Wednesday, 2017-02-08

*** unicell has joined #openstack-kolla00:02
*** unicell has quit IRC00:06
*** unicell has joined #openstack-kolla00:07
klindgren_sdake_, ping?00:10
*** klindgren_ is now known as klindgren00:10
*** sdake_ is now known as sdake00:10
*** huzhengchuan has joined #openstack-kolla00:10
sdakewound me00:10
klindgrenOn the kolla containers, why not make it so where every container is started the exact same way?00:10
sdakeklindgren could you provide an example where that is not hte case00:11
klindgrenSo you have to pass in the command to run in the config.json00:11
klindgrento start the container00:11
*** salv-orlando has quit IRC00:11
klindgrenIE keystone is:  "command": "/usr/sbin/{{ keystone_cmd }}",00:11
sdakeright - so that json is the API - each container started in a standard way?00:12
klindgrenwhy not encapsulate the starting shenanigans in the container and just have it be kolla-start? or something similar?00:12
klindgrenthat way we dont have to know whats running in the container to start it, everything gets called by the same "kolla_container_start_command" which takes care of the real command needed.00:13
sdakeklindgren sounds like a good topic for tomorrow's meeting at 16:00 UTC00:13
harlowjasdake niceee move there ;)00:13
harlowja:)00:13
sdakeharlowja how do you mean?00:14
sdakeharlowja i can't unilaterally make that decision for the rest of the community00:14
harlowjano i know00:14
harlowjanm, i was sorta joking, lol00:14
harlowjajoke fail00:15
harlowjalol00:15
* harlowja runs away00:15
klindgrensorry - I was asking you because I assumed that some discussion had happened around that.00:15
klindgrenbut I coudn't find any in my brief searching00:16
kfox1111klindgren: sometimes you may need to override the settings passed to the command.00:19
klindgrenkfox1111, can you give an example where you would want a container to start with different options thats those were done at container build time?00:20
klindgrenI assume this could also be where if command exists in config do that, otherwise do y by default.00:21
kfox1111klindgren: the tgtd container for example has a listen address in its commandline.00:21
*** yingjun has joined #openstack-kolla00:22
kfox1111its configured in /etc/kolla/global.yaml and generated into tgtd's config.json.00:22
kfox1111so, its really a config thing.00:22
kfox1111yeah, the config.json stuff could be done as a default if not passed. but wasn't done that way. not sure why.00:23
*** saneax-_-|AFK is now known as saneax00:24
klindgrenalso - one more question. How does one handle starting two different containers say glance-api and glance-registry.  Using the same config directory and different command-varriables in config.json?00:25
klindgrenin a manner that wont die in a fire when parrallelism is applied.00:26
*** Pavo has joined #openstack-kolla00:27
klindgrenOr nova-scheduler, nova-conductor?00:27
sdakekfox1111 the reason it wasn't donew that way was to provide flexibility and reduce duplication00:27
sdakeklindgren i understand your question - in k8s we use an emptyDir construct which is a per pod temp dir00:28
sdakeklindgren in ansible on bare metal there is no such thing available00:28
sdakeklindgren i think the emptydir construct could be simulated fairly easily however00:29
klindgrenI mean I could generate a set of configs per service, instead of having everything pointing at /etc/glance/00:29
klindgrenbut that seams.... rough.00:29
sdakeklindgren right bare metal comes with a different set of challenges especially for large scale operator environments that you have00:30
sdakeklindgren if Jeffrey4l is around perhaps he has a clever idea for ya :)00:30
harlowjasdake u are using k8s now to :)00:31
* harlowja didn't know u were on that train00:31
klindgrenI assume that if I pass the command option in via docker - that is going to cause the the container to not run the kolla-start stuff?00:32
klindgrenon startup00:32
sdakeharlowja my goal is to get kolla-kubernetes into the same level of completeness as kolla-ansible00:34
harlowjacool00:34
sdakeharlowja gotta dream big ;-)00:34
sdakeharlowja YOLO and al lthat00:34
harlowjaya, fair00:34
sdakeharlowja we have a solid team to deliver all of our deliverables and inc0 is on board with leading us to victory :)00:35
harlowjalol00:35
harlowjanice00:35
*** adrian_otto has joined #openstack-kolla00:38
kfox1111klindgren: correct. when I override command I end up calling kolla_start at the end.00:38
openstackgerritBorne Mace proposed openstack/kolla-kubernetes master: WIP: Added support for containerized dev bringup  https://review.openstack.org/42929500:39
klindgrenkfox1111, how do you do that?00:40
*** david-lyle has quit IRC00:42
bmace_sdake: almost there.  made a bunch of changes to ceph_workflow_service but there are some odd bits in there, like i think the keystone stuff is getting the wrong admin address for remote calls and i dropped out of my container by mistake so need to test out some of the other wait areas.  will hopefully finish it tomorrow and put a doc page together w/ the review and take out of WIP mode.00:43
bmace_sdake: i'll let you know when it is ready to try end to end.00:44
*** tovin07 has joined #openstack-kolla00:49
openstackgerritDuong Ha-Quang proposed openstack/kolla-ansible master: Change Barbican default secret store to pkcs11  https://review.openstack.org/43013700:50
*** zhurong has joined #openstack-kolla00:51
sdakebmace_ its not the wrong admin address - the admin address is not exposed00:54
sdakebmace_ sounds good duede ;)00:54
bmace_the keystone admin address is exposed in the default setup.. if it isn't exposed how else is a remote host (the dev container) supposed to talk to it?00:54
sdakebmace_ have no idea - its not exposed in the defualt setup - i can gaurantee you00:54
bmace_sdake: well, i didn't change that setting, just ran it, and kubectl get services -a --namespace kolla shows the exposed ip / port, and i was able to hit it.00:55
sdakebmace_ the problem with that setup is that the cloud cannot be bootstrapped (e.g. images loaded, networks created, flavors created, etc)00:55
sdakeit shows a port of 35357?00:55
bmace_yup00:55
sdakewierd, I had to change it in my setup to enable the port00:55
sdakemaybe someone changed the default after my compalining :)00:56
*** sharon_so has left #openstack-kolla00:56
sdakebtw as a shortcut you can to -n kolla00:56
sdaketo cut down on typing ;)00:56
sdaketo/do00:56
*** tonanhngo has quit IRC00:57
bmace_sdake: i do need to set up some aliases too, i think mark does kk to kubectl -n kolla so you can just like kk get services, etc..00:57
SamYapleklindgren: thats actually the opposite approach to the direction things should be going (RE: bake in the start command)00:59
SamYapleklindgren: the startup scripts and things are best moved out of the containers and into the deployment tools now. each deployment tool will need those start commands to be slightly different00:59
*** cuongnv has joined #openstack-kolla01:00
SamYapleadditionally, it makes testing much more difficult if you need to change any command, you have to post a review in kolla, then make sure kolla-ansible and kolla-kubernetes (and kolla-*) work still. that doesnt account for any other project that might use the kolla images (official or not)01:00
klindgrenSamYaple, can you explain the philosophy behind that statement?  It seems like the containers should have the ability to stand on their own?01:01
SamYapleklindgren: as a statement about containers themselves, i tend to agree. that said, the "stand on thier own" part in _most_ contaienrs requires ENV vars to function properly01:02
SamYaplesomething kolla doesnt do (for a long list of reasons)01:02
klindgrenif I am deploying keystone - which now has to be fronted by something running wsgi instead of event let.  Should I have to know that the command to make that container work is /usr/sbin/httpd assuming the container was built via centos?01:02
SamYapleklindgren: the deployment tool needs to know this. yea01:03
*** zhubingbing_ has joined #openstack-kolla01:03
klindgrenbut both the deployment tool and the container creation tool need to know this, seems like its duplicated information?01:03
SamYapleso the part you are leaving out is config.01:03
SamYaplehow do you get the config in there?01:03
klindgrenvolume mount?01:03
SamYapleok. what reads that?01:03
SamYaplewhere do you mount it?01:03
SamYaplewhat format should it be in?01:04
*** david-lyle has joined #openstack-kolla01:04
SamYaplewe dont bind host directly to container, /etc/keystone/keystone.conf:/etc/keystone/keystone.conf isnt bound that way01:04
SamYaplethe entrypoint is going to be a bit of duplication anywhere. what ENV vars do you need to set? any other params?01:05
openstackgerritAnh Tran proposed openstack/kolla master: Typo fix: serivce => service  https://review.openstack.org/42952901:06
SamYapleits really a tricky thing. so the mariadb:10.1 image upstream docker has enough for a _basic_ database. setup a user, database, permissions, and thats about it01:06
SamYapleenough to hook into for the lifecycle of the container01:07
SamYaplepast that, it is recommened you override... the entrypoint script!01:07
klindgrenI am not exactly sure what you are getting at?  The service it self dictates where config files needs to be mounted.  For most openstack services they all take the exact same startup commands.01:07
SamYapleand kolla is "past that"01:07
SamYapleklindgren: what paramaters do you hard code for neutron?01:08
SamYapleml2_conf.ini or openvswitch.ini?01:08
SamYapleor brocade.ini?01:08
SamYaplewhich one are you hard coding?01:08
klindgrenmeh thats the easy part. plugin.ini symlinked to whatever ml2 plugin you want01:08
SamYaplewhat sets up that symlink?01:09
*** duonghq has joined #openstack-kolla01:09
duonghqmorning guys01:09
klindgrenyou do in the external config?01:09
klindgrenjust like how I did it before I had kolla or any containers01:09
SamYapleso you have to know about the internal workings of the container to put in the config, no?01:09
klindgrenYou are blowing this way out of proportion.  lets change this around a little bit.  I take a distro say centos and install rdo on it01:10
klindgrenI start the glacne stuff via systemctl start glance.01:10
klindgrenI dont haveto know all this extra stuff. the specific are in the init script01:10
klindgrenand each service has its own01:10
klindgreninstead a single common file that I needed to overide the "start command" in01:11
SamYapleyou do absolutely need ot know things. you have to configure glance01:11
klindgrenbefore I start the serivce01:11
klindgren*starting* glance is complelty different from configuring glance to work the way I want it to in my env.01:11
SamYapleits not in docker thoguh, because no configuring can be done before you _start_ the contaienr01:12
SamYaplecontainer starts and it must do some initial prep, including move config files into place and set proper permissions01:12
klindgrenI dont see what running something in docker vs's running it as a baremetal service actually changes anything?01:14
*** hfu has quit IRC01:14
klindgrenI mean the way things are setup right now its impossible for me to paralelize start multiple services of the same type at the same time, because the start command comes from config.json01:15
klindgrenwhere it seems like aside from a few services, the start command is fairly well known01:15
klindgrenthose services typically being system services where you are controlling the bind ip's/ports via command line options.01:16
kfox1111klindgren: we start kolla-kubernetes in parallel.01:16
SamYapleklindgren: can you give an example of what you are doing?01:16
kfox1111config.json comes from a configmap.01:16
klindgrenSamYaple, start on the same host at the same time glance-api and glance-registry01:16
*** Pavo has quit IRC01:17
SamYaplethe container creation and configuration has to occur. what does that?01:17
klindgrenor nova-api and nova-scheduler and nova-cells and nova-conductor01:17
kfox1111klindgren: just use a different config.json for each.01:17
kfox1111the files don't have to be the same filename.01:17
*** tonanhngo has joined #openstack-kolla01:17
klindgrenkfox1111, how on baremetal?01:17
SamYapleklindgren: what do you mean how?01:17
klindgrendoesn;t kolla expect config.json as the entry point?01:18
kfox1111how are you trying to start them. with docker run?01:18
klindgrenroughly via docker run - but jsut using the native ansible tooling01:18
kfox1111ah.01:18
klindgrenroughly via docker run - but jsut using the native ansible docker tooling01:18
klindgrendocker_container01:18
kfox1111so, normlally, the configs for a container are in /etc/kolla/<containername>/*01:18
klindgrenright but I am trying to retrofit kolla into an existing baremetal environment01:19
klindgrenwhere openstack had them under /etc/service/01:19
klindgren/etc/<service>/ rather01:19
SamYapleah i see klindgren. i think i may be the only person thats done a conversion like that01:19
kfox1111and its done with an eqiv of docker run -it .... -v /etc/kolla/<containername>:/var/lib/kolla/config01:20
kfox1111SamYaple: I've done the same.01:20
SamYapleso a big problem youll find there is permissions. but now that kolla has static uid/guid you can actually get the correct permissions going and bind from host to contaienr01:20
SamYapleand bypass the config.json config file copying01:20
kfox1111klindgren: the trick is to make each container get its own subdir, then copy the config's from /etc/<service>/ to /etc/kolla/containername/01:21
SamYapleyea thats the thing you dont have to do with a baremetal converstion unless you are trying to migrate _to_ kolla-ansible01:21
*** tonanhngo has quit IRC01:22
klindgrenyea - we are bascially building our own kolla_ansible but we actually want to make use of copy_once01:22
harlowja^ with plans to show it to people to sorta work it through the system ;)01:22
harlowjabecause its approach is slightly different01:22
SamYaplei submitted a talk about that this cycle01:23
harlowja(the upstream system == system)01:23
SamYapleanyway, im sure kfox1111 is willing to walk you though it01:23
klindgrenI was thinking about doing something like:   - /etc/keystone/kolla/keystone.kolla:/var/lib/koll/config_files/config.json01:23
klindgrenas a volume mount01:23
*** sdake has quit IRC01:24
klindgrenbut the other part of me said - that for those services where you didn't need to change startup options.  Having it default to the correct "command" for the container would be a pretty simple optimization01:24
*** crushil has quit IRC01:26
klindgrenseems like doing something like the above will work out what I need it to.  That why I don't have to make our current config management system output that same files multiples times.01:33
klindgrens/why/way01:33
kfox1111could always submit a patch to add the default config's to kolla containers and see if it would fly.01:46
kfox1111you loose backwards compatability though.01:46
kfox1111so probably won't be able to use containers that way before pike.01:46
klindgrenyea - I will plan on doing that.01:47
*** jemcevoy has quit IRC01:47
*** ipsecguy has quit IRC01:51
*** ipsecguy has joined #openstack-kolla01:51
*** MasterOfBugs has quit IRC01:57
*** adrian_otto has quit IRC02:04
sbezverkkfox1111: hey, when you have a chance please check this ps, it is still wip but I would like to know what you think. https://review.openstack.org/#/c/428150/5002:08
*** unicell has quit IRC02:09
*** caowei has joined #openstack-kolla02:09
*** saneax is now known as saneax-_-|AFK02:10
*** sayantani01 has joined #openstack-kolla02:11
*** sayantani01 has quit IRC02:16
*** sayantani01 has joined #openstack-kolla02:16
*** japestinho has joined #openstack-kolla02:17
*** Pavo has joined #openstack-kolla02:25
*** adrian_otto has joined #openstack-kolla02:27
openstackgerritBorne Mace proposed openstack/kolla-kubernetes master: WIP: Added support for containerized dev bringup  https://review.openstack.org/42929502:29
