| sdake | sbezverk idea is to make a common pattern thta people can copy | 00:00 |
|---|---|---|
| sdake | sbezverk so we can get an army of contribs to make microservices just as in done in kolla repo today | 00:00 |
| portdirect | ^^ this would be great | 00:00 |
| sdake | rather packaged microservices | 00:00 |
| sdake | the open source method doesn't f ind the optimal soltuion initially it finds it eventually | 00:01 |
| sdake | jrich523 write that one down ;) | 00:01 |
| sbezverk | sdake: portdirect: we need to confirm the directory structure. so we DO follow native helm's directory structure, right? | 00:01 |
| sdake | sbezverk thats what my patch is going to be doing | 00:01 |
| *** TxGirlGeek has quit IRC | 00:02 | |
| sbezverk | sdake: so parent level is kolla-kubernetes or openstack? | 00:02 |
| sdake | you mean directory structure? | 00:02 |
| sbezverk | sdake: yes | 00:03 |
| sdake | you tell me | 00:03 |
| portdirect | sbezverk: thats what I'm going for as well | 00:03 |
| jascott1 | it currently looks like this in sbezverk's PS | 00:03 |
| jascott1 | helm/kolla-kubernetes/charts/openvswitch/charts/openvswitch-ovsdb/ | 00:03 |
| jrich523 | whoa what now? | 00:03 |
| *** yuanying_ has joined #openstack-kolla | 00:03 | |
| sbezverk | I would suggest to keep: ./helm/kolla-kubernetes/charts/service-1/charts/microservice-1/charts | 00:04 |
| sdake | jrich523 nm - conversation from 2 hrs ago :) | 00:04 |
| sdake | portdirect thoughts? | 00:04 |
| sbezverk | jascott1: :-) we are in sync then.. | 00:04 |
| jascott1 | sdake was droppin knowledge | 00:04 |
| jascott1 | :) | 00:04 |
| sdake | jascott1 hard earned | 00:04 |
| jrich523 | lol its ok, something interesting i should know about? :) | 00:04 |
| jascott1 | "the open source method doesn't f ind the optimal soltuion initially it finds it eventually" | 00:05 |
| jrich523 | lol was just reading that | 00:05 |
| portdirect | im fine with sbezverk's suggestion | 00:05 |
| jrich523 | was going to ask if i should be quoting sdake when i use it or someone else? :) | 00:05 |
| sdake | those are my words | 00:05 |
| jascott1 | i like | 00:05 |
| jascott1 | are there any corner cases with dir structure? | 00:06 |
| * jrich523 saves that one | 00:06 | |
| portdirect | jascott1: almost certainly - though I'm not sure what yet :) | 00:08 |
| jascott1 | staring at it but cant see any now :) | 00:08 |
| *** kristian__ has joined #openstack-kolla | 00:11 | |
| jascott1 | oh hey installing w kubeadm http://paste.openstack.org/show/590729/ <- legit? | 00:12 |
| sdake | fwiw from pervious experience with kolla docker files git mv of directory is easy git mv of files is hard | 00:12 |
| sdake | if there are corner cases its easy to solve with a restructure without a bunch of pain | 00:12 |
| portdirect | jascott1: looks ok | 00:14 |
| jascott1 | thx | 00:14 |
| portdirect | jascott1: https://github.com/att-comdev/halcyon-kubernetes/tree/master/kube-deploy/roles | 00:14 |
| sbezverk | sdake: portdirect: where you want to store global values which is applicable to all microservices? | 00:15 |
| *** kristian__ has quit IRC | 00:15 | |
| sbezverk | things like kolla_tag, kolla_base etc?? | 00:15 |
| portdirect | sbezverk: I'm open to suggestions - where do you think we should have 'em? | 00:16 |
| sdake | do you know how the ascii character set was created | 00:16 |
| sbezverk | portdirect: the only way I could see it now if we go per microservice approach, we need to copy them into values.yaml for each microservice | 00:17 |
| sdake | sbezverk there ya go thats how the ascii character set was created ;) | 00:17 |
| sdake | common DRY = o3 | 00:18 |
| sdake | we can easily make a tool to handle that in the meantime | 00:18 |
| sdake | since we are making all kinds of tools to workaround helm defeciencies | 00:18 |
| portdirect | ok - finish each microservices yaml with @@COMMON@@ and then find/sed it in? | 00:19 |
| sdake | nope that won't compile with helm natively | 00:20 |
| sdake | goal is helm native | 00:20 |
| portdirect | WET is is then | 00:21 |
| sdake | kfox1111 when you appear - you wlll want to read the above ^ | 00:21 |
| sbezverk | sdake: I bet kfox1111 will give -3 on this ps ;-) | 00:23 |
| sdake | this is how the docker containers were developed | 00:23 |
| sdake | it led to a great outcome | 00:24 |
| sdake | it will be vastly sped up this time because we wont have to jerk around and figure out what to do | 00:24 |
| sdake | i already know what to do :) | 00:24 |
| sdake | following same pattern as docker file creation during ramp up of kolla repo | 00:24 |
| sdake | its unfortunate you fellas weren't around for that | 00:25 |
| sdake | however the rest of our community is up to speed on the methodology | 00:26 |
| sdake | just have to take my word for it this will roll well - if not do your own thing :) | 00:26 |
| sdake | jascott1 you joining the party - if so recommend portdirect get you ramped up on the dev env | 00:28 |
| sdake | took me about 45 minutes to get there | 00:29 |
| portdirect | sdake: it working ok for you know? | 00:29 |
| jascott1 | def joining the party | 00:29 |
| *** JRobinson__ is now known as JRobinson__afk | 00:29 | |
| sdake | portdirect seems to be | 00:29 |
| jascott1 | definitely joining that is | 00:30 |
| portdirect | awesome :) what TZ you in? | 00:30 |
| jascott1 | PST | 00:30 |
| *** hfu has joined #openstack-kolla | 00:30 | |
| sdake | portdirect if you guys setup a hangout lemme know i'll join again so i can relive the experience ;) | 00:34 |
| *** jtriley has joined #openstack-kolla | 00:35 | |
| portdirect | np, we should loop inc0 in as well | 00:36 |
| *** yingjun has joined #openstack-kolla | 00:38 | |
| jascott1 | might not be around hes in CST | 00:39 |
| *** jtriley has quit IRC | 00:40 | |
| sdake | portdirect 1:1 is better | 00:40 |
| sdake | portdirect keeps everyone in sync :) | 00:40 |
| sdake | and ya inc0 is not feeling well today i gathered from whta he said this morning in channel | 00:41 |
| sdake | i dont think he will feel well this week was my impression he said soething like "ok guys cut me some slack this week as I dont feel well" or something | 00:41 |
| jascott1 | we are talking about the vagrant one right? | 00:43 |
| sdake | thats the one | 00:43 |
| jascott1 | I had already started the non-vagrant one | 00:43 |
| jascott1 | that was linked to | 00:43 |
| *** vhosakot_ has quit IRC | 00:43 | |
| sbezverk | sdake: I prepped one microservice, just to make sure everybody agrees on approach, I will push it to the queue | 00:43 |
| sdake | sounds good | 00:44 |
| portdirect | anyone here a ceph guru - getting a bit confused about OSD_ID? | 00:45 |
| portdirect | cant work out how the osd container gets it on the 2nd (or subseqent start) https://github.com/openstack/kolla/blob/master/docker/ceph/ceph-osd/extend_start.sh#L30 | 00:46 |
| openstackgerrit | Serguei Bezverkhi proposed openstack/kolla-kubernetes: WIP adding openvswitch-ovsdb microservice https://review.openstack.org/403959 | 00:47 |
| sdake | portdirect no idea how that stuff works - kfox1111 has a good handle on it | 00:47 |
| sdake | sbezverk may as well - pretty sure he demoed it :) | 00:47 |
| sbezverk | portdirect: sdake: it is wip, I will clean it up more tonight.. before doing that I would like to get ok from the community | 00:48 |
| sbezverk | specifically from kfox1111 | 00:48 |
| sdake | sbezverk might as well make it pretty | 00:49 |
| sdake | sbezverk qq, does this follow standard layouts expected of helm? | 00:49 |
| sbezverk | sdake: yes | 00:49 |
| Jeffrey4l | sdake, re gnocchi + ceilometer, it works. i tested it. | 00:50 |
| sdake | Jeffrey4l roger | 00:50 |
| sbezverk | so when cleaned up you should be able to deploy it just by running helm install openvswitch-ovsdb | 00:50 |
| *** jtriley has joined #openstack-kolla | 00:50 | |
| sdake | can you get it to that point so i can test it in my env | 00:50 |
| sbezverk | I will do it, but it is 8pm here so I will concentrator on it mostly later tonight and tomorrow early morning est | 00:51 |
| sdake | where do i run helm install openvswitch-ovsdb from? | 00:51 |
| sdake | sbezverk i need something to copy :) | 00:51 |
| *** JRobinson__afk is now known as JRobinson__ | 00:52 | |
| sbezverk | from openvswitch-ovsdb folder | 00:52 |
| sdake | will it deploy as is? | 00:52 |
| Jeffrey4l | sdake, any issue report? | 00:52 |
| sdake | Jeffrey4l its in today's log - something about continer not starting | 00:52 |
| sbezverk | sdake: yes | 00:52 |
| sdake | didn't have time to dig into it with him | 00:52 |
| sdake | sbezverk good enuf | 00:53 |
| sdake | dec20, we ogt a whole lot of microservices to crank out | 00:54 |
| openstackgerrit | Serguei Bezverkhi proposed openstack/kolla-kubernetes: WIP adding openvswitch-ovsdb microservice https://review.openstack.org/403959 | 00:54 |
| Jeffrey4l | sdake, could u post it out? | 00:55 |
| sdake | Jeffrey4l have to check logs i dont hve any data | 00:56 |
| *** dave-mccowan has quit IRC | 00:56 | |
| Jeffrey4l | sdake, ok. i will try to deploy it later. master branch ,right? | 00:57 |
| sdake | Jeffrey4l no clue - no time to investigate or question what he was using | 00:58 |
| *** Pavo has quit IRC | 00:58 | |
| *** zhubingbing has joined #openstack-kolla | 00:58 | |
| *** tonanhngo has quit IRC | 01:01 | |
| *** jtriley has quit IRC | 01:01 | |
| portdirect | Jeffrey4l: you dont have any insight into my ceph q above do you? | 01:02 |
| *** jtriley has joined #openstack-kolla | 01:02 | |
| jascott1 | haha got that vagrant based k8s going np but then busybox is doing the same thing | 01:03 |
| Jeffrey4l | portdirect, yep. | 01:03 |
| Jeffrey4l | tldr; did u have logs and error message? portdirect | 01:03 |
| *** Pavo has joined #openstack-kolla | 01:03 | |
| portdirect | jascott1: I'll be with you in two mins - could you past the dump of 'kubectl get --all--namespaces pods'? | 01:04 |
| jascott1 | well I think I have it now but will to be sure | 01:04 |
| portdirect | Jeffrey4l: working on kollo-k8s helm stuff and I cant work out how the osd container gets the OSD_ID on the 2nd (or subseqent start) https://github.com/openstack/kolla/blob/master/docker/ceph/ceph-osd/extend_start.sh#L30 | 01:05 |
| Jeffrey4l | portdirect, what did u get when run `ceph osd create`? | 01:06 |
| portdirect | It increments each time | 01:06 |
| Jeffrey4l | portdirect, yes. that how ceph osd create works. | 01:06 |
| jascott1 | portdirect with vagrant its looking good http://paste.openstack.org/show/590734/ | 01:07 |
| openstackgerrit | Serguei Bezverkhi proposed openstack/kolla-kubernetes: WIP adding openvswitch-ovsdb microservice https://review.openstack.org/403959 | 01:07 |
| sdake | sbezverk i'm curious about one thing, can we have a common file for the common code and then just copy that around? | 01:07 |
| sbezverk | sdake: so the latest revision os PS gets rendered by helm correctly | 01:08 |
| portdirect | jascott1: from the busybox pod if you can run 'nslookup kubernetes' then you are good to go :) | 01:08 |
| jascott1 | yes! | 01:08 |
| portdirect | jascott1: the cni driver can take a bit of time to come online, during whihc pod creation will fail | 01:08 |
| jascott1 | it seems to be working | 01:09 |
| portdirect | :D | 01:09 |
| jascott1 | thanks! | 01:09 |
| sbezverk | sdake: wrt your question if values.yaml has ALL variables for all global and per microservice variables then yes | 01:09 |
| *** tonanhngo has joined #openstack-kolla | 01:10 | |
| sdake | sbezverk hmm let me look at your last patch set - i thought there was a common.tpl file somwhere in the mix | 01:10 |
| *** tovin07 has joined #openstack-kolla | 01:10 | |
| *** schwicht has quit IRC | 01:10 | |
| portdirect | Jeffrey4l: whats confusing me is how the variable is populated here: https://github.com/openstack/kolla/blame/master/docker/ceph/ceph-osd/extend_start.sh#L72, on restarts when not bootstrapping: https://github.com/openstack/kolla/blob/master/docker/ceph/ceph-osd/extend_start.sh#L12 | 01:11 |
| *** zhurong has joined #openstack-kolla | 01:11 | |
| sbezverk | sdake: nope all common code was in _helpers.tpl | 01:11 |
| sdake | thats what i'm talking about then :) | 01:12 |
| sbezverk | and then each level has added its variables | 01:12 |
| sdake | so can _helpers.tpl be copied around? | 01:12 |
| sbezverk | sdake: _helpers.tpl has common code and not variables | 01:13 |
| sdake | so that can be copied around? | 01:13 |
| Jeffrey4l | portdirect, https://github.com/openstack/kolla-ansible/blob/master/ansible/roles/ceph/tasks/start_osds.yml#L36 | 01:13 |
| *** JRobinson__ has quit IRC | 01:13 | |
| sbezverk | sdake: yes it can be copied, what exactl you try to solve with this? | 01:14 |
| *** tonanhngo has quit IRC | 01:14 | |
| jascott1 | sbezverk sdake or put in configmap? | 01:14 |
| sdake | making implementation simpler | 01:14 |
| sdake | we need to make the implementation easy - so people can stamp em out | 01:14 |
| sdake | if _helpers.tpl is common across the code base might as well just copy that around | 01:15 |
| jascott1 | would it be bad to say 'install the utils' chart first? tif not we could load that once | 01:15 |
| portdirect | Jeffrey4l: sweet cheers! (now I just need to work out how to k8s that effectively) | 01:15 |
| Jeffrey4l | portdirect, cool | 01:15 |
| sbezverk | sdake: ok I need to step out now. will try to include helpers later today or first thing tomorrow morning.. | 01:16 |
| sdake | roger | 01:16 |
| sdake | jascott1 unknown - i think we want each microservice to standalone tho | 01:17 |
| jascott1 | gotcha | 01:19 |
| jascott1 | i *think* we can load files from root dir of a chart using Files. in templates. we might then only need to have common stuff in the middle level chart | 01:20 |
| jascott1 | oh each MICRO service | 01:21 |
| jascott1 | sorry | 01:21 |
| sdake | another short term hack is symlinks vs copying | 01:21 |
| portdirect | sdake: I think symlinks would be preferable | 01:22 |
| sdake | you think that :) | 01:22 |
| alanmeadows | thought about that but then might break if someone rearranges structure, even if just locally | 01:22 |
| alanmeadows | charts would no longer be "self-contained units" | 01:23 |
| alanmeadows | given the links would be relative in this case | 01:23 |
| sdake | copy sync is easily handled by a 1 liner find operation | 01:23 |
| alanmeadows | micro/templates/_common.tpl -> ../../../util-chart/templates/common.tpl | 01:23 |
