37
*** openstackgerrit_ has joined #kolla14:37
sdakeSamYaple clarkb seemed to indicate it does in some subtle way?14:37
sdakenot sure14:37
sdakeI am not a pypi mirror expert ;)14:37
SamYapleif you are talking about wheel then yea centos doesnt install that by default14:37
sdakebut it is clear cnetos is not pulling from the mirros14:37
SamYaplepypi is not for packages14:38
SamYapleit is _clear_ it is pulling from the pypi repos14:38
*** chandankumar has joined #kolla14:38
sdakeoh right14:38
sdakei am tlalking about precompiled thigns14:38
SamYaplethats not pypi14:38
SamYaplebig difference14:38
sdakeinfra suggested if we used hte precompiled pypi mirrors our gates would go faster14:38
SamYapleyou are talking about an rpm mirror and no that doesnt exist yet14:38
sdakebecause things like lxml wouldn't have to build14:39
sdake(in centos-source case)14:39
sdakepip mirrors14:39
sdakecheck irc log in infra for about t-45 mins ago14:39
SamYapleyup already done ive responded in that channel14:40
*** mandre has joined #kolla14:40
inc0guys, a request, could you pase me your outputs of docker exec -t rabbitmq rabbitmqctl eval 'rabbit_clusterer:status().' ? I wrote some crazy text manipulation jinja2 code and I want to make sure that it will work for others, not just me14:40
SamYapleinc0: why do that? why not write a wrapper script that spits out valid json?14:41
mlimasdake, lvm does not works with kolla, right?14:41
SamYaplei dont want to rely on jinja2 manipulation inc014:41
sdakemlima right14:41
*** iceyao has joined #kolla14:42
sdakeprobablybecuase jinja2 is a parser and inc0 doesn't want to write a parser ;)14:42
mlima not think it would be interesting to try to make it work? sdake14:42
sdakemlima go for it - wont stop it from happening - nobody has made ith appen yet14:42
*** diogogmt has quit IRC14:42
inc0we would need to have this wrapper script inside container14:43
inc0and we don't have it in Liberty14:43
elemoine_inc0: http://paste.openstack.org/show/489337/14:43
elemoine_oops, I have an error there14:43
SamYapleinc0: we can have it in container for 1.1.014:43
SamYaplethats a preq for these upgrades anyway14:43
inc0I don't want to rely THAT much on 1.1.014:44
inc0you could pull off non-1.1.0 upgrade if you use nova-net14:44
SamYapleim -2 on jinja2 manilipulation of that string dude. it breaks all the time with horrible error messages14:44
SamYaple1.1.0 is a requirement for these upgrades14:44
SamYapleand we dont use nova-net anywhere14:45
SamYapleit doesnt work14:45
SamYaple(it never has btw)14:45
inc0well, ok, I don't like it tho. if we ever run into situation like that later on, we're screwed14:45
sdakeinc0 can you review https://review.openstack.org/#/c/278045/614:45
inc0I'll find different solution, for future sake14:46
*** thumpba has quit IRC14:46
SamYapleinc0: a real solution is our own command module. i jsut havent had time to write it14:46
SamYaplewe dont have time to do so before we tag either14:46
SamYaplewould suggest script for now14:46
inc0our own command module? really?14:47
sdakeessential and high work that is incomplete will be settled in the rcs14:47
sdakeantyhing that is medium/low getting bounced14:47
SamYapleinc0: yup14:47
inc0I don't want our own command module...we're slowly removing ansible from ansible14:47
SamYapleinc0: we are executing commands in docker containers, ansible has nothing for that14:48
inc0it's just matter of adding alias for docker exec -t {{ name }} in nodes14:49
inc0I'll figure it out, I don't want to change running containers as requirement for upgrade14:49
inc0as a principle14:49
SamYapleinc0: its already a requirement14:50
sdakewtb review of https://review.openstack.org/#/c/278045/614:50
openstackgerritSam Yaple proposed openstack/kolla: Fix horizon with ssl  https://review.openstack.org/28822514:52
inc0SamYaple, where did we change running containers?14:52
SamYapleinc0: neutron. its already a reguirement for the liberty deploy to be changed to do a proper upgrade14:52
inc0also I do not want to backports stuff that is not completely required to be backported14:52
sdakewtb review of https://review.openstack.org/#/c/288239/214:53
SamYapleits required to to upgrades.14:53
inc0and simple text manipulation imho is not14:53
SamYapleand its our initial relaease that type of backports is expected14:53
SamYaplehell sometimes nova has even had to backport a "feature" to do upgrades properly14:53
SamYaplei think we get slack for our 1.0.0 not being future proof14:53
inc0ok, still, I think there should be better ways to do it14:54
inc0I'll use it as backup solution14:54
inc0and I'd rather create ansible module that will help me with string parsing14:54
inc0so later we can make use of it as well14:54
inc0and actually I think it will be better solution for this problem and any problem like this one14:56
SamYapleinc0: thats not how that works14:56
inc0why? oO14:56
SamYapleyou cant create a module just to parse strings from the output of a task (you cant call a module inside of another module)14:56
inc0ok, I don't intend to14:57
SamYaplewhat you _want_ is an action_plugin that wraps command module. thats what im refering too14:57
SamYaplethen we can string manipulate in python14:57
inc0I'll call module which will take a string as input and push string as output14:57
*** diogogmt has joined #kolla14:57
SamYapleew14:57
inc0and input will be registered stuff from command module14:57
SamYaplethen the failures/changes will be on the wrong task14:57
SamYaplethats no good14:58
sdakewe need a titanium gate14:58
SamYaplethe ansible thing here is action_plugin inc014:58
sdakewe have a glass gate atm :(14:58
SamYaplelook at merge_configs14:58
SamYaplebut we can do that later14:58
inc0ok, action_plugin it is14:58
inc0no, I'm doing it now14:58
inc0I'm not changing running containers and again, this will be useful14:58
sdakeinc0 you ahve 18 days :)14:58
SamYapleim not ok with another action plugin this late inc014:58
SamYaplethats not stable14:58
inc0sdake, I'll do it over this weekend14:58
SamYapleinc0: im -2 on that14:58
SamYapleaction plugins changed between v1 and v2 ansible14:59
*** mandre has quit IRC14:59
inc0god damn it, you're ok by code that will change running container but not ok by code that will be run once/14:59
SamYapleim ok with needing stuff in 1.1.0 to upgrade yes14:59
SamYaplesine thats already a re14:59
SamYaplereq*14:59
SamYaplei would prefer to use what we use everywhere else yes14:59
inc0ok, but this is more scallable solution14:59
SamYapleoutput json from a wrapper script14:59
SamYapleok do it in newton14:59
SamYapleim not arguing its usefulness15:00
*** jtriley has quit IRC15:00
*** iceyao has quit IRC15:02
*** jtriley has joined #kolla15:02
openstackgerritMerged openstack/kolla: Reconfigure nova service  https://review.openstack.org/28814615:03
inc0to me it's same amount of work15:03
inc0maybe a little more15:04
inc0but with better morale during doing it15:04
SamYaplenot even close. youve never done an action plugin15:04
SamYaplenor do you know the changes for v2.0 and how to make sure you dont diverge to much so we can port it15:04
inc0I'm clever guy...or I like to think that I am15:04
inc0we're not supporting 2.015:04
SamYaplein newton we are15:04
inc0when we will, we'll deal with them15:04
SamYapleand that means we have to rewrite that15:04
SamYapleno we wton15:05
SamYaplewe just wont do it now this late in the gate15:05
SamYaplegame*15:05
SamYaplei dont want to rewrite our mostly stable parsing this late in the game.15:05