*** sacharya_ has joined #openstack-kolla02:34
*** sacharya has quit IRC02:36
*** fragatin_ has joined #openstack-kolla02:37
*** Pavo has quit IRC02:38
*** crushil has joined #openstack-kolla02:38
*** fragatina has quit IRC02:40
*** krtaylor has joined #openstack-kolla02:40
*** sdake has joined #openstack-kolla02:41
openstackgerritJeffrey Zhang proposed openstack/kolla-ansible master: Do not check security opt in some case in kolla_docker module  https://review.openstack.org/42945902:46
*** zhurong has quit IRC02:51
*** rwallner has joined #openstack-kolla02:54
*** rwallner has quit IRC02:58
*** sdake has quit IRC03:01
*** sdake has joined #openstack-kolla03:01
*** sudipto_ has joined #openstack-kolla03:03
*** sudipto has joined #openstack-kolla03:03
openstackgerritCao Xuan Hoang proposed openstack/kolla-ansible master: Remove support for py34  https://review.openstack.org/43052503:07
openstackgerritJeffrey Zhang proposed openstack/kolla-ansible master: [wip] test scenario2 in gate  https://review.openstack.org/41227903:08
zhubingbing_hello guys03:10
sdakesup zhenguo03:18
zhenguosdake: hi03:19
sdakesup zhenguo :)03:20
*** fragatin_ has quit IRC03:21
sdakealso hello zhubingbing_ :)03:21
zhubingbing_hi sdake03:21
zhubingbing_i have a question about kolla-k8s03:21
sdakeshoot03:21
*** sayantani01 has quit IRC03:21
zhubingbing_should  i  use kolla-k8S docker images tag 4.0.0 ?03:22
zhubingbing_i see set_gate.sh use 2.0.2  “mistake”03:22
*** sayantani01 has joined #openstack-kolla03:23
*** sayantani01 has quit IRC03:24
sdakesup duonghq03:24
*** fragatina has joined #openstack-kolla03:24
sdakeduonghq how ya been :) - bmace_ almost has a virtualized dev env that is ready to go that is reliable :)03:24
*** fragatina has quit IRC03:24
*** jrobinson is now known as jrobinson-afk03:25
*** fragatina has joined #openstack-kolla03:25
sdakezhubingbing_ I think for dev 2.0.2 is sufficient03:26
zhubingbing_sure03:26
zhubingbing_roger03:27
zhubingbing_;)03:27
sdakezhubingbing_ because of config.json - we have a backwards compatible api03:27
sdakezhubingbing_ if your concerned about the download speed because docker hub perofrmance sucks in china03:27
sdakezhubingbing_ you can use newer versions and that should be fine03:27
zhubingbing_roger03:28
zhubingbing_undestand03:28
sdakezhubingbing_ is that the issue - or am I tilting at windmilsl there?03:28
*** sayantani01 has joined #openstack-kolla03:31
zhubingbing_windmilsl03:31
zhubingbing_what 'a mean ?03:31
openstackgerritZeyu Zhu proposed openstack/kolla-ansible master: Using yaml.safe_load instead of yaml.load  https://review.openstack.org/43053503:32
*** zhurong has joined #openstack-kolla03:33
*** fragatin_ has joined #openstack-kolla03:33
*** cuongnv has quit IRC03:34
*** cuongnv has joined #openstack-kolla03:34
kfox1111sbezverk: that gate refactor looks really nice. :)03:36
*** cuongnv has quit IRC03:36
bmace_sdake: the current wip may work end to end other than a couple of manual steps.  after running setup_dev_env.sh ceph_workflow_service.sh needs to get called with ceph_multi centos <ip_addr> <internal_eth>.  after some tests tomorrow i'll include the workflow call into the setup_dev_env script and have that called at container runtime by default.03:36
*** cuongnv has joined #openstack-kolla03:36
*** fragatina has quit IRC03:37
*** fragatin_ has quit IRC03:37
kfox1111bmace_: have you looked at sbezverk's review?03:37
openstackgerrityuhui proposed openstack/kolla master: [Trivial]Add/Correct some annotation words  https://review.openstack.org/42944203:37
kfox1111he's been reworking the gate scripts to get them more modular/sharable. might be able to reuse some of it with the dev env too.03:38
bmace_kfox1111: not yet.  i'm sure once things are working there is a lot of de-duplication / merging of code that can be done.03:38
openstackgerritSerguei Bezverkhi proposed openstack/kolla-kubernetes master: WIP Adding compute-kit chart  https://review.openstack.org/42815003:38
openstackgerritCao Xuan Hoang proposed openstack/kolla master: Remove support for py34  https://review.openstack.org/43053903:39
kfox1111cool.03:39
sdakebmace_ ok - one last thing I forgot to comment on in your review that needs addressing - you need a reno release note :)03:39
sdakekfox1111 ya at this time I'd have 15 duplicated scripts of the same thing if it got us a working dev env :)03:39
sdakeeven 30 ;-)03:40
bmace_lol03:40
kfox1111:)03:42
bmace_well luckily I don't think it is quite that bad.03:43
*** sudipto has quit IRC03:43
*** sudipto_ has quit IRC03:43
bmace_sdake: and i don't know the structure / etc of the release note stuff.  feel free to toss that comment in the review and i'll look into it.03:44
sdakebmace_ reno new blurb03:45
sdakemodify the blurb file03:45
sdakeit is obvious what to do03:45
bmace_ok :)03:45
sp_portdirect: [Want to start with kolla-k8s bugs fixing] is it possible to allocate me a kolla-k8s dev env prepared? So that I can start.03:46
*** sudipto has joined #openstack-kolla03:46
*** sudipto_ has joined #openstack-kolla03:46
sdakebmace_ that reno new blurb needs to be proceeded with pip install reno ;-)03:47
duonghqsdake, sorry, I just come back from meeting03:50
*** sdake_ has joined #openstack-kolla03:54
*** zhuzeyu has joined #openstack-kolla03:55
*** sdake has quit IRC03:57
*** saneax-_-|AFK is now known as saneax03:58
*** jrobinson-afk is now known as jrobinson03:59
*** mdnadeem has joined #openstack-kolla04:00
openstackgerritjangseon ryu proposed openstack/kolla-ansible master: This feature is supporting for workers config in swift servers.  https://review.openstack.org/43055004:01
*** zhurong has quit IRC04:04
*** fragatina has joined #openstack-kolla04:09
openstackgerritSerguei Bezverkhi proposed openstack/kolla-kubernetes master: WIP Adding compute-kit chart  https://review.openstack.org/42815004:12
*** fragatina has quit IRC04:12
*** fragatina has joined #openstack-kolla04:12
*** fragatina has quit IRC04:13
*** fragatina has joined #openstack-kolla04:13
*** sudipto has quit IRC04:19
*** sudipto_ has quit IRC04:19
*** goldyfruit has joined #openstack-kolla04:22
*** goldyfruit has quit IRC04:33
*** huzhengchuan has quit IRC04:39
*** sacharya has joined #openstack-kolla04:43
*** sacharya_ has quit IRC04:43
openstackgerritSerguei Bezverkhi proposed openstack/kolla-kubernetes master: WIP Adding compute-kit chart  https://review.openstack.org/42815004:47
sdake_portdirect about?04:49
*** huzhengchuan has joined #openstack-kolla04:52
*** sacharya_ has joined #openstack-kolla04:56
*** sacharya has quit IRC04:57
*** crushil has quit IRC04:59
*** sudipto has joined #openstack-kolla05:01
*** sudipto_ has joined #openstack-kolla05:01
*** yingjun has quit IRC05:06
*** sudipto_ has quit IRC05:08
*** sudipto has quit IRC05:08
*** rwallner has joined #openstack-kolla05:10
*** rwallner has quit IRC05:14
*** unicell has joined #openstack-kolla05:16
*** sayantani01 has quit IRC05:19
*** unicell1 has joined #openstack-kolla05:23
*** unicell has quit IRC05:23
openstackgerritJeffrey Zhang proposed openstack/kolla master: Add openstack-magnum-ui package for centos binary  https://review.openstack.org/43056705:24
*** tonanhngo has joined #openstack-kolla05:25
openstackgerritJeffrey Zhang proposed openstack/kolla-ansible master: [wip] test scenario2 in gate  https://review.openstack.org/41227905:25
openstackgerritSteven Dake proposed openstack/kolla-kubernetes master: Check to see if Horizon is working in the gate.  https://review.openstack.org/42602505:28
sdake_kfox1111 are you about05:28
*** sayantani01 has joined #openstack-kolla05:29
*** skramaja has joined #openstack-kolla05:31
*** japestinho has quit IRC05:31
sdake_Jeffrey4l what are your thoughts on this: http://logs.openstack.org/98/426598/9/check/gate-kolla-kubernetes-deploy-ubuntu-binary-2-ceph-nv/077dfba/console.html05:34
*** huzhengchuan has quit IRC05:34
*** huzhengchuan has joined #openstack-kolla05:35
Jeffrey4lsdake_, try to use wget.  wget is more stable to download large file.05:35
sdake_Jeffrey4l thanks didn't know that at all05:36
sdake_Jeffrey4l it is actually kfox1111 's work05:36
Jeffrey4li see the same issue when implement this in kolla-ansible.05:36
openstackgerritJeffrey Zhang proposed openstack/kolla master: Use tini instead of dumb-init  https://review.openstack.org/43010805:37
*** zhangqiankun has joined #openstack-kolla05:42
*** sacharya_ has quit IRC05:43
*** salv-orlando has joined #openstack-kolla05:45
*** unicell1 has quit IRC05:47
*** japestinho has joined #openstack-kolla05:47
*** unicell has joined #openstack-kolla05:48
*** salv-orlando has quit IRC05:50
*** bmace_ has quit IRC06:01
*** yingjun has joined #openstack-kolla06:05
*** qiliang27 has quit IRC06:08
*** unicell1 has joined #openstack-kolla06:09
*** unicell has quit IRC06:09
*** adrian_otto has quit IRC06:13
openstackgerritJeffrey Zhang proposed openstack/kolla master: Fix trove dashboard issue in horizon container  https://review.openstack.org/42918306:17
*** zhurong has joined #openstack-kolla06:22
duonghqsdake_, still around?06:24
sdake_duonghq indeed06:25
duonghqI see that the kolla-ansible is refactored much, currently how does the order of roles is determined?06:25
sdake_duonghq not clear to me - Jeffrey4l or pbourke would know for sure06:26
Jeffrey4lduonghq, the order of roles is not changed.06:26
duonghqsdake_, thanks, Jeffrey4l I don't know but in my setup, the haproxy is deploy before mariadb06:26
duonghqso it failed at waiting for virtual ip06:26
*** sdake_ is now known as sdake06:27
Jeffrey4lhaproxy is deployed always before mariadb.06:27
Jeffrey4lthere may be some other issue.06:27
duonghqfatal: [ubuntu]: FAILED! => {"changed": false, "elapsed": 300, "failed": true, "msg": "Timeout when waiting for 192.168.122.250:3306"}06:27
duonghq06:27
Jeffrey4lhrm. let me check..06:28
duonghqdocker ps at deploy node http://paste.openstack.org/show/598040/06:28
Jeffrey4lduonghq, this task is in mariadb role https://github.com/openstack/kolla-ansible/blob/8b4ae8c2a714a6e7fcf065f4a624ba45d8d3e1ce/ansible/roles/mariadb/tasks/start.yml#L2106:30
duonghqJeffrey4l,  how about this one: https://github.com/openstack/kolla-ansible/blob/8b4ae8c2a714a6e7fcf065f4a624ba45d8d3e1ce/ansible/roles/haproxy/tasks/start.yml#L3206:32
Jeffrey4lduonghq, this just wait for the vip.06:32
Jeffrey4lnot mariadb port.06:32
Jeffrey4lhrm. wait.06:32
Jeffrey4lit should be fine.06:33
Jeffrey4lit connect the vip:database_port06:33
Jeffrey4lthis port is listened by haproxy.06:33
Jeffrey4lit works even though mariadb is not started.06:33
duonghqJeffrey4l, thank you06:33
Jeffrey4lnp06:34
*** huzhengchuan has quit IRC06:34
openstackgerritMerged openstack/kolla-ansible master: Using https instead of http  https://review.openstack.org/43012706:38
*** adrian_otto has joined #openstack-kolla06:43
*** zhurong has quit IRC06:43
*** jrobinson has quit IRC06:45
*** salv-orlando has joined #openstack-kolla06:46
*** huzhengchuan has joined #openstack-kolla06:48
*** zhurong has joined #openstack-kolla06:49
openstackgerritSteven Dake proposed openstack/kolla-kubernetes master: Pull container tallballs into the gate  https://review.openstack.org/42659806:49
sdakeJeffrey4l - I started with kfox1111 's implementation of container pulling - and since you have already sorted this out - i copied alot of your code06:50
Jeffrey4lthat's fine.06:50
sdakeJeffrey4l yes of course that is fine - we are one project ;)06:50
openstackgerritMerged openstack/kolla-ansible master: Add Freezer doc link in README.rst  https://review.openstack.org/42936706:50
sdakeeven copying between other openstack projects is fine06:50
Jeffrey4lnow i think there are lots of duplication code between kolla/kolla-ansible/kolla-k8s06:51
*** satyar has joined #openstack-kolla06:51
*** salv-orlando has quit IRC06:51
sdakei have a request though, could you review the last patchset i submitted there and versus the prior patch set06:51
sdakeand add any commentary you have that I may haev missed06:51
sdakeJeffrey4l fwiw I agree, we need to standardize our gating06:51
Jeffrey4li am thinking move some same code into kolla project. for example, setup docker service, etc06:51
sdakeJeffrey4l i dont mind if it is duplicated - i mind that the logic is completely different06:51
sdakeideally the logic should be very similar so cross-deliverable gating can be done eaisly06:52
Jeffrey4luse a common function script. like "function setup_docker_registry"06:52
sdakeas in one person that works heavily in one deliverable can work in another deliverable's gating code easily06:52
Jeffrey4lthis is optimal. ;)06:52
* Jeffrey4l is checking your code.06:52
sdakeya - we will have some tim to get there06:52
Jeffrey4lye.06:53
Jeffrey4lyep.06:53
sdakeyo uwill notice the images built for kolla-kubernetes gate are different then kolla-ansible06:53
sdake2.0.2, 3.0.2 and master are built for kolla-kubernetes06:53
sdakeonly master is built for kolla-ansible06:53
sdakeso i get the logic needs to be slightly different06:53
sdakethanks for checking Jeffrey4l :)06:53
Jeffrey4lmaybe.06:54
sdakethe fact that one gate was using curl for this operation and you already fixed that problem with wget seems like a problem06:54
Jeffrey4lat least, kolla and kolla-ansible share almost the same code.06:54
sdakewe should be able to learn that lesson once rather than twice06:54
Jeffrey4lyep.06:54
sdakeduonghq did you get sorted out?06:56
Jeffrey4lsdake, why use separate file here ? http://tarballs.openstack.org/kolla-kubernetes/gate/containers/06:58
Jeffrey4lit will be very large to download these files.06:58
duonghqsdake, yup06:59
sdakeJeffrey4l i dont follow your question - partially because i hae no idea how this stuff works :)06:59
duonghqseem that my custom haproxy still hang around, I clean all config file06:59
sdakemind expanding on the specific line # in question06:59
sdakeduonghq got it06:59
Jeffrey4li will.07:00
duonghqcurrently, I got other errors, I'm trying to build another image07:00
*** rwallner has joined #openstack-kolla07:00
openstackgerritSteven Dake proposed openstack/kolla-kubernetes master: Pull container tallballs into the gate  https://review.openstack.org/42659807:01
*** salv-orlando has joined #openstack-kolla07:02
duonghqWhat does the error "matched" mean (in image building result)07:03
Jeffrey4lduonghq, seems like its parent failed?07:04
sdakeduonghq is that an INFO output or an error output?07:04
sdakeJeffrey4l is kolla-ansible upgrade/reconfig/deploy functional at present?07:04
sdake(in master)07:04
duonghqsdake, Jeffrey4l the summary result, just cannot get idea of "matched" word07:05
Jeffrey4lsdake, upgrade has some issue ( nova ceph related )07:05
Jeffrey4ldeploy and reconfigure is fine.07:05
*** salv-orl_ has joined #openstack-kolla07:05
sdakeduonghq its a poor error name07:05
*** rwallner has quit IRC07:05
*** zhurong has quit IRC07:05
Jeffrey4lduonghq, yep. it is bad.07:05
sdakeduonghq rather poor error description07:05
sdakeerrors shoudln't require a phd to decipher07:05
duonghqcan you give me more descriptive meaning of this?07:06
jascott1evening folks07:06
sdakesup jascott107:06
duonghqevening jascott107:06
jascott1how are all the things?07:06
jascott1:)07:06
sdakejascott1 need m0ar reviews07:06
jascott1yes07:07
jascott1need more time period but def CRs07:07
jascott1i havent even got to play with compute kit07:07
jascott1been off in my golangary07:07
sdakejascott1 yar07:08
Jeffrey4lsdake, done i made some comments in patch 10.07:09
sdakethanks Jeffrey4l07:09
Jeffrey4lnp.07:09
*** salv-orlando has quit IRC07:09
Jeffrey4lsdake, btw, centos+source is more stable than others ;)07:10
openstackgerritJeffrey Zhang proposed openstack/kolla-ansible master: [wip] enable ceilometer related project in gate  https://review.openstack.org/41227807:17
openstackgerritSteven Dake proposed openstack/kolla-kubernetes master: Pull container tallballs into the gate  https://review.openstack.org/42659807:20
sdakeJeffrey4l yup not surprising all that much07:20
*** sdake_ has joined #openstack-kolla07:23
*** tonanhngo has quit IRC07:25
*** sdake has quit IRC07:26
*** pcaruana has joined #openstack-kolla07:29
sdake_Jeffrey4l is thi the problem with nova: http://logs.openstack.org/25/426025/24/check/gate-kolla-kubernetes-deploy-centos-binary-3-ceph-multi-nv/d9499c9/console.html#_2017-02-08_05_52_01_05998207:33