| alanmeadows | plus helm tars things up, didn't look closely at how its taring and whether it would follow links or (and this would be weird) preserve links ;-) | 01:24 |
| sdake | find . -name _common.tpl -exec cp /tmp/common.tpl {} \; | 01:24 |
| portdirect | hmm | 01:24 |
| sdake | lets use that one liner for now i think | 01:24 |
| portdirect | i think i may be changing my mind... | 01:24 |
| *** schwicht has joined #openstack-kolla | 01:25 | |
| *** jtriley has quit IRC | 01:25 | |
| alanmeadows | I only found three options to move forward 1) parent chart that all installs must go through, e.g. uberchart, subcharts cant install directly 2) clone globals to every values.yaml (difficult to maintain but helm native today) or 3) some sort of preparatory step taking a common set of defines and placing them within each subchart with my personal hope helm | 01:29 |
| alanmeadows | will have a solution for this in the future | 01:29 |
| sdake | not talking about common values talking about common code in daemonsets/etc | 01:30 |
| sdake | does a cp not work for that? | 01:30 |
| alanmeadows | well 3) for us was about both | 01:30 |
| alanmeadows | you can define anything at the end of the day | 01:31 |
| alanmeadows | common global things you wish to reference everywhere, e.g. fqdn, and common manifest snippets | 01:31 |
| alanmeadows | I thought kfox1111 was doing defines/includes of common code snippets at least within a single chart, e.g. logging in one review I looked at anyways, its just a matter of exposing that to all charts and back to the options above in this case only 1) or 3) | 01:34 |
| alanmeadows | could be it was something else... too much whizzing by the last few days | 01:34 |
| jascott1 | the problem with copying might be that any vars referenced in the common stuff are not going to be at the same hierarchal position in the var tree | 01:37 |
| jascott1 | unless they are global | 01:37 |
| jascott1 | https://review.openstack.org/gitweb?p=openstack/kolla-kubernetes.git;a=blob;f=helm/kolla-kubernetes/templates/_helpers.tpl;h=75977ce54ce5a43a8d0d5eb3b2d4a70f120b2de4;hb=883046b1cfed40595628116cb1cec4867c210c9e | 01:38 |
| alanmeadows | just as sdake was alluding too I think, I think you have to separate out whether the discussion is on common code or common values | 01:38 |
| alanmeadows | even though in some directions, the paths could be the same | 01:38 |
| sdake | so what is happening with the _globals.tpl is being optimized seprately from common values | 01:38 |
| sdake | we can do _globals.tpl now because it is easy | 01:39 |
| sdake | common values harder | 01:39 |
| alanmeadows | we are doing defines for common "global" values | 01:39 |
| alanmeadows | which is the same mechanism for shared snippets | 01:39 |
| sdake | right - this method is helm native off the block | 01:40 |
| sdake | if we really want to add some parser in the midst of o3 we can do that | 01:40 |
| jascott1 | aic is using utils chart tho right? | 01:40 |
| alanmeadows | yes, but as mentioned with kfox1111 | 01:40 |
| alanmeadows | you may be tricked into thinking utils chart can do defines, and you can include it as a dependency in subcharts | 01:41 |
| alanmeadows | depending on the chart load order, it may or may not work | 01:41 |
| alanmeadows | so its not actually safe (if it did work reliably, that would be pretty much helm native) | 01:41 |
| jascott1 | utils folder is gone ;) | 01:42 |
| alanmeadows | its now openstack-base, probably soon to be "common" -- so what we have to do is the utils does some utils things like secrets | 01:42 |
| alanmeadows | but it also is responsible for generating "_common.tpl" | 01:42 |
| alanmeadows | which we copy to all charts | 01:42 |
| alanmeadows | shhackery | 01:42 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/kolla: Updated from global requirements https://review.openstack.org/400770 | 01:43 |
| alanmeadows | i suppose one simple ask to helm devs | 01:44 |
| alanmeadows | would be reliable chart ordering | 01:44 |
| alanmeadows | so you can do defines in a child chart | 01:44 |
| alanmeadows | reliably | 01:44 |
| alanmeadows | obliterates the need for any copying | 01:44 |
| alanmeadows | doing defines isn't 100% optimal but at least its all helm and no "prep" copy step | 01:44 |
| jascott1 | they probably using a map in golang (which is unordered) ;) | 01:44 |
| alanmeadows | yup | 01:44 |
| alanmeadows | but its easy to prove out, do a child chart define, do a helm install --debug --dry-run and grep for your value | 01:45 |
| alanmeadows | hit and miss | 01:45 |
| jascott1 | bummer | 01:45 |
| alanmeadows | that took me a while to realize what was going on ;-) | 01:45 |
| jascott1 | i bet | 01:45 |
| jascott1 | a requirement system without ordering | 01:46 |
| alanmeadows | my plan was to create an issue and see what Lachlan says | 01:47 |
| alanmeadows | I have actually seen a number of charts | 01:47 |
| alanmeadows | where they are incorrectly depending on this behavior | 01:48 |
| jascott1 | yikes | 01:48 |
| jascott1 | ya most everyone seems to end up there | 01:48 |
| alanmeadows | https://github.com/kubernetes/helm/issues/1568 - looks like sbezverk beat me to it | 01:55 |
| *** sdake_ has joined #openstack-kolla | 01:55 | |
| *** sdake_ has quit IRC | 01:55 | |
| *** sdake_ has joined #openstack-kolla | 01:55 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Move glance precheck into its own role https://review.openstack.org/399354 | 01:56 |
| sdake_ | portdirect how abouts do i get my ssh private key into this vagrant environment? | 01:57 |
| *** sdake has quit IRC | 01:58 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Enable cloudkitty ui when enable_cloudkitty is yes https://review.openstack.org/403520 | 01:59 |
| portdirect | sdake_: you need to modify the Vagrantfile to include 'config.ssh.private_key_path' | 02:01 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Enable sahara dashboard when enable_sahara is yes https://review.openstack.org/403504 | 02:01 |
| portdirect | sdake: https://github.com/att-comdev/halcyon-vagrant-kubernetes/blob/master/Vagrantfile#L44 | 02:02 |
| portdirect | though I've never tried it | 02:02 |
| sdake_ | mey I just c&ped it :) | 02:03 |
| portdirect | thats the way :) | 02:04 |
| *** MarMat has quit IRC | 02:04 | |
| *** schwicht has quit IRC | 02:07 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Enable ironic ui when enable_ironic is yes https://review.openstack.org/403446 | 02:07 |
| sdake_ | i cloned sbezverk 's work | 02:07 |
| sdake_ | wondering how i deploy it on my kube cluster | 02:07 |
| sdake_ | portdirect ^^ | 02:07 |
| sdake_ | [vagrant@kube1 charts]$ helm install openvswitch-ovsdb | 02:08 |
| sdake_ | E1129 02:08:23.435628 17426 portforward.go:209] Unable to create listener: Error listen tcp6 [::1]:45307: bind: cannot assign requested address | 02:08 |
| sdake_ | Error: release honest-cardinal failed: namespaces "kolla" not found | 02:08 |
| portdirect | ok the 2nd is easy | 02:09 |
| portdirect | 'kubectl create namespace kolla' | 02:09 |
| sdake_ | ok that deployed it | 02:10 |
| sdake_ | portdirect how do i check it deployed? | 02:10 |
| portdirect | helm ls | 02:10 |
| portdirect | but i dont trust that | 02:10 |
| sdake_ | it says deployed | 02:11 |
| portdirect | kubectl get --namespace=kolla all | 02:11 |
| sdake_ | [vagrant@kube1 charts]$ kubectl get --namespace=kolla all | 02:11 |
| sdake_ | [vagrant@kube1 charts]$ | 02:11 |
| sdake_ | shouldn't that print something out? | 02:12 |
| portdirect | one would hope | 02:12 |
| sbezverk | sdake: kubectl get pods --namespace=kolla | 02:12 |
| sbezverk | should show you running pod | 02:12 |
| sdake_ | nada | 02:12 |
| sbezverk | sdake: which command you use to deploy? | 02:13 |
| sdake_ | [vagrant@kube1 charts]$ helm install openvswitch-ovsdb | 02:13 |
| sdake_ | E1129 02:09:53.775311 18667 portforward.go:209] Unable to create listener: Error listen tcp6 [::1]:44673: bind: cannot assign requested address | 02:13 |
| sdake_ | NAME: angry-ant | 02:13 |
| sdake_ | LAST DEPLOYED: Tue Nov 29 02:09:53 2016 | 02:13 |
| sdake_ | NAMESPACE: default | 02:13 |
| sdake_ | STATUS: DEPLOYED | 02:13 |
| sbezverk | ah it got deployed into default namespace | 02:13 |
| sbezverk | kubectl get pods --namespace=default | 02:13 |
| jascott1 | btw you can use -n | 02:14 |
| sdake_ | [vagrant@kube1 charts]$ kubectl get pods --namespace=default | 02:14 |
| sdake_ | [vagrant@kube1 charts]$ | 02:14 |
| sbezverk | have you changed value for registry? | 02:15 |
| sdake_ | nope | 02:15 |
| *** ipsecguy has joined #openstack-kolla | 02:15 | |
| sbezverk | :-) | 02:16 |
| jascott1 | kubectl get pods --all-namespaces | 02:16 |
| *** duonghq has joined #openstack-kolla | 02:16 | |
| portdirect | sbezverk: told ya that would cause pain :P | 02:16 |
| sdake_ | jascott1 that prints out a slew of kubesystem pods | 02:16 |
| jascott1 | and all other pods | 02:17 |
| duonghq | morning guys | 02:17 |
| sdake_ | dup duonghq | 02:17 |
| sdake_ | duonghq time to join the party dude!! | 02:17 |
| jascott1 | just in case its in a weird namespace | 02:17 |
| portdirect | sdake_: you working from this pr: https://review.openstack.org/#/c/403959/3/helm/kolla-kubernetes/charts/openvswitch/charts/openvswitch-ovsdb/values.yaml | 02:17 |
| sdake_ | we are rolling | 02:17 |
| sdake_ | portdirect roger | 02:17 |
| duonghq | roger | 02:17 |
| portdirect | need to change the docker_registry to docker.io | 02:17 |
| portdirect | then rebuild & deploy | 02:18 |
| duonghq | portdirect, you mean now we use docker registry? or local register which signed with CN: docker.io? | 02:18 |
| sdake_ | nada | 02:19 |
| sdake_ | how do you rebuild? | 02:19 |
| sbezverk | sdake you do not rebuild, you need to delete previous deploy and then start new one | 02:19 |
| sdake_ | yup did that | 02:20 |
| sbezverk | helm ls | 02:20 |
| sbezverk | helm delete {old} | 02:20 |
| portdirect | duonghq: dont quite parse, I'm refering to this line here: https://review.openstack.org/#/c/403959/1/helm/kolla-kubernetes/charts/openvswitch/charts/openvswitch-ovsdb/values.yaml@6, which I think should be changed to docker.io? | 02:20 |
| sdake_ | [vagrant@kube1 charts]$ helm ls | 02:20 |
| sdake_ | E1129 02:20:10.884362 27368 portforward.go:209] Unable to create listener: Error listen tcp6 [::1]:41833: bind: cannot assign requested address | 02:20 |
| sdake_ | [vagrant@kube1 charts]$ | 02:20 |
| sbezverk | sdake_: helm version? | 02:20 |
| portdirect | sbezverk: that error looks nasty | 02:21 |
| sdake_ | Server: &version.Version{SemVer:"v2.0.0", GitCommit:"04f203d2d081417bae8a61a8581f70d5ed652aca", GitTreeState:"clean"} | 02:21 |
| sbezverk | sdake: it looks like your helm installation is not working | 02:21 |
| duonghq | portdirect, got it | 02:21 |
| sdake_ | ya not sure what to make of that error listen tcp6 thing | 02:21 |
| sbezverk | have you done "helm init" | 02:21 |
| sbezverk | ? | 02:21 |
| sdake_ | nope | 02:21 |
| sbezverk | please do it | 02:21 |
| sbezverk | it will start tiller pod | 02:21 |
| sdake_ | where do i do that | 02:21 |
| portdirect | wait - your using my vagrant seup right (as in from portdirect not att-comdev)? | 02:21 |
| sdake_ | [vagrant@kube1 charts]$ helm init | 02:22 |
| jascott1 | wow I see a new error - Error: transport is closing | 02:22 |
| sdake_ | $HELM_HOME has been configured at $HOME/.helm. | 02:22 |
| sdake_ | Warning: Tiller is already installed in the cluster. (Use --client-only to suppress this message.) | 02:22 |
| sdake_ | Happy Helming! | 02:22 |
| sdake_ | portdirect roger | 02:22 |
| sbezverk | sdake: can you do now helm version? | 02:22 |
| sdake_ | Client: &version.Version{SemVer:"v2.0.0", GitCommit:"04f203d2d081417bae8a61a8581f70d5ed652aca", GitTreeState:"clean"} | 02:22 |
| portdirect | thats a nightly build i think | 02:22 |
| sbezverk | hmm, it shold also print server version | 02:23 |
| sdake_ | it does | 02:23 |
| sdake_ | same thing | 02:23 |
| sbezverk | on the second line | 02:23 |
| sdake_ | Server: &version.Version{SemVer:"v2.0.0", GitCommit:"04f203d2d081417bae8a61a8581f70d5ed652aca", GitTreeState:"clean"} | 02:23 |
| sbezverk | ok, now helm ls | 02:23 |
| portdirect | sbezverk: wait 2 mins | 02:23 |
| sdake_ | E1129 02:23:37.770817 30205 portforward.go:209] Unable to create listener: Error listen tcp6 [::1]:42407: bind: cannot assign requested address | 02:23 |
| portdirect | lets get you on a stable helm, thats a canary build that your using | 02:25 |
| portdirect | kubectl delete deployment tiller-deploy --namespace kube-system | 02:25 |
| sbezverk | sdake_: not sure the cause of this message | 02:25 |
| sbezverk | portdirect: I use the same version 2.0.0 | 02:25 |
| sdake_ | deployment "tiller-deploy" deleted | 02:25 |
| portdirect | same git commit? | 02:26 |
| sbezverk | portdirect: that I am not sure | 02:26 |
| sdake_ | happy to use a stable version :) | 02:26 |
| jascott1 | helm init | 02:26 |
| sbezverk | Server: &version.Version{SemVer:"v2.0.0", GitCommit:"51bdad42756dfaf3234f53ef3d3cb6bcd94144c2", GitTreeState:"clean"} | 02:26 |
| portdirect | rm -rf /usr/local/bin/helm | 02:26 |
| jascott1 | oh that just reinstates probaby | 02:26 |
| portdirect | dont do that | 02:27 |
| portdirect | as in the 'helm init' | 02:27 |
| sdake_ | portdirect right- next step after the rm is what? | 02:27 |
| duonghq | have we had docker registry in infra? | 02:28 |
| sdake_ | duonghq nope | 02:28 |
| duonghq | got it | 02:28 |
| sbezverk | here is the link for official 2.0.0 release | 02:28 |
| portdirect | as root: curl -L https://kubernetes-helm.storage.googleapis.com/helm-v2.0.0-linux-amd64.tar.gz | tar zxv --strip 1 -C /tmp; chmod +x /tmp/helm; mv /tmp/helm /usr/local/bin/helm | 02:28 |
| sbezverk | curl http://storage.googleapis.com/kubernetes-helm/helm-v2.0.0-linux-amd64.tar.gz | tar --strip-components 1 -C /usr/bin linux-amd64/helm -zxf - | 02:28 |
| portdirect | lol | 02:28 |
| sbezverk | :-) | 02:29 |
| portdirect | was using: https://kubernetes-helm.storage.googleapis.com/helm-canary-linux-amd64.tar.gz before... | 02:29 |
| portdirect | then run 'helm init' | 02:29 |
| sdake_ | [vagrant@kube1 charts]$ helm init | 02:30 |
| sdake_ | $HELM_HOME has been configured at $HOME/.helm. | 02:30 |
| sdake_ | Tiller (the helm server side component) has been installed into your Kubernetes Cluster. | 02:30 |