*** mbound has quit IRC15:05
inc0I won't rewrite anything...I'm adding new file15:06
inc0and after we decide to support 2.015:06
inc0I volounteer to rewrite it15:06
SamYaplelets stick with the same method we have for everything else im our repo please15:07
SamYaplei dont want a one of for rabbit right at the end here15:07
SamYapleone-off*15:07
elemoine_SamYaple: https://review.openstack.org/#/c/288478/215:09
elemoine_see my comment15:09
elemoine_the error seems to be related to keystone15:09
elemoine_does that ring a bell?15:10
*** mandre has joined #kolla15:10
SamYapleelemoine_: seems like outdated or bad rebase15:11
SamYaplewould suggest checkkout master15:11
elemoine_SamYaple: I do use master!15:11
elemoine_I use my manila branch based on master (for https://review.openstack.org/#/c/288478/)15:12
SamYaplehmmm i mean thats what that error would come from15:12
elemoine_I don't think so15:12
sdakeelemoine_ akwasnie is this done? https://blueprints.launchpad.net/kolla/+spec/central-logging-service as in the state should be implemented?15:12
elemoine_https://review.openstack.org/#/c/286637/15:12
elemoine_manila is not in there15:12
elemoine_sdake: the Heka Elasticsearch integration patch is waiting for reviews15:13
sdakeelemoine_ got link?15:13
sdakei'd like it reviewed folks :)15:13
sdakei'll take bullet on one review15:13
elemoine_sdake: https://review.openstack.org/#/c/284188/15:13
sbezverksdake for swift bug, which solution would you prefer, installing ceilometer middleware into the base image or removing ceilometer from swift proxy server pipeline? it sits there by default.15:14
sdakeelemoine_ it has -1 with comments already15:14
elemoine_there is one comment from SamYaple that I did not take into account because I did not understand it15:14
sdakesbezverk  remove from pipeline15:14
elemoine_sdake: I replied to Sam's question15:14
SamYapleelemoine_: manila is in that review15:14
elemoine_sdake: I don't understand why I should use "optional" forthe heka-elasticsearch.toml file15:15
elemoine_SamYaple: ok sorry15:15
sdakehell if I know15:15
sdakeSamYaple why should optional be used in that heka review?15:16
*** jasonsb has joined #kolla15:16
*** iceyao has joined #kolla15:16
SamYapleidk can i have a link15:16
sdakeeplain to me so i can explain to elemoine_  :)15:16
sdakehttps://review.openstack.org/#/c/284188/15:16
*** vhosakot has quit IRC15:16
sdakethis closes out a blueprint - which i am in the process of doing i nthe tracker atm ;)15:16
elemoine_SamYaple: see https://github.com/openstack/kolla/blame/master/ansible/roles/manila/tasks/register.yml15:16
SamYapleohh elemoine_ instead of the jinja2 bool block15:17
*** mbound has joined #kolla15:17
SamYapleit should use optional rather than that jinja2 bool block15:18
*** blahRus has joined #kolla15:18
*** thumpba has joined #kolla15:18
*** mandre has quit IRC15:19
elemoine_SamYaple: not sure15:20
*** thumpba has quit IRC15:20
sdakeoh ya elemoine_ that makes sense15:20
*** thumpba has joined #kolla15:20
*** thumpba has quit IRC15:20
sdakeelemoine which part are you unsure of15:20
elemoine_If ES is not used (enable_central_logging is disabled) I don't want to copy that file in the Heka config directory15:20
SamYapleelemoine_: you shouldnt have transfered it in then15:21
elemoine_or Heka will pick it up and report error because ES is not there15:21
elemoine_SamYaple: so config.yml should be changed right?15:21
mlimaelemoine, i saw your issue,15:21
SamYapleelemoine_: yea if its not used it shouldnt be copied15:22
*** thumpba_ has joined #kolla15:22
*** thumpba_ has quit IRC15:22
*** mandre has joined #kolla15:22
mlimaand the cause is it url='{# mistral_v2_internal_endpoint #} elemoine15:22
*** thumpba has joined #kolla15:22
elemoine_mlima: but I thought SamYaple fixed it, so I don't understand15:23
elemoine_mlima: this needs investigations…15:23
SamYaplemlima: well that error says that "url" is unsupported not the value of url15:23
elemoine_SamYaple: I may introduce problems if I remove enable_central_logging and use optional: true instead15:25
mlimaSamYaple, but use mistral in a task of manila?15:25
elemoine_SamYaple: think about the case where the operator resets enable_logging to false and re-deploy15:25
mlimamaster have it  - {'interface': 'admin', 'url': '{{ mistral_admin_endpoint }}'}15:26
elemoine_I guess the heka-elasticsearch.yml file will still be in /etc/kollla/heka and the heka container will try to send the logs to ES (which won't be there anymore15:26
SamYapleelemoine_: that is on the operator. there are many things that break if they change stuff like that _affter_ a deploy15:26
mlimashould be {'interface': 'admin', 'url': '{{ manila_admin_endpoint }}'}? SamYaple15:26
elemoine_but by using enable_logging also in config.json I can prevent such errors15:27
mlimaI'll have lunch now15:27
elemoine_that's all I am saying :)15:27
*** iceyao has quit IRC15:28
elemoine_but I agree I could also avoid copying the config file to /etc/kolla/heka when enable_central_logging is false15:28
elemoine_let me change that15:28
*** iceyao has joined #kolla15:29
*** spisarski has joined #kolla15:30
SamYapleelemoine_: i get it, but we should be using optional, the reason we arent in some places is just old code that should be updated15:30
elemoine_SamYaple: yep, adding that as well15:31
sdakeelemoine will this patch make it through the review gauntlet today?15:37
sdakegive me a 90% confidence estimate15:38
*** mbound has quit IRC15:38
*** inc0 has quit IRC15:38
sdakeelemoine_  ^^15:38
*** inc0 has joined #kolla15:39
openstackgerritMerged openstack/kolla: Updated from global requirements  https://review.openstack.org/28835215:44
openstackgerritMerged openstack/kolla: Reconfigure for Swift  https://review.openstack.org/28805915:44
SamYaplelooking for reviews https://review.openstack.org/#/c/288225/15:45
SamYapleneed it for ubuntu tls15:45
SamYaplealso reviews here https://review.openstack.org/#/c/287580/15:46
SamYapleneed that for --tags + ceph15:46
sdakedave-mccowan is https://blueprints.launchpad.net/kolla/+spec/ssl-kolla done?15:51
dave-mccowansdake yes.   i'll find a couple bugs to fix over the next two weeks, but it's feature complete.15:52
sdakedave-mccowan thanks i'll mark it as so15:52
openstackgerritJeffrey Zhang proposed openstack/kolla: Copy the logs out of the container  https://review.openstack.org/28854115:52
dave-mccowansdake i'll also add some doc15:53
sdakeSamYaple the unableto get token happens all the time on centos source15:53
openstackgerritMerged openstack/kolla: Playbook for magnum upgrade  https://review.openstack.org/27579815:53
sdakeSamYaple you said its jsut a slow host issue?15:53
*** dims is now known as dimsum_15:53
*** iceyao has quit IRC15:53
sdakeas in keystone token retrieval times out?15:53
SamYaplesdake: something like that15:54
SamYaplecentos is always a slower gate15:55
SamYapleso it makes sense there15:55
sdakecentos should be pretty speedy i think15:55
SamYaplenope15:55
sdakeslower in terms of downloading all the stuff15:55
sdakebut once buidl the deploy should have same delta as ubuntu no?15:55
SamYapleits a large percentage slower than ubuntu as a storage backend, even with btrfs15:56
SamYaplemy guess is not all the btrfs stuff in the kernel was backported15:56
SamYaplebut thats just a guess15:56
SamYapleits also running xfs not ext415:56
SamYaplethere are seveal factor15:56
sdakemlima or nihilifer one of you sort out who is going to address sam's comment https://review.openstack.org/#/c/278045/16:00