*** adrian_otto has quit IRC07:33
Jeffrey4lsdake_, seems like a bad configuration http://logs.openstack.org/25/426025/24/check/gate-kolla-kubernetes-deploy-centos-binary-3-ceph-multi-nv/d9499c9/logs/openstack/kolla-nova-compute-xz2xd.txt07:34
openstackgerritSteven Dake proposed openstack/kolla-kubernetes master: Pull container tallballs into the gate  https://review.openstack.org/42659807:35
Jeffrey4lsdake_, and this http://logs.openstack.org/25/426025/24/check/gate-kolla-kubernetes-deploy-centos-binary-3-ceph-multi-nv/d9499c9/logs/openstack/kolla-nova-conductor-0.txt07:35
Jeffrey4lseems like nova have a bug. but your nova-compute doesn't work at all.07:36
*** david-lyle_ has joined #openstack-kolla07:37
Jeffrey4lsdake_, what do you think about this https://review.openstack.org/430108 ? use tini instead of dumb-init07:38
sdake_Jeffrey4l interesting07:40
sdake_Jeffrey4l didn't knwo thta was a new feature of docker07:40
Jeffrey4lSamYaple told me yesterday.07:40
sdake_Jeffrey4l I'm not sure we want to pin on docker-engine > 1.13.107:40
sdake_docker-engine 1.10 is pretty much where i"d like to keep our min pin at07:40
Jeffrey4ltbh, docker should handle this.07:40
Jeffrey4lsdake_, maybe we should bump the docker-engine version. but it is too late to do it now.07:41
sdake_I think the issue is that it is possible docker-engine may be replaced by various efforts in the container ecosystem, and then we would be left locked into docker permanently07:41
sdake_or atleast for ocata07:41
Jeffrey4land docker 1.13.1 is not released.07:41
sdake_my take is lets punt until pike and see what happens07:41
Jeffrey4land docker 1.13.0 is buggy with my test.07:41
sdake_docker 1.13 is not high quality sadly07:42
sdake_IIUC google peeps are making a competitor to docker-engine07:42
Jeffrey4lsdake_, i am fine before that happen ( replace docker )07:42
Jeffrey4lyep.07:42
sdake_and it may make sense to support one or the other or both07:42
sdake_really depends on what happens there - i think its premature to lock us into a docker-only world07:43
Jeffrey4lat least, we have no choice, right?  if we have, we can make the possible change. ;)07:43
sdake_coreos has failed imo with their container replacement for docker, however, google may not as they have superior marketing and people-power07:44
*** dciabrin has quit IRC07:44
Jeffrey4lmaybe07:44
sdake_i think we will knwo by the end of pike one way or another07:44
Jeffrey4lpossible.07:45
sdake_if tini requires docker, that seems problematic in a world where there may be multiple container runtimes07:45
sdake_dumb_init does the job, does it not?07:45
Jeffrey4ltini is a stand along app.07:45
Jeffrey4li followed the docker repo.07:46
sdake_i dont follwo the terms stand along app07:46
Jeffrey4lat the beginning docker have its own init process app.07:46
sdake_i guess if we could make it conditoinal that would be ok07:46
Jeffrey4lbut later, they think tini is better.07:46
sdake_the thing is tini depends on docker right?07:46
Jeffrey4lso they just download and compile tini in docker's repo. and ship the tini with docker-engine package.07:47
Jeffrey4lin that package, tini is called "docker-init"07:47
sdake_whatever competitor the community produces to docker-engine I expect dockerfiles as well as the image format will stay comptible07:47
Jeffrey4lsdake_, no tini depends on nothing.07:47
sdake_so tini cn work with docker-engine 1.10?07:47
Jeffrey4lsdake_, yep.07:47
Jeffrey4lbut above docker 1.13. you can use "docker run --init centos:7 bash"07:48
*** matrohon has joined #openstack-kolla07:48
Jeffrey4ldocker will use docker-init( which is tini actuall) to launch bash.07:48
sdake_what happens with docker-engine 1.10?07:49
Jeffrey4lyou image do not have to install any init process in this case.07:49
sdake_are zombie pids handled with tini?07:49
Jeffrey4lyep.07:49
Jeffrey4ldocker engine 1.10 do not support this.07:49
sdake_so tini determines the docker runtime its in?07:49
Jeffrey4lin this patch, we only move from dumb-init to tini.07:49
Jeffrey4lwe do not  pin do docker 1.13's new feature.07:49
sdake_mind linking the github repo07:50
Jeffrey4lsdake_, https://github.com/krallin/tini07:50
*** salv-orlando has joined #openstack-kolla07:50
Jeffrey4ltini is almost the same with dumb-init07:50
Jeffrey4lwe move to it only because docker use it, too.07:50
Jeffrey4lso i think this should be best tested by docker, and will be widely used in the future.07:51
Jeffrey4lonce kolla require docker > 1.13. we can use docker's init feature, then we do not need install tini in dockerfile anymore.07:51
Jeffrey4lthat's future plan.07:52
Jeffrey4lbut i do not know when it will happen and whether it should happen. ;)07:52
Jeffrey4lthis is another topic, imo.07:52
*** salv-orl_ has quit IRC07:52
Jeffrey4lsdake_, ^^07:53
*** dciabrin has joined #openstack-kolla07:54
sdake_Jeffrey4l the code looks good - although this file looks troubling given the fact there was a typo in the license file: https://github.com/krallin/tini/blob/master/src/tiniLicense.h07:54
sdake_Jeffrey4l can you see if the preprocessed license file matches the license in the repo?07:55
sdake_also - license on dumb_init is what?07:55
sdake_I sort of feel like why fix something that isn't broken :)07:55
sdake_especially t-3 weeks prior to a major release07:55
Jeffrey4ldumb-init is The MIT License (MIT)07:55
Jeffrey4lshould be fine.07:55
Jeffrey4ltini is MIT too.07:56
Jeffrey4lhttps://github.com/krallin/tini/blob/master/LICENSE07:56
Jeffrey4lsdake_,07:56
sdake_Jeffrey4l yo usee the git commit message "Fix typo in license"07:56
sdake_you see the preprocessed license file07:56
sdake_we want to ensure those are equivalent07:57
Jeffrey4lsdake_, another one reason we move it because https://review.openstack.org/42990707:57
sdake_and that they are actually the MIT lilcense and not claiming to be so :)07:57
sdake_when someoen makes a typo in a license file, it tells me they customized it in some way07:57
sdake_that is highly dangerous07:57
Jeffrey4lok. let me decrypt it07:58
*** dciabrin has quit IRC07:58
sdake_sounds good07:58
*** dciabrin has joined #openstack-kolla07:58
sdake_and validate it against the OSI approved MIT license07:59
sdake_https://opensource.org/licenses/MIT07:59
*** n0isyn0i1e is now known as n0isyn0ise08:00
Jeffrey4lpreprocessed is the same with LICENSE08:02
*** tonanhngo has joined #openstack-kolla08:03
sdake_and LICENSE matches OSI license?08:03
Jeffrey4lyep.08:04
*** tonanhngo has quit IRC08:04
Jeffrey4lso i think it is fine to use at the license level. sdake_08:05
*** sdake_ is now known as dake08:05
dakeok08:05
dakethe second question is does it actually work :)08:06
dakelooking at that dumb_init 1.2.0 error, it looks like they somehow broke the implementation08:06
dake(the github issue tracker)08:06
dakei do like that arm is available in tini08:06
dakeprobably a topic for tomorrow's team meeting08:07
Jeffrey4ldake, yes it works the gate is fine after changing to tini.08:07
Jeffrey4lyep. i will add this into the meeting agenda.08:07
dakekolla-ansible gate is a leaky sieve :)08:07
Jeffrey4lno. i think it much better ;)08:07
dakeit is probably improving no doubt - i've been pretty much mia from gate work since i retired from PTLship :)08:08
Jeffrey4l;)08:09
*** salv-orlando has quit IRC08:10
Jeffrey4lhttps://review.openstack.org/412279  https://review.openstack.org/412278 you will like this dake08:10
dakeJeffrey4l ++++++08:11
Jeffrey4l;D08:12
*** sdake has joined #openstack-kolla08:14
*** qiliang27 has joined #openstack-kolla08:14
openstackgerritSteven Dake proposed openstack/kolla-kubernetes master: Pull container tallballs into the gate  https://review.openstack.org/42659808:17
*** dake has quit IRC08:17
openstackgerritJeffrey Zhang proposed openstack/kolla-ansible master: Fix copying over nfs_shares task failure  https://review.openstack.org/43063808:27
*** david-lyle_ has quit IRC08:28
openstackgerritSteven Dake proposed openstack/kolla-kubernetes master: Pull container tallballs into the gate  https://review.openstack.org/42659808:31
*** stevemar has quit IRC08:33
*** stevemar has joined #openstack-kolla08:35
*** zioproto has joined #openstack-kolla08:38
sdakethis tarballs in the gate thing looks pretty good08:38
*** huzhengchuan has quit IRC08:39
Jeffrey4ltarballs means?08:39
Jeffrey4ltarballs.openstack.org/kolla/images/ ?08:39
sdakethere is also one for kubernetes as well08:40
Jeffrey4lyep.08:40
sdakebut yup those08:40
Jeffrey4lthe only issue to push large file in tarballs.o.o is downloading them may be slow in some node.08:41
sdaketakes about 2 minutes in the gate08:43
sdakeversus 20 minutes to download from dockerhub08:43
*** prameswar has joined #openstack-kolla08:44
Jeffrey4lhrm. in some time. i saw this once. download 3GB file takes 1 hour ;(08:44
openstackgerritLi Yingjun proposed openstack/kolla-ansible master: Fix telegraf task  https://review.openstack.org/43064808:46
openstackgerritSteven Dake proposed openstack/kolla-kubernetes master: Pull container tallballs into the gate  https://review.openstack.org/42659808:50
*** huzhengchuan has joined #openstack-kolla08:52
openstackgerritSteven Dake proposed openstack/kolla-kubernetes master: Pull container tallballs into the gate  https://review.openstack.org/42659808:54
sdakeJeffrey4l in the gatenodes?08:56
Jeffrey4lyes.08:56
Jeffrey4lbut only once.08:56
Jeffrey4lnever mind.08:56
*** salv-orlando has joined #openstack-kolla08:57
sdakei can tell ya wget is alot better then curl08:57
sdakethanks for the tip there :)08:57
sdakeJeffrey4l do you know why?08:57
Jeffrey4lno ;(08:57
*** egonzalez has joined #openstack-kolla08:58
zioprotoHello all. Looks like it would be much better to start using master than the stable/newton version. At the moment our production (puppet) cloud is on Mitaka. It is possible to use newton kolla containers but kolla-ansible tooling and roles from (future) stable/ocata ? It would work as a migration path ?08:59
Jeffrey4lzioproto, i do not think so.08:59
*** dtturner has quit IRC08:59
zioprotolet's say I am running kolla in stable/newton. What is the upgrade procedure ?09:01
zioprotothere is a spec somewhere ?09:01
zioprotoI did not find too much yet09:01
zioprotoI dont understand what component makes the database migrations09:01
Jeffrey4lupgrade from newton to ocata?09:01
zioprotoyes09:01
zioprotoor mitaka to newton09:01
zioprotois the same09:01
Jeffrey4lthere is no special doc.09:01
zioprotobasically you have three things to do09:01
zioprotoupgrade ansible roles09:01
Jeffrey4lnormally run "kolla-ansible upgrade"09:02
zioprotopull new containers and restart them09:02
zioprotoand do the database migrations09:02
Jeffrey4loh yep.09:02
Jeffrey4l"kolla-ansible pull" then "kolla-ansible upgrade"09:02
Jeffrey4lupgrade will handle the database migrations.09:02
zioprotook, so upgrade would do the database migrations09:02
zioprotoso if I keep in my globals.yml the setting openstack_release: "3.0.2" but I have current master, everything would brake ?09:03
Jeffrey4lalmost. in some case, you need add some items into passwords.yml and your inventory file.09:04
Jeffrey4ltbh, kolla do not handle this very well.09:05
*** Serlex has joined #openstack-kolla09:16
egonzalezJeffrey4l: yep, the bug I opened yesterday is with docker 1.1309:20
*** mgoddard_ has joined #openstack-kolla09:21
egonzalezor maybe because i have devicemapper as docker storage driver09:21
openstackgerritSteven Dake proposed openstack/kolla-kubernetes master: Pull container tallballs into the gate  https://review.openstack.org/42659809:23
Jeffrey4legonzalez, i use btrfs, which has such issue too.09:24
Jeffrey4li think it is worth to install docker 1.13.1rc to have a try. ;)09:24
Jeffrey4lbrb09:24
sdakemorning egonzalez09:25
openstackgerritLi Yingjun proposed openstack/kolla-ansible master: Fix haproxy config for solumn-api  https://review.openstack.org/43067909:25
egonzalezmorning sdake09:26
*** duonghq has quit IRC09:26
openstackgerritSteven Dake proposed openstack/kolla-kubernetes master: Pull container tallballs into the gate  https://review.openstack.org/42659809:27
*** fragatina has quit IRC09:30
*** fragatina has joined #openstack-kolla09:30
*** ipsecguy has quit IRC09:32
*** openstackgerrit has quit IRC09:34
*** mannamne has joined #openstack-kolla09:35
*** hrw has quit IRC09:38
*** japestinho has quit IRC09:38
*** NachoDuck_ has joined #openstack-kolla09:38
*** gfidente has joined #openstack-kolla09:38
*** gfidente has quit IRC09:38
*** gfidente has joined #openstack-kolla09:38
*** Guest66676 has joined #openstack-kolla09:41
*** abc__ has joined #openstack-kolla09:41
*** mandre_ has joined #openstack-kolla09:44
*** hrw has joined #openstack-kolla09:44
*** dims_ has joined #openstack-kolla09:44
*** huzhengchuan has quit IRC09:45
*** yingjun has quit IRC09:45
*** dims has quit IRC09:45
*** cinerama has quit IRC09:45
*** dmsimard has quit IRC09:45
*** timss has quit IRC09:45
*** abc has quit IRC09:45
*** mandre has quit IRC09:45
*** Guest66666 has quit IRC09:45
*** NachoDuck has quit IRC09:45
*** mandre_ is now known as mandre09:45
*** NachoDuck_ is now known as NachoDuck09:49
*** huzhengchuan has joined #openstack-kolla09:51
*** dmsimard has joined #openstack-kolla09:51
*** timss has joined #openstack-kolla09:52
*** openstackgerrit has joined #openstack-kolla09:54
openstackgerritSteven Dake proposed openstack/kolla-kubernetes master: Pull container tallballs into the gate  https://review.openstack.org/42659809:54
openstackgerritSteven Dake proposed openstack/kolla-kubernetes master: Pull container tallballs into the gate  https://review.openstack.org/42659809:55
*** cuongnv has quit IRC10:02
*** japestinho has joined #openstack-kolla10:04
*** zhubingbing_ has quit IRC10:11
*** zhubingbing_ has joined #openstack-kolla10:11
pbourkeanyone having issues getting fluentd to start?10:19
pbourke'service endpoint with name fluentd already exists'10:20
satyarnot yet tried10:20
*** ipsecguy has joined #openstack-kolla10:26
openstackgerritSteven Dake proposed openstack/kolla-kubernetes master: Actually report gate failures for experimental jobs  https://review.openstack.org/43070810:26
*** satyar has quit IRC10:27
*** sambetts|afk is now known as sambetts10:33
openstackgerritWei Cao proposed openstack/kolla-ansible master: Remove Zaqar doc link in README.rst  https://review.openstack.org/43071410:36
*** zhubingbing_ has quit IRC10:42
openstackgerritEduardo Gonzalez proposed openstack/kolla-ansible master: Recover nova reload task to upgrade  https://review.openstack.org/43076210:47
*** salv-orlando has quit IRC10:49
openstackgerritWei Cao proposed openstack/kolla-ansible master: Correct the groups for solum in haproxy.cfg.j2  https://review.openstack.org/42040710:52
*** caowei has quit IRC10:54
openstackgerritSteven Dake proposed openstack/kolla-kubernetes master: Pull container tallballs into the gate  https://review.openstack.org/42659810:57
sdakeJeffrey4l would you mind taking a look at that last patch10:57
portdirecto/10:57
sdakesup portdirect10:58
portdirecthey sdake10:58
sdakeportdirect - if you have the energy can youcheck out that gate patch10:58
sdakeI've been digging into how the gate works a bit and deconstructed atleast the registry mirroring work10:58
*** zhangqiankun has quit IRC10:58
portdirectyeah - its a late start for me - so no excuses today10:58
sdakeportdirect there are also other patches in the queue that could use reviews10:59
portdirectthis one: https://review.openstack.org/42659810:59
sdakeyup that one10:59
sdakeits 4am - going to crash for cuple hours10:59
sdakethen kids off to school10:59
sdake4 hours to learn how that works - however- now  Iunderstand it10:59
sdakeand understand why it can't be the same as kolla-ansible's gating scripts10:59
sdakethe gate scripts need MOAR COMMENTS10:59
*** ipsecguy has quit IRC10:59
sdakesome comments would have cut the 4 hours down to 20 minutes11:00
portdirectlol - man, If you ever win that one I'd be hyper impressed - the lack of comments really makes me sad11:00
sdakeportdirect i'll commet it myself if need be11:01
sdakeportdirect th only way to fix problems is to own them11:01
sdakeexpecting other people to magically fix things you want is unrealistic :)11:01
*** zhangqiankun has joined #openstack-kolla11:01
sdakei'm pretty sure there would not be -1's on comments in the gate scripts11:01