| sdake_ | Happy Helming! | 02:30 |
| sdake_ | [vagrant@kube1 charts]$ helm install openvswitch-ovsdb | 02:30 |
| sdake_ | Error: could not find a ready tiller pod | 02:30 |
| portdirect | give it a min | 02:30 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: TEST_NO_MERGE https://review.openstack.org/403978 | 02:30 |
| portdirect | you can gech the output of 'kubectl get -n kube-system pods' to see if its ready | 02:30 |
| Jeffrey4l | sdake_, could u review this https://review.openstack.org/403516 ? | 02:30 |
| sdake_ | Jeffrey4l in a moment plz | 02:31 |
| sdake_ | portdirect everything in running state | 02:31 |
| portdirect | try the install again | 02:31 |
| sdake_ | E1129 02:32:18.175603 4942 portforward.go:209] Unable to create listener: Error listen tcp6 [::1]:40713: bind: cannot assign requested address | 02:32 |
| sdake_ | this error seeems to be problematic | 02:32 |
| sdake_ | no running pod | 02:33 |
| sbezverk | sdake_: curious why it tries to use ipv6 for communicating with tiller?? | 02:33 |
| sdake_ | no clue - didn't write the dev environmnet :) | 02:33 |
| sbezverk | portdirect: ;-) same question | 02:33 |
| portdirect | its a good one | 02:34 |
| sdake_ | portdirect you said you turned off ipv6 during our conversation last night? | 02:34 |
| sbezverk | I do nto remember seeing much in terms of ipv6 support by kube | 02:34 |
| sbezverk | maybe I just missed it | 02:34 |
| portdirect | yeah - as ceph was trying to bind to it, so i just disabled it to get it out of the way, - thats why I put it down to weirdness in a nightly build of helm | 02:35 |
| sdake_ | wher edid you disable it | 02:36 |
| sdake_ | on the host os or in vagrant? | 02:37 |
| portdirect | ansible: https://github.com/portdirect/halcyon-kubernetes/blob/8d25f2104cca47c1a51aa344f21b38c31f6a0d4c/kube-deploy/roles/upgrade-os/tasks/centos.yml#L45 | 02:38 |
| sbezverk | good night gents, happy helming ;-) | 02:38 |
| portdirect | laters sbezverk | 02:38 |
| sdake_ | ok i ran that | 02:40 |
| sdake_ | late sbezverk | 02:40 |
| sdake_ | now wot :) | 02:40 |
| sdake_ | need to stop tiller? | 02:40 |
| jascott1 | what version of tiller now? | 02:40 |
| sdake_ | kube-system tiller-deploy-2241983194-xcut4 1/1 Running 0 12m | 02:41 |
| portdirect | sdake_: that should already have been active: 'ip a' shouldn't show any ipv6 addrs - which is why that error has me baffled | 02:42 |
| sdake_ | ya - i think tiller is busted | 02:43 |
| sdake_ | how do i remove it | 02:43 |
| portdirect | 'kubectl delete deployment tiller-deploy --namespace kube-system' | 02:44 |
| jascott1 | to get tiller image version you need something like: kubectl get pods -n kube-system -o yaml | grep tiller | 02:44 |
| jascott1 | im using 2.0.0 tiller and client and am having problems on this vagrant so Id like to know what version you have | 02:44 |
| portdirect | but I'm not sure thats gonna bring you joy - as it stores its config in a 3rd party resource afaik - and I have no idea how to delete that | 02:45 |
| portdirect | jascott1: your having issues as well? | 02:45 |
| jascott1 | when I do helm install on that chart I get 'error transport is closing' | 02:46 |
| sdake_ | portdirect what precisely did you do once you turned off ipv6? | 02:46 |
| jascott1 | but this on the new vagrant setup | 02:46 |
| sdake_ | the listen error comes from the server | 02:46 |
| sdake_ | clients don't use listen typically | 02:46 |
| portdirect | sdake_: the link I provided above is the ansible script that vagrant ran to provision the machines. | 02:47 |
| sdake_ | portdirect i just ran the ipv6 disable directly | 02:48 |
| sdake_ | oh that was your branch we deployed my vagrant from | 02:48 |
| sdake_ | i have no ipv6 addresses in my system | 02:48 |
| portdirect | cool that should not have made a difference in that case | 02:48 |
| portdirect | sdake_: yeah | 02:49 |
| portdirect | jascott1: that error you are getting is this: https://github.com/kubernetes/helm/issues/1589 (I'm intlabs, and v1k0d3n is aware of it as well) | 02:50 |
| portdirect | if you change this: https://github.com/att-comdev/halcyon-kubernetes/blob/master/kube-deploy/group_vars/all.yml#L26, to 'v1.4.6' then you'll be fine (i hope) | 02:51 |
| *** JRobinson__ has joined #openstack-kolla | 02:51 | |
| sdake_ | https://github.com/docker/docker/issues/2174 | 02:51 |
| jascott1 | going to eat w kids bbiab. thanks | 02:52 |
| sdake_ | portdirect so i need to tear down my env and reubild it? | 02:53 |
| portdirect | :( thats prob the quickest way | 02:53 |
| * sdake_ groans | 02:53 | |
| sdake_ | in my repo is this: kube_version: v1.4.6 | 02:57 |
| sdake_ | already | 02:57 |
| sdake_ | portdirect ok dude - bit lost | 02:58 |
| portdirect | that comment above was for jascott1 | 02:58 |
| *** Pavo has quit IRC | 02:58 | |
| sdake_ | about the tear down? | 02:59 |
| sdake_ | you said change all.yml to v1.4.6 | 02:59 |
| sdake_ | it looks like it is already 1.4.6 | 02:59 |
| sdake_ | /home/sdake/work/halcyon-vagrant-kubernetes | 02:59 |
| sdake_ | i dont see how halcyon-kubernetes is being pulled in | 03:00 |
| *** yuanying_ has quit IRC | 03:00 | |
| portdirect | jascott1: used the att-comdev repo, not my fork | 03:00 |
| portdirect | you should have: /home/sdake/work/halcyon-kubernetes | 03:00 |
| portdirect | as well | 03:00 |
| sdake_ | i do | 03:00 |
| sdake_ | it contains 1.4.6 | 03:00 |
| sdake_ | on that all.yml version | 03:01 |
| portdirect | thats correct | 03:01 |
| kfox1111 | wow... | 03:01 |
| kfox1111 | long long long thread. | 03:01 |
| kfox1111 | summery please | 03:01 |
| sdake_ | so why tear down? | 03:01 |
| kfox1111 | "I'm up to this point:" | 03:02 |
| kfox1111 | 20:29 < alanmeadows> I only found three options to move forward | 03:02 |
| kfox1111 | and I agree with his option 3. | 03:02 |
| *** sayantan_ has joined #openstack-kolla | 03:02 | |
| sdake_ | i think you didn't read far back enough :) | 03:02 |
| sdake_ | did you read the part where I pinged you? | 03:02 |
| portdirect | sdake_: can you give me 2 mins - sorry to keep you hanging | 03:02 |
| kfox1111 | yeah read that part. | 03:03 |
| sdake_ | portdirect all good | 03:03 |
| sdake_ | kfox1111 lets crank this stuff out - we can reach perfection in o3 | 03:03 |
| kfox1111 | I think sbezverk really wants native helm without any processing no matter what. | 03:03 |
| sdake_ | i want native helm | 03:03 |
| kfox1111 | I really want a clean api | 03:03 |
| kfox1111 | followed by native helm when possible. | 03:03 |
| *** Pavo has joined #openstack-kolla | 03:03 | |
| kfox1111 | but api trumps. | 03:03 |
| sdake_ | agree, api has nothing to do with values.yml being copied around | 03:04 |
| kfox1111 | cause I'm going to deploy this into production and do not want to have an issue when its time to upgrade. | 03:04 |
| sdake_ | rather globals.tpl being copied around | 03:04 |
| kfox1111 | yeah, I don't like globals at all for this use case. | 03:04 |
| kfox1111 | the values are all for configuring a specific microservice. | 03:04 |
| sdake_ | right | 03:05 |
| kfox1111 | when that microservice is nested into a parent service, its config values should hang directly off of the child. | 03:05 |
| sdake_ | we aren't nesting | 03:05 |
| kfox1111 | nova's vars have: | 03:05 |
| kfox1111 | novncproxy.foo = value | 03:05 |
| kfox1111 | not, | 03:05 |
| kfox1111 | globals.foo = value | 03:05 |
| kfox1111 | or globals.novncproxy_foo = value | 03:05 |
| sdake_ | i guess i dont understand that particular problem in detail | 03:06 |
| kfox1111 | each microservice provides a flat value space. | 03:06 |
| kfox1111 | sdake_: look up variable scoping in helm. | 03:06 |
| kfox1111 | global vs package. | 03:06 |
| kfox1111 | I'm aguing global is not a good fit for our use case. it presents what I think is a bad api. | 03:07 |
| sdake_ | we agreed (I believed) that we want a package per microservice | 03:07 |
| kfox1111 | package per microservice, non global variables. | 03:07 |
| kfox1111 | is what I'm arguing for. | 03:07 |
| kfox1111 | templated version numbers, and a common lib. | 03:07 |
| portdirect | sdake_: I now have a working helm setup - when your ready we can go through setting it up - though the easyest way to do that si gonna be a 'vagrant destroy' pulling the mods I've made and then 'vagrant up' again... | 03:07 |
| kfox1111 | howerver we can best meet those goals. | 03:07 |
| sdake_ | ok - cool - so lets sort that out after we have a functional microservice layer | 03:07 |
| *** sp_ has joined #openstack-kolla | 03:07 | |
| sdake_ | kfox1111 we will get you there, yo uhave my word | 03:08 |
| kfox1111 | sdake_: this won't look to be important until someone builds some of the nested packages. | 03:08 |
| kfox1111 | :/ | 03:08 |
| alanmeadows | I still think there is one case for globals and that is microservices defining their service ports - combination from pulling the more 'common' fqdn from a common.tpl and service port from sister chart allows any service to build an endpoint for any other service and allow microservices to control their port binding/exposure | 03:09 |
| sdake_ | kfox1111 this is the spinning out of control build tool i was speaking of ;) | 03:09 |
| *** unicell has quit IRC | 03:10 | |
| kfox1111 | alanmeadows: I don't quite follow. shouldn't the services be looking it up from keystone? | 03:13 |
| kfox1111 | we've been considering configmaps outside of the normal helm space. helm could create them, but it could be something else too. | 03:14 |
| *** v1k0d3n has joined #openstack-kolla | 03:15 | |
| alanmeadows | I see, so something else generates those, not helm interpreted templates in your case (e.g. Keystone endpoint in neutron.conf) | 03:17 |
| kfox1111 | right. | 03:24 |
| kfox1111 | plus, there may be intermediate settings I need to be able to control. | 03:24 |
| kfox1111 | so, endpoint that keystone thinks its listening on, | 03:24 |
| kfox1111 | might be frontended by an external lb, | 03:24 |
| kfox1111 | which nova would talk to. | 03:24 |
| kfox1111 | so a global default, which is overridable at an individual package level would be ok. | 03:24 |
| *** jtriley has joined #openstack-kolla | 03:25 | |
| alanmeadows | I wanted to empower helm to generate endpoints, that helps drive keystone bootstrapping (which requires them as input) and is a lightweight way for things to build them (template vs keystone query) | 03:26 |
| alanmeadows | Granted those endpoints might actually be external e.g lb and that's fine but I wanted helm to be aware | 03:26 |
| kfox1111 | I'm good with that. I don't want to criple helm based orchestration. | 03:27 |
| kfox1111 | I just can't use it myself, so I don't want to be cripled by needing it. | 03:27 |
| kfox1111 | I think we can come up with something that should work for both. | 03:32 |
| kfox1111 | alanmeadows: do you know, with helm, if you instantiate a package with var override,s and then upgrade it, do the overridden vars stick? | 03:34 |
| kfox1111 | or do they need to be respecified? | 03:34 |
| alanmeadows | kfox1111: actually didn't test that interesting to find out -- my guess is it stores them | 03:36 |
| kfox1111 | I hope so. | 03:36 |
| kfox1111 | it would be interesting to make confimap micropacakges too, | 03:36 |
| kfox1111 | then reference a particular version with the microservice package. | 03:37 |
| kfox1111 | for rolling updates, you would upgrade a configmap package, then update the vars on the microservice pacakge to point to the new configmap package. | 03:37 |
| *** jtriley_ has joined #openstack-kolla | 03:38 | |
| *** jtriley has quit IRC | 03:41 | |
| jascott1 | im pretty sure I caught in #helm that it does store the values | 03:41 |
| alanmeadows | Sounds like you don't trust rollback :) isn't this the inherent functionality your suppose to get with helm upgrade and proper manifests that do rolling container upgrades though based on the iterative change, e.g config map update or maybe I'm missing something? | 03:44 |
| jascott1 | thats the sell | 03:44 |
| jascott1 | afaik | 03:44 |
| jascott1 | my helm is busted atm but you can do helm get values <release name> | 03:44 |
| jascott1 | portdirect I am not using 1.5 anymore since I switched to the vagrant deploy | 03:45 |
| jascott1 | oh wai | 03:46 |
| portdirect | jascott1: I've made a gist to go through the setup I'm using: https://gist.github.com/intlabs/9ce71a4a6b93b4e2e90faec500ef70d0 | 03:48 |
| jascott1 | ya I just saw that vagrant is using 1.5 | 03:48 |
| portdirect | alanmeadows: what version of k8s are you running with helm? | 03:49 |
| *** jtriley_ has quit IRC | 03:53 | |
| *** Jeffrey4l has quit IRC | 03:54 | |
| *** Jeffrey4l has joined #openstack-kolla | 03:55 | |
| kfox1111 | alanmeadows: its not that I don't trust rollback. (though, questionable). | 03:56 |
| kfox1111 | its one of who builds/ships/tests what, and what gets upgraded/applied when. | 03:56 |
| kfox1111 | templates should be buildable/testable/shippable as an upstream. | 03:57 |
| kfox1111 | say, I pay (insert name here) to support kolla-kubernetes helm package. | 03:57 |
| kfox1111 | I should be able to helm repo add url-tosupport-repo | 03:57 |
| kfox1111 | upgrades should come from them. | 03:57 |
| kfox1111 | configmaps though, are very particular to my site. | 03:58 |
| kfox1111 | more oddball then helm can ever be customizable to generate. | 03:58 |
| kfox1111 | I do really like the atomicness of k8s. | 03:58 |
| kfox1111 | so it is kind of bad to upload the configmap as x, then update it to x+1 and have half the cluster pods with the old config and half with the new, with no way to know the idfference. | 03:59 |
| *** yuanying has joined #openstack-kolla | 03:59 | |
| alanmeadows | portdirect: 2.0.0rc2 | 04:00 |
| portdirect | alanmeadows: sorry k8s, not helm | 04:01 |
| alanmeadows | 1.5.0beta1 tag - oldest with rbd dynamic pvc support | 04:01 |
| jascott1 | ya 1.5beta1 was causing my pain. 1.4.6 working fine | 04:02 |
| alanmeadows | They changed a lot, including the client interface | 04:02 |
| alanmeadows | Make sure to upgrade kubectl with it or things will be super weird | 04:02 |
| kfox1111 | alanmeadows: did dynamic pvc for ceph make it finally in 1.5? | 04:02 |
| *** mdnadeem has joined #openstack-kolla | 04:03 | |
| alanmeadows | It did, we're using it to run aic-helm mariadb clusters per the readme | 04:03 |
| alanmeadows | Works like a....'charm' ? | 04:03 |
| kfox1111 | nice. :) | 04:03 |