sdakeneed done shortly plz :)16:00
mlimai'll do it now16:01
sdakemlima ack thanks16:01
openstackgerritMerged openstack/kolla: Playbook for Ironic upgrade  https://review.openstack.org/28827416:01
openstackgerritSam Yaple proposed openstack/kolla: Retry token retrival in bootstrap script  https://review.openstack.org/28855116:01
mlimacd ..16:01
SamYaplesdake: try that patch16:01
mlimaops16:01
sdakeSamYaple i am just seeing in gate16:01
sdakebut we will see gate ersults16:01
SamYaplefair enough16:02
sdakeSamYaple don't you need a sleep 1 in there?16:02
sdakeor some time limit16:02
sdakeline 8016:03
SamYapleoops typo16:03
SamYapleit should have a 116:03
SamYaplei broke gerrit!16:04
openstackgerritSam Yaple proposed openstack/kolla: Retry token retrival in keystone bootstrap script  https://review.openstack.org/28855116:04
SamYaplethere we go16:04
openstackgerritHui Kang proposed openstack/kolla: Set mongodb log file to shared volume by heka  https://review.openstack.org/28823816:06
*** daneyon has joined #kolla16:07
openstackgerritSam Yaple proposed openstack/kolla: Bump ansible version to head of devel  https://review.openstack.org/28855316:07
*** mikelk has quit IRC16:09
openstackgerritMerged openstack/kolla: Use use_forwarded_for in manila config  https://review.openstack.org/28847816:10
openstackgerritMauricio Lima proposed openstack/kolla: Playbook for Manila upgrade  https://review.openstack.org/27804516:10
mlimadone, sdake16:11
*** daneyon_ has joined #kolla16:13
openstackgerritMauricio Lima proposed openstack/kolla: Fix Keystone v3 and Horizon for Manila  https://review.openstack.org/28855816:13
ccesariohttp://pastebin.com/wMqSj2uG  does someone has idea about these errors ?16:14
*** daneyon has quit IRC16:16
sdakeccesario we wont be doing support today - we are in the midst of a milestone release16:16
sdakeccesario foundation release team yells at me every time our release is late ;(16:17
*** Marga_ has quit IRC16:17
sdakeI hope you understand and can come back monday, or try hitting of sbezverk - he has recent experience with kolla deployment16:18
SamYaplei dont see why they would care we arent on schduled releases sdake16:18
SamYapleprithiv: hey sorry we had to handle that keystone patch. it was becoming an issue in the gate16:18
*** achanda has joined #kolla16:19
*** achanda has quit IRC16:20
sdakeSamYaple can you review https://review.openstack.org/#/c/288558/16:21
openstackgerritMauricio Lima proposed openstack/kolla: Playbook for Manila upgrade  https://review.openstack.org/27804516:22
sdakeSamYaple the release team wants all openstack projects to release at the same time to prevent confusion which is why they care16:22
mlimaSamYaple, where {{ item.service_type }} is defined?16:26
openstackgerritThiago Gomes proposed openstack/kolla: Add Manila in the list of images provided by Kolla  https://review.openstack.org/28856916:28
openstackgerritMauricio Lima proposed openstack/kolla: Fix Keystone v3 for Manila  https://review.openstack.org/28855816:29
*** vhosakot has joined #kolla16:29
*** ssurana has joined #kolla16:31
*** jtriley_ has joined #kolla16:31
pbourkecan anyone familiar with the new keystone bootstrap script have a quick look at this? http://paste.openstack.org/raw/489353/16:33
*** jtriley has quit IRC16:34
pbourkeah16:35
pbourkeits the pinnging16:35
pbourke*pinning16:35
pbourkeI need to grab a new build-conf16:35
pbourke>_<16:35
SamYaplemlima: its defined per item16:35
SamYapleprithiv: that command is not in liberty16:36
SamYapleso thats that failure16:36
mlimahow item will know that share is the service_type for manila? SamYaple16:37
openstackgerritThiago Gomes proposed openstack/kolla: Add Manila to the list of images provided by Kolla  https://review.openstack.org/28856916:37
openstackgerritEric Lemoine proposed openstack/kolla: Make Heka send logs to Elasticsearch  https://review.openstack.org/28418816:38
openstackgerritEric Lemoine proposed openstack/kolla: Use alphabetical order in cleanup-containers  https://review.openstack.org/28762616:38
prithivprithiv16:38
openstackgerritEric Lemoine proposed openstack/kolla: Make Heka send logs to Elasticsearch  https://review.openstack.org/28418816:40
openstackgerritEric Lemoine proposed openstack/kolla: Use alphabetical order in cleanup-containers  https://review.openstack.org/28762616:40
openstackgerritEric Lemoine proposed openstack/kolla: Make Heka send logs to Elasticsearch  https://review.openstack.org/28418816:43
openstackgerritEric Lemoine proposed openstack/kolla: Use alphabetical order in cleanup-containers  https://review.openstack.org/28762616:43
*** mlima has quit IRC16:43
*** mlima has joined #kolla16:45
*** fthiagogv__ is now known as fthiagogv16:47
openstackgerritMichal Jastrzebski (inc0) proposed openstack/kolla: Playbook for rabbitmq upgrade  https://review.openstack.org/28796916:49
inc0see SamYaple action plugin wasn't that hard:P16:50
*** akwasnie has quit IRC16:51
*** akwasnie has joined #kolla16:52
*** ssurana has quit IRC16:53
openstackgerritMerged openstack/kolla: Set mongodb log file to shared volume by heka  https://review.openstack.org/28823817:01
*** akwasnie has quit IRC17:01
*** daneyon_ has quit IRC17:02
*** sdake_ has joined #kolla17:02
*** Jeffrey4l has quit IRC17:03
*** daneyon has joined #kolla17:03
*** cfarquhar has quit IRC17:03
*** sdake has quit IRC17:03
*** bmace has quit IRC17:06
sdake_SamYaple mind giving https://blueprints.launchpad.net/kolla/+spec/keystone-fernet-token to dave-mccowan ?17:06
*** bmace has joined #kolla17:07
*** sdake_ is now known as sdake17:07
*** inc0 has quit IRC17:09
*** daneyon has quit IRC17:13
*** absubram has joined #kolla17:13
*** prithiv has quit IRC17:13
*** absubram_ has joined #kolla17:14
*** absubram has quit IRC17:18
*** absubram_ is now known as absubram17:18
*** fgimenez has quit IRC17:20
*** Serlex1 has left #kolla17:25
*** jasonsb has quit IRC17:26
g3ekhello all, I have a question. I'm creating new container to add to kolla. I would like my new container who is in ./kolla/docker/NEW_CONTAINER/Dockerfile.j2 to be build when I launch kolla-build. How can I achieve that?17:28
*** haplo37 has joined #kolla17:28
g3ekalready put it in ./kolla/ansible/site.yml and created a ansible role for it17:29
g3eksdake or Samyaple any idea?17:30
sdakeit will build automatically if its in the docker  tree17:31
sdakeif its nto building you probably did a pip install ./kolla17:31
g3ekactually it's not17:31
sdakeundo that by sudo pip install kolla17:31
sdakerather uninstall kolla17:31
*** kproskurin has quit IRC17:31
sdakeif kolla is pip installedit uses the dockerfiles in /usr/share/kolla17:31
sdakeif kolla is not pip installed it uses your cWD17:32
sdakeit should be reversed, don't you think? :)17:32
sdakeit was originally, semone changed it around17:32
g3ekoh ok thanks17:32
g3ekI know why now17:32
openstackgerritMD NADEEM proposed openstack/kolla: Reconfigure for Murano  https://review.openstack.org/28823917:34
sdaketime for some 300!17:34
*** mbound has joined #kolla17:36
sdakei'm going to lock up the review queue this evening17:39
sdakeplese finish all work before that happens :)17:39
sdakeinc0 mariadb goes to rc1 right?17:42
sdakewtb inc017:42
*** gfidente has quit IRC17:42
sdakeelemoine ping re https://review.openstack.org/#/c/284188/12/ansible/roles/kibana/defaults/main.yml17:45
sdakeelemoine the question is shoudl "kibana_elasticsearch_request_timeout" be a variable or not17:46