sdakehowever the gate scripts have grown unwiedly11:02
sdakei hear sbezverk has a refactor somewhere in the review queue :)11:02
sdakei've burned about 40 hours learning how the gate works for kolla-kubernetes - as part of the devenv dev11:02
sdakealthough bmace took over the devenv work and that is compelte as soon as global requirements unfreezes11:03
sdakespeaking of unrealistic expectations, entrypoint is fubared11:04
sdakeit doesn't seem to do what it claims11:04
sdakei see containers enter the error state all the time11:04
sdaketha shouldn't happen if the dep management is done properly11:04
sdakei'm not quite sure how to complain about that as it appears abanonded11:04
portdirectinc) should have some links there i think?11:05
sdakelinks huh?11:05
sdakeyou mean people to do the work11:05
portdirectyeah11:05
sdakeI asked him about it earlier, he said he is out of bullets11:05
portdirectIt's intel folk who look after it these days11:05
portdirect:(11:05
sdakeseems like a topic for the meetign in 5 hours11:06
sdakeor 4 hours11:06
sdakeor whatver it is11:06
sdakeince a whole lot of people depend on it11:06
sdakeperhaps inc0 can get the right poeple on it11:06
sdakei don't know go, although I could learn11:06
sdakehowever, go is not a language that is approved within openstack at this time11:07
sdaketherefore, we can't import it into the openstack namespace11:07
portdirectyeah - I'd file a bug on their github and see if that gets a response11:07
sdakethat doesn't mean we cna't use it tho11:07
sdakei'm not even sure where to begin othe then a screenshot11:07
sdakeand "this shouldn't happen"11:07
des_consoladoso the kolla repo has never tracked files11:07
des_consoladooh wtf11:07
sdakedes_consolado could you expand on your question?11:08
des_consoladosorry premature enter11:08
des_consoladoand I was scrolled up about a day or two uhh... ironic-agent.kernel11:08
des_consoladothe ironic plays are incomplete I guess11:08
sdakeha ironic needs love11:08
sdakeha/ya11:08
sdakesean-k-mooney was working on them at one point - not sure on his current status there11:09
sdakeportdirect the CFP was extended until the8th11:09
sdakeif youcan come up with a good topic for me to present on besides doom and gloom ill submit one since I made you submit one ;)11:09
sdakethanks for that btw portdirect :)11:10
sdakethat is probably 00:00 utc11:10
sdakewhich is 5pm my time iirc11:10
sdakeso 12 hours out11:10
portdirectwill try - but currenlty in the middle of moving house11:11
*** sdake has quit IRC11:11
portdirectmy life is spread across everything11:11
portdirectand just about to head out to the charity shop again11:11
*** sdake has joined #openstack-kolla11:11
hrwlooks like first aarch64 container using kolla gets closer :D11:18
sp_portdirect: facing this error "Could not fetch specs from http://gems.hashicorp.com/" when i run command "vagrant plugin install vagrant-libvirt"11:18
sp_portdirect: I am setting up kolla-k8s dev environment11:19
sp_portdirect: machine is of configuration 32GB RAM, single node, 2 core CPU11:22
*** wagiel has joined #openstack-kolla11:23
Serlexsp_ are you behind a proxy?11:24
portdirectsp_: ^^ thats would be my first guess as well11:25
sp_Serlex: portdirect      when no proxy was set i faced the same issue, as i am facing now after setting this to11:27
sp_ubuntu@openstack-nti-3:~$ env | grep proxy  # just checking11:27
sp_http_proxy=http://192.0.2.1:312811:27
sp_https_proxy=http://192.0.2.1:312811:27
sp_ubuntu@openstack-nti-3:~$11:27
*** wagiel has left #openstack-kolla11:27
portdirectI'm sorry sp_  but I'm not going to be the best help on that today :( you may find this helpful: https://github.com/vagrant-libvirt/vagrant-libvirt#installation11:30
sp_portdirect: thanks11:30
sp_i will check11:30
*** salv-orlando has joined #openstack-kolla11:33
Serlexsp_ stupid question, but 192.0.2.1 is your proxy server right?11:37
sp_Serlex: no, good question. No its not. I got the solution for a blog and I put it there which is wrong11:39
SerlexSo, first check if you are behind a firewall which needs a proxy. if so, find out what is the proxy server and export http_proxy or/and https_proxy to your environment11:41
sp_Serlex: I will catch you after 1:30 hr. BTW thanks11:42
SerlexI asked, because there are few blog and stackoverflow posts with the IP and port above...I thought you might of copied :D11:42
sp_Serlex: You are right.11:42
sp_:D11:42
sp_I didn't modify that before running the same command11:43
*** sdake has quit IRC11:43
*** sp_ has quit IRC11:44
*** ipsecguy has joined #openstack-kolla11:46
*** ipsecguy has quit IRC11:46
des_consoladomliima: are you around today?11:49
mliima\o11:49
mliimades_consolado,11:50
des_consoladomliima: hey - you wrote the ironic config.yml that puts ironic-agent.kernel and .initramfs in the nodes' config dirs11:50
des_consoladowhere can I find out more about these files? they've never been in the repo, do we generate them or have to manually write them?11:50
des_consoladoI can't find too much about them on google11:50
*** cinerama has joined #openstack-kolla11:52
Serlexdes_consolado, you need to generate them locally11:56
des_consoladooh I hadn't even remembered to build the ironic image(s) actually, presuming they're not in the default profile11:58
SerlexI hit the same problem. Something along the lines of "disk-image-create ironic-agent fedora -o ironic-agent"11:59
SerlexWhich will output .kernel and .initramfs files11:59
Serlexhttp://docs.openstack.org/developer/ironic-inspector/install.html12:00
des_consoladooh cool, thanks Serlex12:00
*** huzhengchuan has quit IRC12:02
*** pbourke has quit IRC12:06
*** pbourke has joined #openstack-kolla12:07
*** dave-mccowan has joined #openstack-kolla12:08
*** ipsecguy has joined #openstack-kolla12:09
*** huzhengchuan has joined #openstack-kolla12:14
des_consoladothis should be automated though, right? :\12:14
des_consoladoso there's a new python requirement, python-diskimage-builder, with this12:15
*** ipsecguy has quit IRC12:16
des_consoladobut ironic is optional and not default, so I guess that's why it's not in the requirements.txt - but if you enable ironic, I guess you should use a pip module in one of the playbooks to make sure this is installed...12:16
*** mliima_ has joined #openstack-kolla12:16
*** ipsecguy has joined #openstack-kolla12:17
*** mliima has quit IRC12:20
*** lucker has joined #openstack-kolla12:22
luckerhello12:22
SerlexI know what you mean regarding the automation. But I considering the minimum work required, i dont think its highest priority12:23
des_consoladoyeah I should do a patch, haven't contributed yet12:24
*** david-lyle has quit IRC12:24
*** huzhengchuan has quit IRC12:25
*** salv-orlando has quit IRC12:34
*** ipsecguy has quit IRC12:36
*** ipsecguy has joined #openstack-kolla12:37
*** huzhengchuan has joined #openstack-kolla12:37
*** ipsecguy has quit IRC12:41
zioprotois there a date for the stable/ocata branch ?12:49
openstackgerritPaul Bourke (pbourke) proposed openstack/kolla master: Move fluentd td.repo to base for consistency  https://review.openstack.org/43019312:49
zioprotoI will have this patch thah needs love before the branch of the stable release https://review.openstack.org/#/c/427139/12:50
*** rwallner has joined #openstack-kolla12:50
*** huzhengchuan has quit IRC12:51
*** zhangqiankun has quit IRC12:57
*** japestinho has quit IRC13:01
openstackgerritSerguei Bezverkhi proposed openstack/kolla-kubernetes master: WIP Adding compute-kit chart  https://review.openstack.org/42815013:02
*** rhallisey has joined #openstack-kolla13:05
hrwegonzalez: http://pastebin.com/3x9wzEhM - arm64 ;D13:08
*** goldyfruit has joined #openstack-kolla13:10
*** goldyfruit has quit IRC13:10
*** satyar has joined #openstack-kolla13:12
*** zhubingbing has joined #openstack-kolla13:13
mliima_error in nova container durring deployment -> http://paste.openstack.org/show/598081/13:13
*** mgiles has joined #openstack-kolla13:14
*** mgiles has quit IRC13:16
*** mgiles has joined #openstack-kolla13:16
*** stevemar has left #openstack-kolla13:17
*** sp__ has joined #openstack-kolla13:19
*** salv-orlando has joined #openstack-kolla13:19
kfox1111morning13:21
*** jheroux has joined #openstack-kolla13:21
*** rwallner has quit IRC13:24
*** sp__ has quit IRC13:25
*** sp_ has joined #openstack-kolla13:30
*** saneax is now known as saneax-_-|AFK13:32
*** sudipto has joined #openstack-kolla13:33
*** sudipto_ has joined #openstack-kolla13:33
*** sdake has joined #openstack-kolla13:36
sdakemorning peeps13:36
*** fguillot has joined #openstack-kolla13:36
hrwelo13:39
hrw(venv-for-kolla)14:37 hrw@pinkiepie-centos:kolla$ git review13:39
hrwfatal: ICLA contributor agreement requires current contact information.13:39
hrwso I added ones.13:39
hrwCode Review - Error13:39
hrwServer Error13:39
hrwCannot store contact information13:39
hrwcan someone tell me what is wrng?13:39
mliima_error in nova container durring deployment -> http://paste.openstack.org/show/598081/13:40
mliima_morning sdake13:40
sdakesup mliima_13:41
sdakemliima_ ya - nova is borked13:41
sdakeportdirect can you please review this simple 2 liner: https://review.openstack.org/#/c/430708/113:41
sdakehrw ok - you need to add your contact information on openstac.org's website13:42
sdakewhen you ened to log out of gerrit13:42
mliima_ok sdake13:42
sdakeand log back into gerrit13:42
mliima_:)13:42
hrwsdake: 14:39 < hrw> Cannot store contact information13:42
hrwsdake: that;s what gerrit tells me in browser13:42
sdakehrw in openstack.org, not review.openstack.org13:42
sdakelogin to openstack.org13:42
*** sayantani01 has quit IRC13:43
sdakereview.openstack.org will not letyou store contact information - it comes from review.openstack.org13:43
hrwfucking shit.13:43
* hrw goes13:43
sdakeratheri t comes from the openstack.org website13:43
sdakehrw ya its not very intuitive13:43
sdakethanks portdirect :)13:43
egonzalezhrw: your profile in openstack.o.o need to be as Foundation Member13:44
*** matrohon has quit IRC13:44
hrwegonzalez: iirc it is13:44
openstackgerritSerguei Bezverkhi proposed openstack/kolla-kubernetes master: WIP Adding compute-kit chart  https://review.openstack.org/42815013:44
sdakekfox1111 morning13:45
sdakere rax - I see consistent fialures on rax - as in 100% in the kolla-ansible gate jobs13:45
sdakekfox1111 thanks for the review btw ;)13:45
sp_sdake: morning13:45
sdakekfox1111 looking at your work there helped me figure out how the gate works13:45
sdakesp_ sup13:45
kfox1111sdake: morning13:47
sbezverkkfox1111: it looks there are a couple of systematic failures. 1 - gate-kolla-kubernetes-deploy-centos-binary-3-ceph-multi-nv  always fails with the same error even on a master gate check13:47
hrwR@#!%$@#%13:47
kfox1111sdake: yeah, but kolla-kubernetes does not have that problem.13:47
kfox1111sdake: I think its a failing of the kolla-ansible gate, not rax.13:47
hrwhttps://www.openstack.org/profile/ shows address added. git review says gtfo13:47
sdakekfox1111 i notice alot of failureson rax.iad for kolla-buernetes too13:47
sbezverkkfox1111: 2 - I am suspecting 3.0.2 cinder lvm volume is broken as the same config with 3.0.1 works13:47
sdakehrw now log out of review.openstack.org - then log back in13:47
kfox1111sdake: when debugging issues, we've been logging them for awhile, and looking for patterns.13:48
sdakethis wil sync your contact information with review.openstack.org13:48
hrwsdake: https://review.openstack.org/#/settings/contact shows empty data13:48
sdakekfox1111 fair enough - as i said in the reivew i'l moveremove it13:48
sdakehrw did you log out of review.openstack.org?13:48
hrwsdake: yes, I did13:48
kfox1111sdake: in all the problems sovled, only at the very begining was there a pattern on provider, and that was due to multiple nics. once that was solved we never have seen a rax specific issue again.13:48
sdakehrw i can help you via webex if you ahve a 10 minutes now - otherwise I can help you later13:49
hrwsdake: https://www.openstack.org/community/members/profile/66842/marcin-juszkiewicz is my public profile13:49
sdakekfox1111 sweet, if you founnd a solution to that rax problem, could you please feed it into Jeffrey4l13:49
sdakehrw you need to set your adress information - ican't see that in your public profile13:49
flaper87hey folks, I noticed there's no kolla image for redis, is there a reason for that? Or is it just that it's not being used right now?13:50
*** salv-orl_ has joined #openstack-kolla13:50
kfox1111sdake: I'm afraid it will probably be different for each gate. :/13:50
sdakehrw cisco.webex.com/join/stdake13:50
sdakekfox1111 right- feed in what you know13:50
kfox1111sdake: kolla-kubernetes multinic problem was in our iptables modifying code.13:50
*** matrohon has joined #openstack-kolla13:50
kfox1111kk.13:50
flaper87I'd be happy to work on it if it's ok13:50
sdakekfox1111 just have a conversation with him is all i ask :)13:50
sdakekfox1111 i could be imaginging this problem13:50
sdakeflaper87 wfm :)  needs a blueprint13:50
openstackgerritMerged openstack/kolla-kubernetes master: Actually report gate failures for experimental jobs  https://review.openstack.org/43070813:51
sdakeflaper87 link the blueprint and i'll be happy to approve13:51
sdakeflaper87 or dan can approve13:51
sdakeor any other driver13:51
kfox1111yeah. Ill talk to him about it.13:51
kfox1111I have no doubt rax was a problme at one point for sure, as their nic arangemnt is different from all the rest it seems. but once I fixed the logic that expected only one nic in the vm's, its been stable since.13:52
hrwfukc13:52
kfox1111on the kolla-kubernetes side.13:52
sdakehrw you dont have to swear :)13:52
sdakehrw webex - i can wlak you through it - the signin process is annoying13:52
*** salv-orlando has quit IRC13:52
sdakewebex has no signin proces showever - just join and go13:53
*** sudipto has quit IRC13:53
*** sudipto_ has quit IRC13:53
hrwjoining13:53
sdakethanks hrw13:53
sdakekfox1111 - ok dude thats a huge peice of  infromation13:53
flaper87sdake: sounds perfect, I'll do it then13:53
sdakehrw steven.dake@gmail.com13:55
kfox1111np. :)13:56
*** zhurong has joined #openstack-kolla13:57
sdakehrw can you hear me :)13:57
hrwno13:57
sdakehrw i can hear you13:57
sdakehrw  who knows whats going on with that - i'll type you share :013:57
*** imcsk8_ has joined #openstack-kolla13:57
*** imcsk8 has quit IRC13:58
sdakehrw i dont see your shared screen13:58
sdakehrw ya its annoying al this tech :)13:58
flaper87sdake: https://blueprints.launchpad.net/kolla/+spec/redis-container13:59
sdakehrw i do not13:59
sdakehrw i did seeit originally13:59
flaper87sdake: do I have to write a spec under kolla/specs ?13:59
sdakehrw lets restart the call13:59
sdakehrw maybe a resync will get it going13:59
sdakeflaper87 no spec needed13:59
sdakeflaper87 will be with you in a moment13:59
sdakegood that  worked hrw13:59
flaper87sdake: coolio, thanks14:00
*** caowei has joined #openstack-kolla14:00
sbezverkkfox1111: is there a easy way to access nova-compute.log at the gate?14:01
kfox1111sbezverk: logs/openstack/container-name.txt14:01
sbezverkkfox1111: perfect, exactly what I needed14:03
sbezverkthank you14:03
kfox1111np.14:06
sdakeok flaper87 moment14:06
sdakehrw see not that hard14:06
sdakehrw of course i've done that debug about 100+ timees :)14:06
hrwsdake: it's sick ;d14:06
sbezverkkfox1111: have yu seen this traceback before? http://logs.openstack.org/50/428150/54/check/gate-kolla-kubernetes-deploy-centos-binary-3-ceph-multi-nv/b8c6b77/logs/openstack/kolla-nova-scheduler-0.txt14:07
openstackgerritMarcin Juszkiewicz proposed openstack/kolla master: support non-x86 architectures (aarch64, ppc64le)  https://review.openstack.org/43094014:07
hrwhttps://review.openstack.org/430940 yay!14:07
hrwnow have to add some reviewers14:07
kfox1111sbezverk: no. that one's new to me.14:07
sdakeflaper87 ok your all set to go14:08
kfox1111sbezverk: if I had to guess,14:08
sbezverkkfox1111: there are a couple of trace backs 1 in nova compute and 1 in scheduler, very consistent14:08
kfox1111I'd say it was due to no computes up.14:08
sdakehrw nice arm work :)14:08