| kfox1111 | I'll have to have a look at that. :) | 04:04 |
| portdirect | kfox1111: this image - quay.io/attcomdev/kube-controller-manager:v1.5.0-beta.1 | 04:04 |
| alanmeadows | Requires some further finagling again in the readme | 04:04 |
| kfox1111 | using petsets or individual instances? | 04:04 |
| alanmeadows | Yes that :) | 04:04 |
| portdirect | works well :) | 04:04 |
| *** f13o has quit IRC | 04:05 | |
| alanmeadows | Individual instances, namely helm ranges around values.replicas * pod+pvc+pv to get mariadb-0 and so on | 04:05 |
| portdirect | though I've not manged to get helm working with 1.5 - tiller crashes everytime i try and install a chart | 04:05 |
| portdirect | others have reported the same on slack and github | 04:05 |
| alanmeadows | Pet sets is foobar'd in 1.5 pre ga release as they are renaming to state full set and helm is pinned to 1.4.1 client lib | 04:06 |
| portdirect | i'm talking *any* chart | 04:06 |
| alanmeadows | Hmm working for us.... that's strange I can do a helm version for you tomorrow typing from my phone | 04:06 |
| kfox1111 | alanmeadows: ah. ok. wondered about that. | 04:06 |
| alanmeadows | I have a helm issue open for 1.5/statefulset support :) | 04:07 |
| alanmeadows | I tried my hand at doing it but the interface changed quite a bit | 04:07 |
| portdirect | np - that would be sweet - I've been running your controller=manger image with k8s 1.4.6 without any obvious side effects - but thats a recipy for sadness just waiting to happen. | 04:08 |
| *** tonanhngo has joined #openstack-kolla | 04:08 | |
| alanmeadows | Hehe | 04:08 |
| alanmeadows | This was also our first upgrade experience with kubeadm which has.... 0 upgrade support | 04:09 |
| alanmeadows | That was disappointing | 04:09 |
| *** tonanhngo has quit IRC | 04:09 | |
| portdirect | yeah - i think it'll get there but its super early days. | 04:10 |
| alanmeadows | The price of the bleeding edge! | 04:11 |
| portdirect | mostly painfull... | 04:12 |
| portdirect | anyway - better head - it's so late here it's almost early again :/ | 04:14 |
| alanmeadows | Kfox1111: set aside some time in the AM to give your CR some proper review/comments | 04:14 |
| alanmeadows | Catch you all tomorrow | 04:15 |
| *** zhurong has quit IRC | 04:15 | |
| *** zhurong has joined #openstack-kolla | 04:15 | |
| *** portdirect is now known as portdirect_away | 04:16 | |
| jascott1 | have good ones | 04:17 |
| *** portdirect_away is now known as portdirect | 04:17 | |
| *** portdirect is now known as portdirect_away | 04:17 | |
| *** jgriffith is now known as jgriffith_away | 04:21 | |
| *** tonanhngo has joined #openstack-kolla | 04:21 | |
| *** tonanhngo has quit IRC | 04:22 | |
| kfox1111 | alanmeadows: awesome. thanks. talk to you tomorrow. | 04:31 |
| duonghq | sdake_, ping, can you check the rabbitmq-server version in centos:7 image | 04:36 |
| sdake_ | duonghq lots o people in channel :) - need to do review for jeffrey4l atm | 04:37 |
| sdake_ | duonghq ok - done with that it was a 1 liner moment | 04:39 |
| duonghq | sdake_, I see your comment in the rabbitmq dockerfile, just want to make sure why the rabbitmq server in my setup is 3.3.5, it should be 3.6 | 04:42 |
| sdake_ | rabbitmq-server-3.6.5-1.el7.noarch | 04:42 |
| duonghq | thank sdake_ | 04:42 |
| *** rock_ has quit IRC | 04:47 | |
| duonghq | as I understand, centos:7 is using rabbitmq-server from epel repo, but it's 3.3.5, anybody can verify for me? | 04:52 |
| *** Pavo has quit IRC | 04:58 | |
| *** jrist has quit IRC | 05:03 | |
| *** Pavo has joined #openstack-kolla | 05:03 | |
| *** tonanhngo has joined #openstack-kolla | 05:04 | |
| duonghq | Pavo, ping | 05:05 |
| *** tonanhngo has quit IRC | 05:06 | |
| openstackgerrit | Merged openstack/kolla: Bump Ubuntu cloud-archive repo to Ocata https://review.openstack.org/403516 | 05:09 |
| *** kristian__ has joined #openstack-kolla | 05:13 | |
| *** kristian__ has quit IRC | 05:17 | |
| jascott1 | neato https://github.com/Boostport/kubernetes-vault | 05:24 |
| *** JRobinson__ is now known as JRobinson__afk | 05:33 | |
| *** sdake_ has quit IRC | 05:52 | |
| *** williamcaban has joined #openstack-kolla | 05:54 | |
| *** sayantan_ has quit IRC | 05:57 | |
| *** JRobinson__afk is now known as JRobinson__ | 06:01 | |
| *** Jeffrey4l has quit IRC | 06:12 | |
| *** Jeffrey4l has joined #openstack-kolla | 06:12 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: TEST_NO_MERGE https://review.openstack.org/403978 | 06:13 |
| *** jascott1 has quit IRC | 06:15 | |
| *** jascott1 has joined #openstack-kolla | 06:40 | |
| *** kristian__ has joined #openstack-kolla | 06:42 | |
| *** saneax-_-|AFK is now known as saneax | 06:42 | |
| *** unicell has joined #openstack-kolla | 06:44 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Add manila-ui in horizon image https://review.openstack.org/403689 | 06:44 |
| *** unicell has quit IRC | 06:51 | |
| *** unicell has joined #openstack-kolla | 06:52 | |
| *** kristian__ has quit IRC | 06:52 | |
| *** JRobinson__ has quit IRC | 06:53 | |
| *** unicell1 has joined #openstack-kolla | 06:54 | |
| *** unicell has quit IRC | 06:56 | |
| *** Pavo has quit IRC | 06:58 | |
| *** Pavo has joined #openstack-kolla | 07:03 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Use Dockerfile customization instead of include_header https://review.openstack.org/404049 | 07:09 |
| *** kristian__ has joined #openstack-kolla | 07:13 | |
| openstackgerrit | Merged openstack/kolla: Updated from global requirements https://review.openstack.org/400770 | 07:20 |
| *** yuanying has quit IRC | 07:20 | |
| *** pcaruana has joined #openstack-kolla | 07:21 | |
| *** f13o has joined #openstack-kolla | 07:25 | |
| *** yuanying has joined #openstack-kolla | 07:26 | |
| openstackgerrit | Eduardo Gonzalez proposed openstack/kolla: Tacker Docker configuration https://review.openstack.org/396391 | 07:28 |
| openstackgerrit | Eduardo Gonzalez proposed openstack/kolla: Add strongswan to neutron-vpn-agent source image https://review.openstack.org/404061 | 07:31 |
| *** kristian__ has quit IRC | 07:38 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Install cloudkitty dashboard in horizon image https://review.openstack.org/396535 | 07:38 |
| *** kristian__ has joined #openstack-kolla | 07:40 | |
| *** zhurong has quit IRC | 07:40 | |
| *** zhurong has joined #openstack-kolla | 07:41 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Enable cloudkitty ui when enable_cloudkitty is yes https://review.openstack.org/403520 | 07:42 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Enable cloudkitty ui when enable_cloudkitty is yes https://review.openstack.org/403520 | 07:42 |
| openstackgerrit | Javier Castillo Alcíbar proposed openstack/kolla-ansible: Make sure heka-ceilometer.toml manages correctly apache and openstack ceilometer logs https://review.openstack.org/400716 | 07:43 |
| *** kristian__ has quit IRC | 07:44 | |
| *** magicboiz has joined #openstack-kolla | 07:44 | |
| openstackgerrit | Merged openstack/kolla: Remove --include-header and --include-footer parameter https://review.openstack.org/402309 | 07:50 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Enable cloudkitty ui when enable_cloudkitty is yes https://review.openstack.org/403520 | 07:53 |
| *** zhubingbing has quit IRC | 07:57 | |
| openstackgerrit | Merged openstack/kolla-ansible: Icrease haproxy timeouts for mysql connections https://review.openstack.org/403428 | 07:59 |
| openstackgerrit | Merged openstack/kolla-ansible: Deprecate scheduler_max_attempts option in nova https://review.openstack.org/399011 | 07:59 |
| openstackgerrit | wangwei proposed openstack/kolla: Add the function of docker login before pushing https://review.openstack.org/399426 | 08:01 |
| *** f13o has quit IRC | 08:03 | |
| *** shardy has joined #openstack-kolla | 08:07 | |
| *** matrohon has joined #openstack-kolla | 08:08 | |
| *** rmart04 has joined #openstack-kolla | 08:16 | |
| *** yuanying has quit IRC | 08:17 | |
| *** Serlex has joined #openstack-kolla | 08:23 | |
| *** egonzalez90 has joined #openstack-kolla | 08:39 | |
| egonzalez90 | morning | 08:39 |
| openstackgerrit | Merged openstack/kolla-ansible: use uuidutils instead of uuid.uuid4() https://review.openstack.org/399847 | 08:40 |
| berendt | morning | 08:42 |
| egonzalez90 | magicboiz: ping | 08:45 |
| magicboiz | morning egonzalez90! | 08:46 |
| magicboiz | morning everyone :) | 08:46 |
| duonghq | morning egonzalez90 | 08:48 |
| duonghq | anybody deploy kolla recently and use Horizon? | 08:48 |
| egonzalez90 | btw duonghq , your change "Fix placement of policy.json " was not applied in kolla-ansible repo due to the split | 08:49 |
| duonghq | egonzalez90, moment, maybe I forgot this ps | 08:50 |
| duonghq | egonzalez90, https://review.openstack.org/#/c/394260/ -> it's merged before repo split | 08:51 |
| egonzalez90 | yup, but was merged after repository snapshot, is not present the change at kolla-ansible | 08:53 |
| duonghq | egonzalez90, understood | 08:54 |
| duonghq | egonzalez90, how do I move the bug to kolla-ansible launchpad? | 08:55 |
| egonzalez90 | done | 08:57 |
| egonzalez90 | in also affects project button | 08:57 |
| duonghq | egonzalez90, thank you | 08:58 |
| *** Pavo has quit IRC | 08:58 | |
| openstackgerrit | Duong Ha-Quang proposed openstack/kolla-ansible: Fix placement of policy.json https://review.openstack.org/404092 | 09:00 |
| *** Pavo has joined #openstack-kolla | 09:03 | |
| *** jascott1 has quit IRC | 09:08 | |
| *** jascott1 has joined #openstack-kolla | 09:09 | |
| openstackgerrit | caoyuan proposed openstack/kolla: Remove the "nova-network" container file https://review.openstack.org/404099 | 09:10 |
| *** zhubingbing has joined #openstack-kolla | 09:13 | |
| *** jascott1 has quit IRC | 09:14 | |
| bjolo | morning | 09:14 |
| *** hvlad has joined #openstack-kolla | 09:21 | |
| duonghq | morning bjolo | 09:23 |
| zhubingbing | sup sdake | 09:26 |
| *** gfidente has joined #openstack-kolla | 09:29 | |
| *** kristian__ has joined #openstack-kolla | 09:29 | |
| *** senk has joined #openstack-kolla | 09:32 | |
| *** mkoderer has joined #openstack-kolla | 09:37 | |
| duonghq | hi zhubingbing, long time no see | 09:38 |
| zhubingbing | hi | 09:39 |
| zhubingbing | i alway on irc ;) | 09:39 |
| duonghq | do you deploy Kolla recently? | 09:40 |
| zhubingbing | yes | 09:40 |
| zhubingbing | http://docs.openstack.org/developer/castellan/installation.html | 09:40 |
| *** yingjun has quit IRC | 09:41 | |
| zhubingbing | do you know castellan ? | 09:41 |
| *** yingjun has joined #openstack-kolla | 09:41 | |
| *** yingjun_ has joined #openstack-kolla | 09:41 | |
| *** ram___ has joined #openstack-kolla | 09:41 | |
| zhubingbing | Do you think castellan should put it in docker alone ？ | 09:42 |
| *** hfu has quit IRC | 09:42 | |
| zhubingbing | or put it in barbican docker container ？ | 09:42 |
| *** kristian__ has quit IRC | 09:44 | |
| *** kristian__ has joined #openstack-kolla | 09:44 | |
| *** kristian__ has quit IRC | 09:45 | |
| egonzalez90 | zhubingbing: castillan has not service? | 09:45 |
| egonzalez90 | is only a CLI? | 09:45 |
| zhubingbing | yes | 09:45 |
| *** yingjun_ has quit IRC | 09:45 | |
| *** yingjun has quit IRC | 09:45 | |
| zhubingbing | not service！ | 09:45 |
| zhubingbing | I am very confused；） | 09:46 |
| *** kristian__ has joined #openstack-kolla | 09:46 | |
| egonzalez90 | yup, no info about that | 09:47 |
| duonghq | zhubingbing, sorry, but does your Horizon work normally? | 09:47 |
| zhubingbing | horizon is bad | 09:47 |
| *** kristia__ has joined #openstack-kolla | 09:47 | |
| zhubingbing | Compressing... CommandError: An error occurred during rendering /var/lib/kolla/venv/lib/python2.7/site-packages/openstack_dashboard/templates/_stylesheets.html: Couldn't find anything to import: ../../mixins | 09:48 |
| zhubingbing | Extensions: <NamespaceAdapterExtension>, <DjangoExtension>, <CompassExtension> | 09:48 |
| zhubingbing | Search path: | 09:48 |
| zhubingbing | on line 21 of horizon/lib/roboto_fontface/css/roboto/sass/roboto-fontface-regular.scss | 09:48 |
| zhubingbing | imported from line 21 of horizon/lib/roboto_fontface/css/roboto/sass/roboto-fontface.scss | 09:48 |
| zhubingbing | imported from line 16 of themes/material/bootstrap/_styles.scss | 09:48 |
| zhubingbing | imported from line 16 of themes/material/_styles.scss | 09:48 |
| zhubingbing | imported from line 1 of u'string:c61b5d22d54bf56c:\n // My Themes\n@import "/themes/material/variables";\n\n// Horizon\n@import "/dashboard/scss/horizon' | 09:48 |
| zhubingbing | sorry | 09:48 |
| *** openstackgerrit has quit IRC | 09:48 | |
| duonghq | zhubingbing, same here | 09:48 |
| *** openstackgerrit has joined #openstack-kolla | 09:48 | |
| zhubingbing | yes | 09:49 |
| zhubingbing | I thought it was my problem, so I didn't pay attention to it. | 09:49 |
| zhubingbing | ；（ | 09:49 |
| duonghq | roger | 09:50 |
| *** kristian__ has quit IRC | 09:51 | |
| zhubingbing | duonghq you can make bug in kolla-ansible | 09:52 |
| zhubingbing | ;) | 09:52 |
| *** jrich523 has quit IRC | 09:52 | |
| zhubingbing | egonzalez90 do you think about catillan ? | 09:53 |
| zhubingbing | ;) | 09:53 |
| duonghq | zhubingbing, sure | 09:53 |
| *** jrich523 has joined #openstack-kolla | 09:54 | |
| egonzalez90 | but if is not a service, maybe is on the user side | 09:58 |
| egonzalez90 | zhubingbing: i will look about castillian to give you an opinion | 10:00 |
| zhubingbing | roger | 10:00 |
| duonghq | zhubingbing, https://bugs.launchpad.net/kolla/+bug/1645650 fyi | 10:00 |
| openstack | Launchpad bug 1645650 in kolla "Horizon rendering failed at Compression" [High,New] | 10:00 |
| zhubingbing | ok perfect | 10:00 |
| zhubingbing | think you very much egonzalez90 | 10:00 |
| egonzalez90 | duonghq zhubingbing is an horizon bug https://bugs.launchpad.net/horizon/+bug/1645630 | 10:03 |
| openstack | Launchpad bug 1645630 in OpenStack Dashboard (Horizon) "Material Theme in Newton Appears to be Broken" [Undecided,New] | 10:03 |
| egonzalez90 | not to much to do in our side | 10:03 |
| duonghq | egonzalez90, got it | 10:04 |
| *** zhurong has quit IRC | 10:05 | |
| zhubingbing | thinks | 10:05 |
| *** jascott1 has joined #openstack-kolla | 10:09 | |
| *** jascott1 has quit IRC | 10:14 | |
| *** senk has quit IRC | 10:18 | |
| *** kristian__ has joined #openstack-kolla | 10:20 | |
| *** kristia__ has quit IRC | 10:23 | |