sdakeshould it be hard coded, or configurable17:46
sdakeelemoine any thoughts on this piont?17:46
sdakewonder if elemoine had headed to bed17:48
sdakeneed review on https://review.openstack.org/#/c/288239/317:50
sdakeSamYaple if you have a moment17:50
*** mlima has quit IRC17:50
ajafosdake: sorry maybe it's stupid question but what does it mean "lock up the review queue" there will be no possibility to send any CR(for how long)? what will be with open CR's?17:52
*** salv-orlando has joined #kolla17:52
sdakeajafo it means I will -2 all reviews for a short period, but yuc an still submit them17:53
sdakeajafo I expect the lockout window to be less then 1 day - assuming kolla deploys and works as expected with the merge madness taking place atm ;)17:54
ccesariosdake,  no problem :) .. I'll try investigate :)17:55
openstackgerritMerged openstack/kolla: Reconfigure for Memcached  https://review.openstack.org/28819617:55
*** salv-orl_ has quit IRC17:55
ajafosdake: thx17:56
sdakeif our gate was titanium plated, we could just tag at any time17:57
openstackgerritMerged openstack/kolla: Fix Keystone v3 for Manila  https://review.openstack.org/28855817:57
sdakethat is what is required for the release:managed tag17:57
openstackgerritMerged openstack/kolla: Playbook for Manila upgrade  https://review.openstack.org/27804517:58
sdakethis needs a review rhallisey or SamYaple https://review.openstack.org/#/c/288569/217:59
sdakewtb more core reviewers!!17:59
*** mbound has quit IRC18:02
openstackgerritMerged openstack/kolla: Add Manila to the list of images provided by Kolla  https://review.openstack.org/28856918:02
mdnadeemsdake, not getting your comments in https://review.openstack.org/#/c/288239/3, did you mean we should change name of do_reconfigure.yml to murano_reconfigure.yml ?18:06
SamYapleinc0 you around?18:06
*** chandankumar has quit IRC18:11
mdnadeempbourke, ^^18:11
sdakemdnadeem no the task names18:11
sdakeare inconsistent throughout all the reocnfigure scripts18:12
sdakethe Name of the task18:12
sdakenot the tiask file18:12
openstackgerritSam Yaple proposed openstack/kolla: Retry token retrival in keystone bootstrap script  https://review.openstack.org/28855118:14
openstackgerritMerged openstack/kolla: Add authentication for keepalived  https://review.openstack.org/27708518:15
*** inc0 has joined #kolla18:19
*** mbound has joined #kolla18:20
inc0SamYaple, tell me, why are you -2 to action plugin that is used only there?18:23
inc0what's merit of that?18:23
inc0it won't break anything and may be very well used in other places18:23
inc0also action plugin is simple enough that porting it to 2.0 will be a breeze18:24
inc0and you are blocking upgrade patchset because of that, so let's discuss18:24
*** mbound has quit IRC18:24
inc0because backporting features is something we can do, but I don't want to overdo it18:24
inc0we do it because of dataloss and impossibility of network debugging/downtime during upgrades18:25
inc0and I don't think simplification of string manipulation is enough of a reason to backport it to stable branch18:25
inc0also rabbitmq Dockerfile did change in the process of Mitaka18:26
inc0sorry, I take that last back, it won't be problem18:27
SamYapleinc0: that action plugin isnt even needed18:27
SamYapleare regular plugin will work jsut fine18:28
inc0but unwillingness to backport stuff is still my concern18:28
inc0which plugin is that?18:28
SamYapleyou dont need an action plugin for doing what you are doing18:28
SamYaplea normal plugin is fine18:28
SamYaplelibary/plugin/module18:28
SamYaplethe action_plugin I was refering too would wrap the command module and do string parsing18:28
inc0ok, I'm ok with making it a module18:28
SamYapleim not ok with taht flow though18:29
SamYapleregistering and then parsing is no good18:29
SamYapleand we dont do it elsewhere18:29
inc0why?18:29
*** shakamunyi has joined #kolla18:29
inc0but it's better than scripts inside dockerfiles18:29
SamYaplebecause its two seperate tasks for one piece of info18:29
SamYaplebetter or not ,scripts are what we have this cycle18:29
inc0why is that bad?18:29
SamYaplefailures happen on the wrong task18:29
inc0no they don't18:30
SamYapleif we apply what you are doing to other things, they absolutely do18:30
SamYaplecan you stick with what we have please? i dont have time for this arguement18:30
sdakerelease looking good https://launchpad.net/kolla/+milestone/mitaka-318:30
SamYaplethat is all getting refactored in newton anyway18:30
inc0and if failure happen inside docker container we're looking at these very informative to_json parse errors18:30
SamYapleonly if you write teh script wrong18:31
SamYapleyou can catch it all and ensure it outputs json18:31
openstackgerritMD NADEEM proposed openstack/kolla: Reconfigure for haproxy  https://review.openstack.org/28824818:31
SamYaplebut look i really dont want to do this. i agree its not the best inc0 no argument here. but you should have raised this sisue and wrote that plugin and structure weeks ago18:32
SamYaplenot today18:32
SamYaplethis is late in the cycle please stick with what we have18:32
inc0but it's there and it works18:32
inc0and you ask me to throw it away and rewrite in last day18:32
SamYapleits not what we have elsewhere and im not ok with the differnce18:32
inc0but this is special case, aka upgrades18:32
SamYapleits not a special case18:32
SamYaplewe do this all over teh place18:33
SamYapleit just so happens to be in upgrades18:33
inc0and I'm not ok with backporting this stuff to stable branch18:33
SamYaplebut the case is "extract info fro container"18:33
inc0THAT'S what I'm not ok with18:33
SamYapleits a script htat exracts info, its not backporting anythign that affects the runnign container18:33
inc0and we need it in stable branch if we follow your path18:33
inc0you need to backport image definition18:33
sdakeneed which in stable branch inc0?18:33
SamYapleyou need to backport nothing but the script18:33
inc0which means if we make script to migrate 1.0.0->1.1.018:34
inc0it will be impossible18:34
SamYapleon it wont18:34
inc0sdake, small change to rabbitmq Dockerfile18:34
*** JoseMello has joined #kolla18:34
SamYaplethe 1.0.0 1.1.0 scirpt is already ugly18:34
inc0don't make it worse18:34
SamYaplesmall script addition, its not even a real change18:34
sdakeimo we don't want to do a migration from liberty 1.0.0 to 1.1.018:34
SamYapleit doesnt affect running18:34
inc0moreover, don't make it impossible18:34
SamYapleits impossible i jsut told you that18:34
sdakethe priority here is on upgrades, not on consistency18:35
SamYapleplease change the implemntation to what we have elsewhere inc018:35
sdakeif we can't upgrade a backport, then its a no go from me18:35
SamYaplesdake: im goign to -2 that action plugin patch. this isnt a simple matter18:35
inc0sdake, consistency is what SamYaple -1 my patch on18:35
SamYapleyo uare wrong inc018:35
*** britthouser has quit IRC18:35
SamYaplethere are three things18:35
inc0I'll turn it into module- no problem for me18:35
inc0I'll make changed and failed - no problem for me18:35
SamYapleno module inc0 please use a scirpt like we do elsewhere18:36
inc0but god damned you're saying "don't do it in ansible"18:36
SamYaplei am18:36
SamYapleplease do it like we do elsehwere18:36
inc0and I'm not ok with backporting Dockerfiles because it's an upgrade18:36
sdakeholy christ we have to backport  dockerfiles?18:36
inc0if you would do that I'd -2 it18:36
SamYapleno sdake18:36