flaper87sdake: awesome! Thanks a bunch. I'm on it, should have something up soon14:08
sdakeok gotta jet for awhile to take kid to school and car to shop14:08
sdakeflaper87 note the deadline for newton has pssed14:09
hrwsdake: I based on Sajauddin's work14:09
flaper87sdake: u mean ocata ?14:09
flaper87sdake: it's fine, Pike works14:09
sdakeflaper87 apprently living in the past :)14:09
sdakeok really GTG :)14:09
sbezverkkfox1111: agree about compute, since it is also not too happy: http://logs.openstack.org/50/428150/54/check/gate-kolla-kubernetes-deploy-centos-binary-3-ceph-multi-nv/b8c6b77/logs/openstack/kolla-nova-compute-xv6dj.txt14:09
flaper87sdake: no worries, I don't even know what day it is14:09
sdakeflaper87 it is wednesday :)14:10
flaper87that explains why no one has pinged me saying I'm late for a meeting14:10
flaper87good14:10
kfox1111sbezverk: keystone logs are a bit weird too.14:11
kfox1111sbezverk: Please put a link in the etherpad if you haven't already. This is pretty strange. would be nice if we could correlate it if we see it again.14:13
sbezverkkfox1111: I think I know the problem. keystone internal end point points to keystone-internal:5000 instead of 3535714:14
sbezverkkfox1111: http://logs.openstack.org/50/428150/54/check/gate-kolla-kubernetes-deploy-centos-binary-3-ceph-multi-nv/b8c6b77/14:14
kfox1111hmm.. I think thats the way it always was though?14:15
*** goldyfruit has joined #openstack-kolla14:15
kfox1111I think only the admin server is 3535714:15
sbezverkhm ok maybe you are right, I got mixed up with admin14:15
kfox1111is the error consistent?14:16
sbezverkkfox1111: yes, even on gate check of master..14:19
*** mdnadeem has quit IRC14:19
sbezverkat first I thought I introduced it, but then this monrning I saw exactly the same on sdake's gatecheck run14:19
kfox1111hmm..14:20
kfox1111interesting.14:20
kfox1111so maybe something changed in infra...14:20
kfox1111any pattern to them?14:20
kfox1111seen it on a non multi job?14:20
*** crushil has joined #openstack-kolla14:23
sbezverkkfox1111: nope only branch 314:23
kfox1111weird.14:24
kfox1111we haven't been testing branch 3 very long. could be some race conditions in it still, but would be weird for them to show up more commonly now.14:24
kfox1111maybe something merged recently that made it worse?14:25
*** lamt has joined #openstack-kolla14:26
sp_sdake: I think we should draft a BP in kolla-ansible too w.r.t redis-container BP in kolla14:28
sp_sdake: we can set the target fir pike-2 once redis-container would be completed in pike-1 release14:28
kfox1111so, got prometheus up and running.14:29
sp_sdake: please provide your opinion on this14:29
kfox1111very interesting. :)14:29
openstackgerritSerguei Bezverkhi proposed openstack/kolla-kubernetes master: WIP Adding compute-kit chart  https://review.openstack.org/42815014:30
*** gfidente is now known as gfidente|afk14:32
hrwcan someone tell me was debian support ever working in kolla?14:33
*** sayantani01 has joined #openstack-kolla14:35
sbezverkkfox1111: in kolla-kube or just on your cluster? re prometius14:38
kfox1111just in my cluster.14:39
sbezverkkfox1111: anything interesting you managed to do with it?14:40
kfox1111But, was curious if we could slide this: https://github.com/prometheus/haproxy_exporter14:40
kfox1111as a sidecar intno the api_python_deployments.14:40
*** jtriley has joined #openstack-kolla14:41
kfox1111sbezverk: not too much yet. just got it running last night, plugged into graphite, and collecting metrics from an ingress.14:41
kfox1111just trying to figure it all out at this point. I've used influxdb for a while, and its fairly different.14:41
sbezverkkfox1111: are you looking into use it for accounting purposes? I mean haproxy_exporter?14:42
kfox1111more monitoring I think. should be able to gather stuff like when things die, if latency spikes, etc.14:44
*** crushil has quit IRC14:44
*** crushil has joined #openstack-kolla14:45
openstackgerritMarcin Juszkiewicz proposed openstack/kolla master: support non-x86 architectures (aarch64, ppc64le)  https://review.openstack.org/43094014:48
sbezverkkfox1111: if you check https://review.openstack.org/430708 it fails exactly the same way on that job..14:49
openstackgerritFlavio Percoco proposed openstack/kolla master: Add redis container  https://review.openstack.org/43097614:52
kfox1111ah. yeah.14:54
*** sudipto has joined #openstack-kolla14:54
*** sudipto_ has joined #openstack-kolla14:54
kfox1111sbezverk: added a section to the etherpad at the bootom.14:56
kfox1111lets start recording them there and see if we can spot a pattern.14:56
sbezverkkfox1111: sounds good, could you please paste etherpad again?14:57
kfox1111https://etherpad.openstack.org/p/kolla-kubernetes-gate-failure-tracking14:57
*** mgoddard_ has quit IRC14:57
zioprotoI am trying master, what is this fluentd ? can I disable it ?15:00
*** duonghq has joined #openstack-kolla15:00
*** lamt has quit IRC15:01
*** tonanhngo has joined #openstack-kolla15:01
egonzalezzioproto: nope, fluentd is the replacement for heka15:01
*** tonanhngo has quit IRC15:02
zioprotooh so there is a migration path ?15:02
zioprotoI worked with stable/newton15:02
zioprotoand how I am trying master15:02
zioprotowhat is a valid openstack_release: ?15:02
zioprotoI had 3.0.215:03
egonzalezzioproto: upgrade process remove heka and put fluentd in place15:04
zioprotoI have put openstack_release: auto and I get No module named pbr.version15:04
*** adrian_otto1 has joined #openstack-kolla15:05
*** vcn has joined #openstack-kolla15:08
*** ipsecguy has joined #openstack-kolla15:08
*** vcn has quit IRC15:09
*** vcn has joined #openstack-kolla15:09
*** vcn has quit IRC15:10
*** vcn has joined #openstack-kolla15:11
*** lamt has joined #openstack-kolla15:12
openstackgerritSerguei Bezverkhi proposed openstack/kolla-kubernetes master: WIP Adding compute-kit chart  https://review.openstack.org/42815015:12
*** jmccarthy has quit IRC15:12
sbezverkkfox1111: have you seen any situation when all pods showed running but wait_for_pod was still waiting and then timeing out?15:13
*** vcn has quit IRC15:14
kfox1111hmm..15:15
kfox1111I think it can be in that state if there's a job unfinished too.15:15
sbezverkkfox1111: well but then it would mean bootstrap is not done and one of pods would be in Init state15:18
sbezverkI do not see it15:19
sbezverkI added to wait_for _pod some extra output to see15:19
sbezverkkfox1111: check this log if you have a minute: http://logs.openstack.org/50/428150/56/experimental/gate-kolla-kubernetes-deploy-centos-binary-2-helm-compute-kit-nv/f847804/15:19
*** sacharya has joined #openstack-kolla15:20
*** dave-mccowan has quit IRC15:20
*** adrian_otto1 is now known as adrian_otto15:21
kfox1111got the ceph_exporter working too. :)15:21
kfox1111looking15:22
kfox1111heh. that kind of locked up my browser...15:25
*** saneax-_-|AFK is now known as saneax15:25
kfox1111sbezverk: the Restarts are interesting.,15:26
kfox1111may imply the deps aren't working quite right.15:27
*** krtaylor has quit IRC15:28
*** jmccarthy has joined #openstack-kolla15:29
*** fragatina has quit IRC15:30
*** fragatina has joined #openstack-kolla15:30
*** caowei has quit IRC15:31
*** caowei has joined #openstack-kolla15:31
*** huzhengchuan has joined #openstack-kolla15:31
sbezverkkfox1111: yeah :( when you checked logs for processes which restarted 5 time15:32
sbezverkyou will see that it failed to contact mariadb first few time15:32
sbezverkand it should not happen15:32
openstackgerritSerguei Bezverkhi proposed openstack/kolla-kubernetes master: WIP Adding compute-kit chart  https://review.openstack.org/42815015:34
*** krtaylor has joined #openstack-kolla15:38
*** dave-mccowan has joined #openstack-kolla15:38
*** duonghq has quit IRC15:39
*** saneax is now known as saneax-_-|AFK15:40
openstackgerritMerged openstack/kolla-kubernetes master: python 3.5 is enabled in tox environment  https://review.openstack.org/41688015:46
*** duonghq__ has joined #openstack-kolla15:48
*** duonghq__ is now known as Duonghq15:48
*** Duonghq is now known as duonghq15:49
Jeffrey4legonzalez, around? any progress for  nova upgrade about cellv2 issue?15:49
egonzalezJeffrey4l: have some ideas, but not really sure. i've asked at nova channel but nobody answered15:54
openstackgerritSerguei Bezverkhi proposed openstack/kolla-kubernetes master: WIP Adding compute-kit chart  https://review.openstack.org/42815015:54
Jeffrey4lok. thanks15:54
egonzalezJeffrey4l: i've created a chart with implications of different scenarios https://www.lucidchart.com/publicSegments/view/72b23206-48af-4c8b-826b-5dbe4c9cfac9/image.png15:55
*** huzhengchuan has quit IRC15:56
sp_egonzalez: drafted a BP https://blueprints.launchpad.net/kolla-ansible/+spec/redis-container-ansible-role15:56
sp_egonzalez: please take a look15:57
*** eaguilar has joined #openstack-kolla15:57
*** inc0 has joined #openstack-kolla15:57
*** adrian_otto has quit IRC15:57
egonzalezsp_: for what will be used redis?15:57
inc0good mornign15:57
Jeffrey4legonzalez, have u tried the third one? "create cells with ocata"15:58
sp_egonzalez: redis container would be supported by kolla deliverable15:58
egonzalezdb_sync fails without cells db15:58
sp_egonzalez:  so need to add role too in kolla-ansible deliverable too15:58
inc0meeting time15:58
sp_inc0: morning15:58
sdake'./lsup peeps15:58
sdakesp_ i dont know if redis would be used by ansible deployment or not - generally we try to add containers for generally useful things related to deployment15:59
Jeffrey4lfeel nova leave a bomb there.15:59
sdakeflaper87 may be able to add more context as to the technical need for redis16:00
sdakeJeffrey4l which bomb - you mean mess? :)16:00
Jeffrey4lnova cell creation.16:00
Jeffrey4lnova cell v2 creation during upgrade from newton to ocata.16:00
sp_sdake: confirming with egonzalez for its feasibility16:01
duonghqDo we have weekly meeting?16:01
Jeffrey4lwe may have to handle the database change during upgrade.16:01
*** vhosakot has joined #openstack-kolla16:01
Jeffrey4lduonghq, it is started ;)16:01
*** schwicht has joined #openstack-kolla16:01
sp_duonghq: yes inc0 is here16:01
egonzalezJeffrey4l: i don't know the implications of creating default cell in nova_api_cell0 in upgrade. will that give any issue for future versions?16:02
*** srwilkers has joined #openstack-kolla16:02
Jeffrey4lnova_api_cell0 is the default name in newton and nova_cell0 is the default name in ocata.16:02
Jeffrey4lwe have to change the name during upgrade.16:03
Jeffrey4lit is not confirgurable.16:03
* flaper87 here16:06
flaper87what did I do?16:06
*** unicell has joined #openstack-kolla16:07
sp_flaper87: meeting started16:07
flaper87My need for the redis image is ceilometer and zaqar16:07
sdakeflaper87 you did nothign wrong16:07
sdakeflaper87 cool thanks for adding the context on the reason for redis16:07
sdakeflaper87 would you mind adding that to the blueprint pls16:08
flaper87I can add it on the review/bp if needed16:08
*** penfold has joined #openstack-kolla16:08
*** duonghq_ has joined #openstack-kolla16:08
flaper87yes16:08
flaper87actually, I think I did16:08
flaper87should be in the bp already16:08
sdakeoh i guess i missed it :)16:08
flaper87I forgot to link the bp16:08
* flaper87 dumb16:08
*** unicell1 has quit IRC16:08
sdakeidc if people add stuff to kolla deliverable if it is useful in general - within limits :)16:08
*** mgoddard has joined #openstack-kolla16:08
mgoddard#openstack-meeting-416:08
sdakeothes may feel differently16:08
*** duonghq is now known as duonghq__16:09
sdakewe have our team meeting atm in #openstack-meeting-4 now tho flaper87 so gtg - your wellcoemt o listen in16:09
*** duonghq_ is now known as duonghq16:09
* flaper87 lurks16:09
openstackgerritFlavio Percoco proposed openstack/kolla master: Add redis container  https://review.openstack.org/43097616:09
egonzalezflaper87: good enough, just wanted to know the usage.16:11
flaper87egonzalez: cool, does that address your questions on the review?16:11
egonzalezflaper87: qq, wont need python-redis package, i had some issues before iirc16:11
flaper87Hopefully I didn't misunderstand the use of the profiles in conf.py16:11
flaper87egonzalez: you probably need that in the zaqar image16:12
flaper87and the ceilometer one16:12
flaper87but it's already a dependency for the packages16:12
flaper87so it should be good16:12
flaper87and it's listed in the requirements16:12
flaper87iirc16:12
egonzalezflaper87: cool, tell me under zaqar PS and will match to redis too. I have to rework the whole role and modify docker images too16:13
openstackgerritFlavio Percoco proposed openstack/kolla master: Add redis container  https://review.openstack.org/43097616:14
openstackgerritFlavio Percoco proposed openstack/kolla master: Add redis container  https://review.openstack.org/43097616:15
*** matrohon has quit IRC16:15
*** adrian_otto has joined #openstack-kolla16:16
*** salv-orl_ has quit IRC16:19
*** david-lyle has joined #openstack-kolla16:23
*** Pavo has joined #openstack-kolla16:24
*** zhurong has quit IRC16:25
*** unicell has quit IRC16:25
*** saneax-_-|AFK is now known as saneax16:26
sp_egonzalez: what about redis ansible role ?16:27
egonzalezsp_: good enough, just wanted to know the usage. (missed the name before)16:29
*** fragatina has quit IRC16:29
sp_egonzalez:  please approve the same https://blueprints.launchpad.net/kolla-ansible/+spec/redis-container-ansible-role16:30
*** fragatina has joined #openstack-kolla16:30
flaper87egonzalez: thanks for the review, could you help me understand those configs a bit better?16:33
*** sayantan_ has joined #openstack-kolla16:33
egonzalezflaper87: when you do kolla-build can select a profile to build(default, aux, gate)16:33
egonzalezand don't need to do kolla-build nova haproxy neutron etc16:34
openstackgerritFlavio Percoco proposed openstack/kolla master: Add redis container  https://review.openstack.org/43097616:34
flaper87egonzalez: ah, right! Now I got it #dumb16:34
flaper87egonzalez: ok, I've addressed your comments16:34
flaper87thanks again16:34
*** sdake_ has joined #openstack-kolla16:36
*** sayantani01 has quit IRC16:36
*** sdake has quit IRC16:38
*** caowei has quit IRC16:38
*** sayantan_ has quit IRC16:39
sp_egonzalez: ^^16:40
*** zioproto has quit IRC16:40
*** zhubingbing has quit IRC16:43
*** prameswar has quit IRC16:50
*** david-lyle has quit IRC16:52
*** Guest66676 is now known as Guest666616:53
*** sacharya has quit IRC16:54
*** sacharya has joined #openstack-kolla16:54
*** salv-orlando has joined #openstack-kolla16:54
*** tonanhngo has joined #openstack-kolla16:56
*** tonanhngo has quit IRC16:56
Jeffrey4lgema krtaylor  another question: does binary with static build  works on ppc? ( guess no )16:59
gemaJeffrey4l: I work on ARM64, krtaylor on ppc :D17:00
gemaJeffrey4l: and this patch was done by hrw so i'd say17:00
gemahrw: ^17:00
gemaJeffrey4l: he'll reply when he is back :D17:00
Jeffrey4lthanks.17:00
*** egonzalez has quit IRC17:00
gemaJeffrey4l: but what do you mean by a binary with static build?17:01
gemayou mean built for ppc or for intel?17:01
Jeffrey4lgema, check this https://github.com/krallin/tini#statically-linked-version17:02
krtaylorJeffrey4l, as I understand there are IBMers that have made this work for ppc64, but I have not verified...yet17:02
Jeffrey4lhow can u get openstack packages on arm? ubuntu/debian provide it? or build it by you own.17:02
Jeffrey4lkrtaylor, thanks. just curiosity ;)17:03
gemaJeffrey4l: we have a built and tested openstack package set17:03
gemaJeffrey4l: for centos and debian17:03
gemaJeffrey4l: ubuntu provides their own17:03
Jeffrey4lcool.17:03
krtaylorJeffrey4l, understood, I want to make sure you get an answer17:03
gemaJeffrey4l: https://platforms.linaro.org/documentation/Reference-Platform/Platforms/Enterprise/Documentation/OpenStack/README.md/17:03
hrwJeffrey4l: hi17:04
gemathats for ARM6417:04
*** sayantani01 has joined #openstack-kolla17:04
Jeffrey4lhi hrw17:04
gemaI leave you in good hands, EOD for me :D17:04
krtaylorlikewise for ppc64le, centos/RDO is there, as is ubuntu17:05
hrwJeffrey4l: https://hub.docker.com/u/ppc64le/17:05
hrwkrtaylor: for arm64 we will use rdo too17:05
Jeffrey4lhrw, cool.17:05
krtaylorexcellent17:05