| egonzalez90 | zhubingbing: if I understand right what castellan is, is "like" a olso.key_manager. There is not service | 10:24 |
| egonzalez90 | if you want to use castellan for key management in sahara, you need to pip install castellan in sahara container | 10:24 |
| zhubingbing | yes， | 10:26 |
| duonghq | bye guys | 10:27 |
| zhubingbing | so we should add it in openstack base docker file ? | 10:27 |
| *** duonghq has quit IRC | 10:27 | |
| zhubingbing | bye <duonghq | 10:27 |
| *** kristian__ has quit IRC | 10:27 | |
| *** tovin07 has quit IRC | 10:29 | |
| egonzalez90 | depends, if is going to be use by several services yes, if is only for a service not | 10:30 |
| *** senk has joined #openstack-kolla | 10:31 | |
| *** kristian__ has joined #openstack-kolla | 10:31 | |
| *** jmccarthy has quit IRC | 10:31 | |
| *** jmccarthy has joined #openstack-kolla | 10:32 | |
| *** kristia__ has joined #openstack-kolla | 10:34 | |
| *** kristian__ has quit IRC | 10:34 | |
| *** yingjun has joined #openstack-kolla | 10:35 | |
| *** yingjun has quit IRC | 10:38 | |
| *** yingjun has joined #openstack-kolla | 10:38 | |
| *** senk has quit IRC | 10:41 | |
| zhubingbing | roger | 10:42 |
| *** yingjun has quit IRC | 10:43 | |
| *** ram___ has quit IRC | 10:44 | |
| openstackgerrit | Merged openstack/kolla: Support adding customized plugin to keystone image https://review.openstack.org/401145 | 10:45 |
| *** kristian__ has joined #openstack-kolla | 10:47 | |
| *** kristia__ has quit IRC | 10:47 | |
| *** kristian__ has quit IRC | 10:49 | |
| *** Pavo has quit IRC | 10:58 | |
| *** Pavo has joined #openstack-kolla | 11:03 | |
| *** kristian__ has joined #openstack-kolla | 11:05 | |
| *** kristian__ has quit IRC | 11:06 | |
| *** zhubingbing has quit IRC | 11:07 | |
| *** kristian__ has joined #openstack-kolla | 11:15 | |
| *** kristia__ has joined #openstack-kolla | 11:17 | |
| *** kristian__ has quit IRC | 11:18 | |
| *** kristia__ has quit IRC | 11:21 | |
| mliima_ | morning guys | 11:21 |
| *** kristian__ has joined #openstack-kolla | 11:22 | |
| *** kristian__ has quit IRC | 11:22 | |
| *** ram____ has joined #openstack-kolla | 11:25 | |
| ram____ | Hi all. How do I make sure that kolla is configured to deploy magnum? along with the entire openstack config setup. | 11:25 |
| ram____ | I am new to Kolla. I want to do a small POC . I think magnum is a service we can deploy it through kolla. | 11:28 |
| *** hfu has joined #openstack-kolla | 11:28 | |
| *** ntpttr has quit IRC | 11:30 | |
| *** hfu has quit IRC | 11:30 | |
| *** ntpttr has joined #openstack-kolla | 11:31 | |
| *** YefimG has joined #openstack-kolla | 11:33 | |
| pbourke | ram____: I'm unsure how well magnum works in Kolla | 11:34 |
| *** mdnadeem has quit IRC | 11:34 | |
| egonzalez90 | ram____: you need to enable heat also | 11:35 |
| YefimG | Jeffrey4l: Hello! sdake yesterday pointed, that you have more experience with ceilometer and gnocchi. I am trying to deploy ceilometer with gnocchi backend, but the bootstrap container fails. Is this configuration even supported? Thanks! | 11:36 |
| Jeffrey4l | YefimG, what error message did u get? | 11:36 |
| Jeffrey4l | YefimG, docker logs bootstrap_ceilometer | 11:37 |
| Jeffrey4l | I tried this, it works. | 11:37 |
| ram____ | egonzalez90: Oh. If we run run Kolla-build, then all openstack services can be deployed. May be not magnum service. So how do I change/modify the kolla build config file to enable magnum. can you please provide me a detailed document for that. | 11:38 |
| egonzalez90 | ram____: when do kolla-build all openstack images are built | 11:39 |
| egonzalez90 | then when using kolla-ansible, need to edit /etc/kolla/globals.yml and configure the services as you like | 11:39 |
| *** hvlad has quit IRC | 11:40 | |
| YefimG | After debugging I noticed, that the first container, that fails is gnocchi-statsd. In logs there I found: http://pastebin.com/dvPEg48T | 11:40 |
| Jeffrey4l | YefimG, did u enabled ceph? | 11:41 |
| Jeffrey4l | could u post your globals.yml file out? | 11:41 |
| openstackgerrit | Paul Bourke (pbourke) proposed openstack/kolla: Update Kibana documentation https://review.openstack.org/403729 | 11:41 |
| ram____ | egonzalez90: when we do kolla-build then magnum service image can build automatically? If it is build here only then when we run kolla-ansible deploy then it will deploy magnum service also along with openstack service right | 11:45 |
| YefimG | Jeffrey4l: no, this is a single node all-in-out testing environment with file backend. Here is my globals.yml: http://paste.fedoraproject.org/492625/80419890 | 11:45 |
| Jeffrey4l | YefimG, seems like the gnocchi database is not created. | 11:46 |
| Jeffrey4l | YefimG, have u ever destroy the openstack and try again? | 11:47 |
| YefimG | Jeffrey4l: will do that now and let you know how it goes, thanks | 11:47 |
| Jeffrey4l | YefimG, ok. | 11:47 |
| Jeffrey4l | btw, run kolla-ansible with -vvv parameter and log all ansible s stdout | 11:48 |
| Jeffrey4l | YefimG, ^^ | 11:48 |
| YefimG | Jeffrey4l: ok! | 11:48 |
| egonzalez90 | ram____: yes, but you need to enable_magnum in /etc/kolla/globals.yml, also configure IPs, networks and so on | 11:49 |
| *** mdnadeem has joined #openstack-kolla | 11:50 | |
| Jeffrey4l | pbourke, https://review.openstack.org/403446 please review ;) | 11:51 |
| openstackgerrit | Merged openstack/kolla-ansible: Enable ironic ui when enable_ironic is yes https://review.openstack.org/403446 | 11:54 |
| *** YefimG has quit IRC | 11:59 | |
| ram____ | egonzalez90: Ok. thank you. I will try the setup. | 12:03 |
| *** sdake has joined #openstack-kolla | 12:09 | |
| sdake | morning peeps! | 12:10 |
| openstackgerrit | James McCarthy proposed openstack/kolla-ansible: Add more options for cinder-backup. https://review.openstack.org/402912 | 12:11 |
| *** JoseMello has joined #openstack-kolla | 12:11 | |
| *** jascott1 has joined #openstack-kolla | 12:12 | |
| *** kristian__ has joined #openstack-kolla | 12:12 | |
| *** kristian__ has quit IRC | 12:14 | |
| *** kristian__ has joined #openstack-kolla | 12:14 | |
| *** kristia__ has joined #openstack-kolla | 12:15 | |
| *** jascott1 has quit IRC | 12:17 | |
| *** hfu has joined #openstack-kolla | 12:17 | |
| *** hfu has quit IRC | 12:17 | |
| sdake | The door is jammed... "GET IT OPEN" | 12:18 |
| sdake | V such a fantastic movie, did so poorly in the boxoffice | 12:18 |
| *** kristian__ has quit IRC | 12:19 | |
| sdake | srwilkers ping - you happen to be awake at this godawful hour? | 12:19 |
| *** hvlad has joined #openstack-kolla | 12:19 | |
| *** kristia__ has quit IRC | 12:21 | |
| *** harbor has joined #openstack-kolla | 12:21 | |
| *** harbor is now known as portdirect___ | 12:21 | |
| *** portdirect___ is now known as portdirect__ | 12:22 | |
| *** portdirect__ is now known as portdirect_ | 12:23 | |
| *** senk has joined #openstack-kolla | 12:24 | |
| openstackgerrit | Merged openstack/kolla: Word choice for back end https://review.openstack.org/390788 | 12:25 |
| openstackgerrit | Merged openstack/kolla: shared mount added for /var/lib/nova/mnt to QSG https://review.openstack.org/402191 | 12:26 |
| *** senk has quit IRC | 12:33 | |
| openstackgerrit | James McCarthy proposed openstack/kolla-ansible: Add more options for cinder-backup. https://review.openstack.org/402912 | 12:33 |
| *** egonzalez90 has quit IRC | 12:36 | |
| *** kristian__ has joined #openstack-kolla | 12:36 | |
| openstackgerrit | caoyuan proposed openstack/kolla: Remove the "nova-network" container file https://review.openstack.org/404099 | 12:40 |
| *** pbourke has quit IRC | 12:42 | |
| *** pbourke has joined #openstack-kolla | 12:42 | |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Remove the "nova-network" from the ansible https://review.openstack.org/404200 | 12:44 |
| *** mdnadeem has quit IRC | 12:46 | |
| *** shardy is now known as shardy_lunch | 12:47 | |
| *** athomas has joined #openstack-kolla | 12:49 | |
| *** athomas has quit IRC | 12:49 | |
| *** mnasiadka has joined #openstack-kolla | 12:50 | |
| *** athomas has joined #openstack-kolla | 12:50 | |
| *** hfu has joined #openstack-kolla | 12:55 | |
| *** athomas has quit IRC | 12:57 | |
| *** zhubingbing has joined #openstack-kolla | 12:58 | |
| *** Pavo has quit IRC | 12:58 | |
| *** senk has joined #openstack-kolla | 12:59 | |
| *** kristian__ has quit IRC | 13:00 | |
| *** senk has quit IRC | 13:02 | |
| *** Pavo has joined #openstack-kolla | 13:03 | |
| *** athomas has joined #openstack-kolla | 13:07 | |
| *** eaguilar has joined #openstack-kolla | 13:09 | |
| *** duonghq has joined #openstack-kolla | 13:11 | |
| *** athomas has quit IRC | 13:14 | |
| *** athomas has joined #openstack-kolla | 13:15 | |
| duonghq | evening my friends | 13:21 |
| duonghq | portdirect_, are you around? | 13:21 |
| duonghq | which helm version are we using? | 13:21 |
| portdirect_ | duonghq: hey | 13:21 |
| portdirect_ | I'd recommend the stable 2.0.0 | 13:22 |
| duonghq | portdirect_, got it | 13:22 |
| *** jrist has joined #openstack-kolla | 13:23 | |
| *** yingjun has joined #openstack-kolla | 13:25 | |
| openstackgerrit | James McCarthy proposed openstack/kolla-ansible: Add more options for cinder-backup. https://review.openstack.org/402912 | 13:27 |
| *** shardy_lunch is now known as shardy | 13:28 | |
| *** ayoung has quit IRC | 13:29 | |
| *** mgiles has joined #openstack-kolla | 13:31 | |
| openstackgerrit | James McCarthy proposed openstack/kolla-ansible: Add more options for cinder-backup. https://review.openstack.org/402912 | 13:31 |
| duonghq | portdirect_, did you get "ProvisioningFailedcannot find volume plugin for alpha provisioning" of pvc? | 13:33 |
| duonghq | when install mariadb with helm | 13:33 |
| portdirect_ | yeah | 13:33 |
| duonghq | what reasons can raise this issue? | 13:34 |
| portdirect_ | thats exepcted if you are trying to install mariadb from the helm repo | 13:34 |
| portdirect_ | https://gist.github.com/intlabs/9ce71a4a6b93b4e2e90faec500ef70d0 | 13:34 |
| portdirect_ | it's because there is not support for dynamic provisioning of pvc without cloudprovider atm with stock k8s install with kubeadm | 13:35 |
| duonghq | seem that not only kubeadm :) | 13:35 |
| duonghq | kube-deploy suffer it too | 13:35 |
| duonghq | helm ls work fine for me, so I guess I got dev env | 13:35 |
| portdirect_ | if you run 'kubectl get all' and see a mariadb sevice and a pod stuck in provsioning then you should be good to go | 13:36 |
| *** kristian__ has joined #openstack-kolla | 13:36 | |
| duonghq | portdirect_, http://paste.openstack.org/show/590811/ <- this is what I got | 13:37 |
| portdirect_ | untill we have provisiong of volumes sorted I'd use an 'emptydir' volume as a short term workaround if you need them in your chart | 13:38 |
| duonghq | okay | 13:38 |
| portdirect_ | that paste looks good - it means that helm has successfully deployed an application, but the application it's deployed is broken | 13:39 |
| duonghq | thank youn | 13:40 |
| portdirect_ | you should be able to clean up your env by running 'helm delete wandering-molly' | 13:40 |
| duonghq | already do that :) | 13:41 |
| *** kristian__ has quit IRC | 13:41 | |
| *** fguillot has joined #openstack-kolla | 13:43 | |
| *** mdnadeem has joined #openstack-kolla | 13:43 | |
| *** dave-mccowan has joined #openstack-kolla | 13:43 | |
| *** hfu has quit IRC | 13:45 | |
| *** egonzalez90 has joined #openstack-kolla | 13:45 | |
| *** jtriley has joined #openstack-kolla | 13:46 | |
| *** YefimG has joined #openstack-kolla | 13:46 | |
| sdake | portdirect_ - need help on dev env :) | 13:47 |
| sdake | portdirect_ got any advice for me | 13:47 |
| YefimG | Jeffrey4l: Hi again! I re-deployed Kolla from scratch and got the same error with gnocchi bootstrap container. I have run deploy with -vvv, here is the full deploy.log: http://gasels.com/deploy.log | 13:48 |
| portdirect_ | sdake - lets get you rolling again | 13:48 |
| portdirect_ | whats up with your env? It did not seem happy at all last night | 13:49 |
| Jeffrey4l | YefimG, OK. checking | 13:50 |
| *** kristian__ has joined #openstack-kolla | 13:50 | |
| kristian__ | Hi. Can someone help me with installing custom version of libvirt on ubuntu containers? Container ver 3.0.1 and targer libvirt 1.2.17 or 2.1 | 13:51 |
| kfox1111 | morning | 13:52 |
| *** yingjun has quit IRC | 13:53 | |
| duonghq | morning kfox1111 | 13:53 |
| kfox1111 | hi duonghq | 13:53 |
| *** yingjun has joined #openstack-kolla | 13:54 | |
| *** portdirect_ is now known as portdirect_back_ | 13:54 | |
| duonghq | alright, 3-node k8s and current memory usage is around 60% | 13:55 |
| duonghq | of 16GB RAM | 13:55 |
| *** jtriley has quit IRC | 13:55 | |
| *** lamt has joined #openstack-kolla | 13:56 | |
| *** jheroux has joined #openstack-kolla | 13:58 | |
| *** jtriley has joined #openstack-kolla | 14:01 | |
| sdake | sup kfox1111 | 14:01 |
| sdake | kfox1111 have 1-2 hours to get into the api discussion if sbezverk and portdirect_back_ are about? | 14:02 |
| kfox1111 | yeah | 14:02 |
| sdake | cool lets see if they are around | 14:02 |
| *** portdirect_back_ is now known as portdirect_ | 14:03 | |
| portdirect_ | back | 14:03 |
| *** rmart04_ has joined #openstack-kolla | 14:04 | |
| sdake | well we need sbezverk to make a decision :) | 14:04 |
| *** rmart04 has quit IRC | 14:04 | |
| *** rmart04_ is now known as rmart04 | 14:04 | |
| kfox1111 | yeah. | 14:05 |
| *** senk has joined #openstack-kolla | 14:05 | |
| sbezverk | good morning kfox111: sdake: | 14:05 |
| sdake | sup dude | 14:05 |
| sdake | ok lets get into it | 14:06 |
| kfox1111 | morning. :) | 14:06 |
| kfox1111 | ready.... go! :) | 14:06 |
| sdake | ok where we are - i think is as follows | 14:06 |
| sbezverk | so what is new and exciting? | 14:06 |
| sdake | it is clear to me that people want helm native charts | 14:06 |
| sdake | it is clear to me that people want common code | 14:06 |
| sdake | so yesterday we finished up on the common code values.yaml | 14:07 |
| sdake | is that fair assessment? | 14:07 |
| kfox1111 | not sure about that. | 14:07 |
| sdake | which part | 14:07 |
| kfox1111 | I think we are mostly in agreement on needing a common macros file? | 14:08 |
| sdake | i'm in agreement there | 14:08 |
| portdirect_ | yep | 14:08 |
| sbezverk | wfm | 14:08 |
| kfox1111 | k. | 14:08 |
| sdake | so when i said above people want- i didn't say people agreed ;) | 14:08 |