SamYaplewe dont18:36
SamYapleinc0 is making it seem like we do18:36
sdakecan i get a straight story18:37
sdakelets take the personalization out of it18:37
SamYaplesdake: we jsut need to have teh script availabe in the liberty container. it doesnt affect running container18:37
openstackgerritMerged openstack/kolla: Reconfigure for Murano  https://review.openstack.org/28823918:37
inc0ok18:37
inc0https://review.openstack.org/#/c/287969/218:37
inc0it's about this review18:37
sdakeinc0 what needs to be backported to get it to work sam's way?18:38
inc0problem is, rabbitmq returns gospel node location within a horrible datastructure unlike anything else18:38
inc0sdake, new script has to be inside rabbitmq container18:38
*** kproskurin has joined #kolla18:38
inc0one we upgrade *from*18:38
sdakedo we have to do that for all containers?18:38
inc0just this one18:38
inc0it's a small string parsing stuff18:38
SamYaplesingle file has to exist in liberty container. thats the only "backport" sdake18:39
*** mlima has joined #kolla18:39
sdakeSamYaple is this the only container we would need to do this script backport  for, or are there more?18:39
inc0it's to extract hostname18:39
SamYaplesdake: jsut taht one18:39
SamYaplemind you that file doesnt even affect operation at all18:39
inc0but I made it working without backport at all18:39
SamYapleit just needs to eixst18:39
sdakewell you guys almoset gave mea heart attack that every dockerfile had to be backported18:39
SamYapleinc0: a way thats not consitent and prone to who knows what errors18:39
sdakeok so lets think forward18:39
inc0using ansible and new action_plugin18:39
SamYaple-2 on action_plugin, take that off the table18:40
sdakewhen we are in newton, and we need to ugprade from this action plugin mechanism18:40
sdakewhat doess that look like18:40
SamYapleim -2 on action plugins. they are going to block migration to ansible 2.018:40
inc0no they don't oO18:41
SamYapleinc0: they will. lots changed there18:41
inc0and I can make it module, not action_plugin18:41
SamYaplethats fine. lets talk about it like a module then18:41
inc0and it's a action_plugin that took me 1hr to write18:41
inc0which I assume will be less in 2.018:41
SamYapleone that has no error checking and is bad inc018:41
sdakelets forget sunk cost arguments18:41
sdakebecausethose don't fly with me18:41
sdakethe upgrade to newton18:42
inc0I'm ok with making it module with more error handling18:42
sdakewill it be a clusterfuck with this action plugin mechanism?18:42
inc0no, it will not18:42
inc0it will work just as well18:42
SamYaplesdake: do you remember all the issue we had with json parsing? would you like to introduce who knows what issue or stick with our solution of script in container?18:42
SamYaplethats the question18:42
inc0moreover, we can refactor other stuff that deal with strange strings to use it18:42
sdakeok guys18:43
sdakehere is what i'm thinking18:43
SamYaplethats alot of work instead of just using what we have and doing what you purpose in newton inc018:43
inc0which will have to be parsing errors just as much18:43
sdakei'm thinking we push this upgrade to rc118:43
sdakelike was done with heat18:43
SamYaplesdake: for the recond i want to do what inc0 suggests (kinda).... IN NEWTON18:43
inc0SamYaple, but that's twice the work18:43
SamYapleits a big change18:43
SamYapleyea but i want to solve this everywhere18:43
sdakewhich will give us more time to reasonably sort it out18:43
SamYaplescripts are what we have for this cucle18:43
inc0it is not if we DONT refactor all at once18:43
SamYaplecycle*18:43
SamYapleinc0: so partial refactor for end of cycle18:44
SamYaplesounds fantastic :/18:44
inc0it's not a refactor18:44
inc0it's a new code18:44
*** achanda has joined #kolla18:44
inc0this code IS NOT THERE YET18:44
SamYaplefine new mechanism for end of cycle :/18:44
SamYapleuse what we have please18:44
inc0so we might as well do it correctly18:44
inc0and without backport18:44
SamYapleor do what we know works....18:44
inc0SamYaple, plugins inside will have exaclty the same error handling issues18:45
inc0moreso even18:45
sdakedid my idea of pushing this blueprint to rc1 not hit anyone's radar?18:45
inc0because now I can use educated failed=True logic18:45
SamYaplesdake: yea i heard it but im goign to be arguing the same thing18:45
SamYapleim not for adding this new mechanism at the end of the cycle18:45
SamYapleim very strongly against it18:45
SamYaplethis is work for netwon18:45
inc0and I am strongly against having to backport this script to 1.1.018:46
sdakeSamYaple is the script mechanism the json parser you ahve for keystone?18:46
inc0so I guess we need tiebreaker18:46
inc0sdake, no, it's a new one18:46
SamYapleyes sdake (and simliar things for haproxy and the like)18:46
SamYapleinc0: no its what i have for keystone18:46
inc0but with new logic18:46
sdakeso these scripts need to be bckported?18:46
SamYaplejust the rabbitmq one18:46
SamYapleand it isnt used in the branch, the file jsut needs to exist in the container18:47
sdakebecause bootstrap doesn't exist in keystone liberty?18:47
SamYapleno because thats a different issue18:47
inc0sdake, we don't need to bootstrap liberty18:47
inc0it's a different issue18:47
SamYapleim jsut talking about the method of extracting the data18:47
sdakei know we dont neeed to bootstrap liberty - i'm nott a total moron :)18:47
inc0I need to check out status of running cluster and extract hostname from it18:47
SamYaplewe do script in container in other places18:47
inc0but we didnt backport it18:48
inc0and we don't require them to be in previous version we upgrade from18:48
inc0this is feature backport18:48
sdakewe have already delcared 1.0.0 DOA and only usable for  evaluation18:48
inc0and I'm againt feature backport if we don't need it18:48
inc0simple feature, but feature nonetheless18:48
inc0and yes, we can avoid it by making a module18:49
SamYaplethis isnt a feature18:49
SamYapleyou keep saying that18:49
SamYapleits a single file that needs to exist in the container18:49
inc0is it bug SamYaple ?18:49
SamYapleno wait. we still need all this anyway!18:49
inc0if it's not, it's a feature18:49
SamYaple1.1.0 -- 1.1.1 still needs an upgrade18:49
SamYaplethis is all still required.18:49
inc0no18:50
SamYapleyes18:50
inc0we don't support upgrade 1.1.118:50
inc0I'm NOT backporting upgrade code to 1.1.018:50
inc0or 1.1.1 in that matter18:50
SamYaplethese same scripts are needed to move from z to z.....18:50
SamYaplehow are you purposing thats solved18:50
inc0yup. 2.0.0 -> 2.0.118:50
inc0earliest18:50
inc0we didnt have upgrades in liberty18:50
SamYapleso no z upgrades on liberty?18:50
sdakeagree with inc0 on thi point18:51
inc0we're not going to backport upgrades to liberty18:51
SamYaplewhy even pump out a 1.1.0 then?18:51
SamYapleif we dont support it whats the point18:51
sdakewe support upgrades from 1.1.0 to 2.0.018:51
inc01.1.0 -> 2.0.018:51
SamYapleif there are any bugs 1.1.0 is unstable18:51
SamYapleand no z upgrades means its worthless18:51
SamYaplewhats the point of having 1.1.0 if we cant pop out a 1.1.1 if needed18:51
*** athomas has quit IRC18:51
inc0we're not backporting upgrades to liberty18:52
SamYapleno one will use 1.1.0 if we have no way to even do security updates18:52
sdakethis is a complex issue18:52
inc0that will mean we need to split bootstraping  in liberty18:52
SamYapleim certainly not going to do any 1.1.0 work if we cant even do security updates after18:52
inc0and basically re-check all the stuff we did last 4 months18:52