hrwJeffrey4l: I am waiting for official aarch64/centos7 image - will be provided on days17:06
Jeffrey4lthen {{ base_arch }}/{{ base_image }}:{{ base_distro_tag }} is not good imo.17:06
krtayloryeah, I was just looking at that also17:07
*** duonghq has quit IRC17:08
hrwJeffrey4l: I used that from ppc64le patch17:08
*** sayantani01 has quit IRC17:09
hrwand with clean docker and that patch I build 25 containers from base17:09
Jeffrey4lhrw, krtaylor i left another comment about this.17:10
krtayloryeah, thats the other folks that popped up and did this for ppc64le, not part of our openstack ppc64 team17:10
hrwJeffrey4l: not agree with comment17:10
krtaylorI need to review that one also, on my list17:10
krtaylorgetting basic understand of the project atm17:10
krtaylorunderstanding17:11
hrwJeffrey4l: current state allows you to say --base-arch aarch64 and then you get aarch64. if someone wants to add arm (as arm 32bit) then it does nt have to play with dockerfile.j2 at all as it will just be used.17:11
hrwor any other arch17:11
*** sayantani01 has joined #openstack-kolla17:11
*** david-lyle has joined #openstack-kolla17:12
hrwJeffrey4l: and setting conf.base_image is just wrong imho.17:12
krtaylorI need to look at how we did this for other projects17:12
Jeffrey4lhrm. ok...17:12
krtaylorjust trying to get a sandbox working and get up to speed for  longer-term kolla involvement17:12
hrwI am against hardcoding things17:12
Jeffrey4las far as i know, kolla do not support 32bit.17:13
hrwJeffrey4l: so make it s390x?17:13
*** sayantani01 has quit IRC17:13
hrwor any other 64bit? mips64? ia64? riscv64?17:13
Jeffrey4li do not mark it hard coding.17:14
Jeffrey4lbut "FROM {{ base_arch }}/{{ base_image }}:{{ base_distro_tag }}" does17:14
*** sayantani01 has joined #openstack-kolla17:14
hrwless than code from your last comment17:14
Jeffrey4lwhy my last comment does?17:14
openstackgerritSerguei Bezverkhi proposed openstack/kolla-kubernetes master: WIP Adding compute-kit chart  https://review.openstack.org/42815017:14
Jeffrey4lbase_image is configurable. base_arch is too.17:15
hrwok, let me look other way at it17:16
hrwstill not fully used to that code17:16
Jeffrey4lbase_image parameter is introduced to support power cpu.17:17
*** sayantani01 has quit IRC17:20
hrwJeffrey4l: so how would you use it?17:20
hrwkolla --base centos --base_image ppc64le/centos?17:20
hrw(assuming that ppc64le/centos exists)17:21
*** dciabrin has quit IRC17:21
Jeffrey4lhrw, yep.17:22
Jeffrey4lhrw check http://lists.openstack.org/pipermail/openstack-dev/2016-April/093057.html17:22
*** Serlex has quit IRC17:22
hrwJeffrey4l: ok, sound fine for me. just have to expand --base-image help entry to mark that17:24
hrw"kolla --base centos --base_image aarch64/centos7 --base_arch aarch64"17:24
Jeffrey4lyep.17:24
clilesanyone know why the designate-ui from horizon was reverted?17:24
hrwJeffrey4l: thanks for comments17:24
Jeffrey4lnp.17:24
* krtaylor reads scrollback17:25
Jeffrey4lcliles, long story. but you can add it now again. :)17:25
clilesJeffrey4l: yes, if I add it back it works fine17:26
clileswith a source build....17:26
Jeffrey4lcliles, have u put the patch?17:26
Jeffrey4lwe can merge it now.17:26
krtaylorJeffrey4l, thanks for the link, that gives me someone to go talk to about getting my ppc64 env up and going :)17:27
Jeffrey4lkrtaylor, np. lol17:27
Jeffrey4lseems they have set ppc64 up17:27
krtayloryes, now I need to make sure it stays that way, so I'll be around :)17:29
hrw  --base-image BASE_IMAGE17:30
hrw                        The base image name. Default is the same with base.17:30
hrw                        For non-x86 architectures use full name like17:30
hrw                        "aarch64/debian".17:30
hrwI can work with that ;D17:30
openstackgerritMarcin Juszkiewicz proposed openstack/kolla master: support non-x86 architectures (aarch64, ppc64le)  https://review.openstack.org/43094017:32
*** bmace_ has joined #openstack-kolla17:32
hrwJeffrey4l: your comments implemented. patch got even smaller now17:33
Jeffrey4lnice17:33
*** sayantani01 has joined #openstack-kolla17:34
Jeffrey4lcould u add the default base_image for ppc? like what my comment proposed?17:35
Jeffrey4lhrw17:35
hrwJeffrey4l: ppc64le support for ubuntu required more than just setting default to ubuntu (at least according to original pp64le patch)17:37
hrwJeffrey4l: I prefer to make smallest steps. if ppc64le needs default to be !default (==centos) then they are free to put patch after that one17:38
Jeffrey4ladd default base-image make it works with simple "tools/build.py"17:38
*** adrian_otto1 has joined #openstack-kolla17:39
Jeffrey4ladd these logical, the end-user still free to use his own base-image.17:39
Jeffrey4lsee the outer if not conf.base_image condition?17:39
openstackgerritSayantani Goswami proposed openstack/kolla master: Fix pushing images to local registry  https://review.openstack.org/42633817:40
hrwJeffrey4l: the proper play for that would be rather setting of _CLI_OPTS as this is where centos is set as default17:40
*** mgoddard has quit IRC17:40
hrwJeffrey4l: and that's also changing tool defaults just because someone prefers other values17:41
hrwJeffrey4l: should I add 'if aarch64 then go for debian' etc?17:41
Jeffrey4lhrm i  think so.17:41
Jeffrey4lthis make arch64 more easier to use and have nothing else side effect.17:42
hrwuser X will install kolla, read docs ("default base is centos"), run tool and go for other tasks. few hours later will see wrong distro built17:42
*** adrian_otto has quit IRC17:42
hrwJeffrey4l: why not hardcode "from mindreading import userdefaults"?17:43
Jeffrey4lsorry?17:43
hrwJeffrey4l: x86-64 goes for centos, ppc64le goes for ubuntu, aarch64 goes debian, mips64 will went for yetanotherthing. Hate to be the one who has to write docs for it17:44
Jeffrey4lthis is how kolla works now.17:45
hrwJeffrey4l: kolla knows only x86-64.17:45
hrwNOTHING else17:45
hrwif potental ppc64le maintainer wants to have own defaults then I can +1 his changes. but do not want to have it in my patch17:46
Jeffrey4lif no. the writer will add" if you are build ppc64le, try "build.py --base-image xxx --base-arch ppc64"  if you are building arm, try "build.py --base-image yyy --base-arch arm"17:46
*** sudipto has quit IRC17:46
*** sudipto_ has quit IRC17:46
*** Pavo has quit IRC17:46
Jeffrey4lhrw, aren't u trying to make kolla know ppc or arm?17:47
hrwnow you run kolla and you have centos/x86-64 as defaults. if you run on aarch64 you get 'gtfo with this arch'. same on any other17:47
hrwJeffrey4l: not arm. aarch64 (aka arm64)17:47
hrwJeffrey4l: ppc64le is listed in ONE place in whole patch. just because someone started working on it.17:47
hrwaarch64 is mentioned in two. one of them is help text17:48
hrwI would write that patch same way for any other architecture17:48
Jeffrey4lanyway, just give u some advice ;)17:50
hrwJeffrey4l: and -1 too ;D17:50
Jeffrey4lremoved it.17:51
hrwthx17:51
*** adrian_otto1 has quit IRC17:51
hrwJeffrey4l: if I would like to make aarch64 == debian by default I would do that in separate patch as well.17:51
Jeffrey4lok.17:52
*** adrian_otto has joined #openstack-kolla17:53
hrwadding debian support will be interesting...17:54
sdake_portdirect i am transitioning away from lastpass even though all the cool kids are using it and instead transitioning to google passwords since it theoretically should be more secure than the plaintext lastpss stores their databases in and the sloopy technical support nerds who give out passwords to random strangers for non-essential services18:00
bmace_sdake_: if there were major security concerns about lastpass i would be surprised if i wouldn't have read about it on slashdot / reddit.  have any links with relevant data?18:02
sdake_bmace_ just hearsay18:03
jascott1centralized storage of everyones passwords... what could go wrong18:03
sdake_bmace_ i also disliek how it forces all passwords to be saved and doesn't offer a no-save option in the latest upgrade - which i find mandatory for very specific services to which I need my password memoried18:04
sdake_jascott1 right ;)18:04
bmace_sdake_: lol, well, don't believe everything you hear :)  with how many people use it, if there were real concerns i think it would be a little more well known.18:04
*** mannamne has quit IRC18:04
sdake_bmace_ my other concern remains unanswered -and google passwords does permit a "never save" option :)18:04
sdake_bmace_ i'd like to transition off of all these password rememberance services and just have 3 paswords - 1 = critical services, 2 = important services, 3 = junk services18:05
sbezverkjascott1: qq in go struct, is there a way easily to get element label?18:05
sdake_for the 1 category i memorize my passwords18:05
bmace_sdake_: sure.. then if somehow your critical services password is discovered, keylogger, whatever.. they can now get into all your critical services..18:05
jascott1i only stress over ones that dont have recovery18:05
sdake_and use strong passwords as per https://www.usenix.org/system/files/1401_08-12_mickens.pdf18:06
sdake_jascott1 there are things i need secured (e.g. banking, medical records, etc) that i dont want in any centralized password store18:07
portdirectMossad, sdake, doing mossad things...18:07
jascott1i hear ya18:07
sdake_portdirect media FTL :(18:07
sdake_bmace_ you assume i use the same password for each category of #1 :)18:07
jascott1sdake_ element label? you mean a struct property name?18:08
sdake_jascott1 was that a misfire :)18:08
jascott1if its unknown struct prop reflect pg18:08
jascott1pg18:08
jascott1!@#$18:08
openstackjascott1: Error: "@#$" is not a valid command.18:08
jascott1pkg18:08
bmace_sdake_: you said you wanted 3 passwords, so not a terrible assumption on my part :P18:08
jascott1have to agree the 3 passwds sounds risky18:09
portdirectI've not reviewed last pass sec policy, but I remember going meh last time I did. For all my really important things I've a magical amulet and use the super secure combination "Password123"18:09
sdake_bmace_ oh right, well I guess I didn't want to give away all my secrets :)18:09
*** unicell has joined #openstack-kolla18:09
sdake_portdirect really does use Password123 - I've seen it :)18:09
jascott1i like to use my own algorithm18:10
jascott1for pass gen18:10
portdirectDude, don't lie: that was password1234 I gave you18:10
sdake_portdirect oh right - my memory is fuzzy :)18:10
portdirect(It actually was as well)18:10
bmace_the 4 foils the hackers every time.. smart! :)18:10
portdirectYeah, my policy is either "don't care" or your gonna have a really bad time18:11
bmace_even better, make it 125.. a little monty python in there!18:12
*** srwilkers has quit IRC18:12
portdirectbmace_: you know the score :)18:13
jascott1remember to password storage systems can keep your password a secret if one of them doesnt exist18:15
sbezverkjascott1: example type a struct { a string / b int} I would like to be able to get 'a'18:15
sbezverknot it is value but property name not sure how correctly call it18:15
jascott1sbezverk but you are asking ' i would like to know a exists though I dont know what props the struct has' or 'how do I access prop a on my struct'?18:16
jascott1s := &mysruct{a: "asd", b: 1}18:17
jascott1fmt.Printf("a %v", mysruct.a)18:18
sbezverkjascott1: man I want to use "a" as a key in the map to retrive from the map value which need to go into the struct. I do not want to type a: map["a"]18:18
jascott1you want to dynaically name struct properties based on map key?18:18
sbezverkI would like to use map[varA] where varA would have "a" as a value18:18
sbezverkyep18:18
jascott1need codegen maybe18:19
sbezverkoh so there is no known easy solution :(18:19
jascott1idk I havent tried to make struct on the fly18:19
jascott1wouldnt take my word for it18:19
jascott1i will look into it, is there a PS dealing with this i can look at?18:20
sbezverkI would think when you deal with large structures, it would have been already solved ;)18:20
jascott1why does the struct have to be dynamic? its basically a  class in oop parlance would you use a class todo what you are trying to accomplish?18:20
sdake_kfox1111 are you about18:21
krtaylorhrw, just got a chance to read back through the follow-on discussion, I am fine with your proposed approach from what I know so far (yes, I'll add that to my review)18:21
sbezverkjascott1: example when you instantiate struct record18:21
sbezverkif you have 20-30 elements in struct, that would take 30 line18:22
sbezverkwould be more elegant to do it in loop18:22
sdake_kfox1111 i hope you dont mind me taking over that gate patch, i think it will help take some burden off you18:22
jascott1why so many elements?18:22
sdake_kfox1111 and also help me learn how the gate works :)18:22
jascott1sounds like you need a .values that holds a map18:22
sbezverkjascott1: ok thanks I will think about it18:23
sdake_kfox1111 you wlil notice i went thorugh about 10 patchsets before i figured out the gate doesn't use a registry at all18:23
sdake_kfox1111 good way to learn how the software works18:23
openstackgerritMerged openstack/kolla-ansible master: Remove Zaqar doc link in README.rst  https://review.openstack.org/43071418:27
openstackgerritSerguei Bezverkhi proposed openstack/kolla-kubernetes master: Adding compute-kit chart  https://review.openstack.org/42815018:28
openstackgerritMerged openstack/kolla-ansible master: Fix BASEDIR to package Kolla with snapcraft.io  https://review.openstack.org/42670318:30
openstackgerritMerged openstack/kolla-ansible master: Using yaml.safe_load instead of yaml.load  https://review.openstack.org/43053518:31
openstackgerritSteven Dake proposed openstack/kolla-kubernetes master: Pull container tallballs into the gate  https://review.openstack.org/42659818:33
sbezverksdake_: I would appreciate if you could hold with gate modifications. I am almost done with my refactor.18:34
sdake_sbezverk i'll be happy to rebase18:34
sdake_sbezverk my changes are small18:34
sdake_sbezverk i would appreciate reviews however18:34
sbezverksdake_: the code you put in setup_gate is gone in my version18:34
sdake_sbezverk again i'll be happy to rebase18:34
*** rwallner has joined #openstack-kolla18:35
sdake_sbezverk rebase  to me means taking a peice of work and making it work in whatever is in master18:35
*** qwang has quit IRC18:35
*** portdirect has quit IRC18:35
*** robcresswell has quit IRC18:35
*** duonghq__ has quit IRC18:35
*** erlon has quit IRC18:35
sdake_sbezverk would still appreciate reviews :)18:35
sdake_sbezverk +1 is sufficient18:35
sdake_(or -1)18:35
*** sindhu has quit IRC18:35
sdake_irccloud just crashed - roflcopter ;)18:36
rwallnerquestion for the channel, deploying kolla newton branch with cinder+ceph, i noticed cinder-volume gets deployed on every [storage] inventory node, are multiple cinder-volume processes normal or is it using any of the active/active cinder config18:36
sdake_sbezverk you will notice i added many comments to the review when you review it18:36
sdake_sbezverk comments are a thing and help people understand the code :)18:36
sdake_sbezverk if it takes all your brainpower to write it, it takes double that from someone else to understand it18:37
*** SaMnCo has quit IRC18:37
*** ediardo has quit IRC18:37
*** DuncanT has quit IRC18:37
*** macsz has quit IRC18:37
*** akwasnie has quit IRC18:37
*** kong has quit IRC18:37
*** turvey has quit IRC18:37
sdake_wow lots of pplz use irccloud18:40
sdake_sbezverk by commenting, you reduce the brainpower to understand to be equivalent to the brainpower of the author (in theory)18:41
sdake_sbezverk the net effect is the community grows as a result18:41
openstackgerritMichal Jastrzebski (inc0) proposed openstack/kolla-ansible master: Release note for reconfiguration optimizaiton  https://review.openstack.org/43110718:42
*** greenlight_ has joined #openstack-kolla18:43
*** berendt has quit IRC18:45
*** serverascode has quit IRC18:45
*** aimeeu has quit IRC18:45
*** NachoDuck has quit IRC18:45
*** NobodyCam has quit IRC18:45
*** hogepodge has quit IRC18:45
*** tonanhngo has joined #openstack-kolla18:45
sdake_sbezverk note this gate failure: http://logs.openstack.org/08/430708/1/check/gate-kolla-kubernetes-deploy-centos-binary-3-ceph-multi-nv/c7cf47d/console.html18:45