| kfox1111 | what about common vars to go along with the common macros? | 14:09 |
| sdake | ok common vars | 14:09 |
| sdake | this is values.yaml right? | 14:09 |
| portdirect_ | yes - afaik | 14:09 |
| kfox1111 | yeah. values.yaml combining multiple files. | 14:10 |
| sdake | coudl you expand on that combinging part | 14:10 |
| kfox1111 | yeah. sec | 14:10 |
| sbezverk | kfox1111: could you clarify which multiple files you refer to? | 14:10 |
| kristian__ | Hi where can I fix this error when I compiled libvirt 1.2.16? http://paste.openstack.org/show/590817/ | 14:11 |
| portdirect_ | kfox1111: could I try and then you correct me? | 14:11 |
| kfox1111 | https://review.openstack.org/#/c/396296/27/helm/src/common_values.yaml | 14:11 |
| sdake | portdirect_ dont sk for permission dude ask for forgiveness :) | 14:11 |
| NachoDuck | Hi all, is there still support for LDAP auth in Kolla, has anyone used it? | 14:11 |
| kfox1111 | for example, that file is appended to: | 14:11 |
| kfox1111 | https://review.openstack.org/#/c/396296/27/helm/src/neutron/values/l3_agent_daemonset.yaml | 14:11 |
| *** jtriley has quit IRC | 14:11 | |
| kfox1111 | when the l3_agent package is built. | 14:11 |
| kfox1111 | and | 14:11 |
| kfox1111 | https://review.openstack.org/#/c/396296/27/helm/src/neutron/values/openvswitch_agent_daemonset.yaml | 14:12 |
| kfox1111 | when openvswitch-agent is built. | 14:12 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Enable cloudkitty ui when enable_cloudkitty is yes https://review.openstack.org/403520 | 14:12 |
| kfox1111 | so the vars for the common stuff is defined just once, in the common vars file. | 14:12 |
| sdake | ok seems like a worthwhile thing to have | 14:12 |
| sdake | the common vars part ;) | 14:13 |
| sbezverk | kfox1111: in general there will be just 2 logical type of variables: common and microservice specific right? | 14:13 |
| kfox1111 | sbezverk: at the microservice level, yes. | 14:13 |
| kfox1111 | when combined into service packages, no. | 14:13 |
| *** senk has quit IRC | 14:13 | |
| sdake | ok lets focus on microservice level at present | 14:14 |
| sdake | thinking about all the rpoblems is npcomplete ;) | 14:14 |
| kfox1111 | sdake: the distinction is important as it determines if package vars, or globals should be used. | 14:14 |
| *** jascott1 has joined #openstack-kolla | 14:14 | |
| *** yingjun has quit IRC | 14:14 | |
| Jeffrey4l | YefimG, which branch code are u using? | 14:14 |
| *** yingjun has joined #openstack-kolla | 14:15 | |
| *** v1k0d3n has quit IRC | 14:15 | |
| *** v1k0d3n has joined #openstack-kolla | 14:15 | |
| sbezverk | kfox1111: I would agree with sdake to go by layer, it is easier imho | 14:17 |
| kfox1111 | k | 14:17 |
| sbezverk | starting from ground up | 14:17 |
| sdake | kfox1111 tbh you just melted my brain with that statement ;) | 14:17 |
| sdake | kfox1111 so lets learn this together - since you have context I dont :) | 14:18 |
| sdake | one layer at a time | 14:18 |
| kfox1111 | we'll table it for now. | 14:18 |
| sdake | sounds good | 14:18 |
| kfox1111 | k. | 14:18 |
| YefimG | Jeffrey4l: I am using 3.0.1 from docker hub | 14:18 |
| sdake | ok so back on the variables | 14:18 |
| sdake | common and microservice specific | 14:18 |
| Jeffrey4l | YefimG, docker logs bootstrap_gnocchi, what u get? | 14:18 |
| *** jascott1 has quit IRC | 14:19 | |
| *** yingjun has quit IRC | 14:19 | |
| sdake | seems like this is doable in many different ways | 14:19 |
| YefimG | Jeffrey4l: http://paste.fedoraproject.org/492994/80429173 (I am rebuilding with Ceph now) | 14:19 |
| sdake | i'd like to keep a helm-native packaging approach if at all possible | 14:19 |
| sdake | this is something i personally want | 14:19 |
| kfox1111 | sdake: not sure thats true. | 14:20 |
| kfox1111 | if they are to be done as package vars, they have to be in variables.yaml. | 14:20 |
| sdake | kfox1111 not sure which is true | 14:20 |
| kfox1111 | and there is only one file. | 14:20 |
| kfox1111 | and its not templated. :/ | 14:20 |
| sdake | ok templating - we can tackle next :) | 14:20 |
| pbourke | berendt: ping | 14:20 |
| sdake | alanmeadows listed out 3 approaches yesterday | 14:21 |
| berendt | pbourke: pong | 14:21 |
| sdake | are there any others? | 14:21 |
| pbourke | berendt: could you fix up https://review.openstack.org/#/c/329487/ for merge? | 14:21 |
| portdirect_ | not that im aware of | 14:21 |
| sdake | none of those approaches provide native helm implementation | 14:21 |
| Jeffrey4l | YefimG seems there are some bug in centos+binary. checking | 14:21 |
| kfox1111 | sdake: yeah. | 14:21 |
| kfox1111 | but, | 14:21 |
| portdirect_ | no :( the makefile solution they are using is probably the closest | 14:21 |
| kfox1111 | something that could pretty easily be added to helm though I think. | 14:22 |
| kfox1111 | all they need is like a vars.d magicdir. | 14:22 |
| sdake | so i am working on combining native helm + our requirements for global and local variables | 14:22 |
| sdake | yup but we can't wait on our upstreams | 14:22 |
| kfox1111 | right. | 14:22 |
| sdake | we chose our dependencies, we have to suck it up with what they deliver to us | 14:22 |
| kfox1111 | I'm pretty sure its either or. :/ | 14:22 |
| kfox1111 | until helm matures. | 14:22 |
| sdake | ok so I kind of hda this wierd drema last night | 14:22 |
| *** huzhengchuan_ has joined #openstack-kolla | 14:23 | |
| sdake | its a rift on your idea kfox1111 | 14:23 |
| sdake | i wont get into the wierd part of it | 14:23 |
| sdake | since thats rated R:) | 14:23 |
| kfox1111 | :) | 14:23 |
| portdirect_ | oh god | 14:23 |
| sdake | or probably X:) | 14:23 |
| openstackgerrit | Christian Berendt proposed openstack/kolla: Install neutron-lbaas plugin in neutron-server container https://review.openstack.org/329487 | 14:23 |
| Jeffrey4l | YefimG, it is a bug in 3.0.1 . it is fixed on master ;( | 14:23 |
| sdake | anyway | 14:23 |
| sdake | back on topic | 14:23 |
| *** mdnadeem has quit IRC | 14:23 | |
| sdake | part of that was the idea of a compiler | 14:23 |
| berendt | pbourke: done | 14:24 |
| pbourke | berendt: thanks! | 14:24 |
| pbourke | Jeffrey4l: sdake: small review for approval https://review.openstack.org/329487 | 14:24 |
| sdake | something that reads the helm files (pre-checkin) and builds the values.yaml file | 14:24 |
| sdake | this would be a new option that is unexplored | 14:24 |
| Jeffrey4l | YefimG, two option. 1 build centos+binary using branch code 2. use centos+source. | 14:24 |
| kfox1111 | pre-checkin? | 14:24 |
| sdake | it would result in duplication | 14:24 |
| kfox1111 | you mean git? | 14:24 |
| sbezverk | sdake: I think this could work with native charts too | 14:25 |
| sdake | ya you run the compiler - it reads the template - it spits out a values.yaml - you check that into git | 14:25 |
| Jeffrey4l | pbourke, done. waiting for the gate. | 14:25 |
| kfox1111 | oh.... | 14:25 |
| sbezverk | sdake: when you need to change variable? | 14:25 |
| portdirect_ | ok - why pre git? | 14:26 |
| kfox1111 | thats putting built things in git. | 14:26 |
| sdake | portdirect_ that provides helm-native | 14:26 |
| sdake | just values.yaml | 14:26 |
| sdake | not hte whole pakage itself | 14:26 |
| kfox1111 | yeah, helm native, but breaks a cardinal source code revision rule. | 14:26 |
| YefimG | Jeffrey4l: good, thanks, will clone the master and build my own images and deploy kolla. Thanks! By the way, speaking of gnocchi, is it faster then ceilometer + mongodb? The ceilometer + mongodb was terribly slow last time I used it :( | 14:27 |
| portdirect_ | i see - this works for that but sbezverk's point is where my concern come from here | 14:27 |
| Jeffrey4l | YefimG, yes. it is faster. and you can try gnocchi+ceph , it is more better ;) | 14:27 |
| YefimG | Jeffrey4l: Thank you very much! | 14:27 |
| Jeffrey4l | ;) | 14:27 |
| kfox1111 | hmm... | 14:28 |
| *** krtaylor has quit IRC | 14:28 | |
| sdake | portdirect_ i dont follow | 14:28 |
| kfox1111 | my initial reaction is it feels a bit slimy. but, it would be a stop gap until helm supports templating. | 14:28 |
| sbezverk | portdirect_: sorry man which concern you are refering too, I had and lost so many it is hard to track ;-) | 14:28 |
| portdirect_ | "sdake: when you need to change variable?" | 14:29 |
| kfox1111 | so, I'm not against it. | 14:29 |
| *** fguillot has quit IRC | 14:29 | |
| sbezverk | portdirect_: got it, it was not a concenr, just question | 14:29 |
| kfox1111 | sdake: question. would it be good to do the same for the docker containers? | 14:29 |
| portdirect_ | sorry phrasing :) | 14:29 |
| kfox1111 | prebuild the dockerfiles and check them into kolla so docker builds just work? | 14:29 |
| sdake | kfox1111 actually we did this same thing for docker containers :) | 14:29 |
| sbezverk | I think since this file is checked in, you could edit it manually and then check back in | 14:29 |
| sdake | at one point in our history | 14:30 |
| kfox1111 | sbezverk: thats the danger in the aproach. | 14:30 |
| sdake | nah rerun compiler | 14:30 |
| kfox1111 | sdake: yeah, but if someone does it not knowing its a built file, then the nextg person rebuilding might drop the changes. | 14:30 |
| kfox1111 | by mistake | 14:30 |
| sbezverk | kfox1111: that was just a thought, re-run looks doable from my point of view too | 14:30 |
| sdake | that is what the 2 +2 gating is all about :) | 14:31 |
| kfox1111 | sdake: k | 14:31 |
| kfox1111 | sbezverk: yeah. | 14:31 |
| portdirect_ | oh if we can do that - and changes ripple thorugh a checked out repo, then I'm down with it | 14:31 |
| portdirect_ | could we create a gate that checked the default values dont change? | 14:31 |
| kfox1111 | portdirect_: probably a gate taht does a 'make clean; make; git diff' | 14:32 |
| sdake | ok so there is helm native + global variables + local variables | 14:32 |
| kfox1111 | if values change, then someone edited the wrong file. | 14:32 |
| sdake | ya linting that is easy -we also followed htis pattern in kolla early on (the linting part) | 14:32 |
| kfox1111 | sdake: built charts, but prebuilt in the repo so helm native works out of the box. | 14:32 |
| sbezverk | portdirect_: what if we need to change defaults? would it not be more important to check if with new global variables it gets built/deployed? | 14:32 |
| kfox1111 | sdake: but. | 14:32 |
| kfox1111 | I'm not agreeing fully to it yet. just that this one use case is covered by that idea. | 14:33 |
| kfox1111 | lets make sure it works with the rest. | 14:33 |
| *** inc0 has joined #openstack-kolla | 14:33 | |
| sdake | which rest, you mean services? | 14:33 |
| inc0 | good morning | 14:33 |
| v1k0d3n | portdirect_: you there? | 14:33 |
| sdake | sup inc0 | 14:33 |
| portdirect_ | sbezverk: then we would need to change the gate - a bit crusty but they should not change oftern | 14:34 |
| kfox1111 | sdake: no. just the other pieces of the puzzle. we need to poc it and make sure it doesn't have any hiden gotchas. | 14:34 |
| v1k0d3n | looking at your changes right now, and going to going to dive in a bit on this today. | 14:34 |
| portdirect_ | v1k0d3n: hey man | 14:34 |
| sdake | kfox1111 right - so compiler idea | 14:34 |
| sdake | complicated | 14:34 |
| sdake | is there a gotpl parser for python | 14:34 |
| sdake | doubt i t | 14:34 |
| kfox1111 | not taht I know of. | 14:34 |
| sdake | i could write a parser | 14:34 |
| sdake | i could write a compioelr | 14:34 |
| inc0 | kfox1111, sdake v1k0d3n can you bried me on problem please? | 14:35 |
| sdake | it would tkae a couple weeks | 14:35 |
| kfox1111 | we could build one, but then would need go source in our repo. | 14:35 |
| sdake | i'd rather move faster | 14:35 |
| inc0 | no...let's *not* write template parser | 14:35 |
| sdake | kfox1111 right - not going there | 14:35 |
| sdake | inc0 please read scrollback | 14:35 |
| kfox1111 | yeah. don't think we want to go there. | 14:35 |
| v1k0d3n | inc0: what's up man? | 14:35 |
| kfox1111 | don't think we need much out of the build step either. | 14:35 |
| sbezverk | sdake: question, if the compiler only deals with values.yaml which is native yaml, why would you need gotpl parser? | 14:35 |
| kfox1111 | right now I was able to do everything I need to with a simple append or a str replace. | 14:35 |
| sdake | sbezverk to read the daemonset.yaml gotpl stuff | 14:36 |
| kfox1111 | values, and chart too. | 14:36 |
| *** krtaylor has joined #openstack-kolla | 14:36 | |
| kfox1111 | some kind of easy to manage version numbering system will be important too I think. | 14:36 |
| v1k0d3n | inc0: what do you mean by " bried me on problem please"? | 14:36 |
| sbezverk | sdake: kfox1111: to get list of varibles used in the template and then build values.yaml for these varibles? | 14:37 |
| inc0 | v1k0d3n, writing is hard - brief me on problem | 14:37 |
| *** jascott1 has joined #openstack-kolla | 14:37 | |
| v1k0d3n | portdirect_: have a feeling that the changes may not work on ubuntu...but i can change that. | 14:37 |
| kfox1111 | sbezverk: yeah, for values. (if I understand the question right) | 14:37 |
| inc0 | I barely waken up but already see argument | 14:37 |
| v1k0d3n | inc0: not sure of a problem. is there a problem? | 14:37 |
| v1k0d3n | i just got on. | 14:37 |
| sdake | inc0 we are not arguing at all | 14:37 |
| v1k0d3n | well...just started getting spun up i mean. | 14:38 |
| kfox1111 | inc0: not argument. working on concensus on an architectural issue. | 14:38 |
| portdirect_ | v1k0d3n: no probs - thought there might be a few issues | 14:38 |
| sdake | inc0 we are working together -there is a difference | 14:38 |
| sbezverk | kfox1111: that is right, that was my question.. | 14:38 |
| v1k0d3n | portdirect_: should be pretty easy. | 14:38 |
| v1k0d3n | should have in a little bit. | 14:38 |
| inc0 | fine...what are you working together on then? can I join?;) | 14:38 |
| sdake | inc0 absolutelly | 14:38 |
| sdake | read scrollback :) | 14:38 |
| v1k0d3n | inc0 kfox1111 sdake what is the issue? | 14:38 |
| sdake | inc0 peoples brains are loaded with state atm | 14:38 |
| sdake | you need to load yours | 14:38 |
| sdake | v1k0d3n scrollback ;) | 14:39 |
| sbezverk | kfox1111: I like this idea, it will generate very optimal values.yaml. I just hope it will not be super complicated to write it.. | 14:39 |