sdakewe never claimed we would upgrade 1.1.018:52
SamYaplethen its a worthless branch18:53
sdakeif someone wnats security udates they can redeploy 1.1.118:53
SamYaplethat is the worst thing ive ever heard sdake18:53
SamYaplethats not a stbale branch18:53
SamYaplethats a tech preview18:53
SamYapleno one will ever go for that and certainly would never recommend an operator use it18:53
inc0no, it's not willingness to backport Mitaka to Liberty...full Mitaka to Liberty18:53
SamYaplei will be vocally outspoken if we dont have the ability to z upgrade the liberty branch18:54
sdakewell the 3 of us can't decide to offer backports in liberty18:54
inc0we can deploy mitaka kolla with liberty code of nova and such18:54
sdakethat needs to be a wider audience18:54
SamYaplei will be telling people to stay away18:54
inc0but that's not liberty kolal18:54
inc0we did fight for 1.1.0 because of problems we know of18:55
SamYaplewe were supose to have z upgrade for liberty all along.....18:55
inc0not from liberty to liberty18:55
SamYaplenow we are saying we are abandoning it18:55
SamYapleyes18:55
SamYaplethat was always the plan18:55
SamYapleeven before you were here inc018:55
inc0we don't have upgrade play in liberty18:55
inc0never occured to anyone to have it18:55
SamYapleyes.. yes it did18:55
SamYaplewe were always going to be solving z upgrades with the rest18:56
SamYaplethis was always the plan18:56
inc0we can try to make 1.2.018:56
SamYapleno upgrades == abandoned branhc18:56
inc0with all the upgrades in it18:56
inc0but not 1.1.018:56
SamYaplewell good luck finding anyone to use it18:56
SamYapleill tell every operator i know not to18:56
SamYaplethats a horrible thing yo upurpose18:57
inc0and then 1.2.0 will really be 2.0.0 with liberty branch in containers18:57
SamYaplewow these are some bad ideas flying around18:57
inc0why? if we add upgrade code in 1.2.0 we will still be able to upgrade 1.1.0 -> 1.2.018:57
SamYapleplanning two y releases, really?18:57
inc0(minus this rabbitmq, but what the hell)18:57
sdakeSamYaple first off, what you propose, telling operators not to use a branch of kolla is harmful to the project18:58
sdakeso I really hope you dont do that18:58
inc0we agreed that 1.1.0 is liberty+named volumes+thin containers18:58
sdakesecond off, may I remind everyone involved in this conversation18:58
sdakeANYONE CAN TRIGGER A FEATURE BACKPORT REQUEST BY ASKING THE PTL18:58
SamYaplesdake: i will not tell operators to use a project that cannot do security upgrades. im stating that right now18:58
SamYapleit should be harmful to the project if thats the type of decsion we make18:59
sdaketellign and not  telling are two different things18:59
SamYaplei will tell operators they cannot do updates, even security and warn them away from it18:59
SamYaplethis project is about operators and you are screwing them with no updates18:59
sdake: i will not tell operators18:59
SamYaplei will tell them19:00
SamYapleyou can not19:00
sdakeback and forth19:00
sdakemake up your mind19:00
sdakelisten cats, if you want a vote, ask forone19:00
SamYapleits not back and forth sdake19:00
sdakeif you want  to argue all day dont ask19:00
sdakeup to you19:00
SamYapleits reading comprehenssion19:00
SamYaplei will not tell operators to use a project without securit updates19:00
SamYaplei will tell them to not use it19:00
inc0Sam...1.2.019:00
inc0this is least I can agree to19:01
SamYapleinc0: im not plannign two y releases thats madness19:01
inc0and it will handle security upgrades from 1.1.019:01
SamYaplethe work needed for backporting and pulling in upgrades is about the same amount of wrok19:01
inc0but we don't have 1.1.0 and I have people waiting for it19:01
inc0and I won't vouch for this code if we just copy it19:02
SamYapleweve had people waiting for it since dec inc019:02
inc0lots have happend in upgrades and it might not be so simple to backport it19:02
SamYapleit will be identical, minus the neturon which needs small tweak19:02
SamYaplei know because ive already looked into it because i was backporting it for 1.1.019:03
inc0SamYaple, 1.2.0 is least I can agree to, and 1.2.0 will effectively be almost mitaka19:03
SamYapleno19:03
SamYapleim not plannign two y releases19:03
inc0splitting of bootstraps19:03
SamYaplethats ridiculous19:03
inc0is one thing that needs to be changed19:03
sdaketwo y releases is off the table imo19:03
SamYapleagreed sdake19:03
sdakei want to get out one y release that operators can move to mitaka with19:03
inc0sdake, then call a vote for upgrades in 1.1.019:03
sdakeinc0 will do you got it ;)19:04
SamYapleim not doign any work on 1.1.0 at all unless you can also include security updates to libery branch19:04
inc0we did agree that 1.1.0 is backporting of named volumes and thin containers and nothing else19:04
sdakeinc0 yes we did agree that19:04
inc0SamYaple, you did assign yourself to backporting of named volumes and thin containers19:04
sdakenow you see why i didn't want  to backport thin containers :)19:04
sdakeexceptions suck19:04
SamYapleand i refuse if there are no seucirty updates.19:05
sdakethey lead to the dark side19:05
SamYaplei was under the impression upgraes were going to19:05
SamYaplethat was the whole idea19:05
inc0SamYaple, that was never brought up19:05
sdakeagree never discussed19:05
SamYaplewe needed upgrades for liberty branch for long before oyu were around inc019:05
SamYapleive been waiting for liberty upgrades19:05
sdakerequires a vote19:05
SamYaplei though that was the whole hting here19:05
inc0SamYaple, I don't know what you discussed before I was around19:05
SamYapleexactly19:05
SamYapleit was the whole community though19:06
inc0but since I'm around...and all the time I was around, that was never discussed19:06
sdakewe pushed upgrades to mitaka because we were out of capacity for liberty19:06
SamYaplenow you are saying no upgrades on the liberty branch at all19:06
sdakethat is what was discussed19:06
sdakethat and only that19:06
SamYaplethats not true19:06
SamYaplewe had z upgrades planned for liberty19:06
SamYaplethats why we backported things19:06
SamYapleno reason to backport if no upgrades were going to happen ;)19:06
openstackgerritVikram Hosakote proposed openstack/kolla: Reconfigure ceph service  https://review.openstack.org/28825219:06
inc0SamYaple, no, we backport things because rerun of deployment will destroy data container19:06
inc0kolla-ansible deploy19:07
SamYaplewe backported things long before we agreed on a 1.1.019:07
inc0and upgrade from 1.0.0 to 2.0.0 will destroy data19:07
sdakeyes that is the reason we agreed to backport, because of data loss19:07
SamYaplebut we _always_ needed an upgrade path from 1.0.0 to the next version19:07
SamYapleway to rewrite history guys19:07
inc01.0.0 => 1.0.1 was never metioned19:07
sdakewe defined the next version originally as mitaka19:07
SamYapleinc0: it absolutly was19:07
inc0or 1.1.0 -> 1.1.1 in that regard19:07
SamYapleinc0: many times19:07
SamYapleok thats fine19:08
SamYaple1.1.0 -> 1.1.1 my not have been19:08
sdakewe talked about 1.0 to 1.1 and everyone was "MEH"19:08
SamYaplebut what was the plan for 1.0.0 -> 1.0.1? we were always goign to have upgrades19:08
SamYaplesdake: no they werent at all19:08
SamYaplesdake: we were pushing toward it and you disappeared for a month19:08
SamYaplethen intrested died because people left the community19:08
sdakenothing i can do aabout health problems SamYaple19:08
SamYapleor stopped being active19:08