inc0sdake_,yeah, lots of people;)18:45
sdake_it seems as though 3.0.2 image's config is broken with nova18:45
sbezverksdake_: what about it?18:45
sdake_I asked Jeffrey4l to analyze it yesterday to which he did and indicated its a nova configuration problem18:46
sdake_sbezverk any idea where I should begin to sort it out?18:46
*** vhosakot has quit IRC18:46
sbezverksdake_: kfox1111 and I chatted about it earler. it seems keystone url in nova config is broken18:47
sdake_sbezverk cool can you chat about it on the channel so I can read about it :)18:47
*** sambetts is now known as sambetts|afk18:47
*** tonanhngo has quit IRC18:47
sdake_here is the irc log from Jeffrey4l 's analysis: http://eavesdrop.openstack.org/irclogs/%23openstack-kolla/%23openstack-kolla.2017-02-08.log.html#t2017-02-08T07:29:3018:47
sbezverksdake_: it should be in the scroll back18:48
*** tonanhngo has joined #openstack-kolla18:48
sdake_sbezverk ahh ok cool - i guess i missed it this morning - my bad18:48
sdake_this morning was super hectic, hd to drop off vehicle nd get two kids to school on time in a 1 hour window18:48
sdake_note i demoted the trunk gate job to experimental here: https://review.openstack.org/#/c/430704/18:49
sdake_as it is unimplemented18:49
sdake_sbezverk where is your gate rework patch - i'll be happy to review (if I can understand it)18:50
*** vhosakot has joined #openstack-kolla18:50
sdake_sbezverk strugglign to find your gate refactor patch - you mention it in review https://review.openstack.org/#/c/426598/23 however I don't see a link to it nor the change ID18:53
openstackgerritMerged openstack/kolla-ansible master: Use neutron user in neutron_lbaas.conf and overwrite defaults  https://review.openstack.org/42515018:55
bmace_sdake_: for some reason the bash script ceph_workflow_service.sh exits right after the endpoint_test.sh line even though it was successful.. i'm not like a super bash guy.. no clue why that might be happening.18:55
bmace_sdake_: other than that i think the review is there, other than writing up a doc.18:56
openstackgerritMerged openstack/kolla-ansible master: adds ovs-dpdkctl tool  https://review.openstack.org/40887118:56
sdake_bmace_ i'm not really a bash dude either18:56
*** greenlight_ has left #openstack-kolla18:56
openstackgerritBorne Mace proposed openstack/kolla-kubernetes master: WIP: Added support for containerized dev bringup  https://review.openstack.org/42929518:56
sdake_ther eis a way to get debug however18:56
sdake_LMTGFU :)18:56
bmace_sdake_: i had a similar issue with pkill helm stopping the whole script... i just nuked the pkill since we should be starting up in a fresh environment anyway and that fixed it.. i can't really do the same with the endpoint test.. i mean i could, but would rather have it in there as it seems useful.18:57
sdake_let me look at the code first - got a line # in particular18:57
bmace_sdake_: 119 in the latest ceph_workflow_service.sh in my review18:58
bmace_sdake_: no real changes around there or even in that script vs. upstream so... odd18:58
sdake_bmace_ so endpoint_test.sh fails?18:59
sdake_moment - let me look at repo19:00
bmace_sdake_: no, the test succeeds, then the script exits before the next line :/19:00
sbezverksdake_: https://review.openstack.org/#/c/428150/6119:00
*** sayantani01 has quit IRC19:03
bmace_sdake_: hrm.. looks like there is an exit -1 in the method in that script... a couple of exit -1s in there.. will have to look deeper into what this script is really trying to test.19:05
openstackgerritMerged openstack/kolla-ansible master: Modify the version_info in pbr  https://review.openstack.org/42496519:07
sdake_bmace_ ^^19:08
bmace_sdake_: what?  the pbr change?19:08
sdake_moment bmace_19:08
inc0so guys, we have 119 contributors to Ocata Kolla:) even tho Ocata was only 3 or so months with holidays and spring festival19:09
openstackgerritSteven Dake proposed openstack/kolla-kubernetes master: Document the magic in endpoint_test.sh  https://review.openstack.org/43111419:10
sdake_which spring festival19:11
sdake_therer was also the chinese new year :)19:11
inc0spring festival == chinese new year19:11
sdake_bmace_ that should help get you going19:11
sdake_oh i see :)19:11
inc0and we actuyally have more unique reviewers than we did in newton19:11
sdake_bmace_ the issue I think the cp operation isn't actually working19:12
inc0so our numbers looks similar to newtons even tho as I said, it was short and cut off cycle19:12
sdake_inc0 the committers will go down this cycle as well as less people gaming the ATC passes19:12
inc0Pike will be glorious:)19:12
sdake_so I think those numbers are actually better than you think :)19:12
inc0maybe19:12
inc0yeah19:13
sdake_bmace_ did that patch help19:13
inc0well atc gaming folk usually rush commits closer to summit19:13
sdake_inc0 my point is everyone knows there is no atc passes to game19:13
bmace_sdake_: yeah, that may be it.. i don't even see anywhere that sets $WORKSPACE19:13
inc0sdake_, yeah, so people commiting actually care19:13
sdake_bmace_ that gets set in setup_gate.sh19:13
sdake_inc0 people gaming the atc passes care typically19:14
sdake_inc0 they care to make a contribution and make it worthwhile19:14
bmace_i'll take a look at that and maybe do it in setup_dev_env19:14
sdake_inc0 they are just not going to be repeat contributors19:14
inc0sdake_, well what I'm saying reasoning behind cotnributions without atc gaming is only to improve project19:14
inc0which is great19:14
sdake_bmace_ if you like my patch put a +1 on it, the gate needs to be documented19:14
sdake_inc0 agreed, although atc gaming is dual purpose - to improve project and obtain atc pass :)19:15
bmace_sdake_: looks like even setup_gate assumes workspace is defined already19:15
*** saneax is now known as saneax-_-|AFK19:15
bmace_sdake_: how about an optional arg to that script that doesn't do the workspace stuff that really assumes this is being done in the gate?19:16
sdake_bmace_ whatever makes it through the core reviewers works for me :)19:17
bmace_kk19:17
bmace_commented on review.19:17
openstackgerritSerguei Bezverkhi proposed openstack/kolla-kubernetes master: Adding compute-kit chart  https://review.openstack.org/42815019:18
harlowjahave u guys seen http://paste.openstack.org/show/598133/ ?19:19
harlowjapretty sure its the setuptools is messed up issue i've seen, but just wnted to check19:19
harlowjaie https://github.com/pypa/setuptools/issues/94219:19
*** NachoDuck has joined #openstack-kolla19:21
*** rwallner has quit IRC19:22
*** MasterOfBugs has joined #openstack-kolla19:22
openstackgerritSteven Dake proposed openstack/kolla-kubernetes master: Pull container tallballs into the gate  https://review.openstack.org/42659819:23
sbezverkbmace_: wrt WORKSPACE, I think it gets set by the gate when they give us VM19:24
openstackgerritSteven Dake proposed openstack/kolla-kubernetes master: Pull container tallballs into the gate  https://review.openstack.org/42659819:24
*** erlon has joined #openstack-kolla19:25
bmace_sdake_: sbezverk: it looks like the cp isn't even running, since workspace isn't set.. i think the last command return is -1 from the directory check for workspace/logs.  how about just an exit 0 at the end, so if it doesn't exit -1 with a failure you get a nice clean 0 exit at the end?19:26
bmace_i tested it and it seems to work well enough19:26
bmace_not sure if there is a better / cleaner bash way19:26
sdake_bmace_ leave a comment in my review and i'll fix it19:27
sdake_bmace_ i had exit 0 in there originally19:27
sdake_however before I pushed git review i used dd to remove the exit 019:27
bmace_aaah, kk, i'll add the comment and change it locally for testing19:28
sdake_the review is: https://review.openstack.org/43111419:28
bmace_hopefully just one more pass through this process! :)19:28
sbezverkpersonally I do not like -1 as exit, because in bash 0 is good 1 is bad. not sure why -1 came to the picture19:28
sdake_sbezverk no idea - thats what is in master atm19:28
sbezverksdake_: hm I thought it was pushed by you and got merged this morning, no?19:29
sdake_sbezverk no here is the original commit: https://github.com/openstack/kolla-kubernetes/commit/bc9ccf624dc26173f4580df00aab7b0d5eb5085d19:29
sdake_I will change it to 1 in a stacked patch set, indeeed it should not be -1 according to posix standards19:30
sdake_sbezverk i may have done a -1 review this morning tho, and if so, I'll change that to exit 119:30
sdake_one more pass ;)19:31
*** DuncanT has joined #openstack-kolla19:31
*** serverascode has joined #openstack-kolla19:31
sbezverksdake_: well it got merged before I could react.. sorry19:32
openstackgerritSteven Dake proposed openstack/kolla-kubernetes master: Remove all -1 from setup_gate.sh  https://review.openstack.org/43111919:32
sdake_sbezverk that happens19:32
sdake_sbezverk and there is a whole new patch to fix it :)19:32
sdake_sbezverk we don't need each patch to be totally perfect19:32
sdake_sbezverk iterate iterate iterate :)19:33
sdake_optimizing for perfection is a bad practice imo :)19:33
sdake_we should be optimizing for velocity19:33
sdake_through velocity perfection comes naturally19:33
*** DuncanT has quit IRC19:35
*** fragatina has quit IRC19:37
*** fragatina has joined #openstack-kolla19:37
openstackgerritMerged openstack/kolla-ansible master: Optimize reconfiguration for elasticsearch  https://review.openstack.org/42413919:38
*** skuda has quit IRC19:40
hrwkrtaylor: thx19:40
openstackgerritSteven Dake proposed openstack/kolla-kubernetes master: Document the magic in endpoint_test.sh  https://review.openstack.org/43111419:44
openstackgerritSteven Dake proposed openstack/kolla-kubernetes master: Make endpoint_test return 1 instead of -1 on fail  https://review.openstack.org/43112619:44
*** krtaylor has quit IRC19:45
*** haplo37 has quit IRC19:46
*** haplo37_ is now known as haplo3719:46
sdake_bmace_ you should add a dependson to sbezverk 'z gate rework patch19:47
sdake_bmace_ since your review is blocked anyway as a result of the kubernetes dependency19:47
*** sacharya has quit IRC19:47
*** salv-orlando has quit IRC19:48
*** sacharya has joined #openstack-kolla19:48
sdake_hpoefully sbezverk 'z review wont sit aorund forever if he splits out the gate changes which are hard to digest from the compute kit19:48
*** imcsk8_ has quit IRC19:51
sbezverksdake_: wow 3 days is forever for you ;) I suspect is hard to live with such high expections19:51
*** imcsk8 has joined #openstack-kolla19:51
*** DuncanT has joined #openstack-kolla19:51
sdake_sbezverk how do you mean19:52
*** Pavo has joined #openstack-kolla19:53
*** qwang has joined #openstack-kolla19:53
sdake_sbezverk 3 days of blocking 6 other patches is a problem19:53
sdake_sbezverk those kinds of changes need to be fast tracked19:53
sdake_sbezverk that results in compounding rebases which apparently people don't like to do :)19:53
sbezverksdake_: well when we talk about changes the gate, 3 days is not much.. I would agree with you if it took me 3 days to change a comment or spell or something of that sort19:54
sdake_sbezverk i'm not saying your slow19:54
sdake_i'm say8ing the review process is slow19:54
*** tonanhngo has quit IRC19:54
sdake_sbezverk in fact your quite fast :)19:54
sdake_i on the other hand am slow to attend to patches19:55
sbezverksdake_: if you'd examin that review carefully, you would see that compute part actually is not changing and is very small.19:55
sdake_and can only really work in short spurts swhen I have free time available - so don't like to sit around being blocked -because saturday i will be prepping for a 2 hour seminar on kolla19:55
sbezverkso if you need to merge, it can be done now19:55
sbezverkas the main gate part is workign19:55
*** robcresswell has joined #openstack-kolla19:55
sdake_sbezverk right - patches should be self-contained units not massive things19:55
sbezverkI am just making compute kit to work in iscsi env..19:56
sdake_need to run for a bit19:56
*** sacharya_ has joined #openstack-kolla19:56
sdake_sbezverk cool - just shard off the gate refactor so other people can get unblocked :)19:56
*** portdirect has joined #openstack-kolla19:57
sdake_sup portdirect19:57
sdake_i see irccloud is back online :)19:57
*** haplo37_ has joined #openstack-kolla19:57
*** david-lyle has quit IRC19:57
*** sacharya has quit IRC19:57
sdake_bmace_ your goin g to have to rebase your patch on top of sbezverk's work - along with the other 6 gate patches pending19:58
sdake_ok - bbifew hours19:58
*** rwallner has joined #openstack-kolla19:58
*** kong has joined #openstack-kolla19:59
bmace_sdake_: yup, well, waiting on kubernetes to be allowed anyway, so i expected some merges would be needed.19:59
*** akwasnie has joined #openstack-kolla20:00
*** NobodyCam has joined #openstack-kolla20:02
*** aimeeu has joined #openstack-kolla20:02
klindgrenIs it on the road map to run all api's in the same manner?  Like instead of some under eventlet and other under apache?20:05
klindgrenwe were working on this internally but using uswgi instead of apache.20:05
openstackgerritMarcin Juszkiewicz proposed openstack/kolla master: support non-x86 architectures (aarch64, ppc64le)  https://review.openstack.org/43094020:07
Pavoquestion, how can I do ovs-vsctl cmds with using kolla-ansible?20:07
*** mannamne has joined #openstack-kolla20:07
*** tonanhngo has joined #openstack-kolla20:08
openstackgerritSerguei Bezverkhi proposed openstack/kolla-kubernetes master: Gate refactoring  https://review.openstack.org/42815020:10
*** macsz has joined #openstack-kolla20:11
sbezverksdake_: here is gate refactor PS20:11
sbezverkI removed compute kit for now to get it merged so it will not block other folks..20:11
*** salv-orlando has joined #openstack-kolla20:11
*** ediardo has joined #openstack-kolla20:12
*** SaMnCo has joined #openstack-kolla20:13
openstackgerritSerguei Bezverkhi proposed openstack/kolla-kubernetes master: Gate refactoring  https://review.openstack.org/42815020:14
*** david-lyle has joined #openstack-kolla20:15
*** berendt has joined #openstack-kolla20:15
*** jtriley_ has joined #openstack-kolla20:16
*** sindhu has joined #openstack-kolla20:16
*** turvey has joined #openstack-kolla20:18
*** hogepodge has joined #openstack-kolla20:19
*** Pavo has quit IRC20:19
*** jtriley has quit IRC20:20
*** david-lyle has quit IRC20:20
*** mgkwill has quit IRC20:21
*** mgkwill has joined #openstack-kolla20:21
*** Pavo has joined #openstack-kolla20:21
*** jtriley_ has quit IRC20:24
*** fragatina has quit IRC20:27
*** krtaylor has joined #openstack-kolla20:28
*** matrohon has joined #openstack-kolla20:28
openstackgerritSerguei Bezverkhi proposed openstack/kolla-kubernetes master: Gate refactoring  https://review.openstack.org/42815020:30
openstackgerritSerguei Bezverkhi proposed openstack/kolla-kubernetes master: Adding compute kit based on iscsi backend  https://review.openstack.org/43115620:30
openstackgerritSteven Dake proposed openstack/kolla-kubernetes master: Change endpoint-test -1 to 1 return code  https://review.openstack.org/43112620:32
openstackgerritSteven Dake proposed openstack/kolla-kubernetes master: Document the magic in endpoint_test.sh  https://review.openstack.org/43111420:32
*** jtriley has joined #openstack-kolla20:32
sdake_thanks sbezverk - you left the release note in - but i'll do a thorough review after my meeting in 30 mins20:33
*** pcaruana has quit IRC20:43
openstackgerritSerguei Bezverkhi proposed openstack/kolla-kubernetes master: Adding compute kit based on iscsi backend  https://review.openstack.org/43115620:46
*** Pavo has quit IRC20:47
*** penfold has quit IRC20:48
bmace_sbezverk: i'm curious, in what environment does --image-feature layering actually work, because if you don't pass in the arg "2" and have that arg set i get an error about the argument being invalid or no longer supported or some such20:49
*** crushil has quit IRC20:50
*** sdake has joined #openstack-kolla20:54
kfox1111bmace_: kolla 2.x doesn't support the flag, and 3.x requires it.20:54
kfox1111really, its a ceph hammer vs jewel thing.20:55
SamYapleyou can set that in ceph.conf20:55
bmace_kfox1111: thanks much.20:56
kfox1111np20:56