| v1k0d3n | kfox1111 are you talking with alanmeadows on our side? | 14:39 |
| *** dasm has quit IRC | 14:39 | |
| sdake | ok guys - compiler = pita | 14:39 |
| sdake | anything simpler? | 14:39 |
| v1k0d3n | i think a lot of the issues we have worked out, or have workarounds for now...knowing that some things will clean up soon. | 14:39 |
| *** rhallisey has joined #openstack-kolla | 14:39 | |
| kfox1111 | v1k0d3n: yeah. talked with him for a while. both on the same page. | 14:40 |
| v1k0d3n | ok cool | 14:40 |
| sdake | v1k0d3n we are on to a new pattern alanmeadows hasn't explored ;) | 14:40 |
| sbezverk | what about to have global values.yaml with sections per microservice? | 14:40 |
| v1k0d3n | hopefully we're working similar paths? | 14:40 |
| sbezverk | example: | 14:40 |
| sbezverk | openvswitch-ovsdb: | 14:40 |
| kfox1111 | sdake: I'd say its the same pattern but a slightly different workflow. | 14:40 |
| sbezverk | - variable-1: "blah" | 14:40 |
| kfox1111 | the build step is pushed before checkin rather hten after checkout. | 14:40 |
| v1k0d3n | sdake: that's why i asked if people are syncing up with him... | 14:40 |
| sdake | kfox1111 fair enough | 14:40 |
| portdirect_ | v1k0d3n: yeah allanmeadows was really helpful | 14:41 |
| kfox1111 | but architecturally is the same. | 14:41 |
| inc0 | sbezverk, this file will be hard to maintain | 14:41 |
| sdake | its helm native - i think that is one big architectural difference ;) | 14:41 |
| sdake | but thats besides the point :) | 14:41 |
| *** Marcellin__ has joined #openstack-kolla | 14:41 | |
| kfox1111 | sbezverk: thats possible, but might require a more complicated engine to parse out the irelevent bits? | 14:41 |
| sbezverk | inc0: we already have something similar in kolla-kubernetes service registry file | 14:41 |
| v1k0d3n | sdake: that is only a small part of a much larger open field honestly. | 14:41 |
| v1k0d3n | it's a good start. | 14:41 |
| YefimG | Jeffrey4l: I am sorry for the lame question, but how do I correctly install kolla from the master branch? I have cloned it, installed dependencies, installed kolla (pip install kolla/), but it doesn't have anything else but kolla-build and it doesn't work. | 14:42 |
| kristian__ | How can I install libvirt 2.1 on ubuntu kolla? | 14:42 |
| *** jascott1 has quit IRC | 14:42 | |
| v1k0d3n | but after that...it can sprawl pretty significantly. | 14:42 |
| Jeffrey4l | YefimG, you can try centos+source + 3.0.1 now. | 14:42 |
| Jeffrey4l | normally, source is well tested. | 14:42 |
| sdake | sbezverk ok file - lets focus on digging into that | 14:42 |
| inc0 | YefimG, kolla will only have kolla-build, kolla-ansible will have deploy | 14:43 |
| sdake | hard to maintain - who cares | 14:43 |
| inc0 | separate project now | 14:43 |
| Jeffrey4l | YefimG, for install from master branch, you can use ./tools/kolla-build.py script. | 14:43 |
| sbezverk | kfox1111: I am not sure it will add complexity, because to generate per microservice values all you need is to copy relevat section | 14:43 |
| sdake | inc0 separate deliverable | 14:43 |
| sbezverk | to local micsroservice values.yaml | 14:43 |
| kfox1111 | sbezverk: the complexity is in parsing. right now, I just append some files together and do str replace. thats simpler then doing some kind of switch in the file? | 14:44 |
| YefimG | Jeffrey4l, inc0: Thank you, will try that. | 14:44 |
| kfox1111 | we could use jinja2 for that. but thats more complicated. | 14:44 |
| portdirect_ | but we would only need to modify the values.yaml on a 2nd run wouldnt we? | 14:44 |
| kfox1111 | I'm ok with it. just want people to understand/accept the complexity. | 14:45 |
| kfox1111 | as people were worried about the append/sed being complex. | 14:45 |
| sdake | kfox1111 ok - so helm native | 14:46 |
| sbezverk | kfox1111: sdake: maybe we could list possible scenarios of workflow to better understand complexities. | 14:46 |
| kfox1111 | sbezverk: yeah. good idea | 14:46 |
| sdake | ok workflow I see is this | 14:46 |
| sdake | 1. dev creates a new chart | 14:47 |
| sdake | 2. dev adds values to some global file somewhere that contains the values that should be yaml.loads()'ed and chopped up into a values.yaml in 1 | 14:47 |
| sdake | 3. dev runs some type of tool that does the chopping | 14:47 |
| sdake | 3 = not a compiler but a simple parser | 14:48 |
| sdake | another workflow | 14:48 |
| sdake | 1. dev modifies a chart | 14:48 |
| sdake | 2. dev adds values to some global file somewhere just like 2 | 14:48 |
| sdake | 3. dev runs some type of tool thatdoes the chopping | 14:48 |
| kfox1111 | yeah. | 14:49 |
| sdake | so its the same workflow? | 14:49 |
| sbezverk | sdake: make sense, in 3 chopping is done only on values.yaml file right? | 14:49 |
| sdake | with 1 being a delta | 14:49 |
| kfox1111 | yeah. if you make any changes to vars, then you run the tool. | 14:49 |
| *** srwilkers_ has joined #openstack-kolla | 14:50 | |
| kfox1111 | sbezverk: I'd kind of like to see it done with charts too. | 14:50 |
| sdake | and if someone makes changes to vars manually the linting -1's their patch | 14:50 |
| kfox1111 | bumping a version should be easy for an op to do. | 14:50 |
| sbezverk | kfox1111: not sure about you but I could live with running tool when value needs to be changes since it does not change too often | 14:50 |
| kfox1111 | I like your single file idea. what about putting all the package version numbers in one file? | 14:50 |
| kfox1111 | sbezverk: me too. | 14:50 |
| portdirect_ | I could live with that as well | 14:51 |
| kfox1111 | sbezverk: it being yaml solves the parsing complexity too. | 14:51 |
| *** rmart04 has quit IRC | 14:51 | |
| sbezverk | wow we are getting somewher ;-) | 14:51 |
| kfox1111 | you just select the root level key for your package, and common. | 14:51 |
| kfox1111 | easy to parse... | 14:51 |
| *** lrensing has joined #openstack-kolla | 14:51 | |
| v1k0d3n | sdake: to be honest...we're doing everything via jenkins builds...and we're providing that as a helm chart. | 14:52 |
| kfox1111 | v1k0d3n: yeah. thats one of hte use cases I'm shooting for. | 14:52 |
| kfox1111 | having it easy for a ci/cd bump versions of charts. | 14:52 |
| v1k0d3n | so jenkins fires up a kubernetes node for testing, building, etc. | 14:52 |
| sdake | v1k0d3n np-complete problem :) | 14:52 |
| YefimG | Jeffrey4l: sorry, but what do mean by centos+source? | 14:52 |
| *** rmart04 has joined #openstack-kolla | 14:52 | |
| sdake | v1k0d3n i am solving an O(n) problem atm :) | 14:52 |
| *** srwilkers has quit IRC | 14:53 | |
| kfox1111 | thats why I think we need to template the chart.yaml too | 14:53 |
| *** srwilkers has joined #openstack-kolla | 14:53 | |
| v1k0d3n | ok...i'll back down then. | 14:53 |
| sdake | sup srwilkers | 14:53 |
| kfox1111 | (until helm gets a command line override for version number) | 14:53 |
| sdake | Marcellin__ was looking for ya | 14:53 |
| v1k0d3n | i'll be around if anyone pings me. can't watch IRC though... | 14:53 |
| sdake | ok so back on the api | 14:53 |
| sdake | v1k0d3n roger that - we should be done shortly nayway | 14:53 |
| sdake | does this deliver an api that is stable? | 14:54 |
| *** srwilkers_ has quit IRC | 14:54 | |
| portdirect_ | i think it could | 14:54 |
| kfox1111 | sdake: I think so, provided we're careful not to rename vars too. | 14:54 |
| *** srwilkers_ has joined #openstack-kolla | 14:54 | |
| sdake | api wins all arguments here | 14:54 |
| sdake | so if it doesn't this solution isn't viable | 14:55 |
| sbezverk | kfox1111: what exactly you want to change in chart? | 14:55 |
| kfox1111 | sbezverk: just the version number. | 14:55 |
| sbezverk | kfox1111: let me re-phrase do you mean Chart.yaml ot template? | 14:56 |
| kfox1111 | I'd like it to be very easy to release updated versions. | 14:56 |
| *** srwilkers has quit IRC | 14:56 | |
| kfox1111 | Chart.yaml contains the version number. | 14:56 |
| *** srwilkers has joined #openstack-kolla | 14:56 | |
| srwilkers_ | hey everyone -- hope you had a decent enough holiday | 14:56 |
| sdake | srwilkers_ welcome back | 14:56 |
| sdake | srwilkers_ Marcellin__ was looking for you | 14:56 |
| sdake | srwilkers_ i think he mentioned something about taking over the documentation blueprint you had | 14:57 |
| sbezverk | kfox1111: ah, got it, sure thing.. if it makes your life easier and not break any native support I do not see why not! | 14:57 |
| sdake | not sure, you will have to tlak to him - feel freee to use the channel :) | 14:57 |
| kfox1111 | cool. | 14:57 |
| sdake | ok well i'm not totally up to speed on the Chart.yaml thing | 14:58 |
| portdirect_ | I concurr - so we limit the files the parser touches to varibles.yaml and Chart.yaml? | 14:58 |
| sdake | i'll take on implementing this parser | 14:58 |
| sdake | since i'm dead in the water without a dev environment anyway | 14:58 |
| *** Pavo has quit IRC | 14:58 | |
| sdake | what shall we call the tool | 14:58 |
| sdake | "chop"? :) | 14:58 |
| kfox1111 | portdirect_: +1. I think thats all thats needed. the rest helm can deal with. | 14:59 |
| kfox1111 | sdake: I think the build tool from my review is pretty close already. | 14:59 |
| sbezverk | portdirect_ +1 | 14:59 |
| sdake | kfox1111 you want to tackle it then? | 14:59 |
| kfox1111 | I think theres one more thing to talk through? | 15:00 |
| sdake | ok before we get on to the next thing | 15:00 |
| sdake | whatever tht may be ;) | 15:00 |
| kfox1111 | k. | 15:00 |
| sdake | are we in general agreement this approach is better then others - it sort of checks all the tickboxes everyone wants | 15:00 |
| kfox1111 | sdake: yeah. I can make a few tweaks and get it into the agreed aproach this morning I think. | 15:00 |
| sdake | kfox1111 fantastic | 15:01 |
| portdirect_ | sdake: I'm a happy camper atm | 15:01 |
| sdake | it would take me a couple days | 15:01 |
| sdake | sbezverk do you agree with approach? | 15:01 |
| kfox1111 | sdake: I'll let you know when its ready, then you can double check it for me? | 15:01 |
| sbezverk | kfox1111: you will change templates to be native templates and not macros right? | 15:01 |
| sdake | kfox1111 just push it in review queue | 15:01 |
| sdake | kfox1111 then everyone can doublecheck it :) | 15:02 |
| kfox1111 | sbezverk: want to be extra sure what your talking about. you talking about the common_main thing from my review? | 15:02 |
| *** Pavo has joined #openstack-kolla | 15:03 | |
| portdirect_ | sbezverk, kfox1111: yeah it would need to use common templates and dir structure, so the only 'non' standard part was values.yaml | 15:03 |
| sbezverk | kfox1111: each your microservice template starts with {{ define "blah" }} which makes is as a macro, helm does not use it in this way as it needs to be included somewhere | 15:04 |
| sdake | portdirect_ nah values.yaml would be standard because it was checked into rpeo that way | 15:04 |
| kfox1111 | sbezverk: that is actually a helm native thing. | 15:04 |
| portdirect_ | and even that should be standard (damn stake you beat me) | 15:04 |
| sdake | agree helm native | 15:04 |
| *** TxGirlGeek has joined #openstack-kolla | 15:04 | |
| sbezverk | kfox1111: right but it requires additinal manipulation by your build tool | 15:05 |
| kfox1111 | helm is parsing the macro. | 15:05 |
| kfox1111 | it requires one file, common_main to be added to the chart. | 15:05 |
| kfox1111 | and that one file can contain all the common var building logic. | 15:05 |
| kfox1111 | so, for example, kollaToolboxImageFull needs to be built and used in a lot of templates. | 15:05 |
| kfox1111 | that var building can either be copy pasted into all of the templates, or put in one place. | 15:06 |
| sbezverk | kfox1111: I thought we agreedd that all common variables will be in values.yaml file | 15:06 |
| kfox1111 | sbezverk: there are two parts to it. | 15:06 |
| sdake | sbezverk you mean a global-values.yaml | 15:06 |
| kfox1111 | variable defintions. thats what variables.yaml is for in helm. | 15:06 |
| kfox1111 | then there is logic. like: | 15:07 |
| kfox1111 | {{- $kollaToolboxImageBuilt := printf "%s/%s/%s-%s-kolla-toolbox:%s" .Values.docker_registry .Values.docker_namespace .Values.kolla_base_distro .Values.kolla_install_type .Values.kolla_toolbox_image_tag }} | 15:07 |
| kfox1111 | {{- $kollaToolboxImageFull := .Values.kolla_toolbox_image_full | default $kollaToolboxImageBuilt }} | 15:07 |
| sbezverk | sdake: the problem to make kfox1111 temaplate to be deployable by helm it rquires manipulation with temaplte | 15:07 |
| kfox1111 | kind of a common-lib kind of thing, | 15:08 |
| kfox1111 | but helm does not allow a macro to return anything. | 15:08 |
| kfox1111 | :/ | 15:08 |
| sbezverk | kfox1111: why we cannot use _helper.tpl for common code? | 15:08 |
| portdirect_ | ^^ | 15:08 |
| kfox1111 | sbezverk: because vars are one directional. | 15:08 |
| kfox1111 | from caller to callee. | 15:08 |
| kfox1111 | so if you define kollaToolboxImageFull in _helper.tpl, | 15:08 |
| kfox1111 | then the main template can't use it for its init container. | 15:09 |
| kfox1111 | it has to build the variable itself. | 15:09 |
| *** jtriley has joined #openstack-kolla | 15:09 | |
| kfox1111 | its a current helm limitation. :/ | 15:09 |
| sdake | ok so does that mean your proposing having a diferent workflow then the two i listed above? | 15:09 |
| kfox1111 | uness we can find a different workaround. | 15:09 |
| *** Jeffrey4l has quit IRC | 15:09 | |
| portdirect_ | I think we just need to accept that - and have templates that are a bit less DRY than you would like - as otherwise we are moving awa from sdakes suggested workflow i think | 15:10 |
| kfox1111 | sdake: not really? its just a proposal to have the build tool drop in one more file. | 15:10 |
| sdake | ok lets dig into that kfox1111 | 15:10 |
| kfox1111 | it has the build tool drop in https://review.openstack.org/#/c/396296/27/helm/src/common_main.yaml | 15:10 |
| sdake | sounds promising | 15:10 |
| sbezverk | kfox1111: but this file is code file not variables/values | 15:10 |
| kfox1111 | and then wrap the main template in a macro. | 15:11 |
| kfox1111 | everything else is the same. | 15:11 |
| kfox1111 | sbezverk: true. | 15:11 |
| sbezverk | it means any changes in code would require to run tool again | 15:11 |
| kfox1111 | but its copied verbatim. | 15:11 |
| kfox1111 | true. | 15:11 |
| sdake | changes in which code | 15:11 |
| kfox1111 | but would we do the same for common lib? | 15:11 |