inc0who left community?19:08
SamYaplekfox1111: you remember wanting working stuff ofr liberty in Dec?19:09
inc0and guys STOP doing it ad-personam19:09
sdakeya who precisely left?19:09
inc0it's not helping19:09
inc0to anyone19:09
inc0worse of community or project19:09
inc0SamYaple, sdake let's call a vote for backporting upgrades to 1.1.019:09
inc0I might even vote +1 but I'd rather have it 1.2.019:10
inc0to have 1.1.0 earlier19:10
SamYaplelook ive got other stuff to do guys. i refuse any work on 1.1.0 unless we can at least promise seucrity updates of z release of it. otherwise liberty is unstable19:10
inc0and then 1.2.0 a month or so from now19:10
SamYaplecalling it anything but unstable is a lie19:10
sdakelets change topics for a moment19:10
sdakei need some information19:10
vhosakotsdake: SamYaple: new patch set for ceph reconfig uploaded for review  -  https://review.openstack.org/28825219:10
SamYaplethanks vhosakot19:10
g3ekhello, what's the default value of thread while building containers whith kolla-build?19:10
sdakeWHEN would 1.1.0 with updates be completed?19:10
sdakeg3ek 819:11
SamYaplesdake: same time frame19:11
sdakewhich is what?19:11
g3eksdake: thx19:11
sdakeI set arbitrary date of 15th in the tracker19:11
SamYaplesdake: i always gave estimates with upgrades in mind19:11
sdakeapril 15th19:11
sdakei never heard the estimate mind repeating19:11
SamYaplesdake: end of mar was what was discussed19:11
SamYapleapril 15th is even better19:11
SamYaplework isnt going to take long, reviews might take a while19:11
sdakethanks i'll include that info in the vote19:11
sdakecommunity can decide19:12
sdakeboth of you can chime in if i got anything wrong19:12
sdakei'll include irc link discussion19:12
sdakesound good?19:12
inc0ok19:12
SamYaplesure, but that doesnt solve the rabbit issue19:12
SamYaplewe devolved19:12
sdakeno we escalated19:13
inc0I'll go back to rabbit issue next week19:13
inc0I'm out of juice today.19:13
SamYapleinc0: i get that :/19:13
SamYaplepff19:13
inc0we can discuss that19:13
inc0then19:13
g3ekIs there a place where I can see the roadmap that kolla team try to aim for the release of mitaka?19:13
SamYaplethats fine with me19:13
sdakeif upgrades are in, then the file can be backported19:13
sdakeif upgrades are out, then the issue i still open19:13
sdakeg3ek launchpad.net/kolla19:14
SamYapleif upgrades are out im not doing any 1.1.0 work, just fyi19:14
SamYapleothers can do it, i wont19:14
sdakeclick mitaka-319:14
inc0SamYaple, then commit whatever you have19:14
inc0please19:14
inc0so someone else can pick it up19:14
SamYapleinc0: feel free to review whatevers up there19:14
SamYapleim not going to go abandoning work out of spite19:15
sdakeinc0 I think all of SamYaple's work is in the queue already19:15
SamYaplethats not very community like19:15
openstackgerritThiago Gomes proposed openstack/kolla: Fix rst syntax in the list of images provided by Kolla  https://review.openstack.org/28866719:22
*** fthiagogv has quit IRC19:22
sbezverkwho was taken care of memcached reconfiugre?19:28
SamYaplesbezverk: its in the queue somewhere19:29
sbezverkSamYaple I am getting an error when I run reconfig19:31
sbezverkwanted to make sure it was tested in all in one scenario19:31
SamYaplesbezverk: im not sure how much testing has been done19:31
openstackgerritThiago Gomes proposed openstack/kolla: Fix rst syntax in the list of images provided by Kolla  https://review.openstack.org/28866719:32
sbezverkSamYaple ok will try to figure out19:32
*** sdake has quit IRC19:36
*** sdake has joined #kolla19:37
sbezverkSamYaple, one more thing, could you show how to do conditional for swift-rsyncd, since it is not running in all-in-one, and it has to be addressed..19:38
SamYaplesbezverk: use the when conditional that matches the other tasks19:40
SamYaplenot the negative, launch it on all nodes that needed it19:40
sbezverkSamYaple I am a bit slow, please bare with me..  first condition we evaluate in reconfigure is if containers are running, do you want me remove the check for swift-rsyncd from there?19:42
sbezverkor you suggest to launch swift-rsyncd in all scenarios?19:43
SamYaplewhich review is it sbezverk ?19:45
sbezverkSamYaple it is too late for review it has been merged19:46
sbezverkhttps://review.openstack.org/#/c/288059/19:46
openstackgerritSaisrikiran Mudigonda proposed openstack/kolla: Nit: Occurances of OpenStack  https://review.openstack.org/28870919:52
mlimaTASK: [ceph | Looking up hostname] failed: [localhost] => {"changed": false, "cmd": ["getent", "hosts", "10.102.22.89"], "delta": "0:00:00.003176", "end": "2016-02-24 11:17:29.999575", "rc": 2, "start": "2016-02-24 11:17:29.996399", "stdout_lines": [], "warnings": []}19:57
mlimaWhat did I do wrong?19:58
*** elmiko is now known as _elmiko20:00
*** mlima has quit IRC20:02
SamYaplemdnadeem: your hostname is not defined20:03
SamYaplemlima*20:03
SamYaplesorry mdnadeem wrong name20:03
openstackgerritThiago Gomes proposed openstack/kolla: Deleting only images from kolla build  https://review.openstack.org/28682320:04
*** Marga_ has joined #kolla20:10
g3ekhello20:18
*** akwasnie has joined #kolla20:18
g3ekis it normal : TASK: [common | Starting heka container] **************************************20:18
g3ekfailed: [localhost] => {"failed": true}20:18
g3ekmsg: The requested image does not exist: kollaglue/ubuntu-source-heka:2.0.020:18
g3ekroot@kolla-operator:~/kolla/ansible# docker images | grep kollaglue/ubuntu-source-heka20:18
g3ek192.168.2.4:4000/kollaglue/ubuntu-source-heka                        2.0.0               d3b68b4e24c8        About an hour ago   282.4 MB20:18
*** ravla has left #kolla20:19
SamYapleg3ek: you forgot to set your registry in globals.yml20:20
SamYaplethe image kollaglue/ubuntu-source-heka:2.0.0 does not exist, 192.168.2.4:4000/kollaglue/ubuntu-source-heka does20:20
*** akwasnie has quit IRC20:22
g3ekdamn, my bad. I forget to sed the registry in my auto install ^^20:25
*** inc0 has quit IRC20:27
ccesariokolla-build does support --type binary when --base ubuntu !?20:30
*** salv-orlando has quit IRC20:33
*** salv-orlando has joined #kolla20:33
kproskurinccesario, not yet, but we working on it20:33
ccesariokproskurin, thanks :)20:33
SamYaplekproskurin: re rabbitmq20:35
SamYapleits multinode that breaks fyi20:35
kproskurinSamYaple, yeah, I assume it multinode problem, since AIO works ok20:36
kproskurinIm currently creating multinode env and will look into it, but, looks like  in kolla-mesos we switched to ips as a anme of nodes in RBMQ20:37
SamYaplei think it works fine with ips, but that wasnt part of the bug20:37
SamYaplethe bug was '-' in hostname20:38
SamYaplewhich oddly endough ubunt udoesnt have a problem with20:38
kproskurinyes, just saying20:38
kproskurinmaybe different versin of erlang?20:38
*** sdake has quit IRC20:39
*** cfarquhar has joined #kolla20:40
*** cfarquhar has quit IRC20:40
*** cfarquhar has joined #kolla20:40
sbezverkSamYaple sdake please check this paste: http://paste.openstack.org/show/489380/ I think it should address swift-rsyncd solution. appreciate your thoughts20:41
SamYaplekproskurin: but how20:43
SamYaplekproskurin: we build the smae images no?20:43
kproskurinSamYaple, I mean between Centos and Ubuntu20:43
kproskurinAs far as I understand one works and one dont20:44
SamYaplekproskurin: do you have this bug on ubuntu or were you just bug squashing?20:44
kproskurinSamYaple, nah, I just found this error while deploying AIO, add quotes, works for me, though it is TrivialFix and post CR20:45