kfox1111SamYaple: yeah. that woudl be another way to solve it.20:56
*** salv-orlando has quit IRC20:56
*** sdake_ has quit IRC20:58
sbezverkbmace_: when we depricate hammer ;) if it happenes, that part can be removed20:59
*** sp_ has quit IRC21:01
*** sacharya_ has quit IRC21:05
bmace_sbezverk: yup yup, just need to tweak my review a bit and pass in the $BRANCH like the gate script does.21:06
*** sacharya has joined #openstack-kolla21:06
*** mannamne has quit IRC21:09
*** mannamne has joined #openstack-kolla21:10
*** rwallner has quit IRC21:12
*** rwallner has joined #openstack-kolla21:13
*** rwallner has quit IRC21:13
*** rwallner has joined #openstack-kolla21:14
*** rwallner has quit IRC21:14
*** mannamne has quit IRC21:20
*** matrohon has quit IRC21:22
*** rwallner has joined #openstack-kolla21:22
inc0sbezverk, that's why I'm saying that kolla-k8s should use up to date images;)21:25
*** dave-mccowan has quit IRC21:26
*** rwallner has quit IRC21:27
sbezverkinc0: agree :) it is just up date images has tendency to be more broken ;)21:27
inc0stable/newton don't21:28
inc0stable/ocata will be available in few days21:28
inc0and well...cross repo gating21:28
*** sacharya_ has joined #openstack-kolla21:29
*** adrian_otto has quit IRC21:29
*** sacharya has quit IRC21:29
*** Jeffrey4l has quit IRC21:34
*** Jeffrey4l has joined #openstack-kolla21:34
*** penfold has joined #openstack-kolla21:38
*** salv-orlando has joined #openstack-kolla21:40
*** vhosakot has quit IRC21:46
*** dave-mccowan has joined #openstack-kolla21:46
*** vhosakot has joined #openstack-kolla21:46
*** vhosakot has quit IRC21:47
*** adrian_otto has joined #openstack-kolla21:47
*** vhosakot has joined #openstack-kolla21:48
portdirectI'd love it if we used master tbh, that way as we add them we can use them - though I'd also love it if we could start culling some for upstream's here and there (looking at you memcached for some ults low hanging fruit)21:48
inc0portdirect, yisss, when I proposed this as ptg session https://etherpad.openstack.org/p/kolla-pike-ptg-schedule #721:50
inc0you can put your support there;)21:51
inc0for this session21:51
* inc0 really wants kolla to limit focus just to OpenStack as much as it's reasonable21:51
portdirectawesome - cheers bro :)21:53
*** fguillot has quit IRC21:54
*** skuda has joined #openstack-kolla21:59
*** vhosakot has quit IRC22:01
*** vhosakot has joined #openstack-kolla22:01
*** sdake has quit IRC22:02
*** Pavo has joined #openstack-kolla22:02
*** sdake has joined #openstack-kolla22:02
sdakeportdirect inc0 in compelte agreemenet wrt adding master to the gate22:02
sdakeportdirect inc0 after sbezverk 's patch merges, I'll take al ook at getting a gate setup for that22:03
sdakesince bmace_ has the dev env wrapped up - i'm out of things to do :)22:03
sdakeinc0 what i do need is some wayt o get the master images into tarballs.oo22:03
inc0they're there22:03
inc0they're being uploaded every commit22:04
inc0to kolla22:04
sdakeok - i could pull them from there22:04
inc0not that easily22:04
inc0http://tarballs.openstack.org/kolla/images/22:04
inc0you download tar.gz22:04
sdakeinc0 i check this out:22:04
inc0untar it22:04
inc0and then docker run registry -v /var/lib/registry:/path/to/untar22:05
inc0and that's it22:05
sdakehttps://review.openstack.org/#/c/426598/22:05
sdakeicn0 that wont work with kubernetes22:05
sdakeyou have to use docker load22:05
sdakebut its basically teh same concept22:05
inc0kubernetes knows how to do registry...22:05
sdakeright - althoug hthe gate code at present doesn't use a registry22:05
sdakeit coudl be made to do so - baby steps first, would liek to get those tarballs being used22:06
inc0https://review.openstack.org/#/c/418550/ this used to work...22:06
sdakethat patch was wroking prior to merge conflict22:06
portdirectoff the top of my head adding a registry would be the easiest way to do that - unless setting a custom image namespace is still a hassle22:06
inc0shouldn't be22:07
sdakei've got a  working review that uses docker load based upon kfox1111 'z work (I just reworked it slightly and learned about the gate)22:08
sdakeafter that I could add a registry22:08
sdakethe issue is multinode22:08
sdakeand sorting that out with a registry is semi-irritating22:08
inc0you can run registry on every node if that's the issue22:08
sdakedocker load is easy first step22:08
inc0and always pull 127.0.0.122:08
sdakedocker load loads the images into the cache22:08
sdakeso it works prefectly22:08
inc0no it's not unless you're willing to break our build gates byu taring every-single-image22:08
inc0and uploading them22:09
inc0with no regard to layering22:09
inc0so full size image every time22:09
sdakehrm, does docker load ignore layering?22:09
*** vhosakot_ has joined #openstack-kolla22:09
*** p6 has joined #openstack-kolla22:09
inc0indeed it does22:09
sdakeok well let me get the gate stablized at first22:10
sdakethen ill sort out a registry next22:10
sdakei know how to do the job now (pre sbezverk 's rebase)22:10
sdakerather pre sbezverk 's refactor of the gate22:10
p6interesting question- doing a lab deployment test of kolla here, keepalived won't bring up the internal VIP IP because haproxy is failed, haproxy won't start because it can't bind to the VIP...22:10
*** vhosakot_ has quit IRC22:10
openstackgerritMerged openstack/kolla master: Add redis container  https://review.openstack.org/43097622:10
inc0p6, keepalived shouldn't fail because of haproxy22:11
inc0what's actual error?22:11
portdirectnice - if you need a hand with k8s multinode registry give me a holler - we may be able to do it super easy with a single k8s service, as thats what a few of my clusters are running atm22:11
*** vhosakot_ has joined #openstack-kolla22:11
inc0(thing is keepalived deploy is part of haproxy role, so that might be misleading)22:11
openstackgerritMerged openstack/kolla master: Add openstack-magnum-ui package for centos binary  https://review.openstack.org/43056722:11
p6inc0: VRRP_Instance(kolla_internal_vip_51) Entering Backup, then almost immediately after that- entering FAULT22:11
p6i'm presuming FAULT is because the check failed, the actual interface looks up from the container22:12
sdakeinc0 also the problem with limiting focus to just openstack is that we break the ability to have cross-distro and cross-arch support22:12
inc0p6 keepalived doesn't check anything with haproxy22:12
inc0sdake, which I am strongly considering22:12
inc0as long as we enable people to override default image22:12
*** vhosakot has quit IRC22:13
inc0if someone cares about centos mariadb image and it's not available22:13
inc0I'd be happy if this person upload it, or maybe rht could?22:13
sdakeas i said in the review - i am -2 on that technical approach22:13
inc0thing is, I don't think *we* need to22:13
inc0as you said to me on multiple occasions, you can't -2 specs;)22:13
sdakeright - id be -1 on a rollcallvote then :)22:13
sdakeand I think you wuld struggle to find consensus but welcome to try22:14
inc0fair enough, but don't worry, I want to have lenghty conversation22:14
inc0I just think our model is impossible to maintain22:14
sdakeits being maintained just fine now22:14
inc0on a long run22:14
p6inc0: it has vrrp_script /check_alive defined which is doing a show info against /var/lib/kolla/haproxy/haproxy.sock22:14
sdakewith no struggle22:14
*** vhosakot_ has quit IRC22:14
inc0sdake, ubuntu was broken for 2 months22:14
sdakeinc0 that is because nobody cares about ubuntu and the gating is weak22:15
inc0and I have no idea if our rabbitmq doesn't miss any security patch or whatnot22:15
portdirectyeah - the only people i see really caring about that are RDO/RH - and we could easily help them put the framework in place to take on that tech debt22:15
inc0well "nobody cares" is a bit of a strong given it's one of 3 main distros we support22:15
sdakeperhaps22:15
sdakethe fat it was broken in master for 2 months shows nobody tests it22:16
inc0also, rabbitmq is equally true to centos22:16
sdakethe solution is gating22:16
portdirector uses it :(22:16
inc0portdirect, people use it22:16
inc0just not necessarily main body of contributors22:16
sdakeportdirect most of the community uses centos not ubuntu22:16
inc0it was broken in master22:16
*** vhosakot has joined #openstack-kolla22:16
inc060% of operational openstack in the wild runs on ubuntu22:16
sdakeanyway sounds like a fine ptg debate right before dinner :)22:16
inc0according to survey22:16
sdakeso we can get drunk afterwards ;-)22:17
portdirectyeah - thats a fair point inc0, we need to find a way to close that feedback loop22:17
* portdirect typing lagg :(22:17
inc0portdirect, true, and true that we need gating22:17
*** jtriley has quit IRC22:17
inc0but no gating ever will ensure that we have correct versions of stuff22:17
sdakeif we had effective gating ubuntu maintenance would be no sewat22:17
inc0and not miss some security updates somewhere22:17
sdakeand jeffrey4l is working on that with others22:17
sdakei'd trust a distro such as red hat prior to trusting an upstream such as mariadb for maintenance22:18
sdakered hat's bread and butter is making sure mariadb and rabbitmq work well22:18
inc0remember issues we had with rabbitmq versions?22:18
sdakeyes because we were using the rabbitmq from rabbitmq, rather then the  rabbitmq from RDO :)22:19
sdakethat got fixed22:19
sdakered hat has 30+ people maintaing rabbitmq and mariadb just for their distro22:19
portdirectsdake: i suppose that's where we differ (and drinks need to be involved in the resolution...) I agree in a package world. but in a container world I kinda trust the project more than a packager22:19
sdakethat amplifies the effect that the mariadb contribs have22:19
sdakeportdirect knowing how the sausage is made at rht, and knowing how the sausage is made upstream, the distros do add significant value even if they give it away for free22:20
sdakeatleast the rht case22:20
sdakenot sure about cononcial22:20
sdakei think their maintenacei sk ind of laggy adn probably where we have the disconnect22:20
inc0sdake, then let rht support images22:21
sdakethe questoin is will they22:21
sdakeif they wont then we have to do so22:21
portdirectsdake: and *this* is why we need shots involved as we fire shots at each other ;)22:21
sdakewe are making inroads there bu twe aren't there yet22:21
inc0if image is supported by someone even outside kolla22:21
sdakeportdirect only if its rum :)22:21
inc0(and that can be our community as people)22:21
inc0it can be used and provide feedback by wider audience22:22
sdakei really dont see what that change buys us22:22
inc0there is no value in having images for that stuff in kolla imho22:22
sdakebut i've got real work to do - we can discuss at ptg22:22
sdakei would agree with your position if red hat and cononccal agreeed to maintain containers to our standards22:23
sdake(and oracle)22:23
inc0dmsimard, ^ ideas?22:23
sdakeI doubt we can get all 3 of those bohemeths to do that22:23
portdirectRH will soon i think22:23
sdakewhat woudl happen is they would maintain them to *their* standards22:23
inc0you think you guys would be willing to run containers for stuff like mariadb, rabbitmq and other non-openstack services?22:23
inc0sdake, we *just* need packages in container22:24
inc0other than that, we can handle this with layer above them22:24
sdakerabbitmq has a huge dep chain on erlang22:24
inc0all I need is docker image with every package service needs22:24
sdakeerlang has a huge mainteance team at red hat22:24
sdakerelying on pivotal to maintain that is painful22:24
sdakewe tried that and it led to near disaster22:25
inc0that's exactly why we can't hope to keep up with them22:25
sdakeinstead we rely on rdo - and they do a good job22:25
sdakeanway - real work :) bbl22:25
inc0only we don't really https://github.com/openstack/kolla/blob/master/docker/rabbitmq/Dockerfile.j2#L4722:26
*** rhallisey has quit IRC22:26
*** fragatina has joined #openstack-kolla22:30
*** gfidente|afk has quit IRC22:31
*** dave-mccowan has quit IRC22:31
*** sdake_ has joined #openstack-kolla22:32
dmsimardinc0: need to give me a tl;dr please22:32
inc0dmsimard, so, we are thinking of loosing support for non-openstack services like mariadb22:33
inc0but if we'd like to keep all centos stuff...well...centos22:33
inc0that would require us to have access to centos-based mariadb container somewhere...dockerhub22:34
inc0wouldn't rht be interested to host images like that?22:34
inc0a) rh stuff would get more exposure as people would just install mariadb with centos base22:34
inc0not only openstack people, anyone wanting mariadb22:34
*** erlon has quit IRC22:35
inc0b) it would help us drop this code from our repo, which some people, including myself, would like to see one day22:35
*** vhosakot has quit IRC22:35
inc0not to mention openshift could make use of these things22:35
*** sdake has quit IRC22:36
*** jheroux has quit IRC22:36
dmsimardSo basically instead of having kolla build a rabbitmq container, a mariadb container, etc.. use whatever is on dockerhub ?22:39
portdirectby default yes - though pinned to a version22:39
inc0dmsimard, poc of what I'd do https://review.openstack.org/#/c/427931/22:39
inc0portdirect, ^22:40
portdirectinc), I'd love it if we could get rid of even that, though I understand that for ansible that wouldn't be possible22:41
inc0portdirect, well there are issues this solves even in k8s layer22:41
inc0like init to reap properly or uid fixing for upgrades22:41
dmsimardI don't have a strong opinion, but I can ask people that might22:42
inc0dmsimard, I'm not asking for answers, just throwing it out there22:42
inc0idea that lots of people could benefit from22:42
portdirectyeah - this is deff one to mull over for a bit :)22:43
dmsimardas far as I'm concerned, I'd at least make it configurable so that we can pull the mariadb container from another source22:45
dmsimardI have to take off, will you start a ML thread so that people can chime in ?22:46
inc0dmsimard, we'll discuss this on ptg, and source of mariadb will be configurable22:47
dmsimarddiscuss this on openstack-dev too :)22:48
inc0yeha probably too, just want to get some initial thoughts22:48
*** tvignaud has joined #openstack-kolla22:48
*** rwallner has joined #openstack-kolla22:49
*** inc0 has quit IRC22:52
*** huzhengchuan has joined #openstack-kolla22:52
*** rwallner has quit IRC22:53
bmace_any of the kolla-k8s folks have an issue w/ running branch 3.0.2 by default in devenv?  if someone wants to run 2.0.2 they could always tweak the code and rebuild the container.22:53
*** mannamne has joined #openstack-kolla22:55
portdirectbmace_: I'l like us to be as close to the tip as possible for kolla-k8s dev as possible22:59
portdirectif you can get 3.0.2 working then that gets a +2 from me23:00
bmace_portdirect: yup, going to try it out and see what breaks23:00
*** skuda has quit IRC23:03
openstackgerritSerguei Bezverkhi proposed openstack/kolla-kubernetes master: Gate refactoring  https://review.openstack.org/42815023:04
*** mannamne has quit IRC23:11
*** tonanhngo has quit IRC23:12
*** tvignaud has quit IRC23:13
*** satyar has quit IRC23:15
*** crushil has joined #openstack-kolla23:19
*** goldyfruit has quit IRC23:22
openstackgerritSerguei Bezverkhi proposed openstack/kolla-kubernetes master: Gate refactoring  https://review.openstack.org/42815023:22
*** krtaylor has quit IRC23:25
*** huzhengchuan has quit IRC23:26
*** jrobinson has joined #openstack-kolla23:26
*** fragatina has quit IRC23:29
*** fragatina has joined #openstack-kolla23:30
*** zhurong has joined #openstack-kolla23:30
*** jrobinson has quit IRC23:30
*** jrobinson has joined #openstack-kolla23:32
*** schwicht has quit IRC23:32
*** zhurong has quit IRC23:36
sdake_@bmace_ 3.0.2 from earlier converstions with sbezverk is busted in the gate so its likely busted outside the gate :)23:56
bmace_sdake_: ah... what about it is busted?23:56
sdake_@sbezverk said the endpoint isn't registered correctly23:56
sdake_for nova23:56
sdake_sbezverk says its in the scrollback althogh i don't see it in the archives and didn't persoanlly witness the conversation :)23:58
sdake_that needs fixing sap tho23:59

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