SamYaplesbezverk: no you need '    - { name: swift_rsyncd, group: swift-account-server }'20:45
SamYapleone for account-server,container-server,object-server20:45
SamYaplekproskurin: i could never replicate it but it absoluely broke my ubuntu stuff20:46
kproskurinSamYaple, Ill recheck it on multinode and update CR or post it here20:47
*** inc0 has joined #kolla20:50
sbezverkSamYaple, so you are ok if container up check will be split in two, I just need to change the line you mentioned, right?20:53
*** thumpba has quit IRC20:54
*** thumpba has joined #kolla20:55
SamYaplesbezverk: no reason to split the check up20:56
SamYaplealso that might break the check so dont break up the check20:56
sbezverkSamYaple, then could you exaplain how to do two checks in the same statement?20:58
SamYapleyou dont need two checks20:59
*** thumpba has quit IRC21:00
sbezverkok If it is all in one then host name will automaticaly be in swift-proxy-server group and we do not need to check for swift_rsycnd container, otherwise we need to check..21:00
sbezverkI am confused how to do "otherwise"21:01
SamYaplewe always need to check, there is no otherwise21:02
SamYaplewe dont do special things for AIO21:02
sbezverkbut in this case if we always check and container is NOT running because it does not get launched then it will ALWAYS fail?!?!?21:02
sbezverkin start.yml swift-rsyncd started only if hostname is not member of swift-proxy-server group21:03
sbezverklike in my setup swift-rsyncd is not running at all!!21:04
sbezverkthat is why I still think second check is needed21:04
SamYaplesbezverk: i keep saying THIS is what you need to change "in start.yml swift-rsyncd started only if hostname is not member of swift-proxy-server group"21:05
SamYaplethats the issue here21:05
sbezverkunless of course if rsyncd will always be running regardless of deployment model21:05
sbezverkok sorry21:05
*** sdake has joined #kolla21:05
sbezverkso you want rsyncd always running, now I got it, I told you I am slow.21:05
SamYapleno the when conditional hsould start rsyncd on container,account, and object server21:07
SamYapleit should not be a "not proxy-server" statement21:08
*** JoseMello has quit IRC21:08
openstackgerritLingfeng Xiong proposed openstack/kolla: fix mariadb failed to bootstrap  https://review.openstack.org/28156121:09
vhosakotsbezverk: "inventory_hostname not in groups['swift-proxy-server']"   means that the swift_rsyncd is started on any host that is _not_ in the "swift-proxy-server" group meaning it will be started for hosts in the other swift groups (like swift-account-server, swift-account-server and swift-object-server)21:14
SamYaplevhosakot: the problem with that is if proxy server is on the same host as account server for instance21:16
sbezverkvhosakot not in swift-proxy-group in all in one deployment automatically exclude all in one host from rsyncd container. I am changing start.yml to start on a positive condition, when host is a memober of account/container/object21:17
vhosakotSamYaple: cool... I agree about your statement....    it should not be a "not proxy-server" statement21:17
sbezverk  when: inventory_hostname in groups['swift-account-server'] or21:18
sbezverk        inventory_hostname in groups['swift-container-server'] or21:18
sbezverk        inventory_hostname in groups['swift-object-server']21:18
*** shardy has quit IRC21:18
sbezverknow it should be ok21:18
SamYaplesbezverk: rearange the or statements to match the other when statements, but yea thats it21:19
*** jtriley_ has quit IRC21:20
sbezverkok I copied it from start.yml just another container. everywhere I see it goes in this order, which order do you prefer?21:20
vhosakotsbezverk: SamYaple yes, i agree21:21
*** jtriley has joined #kolla21:21
*** dmsimard has quit IRC21:23
*** kproskurin has quit IRC21:35
*** kproskurin has joined #kolla21:36
*** inc0 has quit IRC21:36
SamYaplesbezverk: dont worry about it itll get worked out later21:37
SamYaplejust get the functionality right21:37
*** salv-orlando has quit IRC21:47
*** salv-orlando has joined #kolla21:48
*** dwalsh has quit IRC21:49
*** salv-orlando has quit IRC21:51
*** salv-orlando has joined #kolla21:52
*** gfidente has joined #kolla21:55
*** skape has joined #kolla21:56
skapehi!!21:59
skapewhen I run ./tools/kolla-ansible  prechecks   it fails at TASK [prechecks : Checking Ansible version]     my version is ansible 2.0.1.0   as I understand is the latest22:01
SamYapleskape: for mitaka you can only use 1.9.x branch, please look at the quickstart docs online22:01
skapeok Can I downgrade it with pip ?22:03
SamYapleyou can skape22:05
sbezverkSamYaple now it working with swift_rsyncd properly, should I send review or wait for later?22:07
SamYapleif oyu have something to fix, send a review22:07
skapeSamYaple : how? I've archieved it by removing and installing with yum, but not with pip22:10
SamYapleskape: im not sure what you are asking. with pip you can `pip install 'ansible<2'`22:11
skapetha's it, it was just curiosity because I got the correct version  using yum, and the precheck finished fine and the deploy is running in the moment22:14
*** jtriley has quit IRC22:20
sdakehey look another lets correctly capitalize openstack in register.yml patch22:25
sdakeSamYaple we should just leave that in the queue so peopel stop doing that nonsense22:25
sdakeskape th erpm version (on centos) is fine, wfm all the time22:26
sdakeon ubuntu i think its old22:26
vhosakotsdake: I saw the register.yml patch.. I agree arguments passed to kolla_keystone_service in register.yml should not be touched22:27
skapethx it's already working22:27
sdakerhallisey if your around can you look at https://review.openstack.org/#/c/288551/422:29
openstackgerritMerged openstack/kolla: Reconfigure ceph service  https://review.openstack.org/28825222:35
vhosakotcool, ceph reconfig is merged! :)22:36
*** rhallisey has quit IRC22:42
sdakejpeeler if your around can you look at https://review.openstack.org/#/c/288551/422:42
sdakethis fixes a gating problem and is an easy review22:43
jpeelerit's already approved22:43
sdakeoh sweet  thanks :)22:43
jpeelerha np22:43
openstackgerritMerged openstack/kolla: Retry token retrival in keystone bootstrap script  https://review.openstack.org/28855122:51
*** haplo37 has quit IRC22:57
*** skape has quit IRC22:59
*** mbound has joined #kolla23:13
*** blahRus has quit IRC23:13
openstackgerritMerged openstack/kolla: Fix rst syntax in the list of images provided by Kolla  https://review.openstack.org/28866723:14
SamYaplesdake: well be able to update service names in a little bit23:16
SamYaplesdake: v3 supports updates, and i have a shade and ansible patch to allow this23:16
SamYaplesdake: https://review.openstack.org/#/c/288225/ and https://review.openstack.org/#/c/288553/23:17
openstackgerritMerged openstack/kolla: Fix horizon with ssl  https://review.openstack.org/28822523:17
*** sdake has quit IRC23:25
*** vhosakot has quit IRC23:27
*** spisarski has quit IRC23:31
*** Jeffrey4l has joined #kolla23:43
*** kproskurin has quit IRC23:48
*** jtriley has joined #kolla23:48
*** daneyon has joined #kolla23:49
*** daneyon_ has joined #kolla23:52
*** salv-orl_ has joined #kolla23:52
*** daneyon has quit IRC23:55
*** salv-orlando has quit IRC23:55
*** akwasnie has joined #kolla23:56
*** mbound has quit IRC23:59
*** akwasnie has quit IRC23:59
« Thursday, 2016-03-03 Index Saturday, 2016-03-05 »

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!