| sbezverk | how then it is native to helm?? I fail to see | 15:11 |
| alanmeadows | is this just about yaml merging some top level input? | 15:11 |
| sdake | run which tool | 15:11 |
| kfox1111 | we have to copy _helper.tpl in to the package too, which then has the same issue there too. | 15:12 |
| kfox1111 | if you change the common code, you have to run the build tool again to copy it everywhere. | 15:12 |
| sbezverk | kfox1111: it is done only once | 15:12 |
| sdake | kfox1111 right - to copy that _helper.tpl is easy - its a find operaiton away to resync the whole tree | 15:12 |
| kfox1111 | sbezverk: same with common_main.yaml? | 15:12 |
| *** jtriley has quit IRC | 15:12 | |
| inc0 | alanmeadows, as I'm trying to follow this discussion, it seems that's it's about reusing a variables across multiple charts | 15:12 |
| inc0 | which helm doesn't feel like doing | 15:13 |
| kfox1111 | the tool builds variables.yaml and charts.yaml (version only), then drops in common_main and common_lib ? | 15:13 |
| kfox1111 | inc0: right. | 15:13 |
| alanmeadows | again separating out values from defines you can merge even in a helm native way without an uber chart | 15:13 |
| inc0 | can helm templates be rendered with more than one yaml or that's the problem? | 15:13 |
| alanmeadows | helm install --values=global-uber-values-to-merge-in.yaml foo/bar --namespace=blah | 15:13 |
| sbezverk | building variables/values/code file on the fly does not sounds as good approach | 15:13 |
| kfox1111 | inc0: helm templates can, but vars are one direction. | 15:14 |
| inc0 | kfox1111, what do you mean by one direction? | 15:14 |
| sdake | kfox1111 that seems workable - if the tool only needs to be run once - pre commit | 15:14 |
| *** jtriley has joined #openstack-kolla | 15:14 | |
| kfox1111 | inc0: in gotl a template can define vars, but can't return them to the parent. | 15:14 |
| sdake | kfox1111 i'd encourage you to simplify things into multiiple patches if your going to do the implementation | 15:14 |
| kfox1111 | sdake: yeah. onece on precommit. | 15:14 |
| kfox1111 | sdake: I can do that. | 15:15 |
| sdake | kfox1111 one for values.yml | 15:15 |
| sdake | then add on each patch from there | 15:15 |
| sdake | i understand your idea | 15:15 |
| inc0 | kfox1111, but don't we define variables in yaml? | 15:15 |
| sdake | when its mixed up in one big ball of tool its hard to understand ;) | 15:15 |
| kfox1111 | inc0: two different pieces. | 15:15 |
| kfox1111 | vars are defined/defaulted in variables.yaml. | 15:15 |
| kfox1111 | logic in the templates can devine other vars based on them. | 15:16 |
| kfox1111 | so, templates can do this: | 15:16 |
| kfox1111 | {- $kollaToolboxImageBuilt := printf "%s/%s/%s-%s-kolla-toolbox:%s" .Values.docker_registry .Values.docker_namespace .Values.kolla_base_distro .Values.kolla_install_type .Values.kolla_toolbox_image_tag }} | 15:16 |
| kfox1111 | {{- $kollaToolboxImageFull := .Values.kolla_toolbox_image_full | default $kollaToolboxImageBuilt }} | 15:16 |
| kfox1111 | and then use $kollaToolboxImageFull | 15:16 |
| inc0 | ahh I understand, in ansible we do have jinja2-based yamls | 15:16 |
| kfox1111 | but that can't be done in an included file. | 15:16 |
| kfox1111 | right. | 15:16 |
| kfox1111 | hmm.... | 15:16 |
| kfox1111 | though... | 15:17 |
| inc0 | we could use genconfig | 15:17 |
| kfox1111 | a macro can return just some text... | 15:17 |
| inc0 | kinda | 15:17 |
| sdake | lets keep it simple to start plz :) | 15:17 |
| kfox1111 | and I don't think we really need the var to be non text for the common bits... | 15:17 |
| alanmeadows | kfox1111: thats what we''re doing | 15:17 |
| kfox1111 | so we could maybe just make a macro | 15:17 |
| kfox1111 | that returns the kollaToolboxImageFull value, | 15:17 |
| kfox1111 | then call it inline. | 15:17 |
| sdake | kfox1111 question | 15:18 |
| kfox1111 | the template will look a bit uglier, as its calling a macro, not a var. | 15:18 |
| kfox1111 | but wouldn';t need the common_main thingy. | 15:18 |
| kfox1111 | I can prototype that... | 15:18 |
| kfox1111 | sdake: sure. | 15:18 |
| sdake | kfox1111 i htink we are in agreemenet on a global-variables.yml file that gts chopped up | 15:18 |
| kfox1111 | alanmeadows: have an example? | 15:18 |
| alanmeadows | to be sure though what I am thinking is _common.tpl includes common macros that really are macros, for actual strings that belong in yaml for chart straddling values I am leaning toward requiring helm install --values=aic-helm-big-default.yaml ... | 15:19 |
| kfox1111 | sdake: yeah. | 15:19 |
| alanmeadows | https://github.com/att-comdev/aic-helm/blob/master/openstack-base/templates/_hosts.tpl | 15:19 |
| alanmeadows | fqdn in that case | 15:19 |
| alanmeadows | if I am following what you are saying | 15:19 |
| alanmeadows | but there is no reason that cant be in | 15:19 |
| sdake | kfox1111 the idea is helm native - what i'm not clera on with your approach is do you plan to render a complete helm file? | 15:19 |
| kfox1111 | alanmeadows: then you include them? | 15:19 |
| sbezverk | alanmeadows: multiple file can be specified in command line? | 15:19 |
| sdake | a complete helm subtree | 15:20 |
| kfox1111 | brb. | 15:20 |
| alanmeadows | kfox1111: right | 15:20 |
| alanmeadows | sbezverk: reading through https://github.com/kubernetes/helm/blob/62df1ad3588e8d4791423aec19d46bc383438c4f/pkg/chartutil/values.go#L131-L163 | 15:20 |
| alanmeadows | seems to merge, priority given to --values=... not sure about multiples | 15:20 |
| *** senk has joined #openstack-kolla | 15:20 | |
| sbezverk | alanmeadows: that is for multiple layers | 15:21 |
| sbezverk | alanmeadows: but then to access them they must be defined as global, is it not the case ? | 15:21 |
| *** huzhengchuan_ has quit IRC | 15:22 | |
| alanmeadows | why would they need to be globals? | 15:22 |
| kfox1111 | alanmeadows: nice. | 15:22 |
| alanmeadows | coalesce seems to me to merge it directly into the namespace | 15:22 |
| kfox1111 | alanmeadows: that eleviates the need to do the common_main thing I was doing. | 15:22 |
| kfox1111 | coalesce? | 15:22 |
| sdake | kfox1111 ok so before we go chase more shinies :) | 15:23 |
| sbezverk | alanmeadows: when I tried to do that, it was not rendering values if variable was not defined at the same level as template | 15:23 |
| sdake | (my daughter is a fan of pokemon if you didn't get the shinies rreference) | 15:23 |
| *** saneax is now known as saneax-_-|AFK | 15:23 | |
| *** mnasiadka has quit IRC | 15:24 | |
| sdake | lets focus on our immediate problem - which is helm native + API | 15:24 |
| *** cliles has quit IRC | 15:24 | |
| *** bmace has quit IRC | 15:24 | |
| sdake | the API problem raised is solved by chopping up a global-values.json | 15:25 |
| kfox1111 | sbezverk: he found a way to do it with amacro instead of a var. | 15:25 |
| sdake | this also solves the helm native problem | 15:25 |
| kfox1111 | with an {{include macroname}} instead of "{{$varname}}" | 15:25 |
| sbezverk | kfox1111: Just to be clear, the only reason I am being pain is because I want to preserve two options of deploying chart without any additional twiking. your approach force to do it, which makes me feel very uncomfortable.. | 15:25 |
| *** cliles has joined #openstack-kolla | 15:25 | |
| sdake | sbezverk hows that? | 15:26 |
| kfox1111 | I think we're down to just common_lib now. | 15:26 |
| kfox1111 | prebuild tool does: variables.yaml, chart.yaml (version) and drops in common_lib. | 15:26 |
| sbezverk | kfox1111's template is not a template which can be deploy by just runing helm install | 15:26 |
| *** hvlad has quit IRC | 15:26 | |
| kfox1111 | sbezverk: yeah. you want to skip the helm pacakge step. | 15:27 |
| sbezverk | it always requires some manipulation to be done with CODE, not just varaibles | 15:27 |
| kfox1111 | I get that. | 15:27 |
| sdake | ok i think there is another requirement that is hidden around in here | 15:27 |
| sbezverk | If we could some to an option where we could have both, that would be fantastic using sdake parlor | 15:28 |
| sdake | besides helm native + helm API | 15:28 |
| sbezverk | some/come | 15:28 |
| kfox1111 | sbezverk: I think the prebuild tool covers the deploy from raw chart case. | 15:28 |
| alanmeadows | To be clear, that _hosts.tpl example is an example of how we're doing it now, I wanted to experiment with requiring operators to do helm install --values=required-big-cross-chart-values.yaml (effect of uber chart without the pain). sbezverk mentions there may be namespace problems but its possible to work within that limitation | 15:28 |
| sdake | and that is whaty ou just said kfox1111 :) | 15:28 |
| *** hvlad has joined #openstack-kolla | 15:29 | |
| sdake | so helm native + helm API + deploy from raw chart | 15:29 |
| sbezverk | kfox1111: again pre-build for var/val works for me, pre-built for code does not.. | 15:29 |
| *** jtriley has quit IRC | 15:30 | |
| sdake | sbezverk mind digging in to the rationale there of the prebuilt for code does not | 15:30 |
| kfox1111 | sbezverk: arae you completely against a common lib then? | 15:30 |
| portdirect_ | kfox1111: I think its important that what we have in git can be built into a chart using only helm from a checkout | 15:30 |
| sdake | portdirect_ ++ | 15:30 |
| portdirect_ | at the start of this convo felt otherwise but this has convinced me that this is a criteria we should have | 15:30 |
| kfox1111 | portdirect_: not sure that will work well with workflows, as you typicaly will want to bump versoins/revisions etc. | 15:31 |
| kfox1111 | but sure. its fine to be able to. | 15:31 |
| kfox1111 | so: | 15:31 |
| kfox1111 | prebuild generates varaibles and charts, and drops in _common_lib.tpl. | 15:32 |
| kfox1111 | thjen it gets git checked in. | 15:32 |
| *** msimonin has joined #openstack-kolla | 15:32 | |
| sbezverk | sdake: when you develop a template, we need to test it so you change code more often than change data. Running a tool for every change you do on template to see looks very not optimal to me.. | 15:32 |
| sdake | is _common_lib.tpl manipulated in any way? | 15:32 |
| kfox1111 | then on checkout, they should all be valid charts. | 15:32 |
| kfox1111 | no manipulation. just copied. | 15:32 |
| sdake | sbezverk so the workflow would be as crazy ass this sounds | 15:33 |
| sdake | 1. tool | 15:33 |
| kfox1111 | sbezverk: don't you have to bump the version to redeploy your test code anyway? | 15:33 |
| *** mdnadeem has joined #openstack-kolla | 15:33 | |
| sdake | 2. add stuff to global-values.yaml | 15:33 |
| *** tonanhngo has joined #openstack-kolla | 15:33 | |
| sdake | 3. tool | 15:33 |
| sdake | 4. deploy | 15:34 |
| sbezverk | kfox1111: if your template defained as macro, it has to be included soemwhere right? I guess you do it in common_main. Would it not this operation considered as on the fly code modification before the template gets deployed?? | 15:34 |
| sdake | rather git checkout for 4 :) | 15:34 |
| *** huzhengchuan_ has joined #openstack-kolla | 15:34 | |
| kfox1111 | sdake: no, I think we're past that. no common_main. | 15:34 |
| sdake | self-modifying code is very very very very bad ;) | 15:34 |
| kfox1111 | just _commmon_lib.tpl/_helpers.tpl for the shared code that | 15:35 |
| kfox1111 | gets included | 15:35 |
| *** jtriley has joined #openstack-kolla | 15:35 | |
| sdake | so lets look at the developer workflow for a minute | 15:36 |
| sdake | kfox1111 mind defining that for us | 15:36 |
| sbezverk | kfox1111: could you answer my question? | 15:36 |
| sdake | kfox1111 on firing line :) | 15:36 |
| kfox1111 | sbezverk: with the old poc it was staring with a raw included file and macro'ed in. not self modifying code. | 15:37 |
| kfox1111 | but alanmeadows has a better workaround then that. | 15:37 |
| kfox1111 | so not important anymore. | 15:37 |
| kfox1111 | we just include the common lib like https://review.openstack.org/#/c/401394/8/helm/kolla-kubernetes/charts/openvswitch/charts/openvswitch-vswitchd/templates/openvswitch-vswitchd-daemonset.yaml | 15:38 |
| * alanmeadows backs away slowly. | 15:38 | |
| kfox1111 | sbezverk: so, doing it your wya. | 15:38 |
| kfox1111 | way | 15:38 |
| sdake | alanmeadows backing away from an audience isn't good alan :) | 15:38 |
| alanmeadows | ;-) | 15:38 |
| *** senk has quit IRC | 15:38 | |
| sdake | kfox1111 then the workflow for developers looks like what? | 15:39 |
| sdake | here is where i'm struggling and maybe you guys are thinking :) | 15:41 |
| sdake | i hear sbezverk question the concept of having a globals.tpl file, and i hear kevin say to have a globals.tpl file | 15:42 |
| sdake | is that an accurate refelction? | 15:42 |
| kfox1111 | sorry. internet crapped out. | 15:42 |
| sdake | or is it more a matter of the tool should not copy it | 15:42 |
| kfox1111 | dev checks out kolla-kubernetes, | 15:42 |
| kfox1111 | goes to chart dir, | 15:42 |
| kfox1111 | edits l3-agent template. | 15:43 |
| kfox1111 | tweaks someting. | 15:43 |
| kfox1111 | runs helm install on it. | 15:43 |
| *** JoseMello has quit IRC | 15:43 | |
| kfox1111 | tests it. | 15:43 |
| kfox1111 | git commits. | 15:43 |
| portdirect_ | that misses a few steps? | 15:43 |
| sdake | ok that sounds good to me ifthose are the steps for an existing code base | 15:43 |
| kfox1111 | if they need to edit common code, | 15:43 |
| kfox1111 | the tweak that, | 15:43 |
| kfox1111 | run prebuild. then helm install | 15:43 |
| *** Jeffrey4l has joined #openstack-kolla | 15:44 | |
| kfox1111 | if tehy want to add a brand new microservice, | 15:44 |
| kfox1111 | they add anything needed ito global-vars.yaml, | 15:44 |
| *** jtriley has quit IRC | 15:44 | |
| kfox1111 | run prebuild, | 15:44 |
| sbezverk | kfox1111: what does prebuilt? | 15:44 |
| kfox1111 | add their template yaml, | 15:44 |
| kfox1111 | and run helm install | 15:44 |
| kfox1111 | then when happy, git review. | 15:44 |
| kfox1111 | prebuild does: creates variables.yaml from globals.yhaml, sets version in charts.yaml, and copies in common_lib. | 15:45 |
| kfox1111 | I've gota to in 5 min. | 15:45 |
| sdake | kfox1111 me too | 15:45 |
| sdake | in about 15 | 15:45 |
| sbezverk | kfox1111: cool it is something like helm create with extra | 15:45 |
| sdake | so lets wrap this up | 15:45 |
| kfox1111 | does that sound like it covers all the workflow? | 15:45 |
| sdake | i think that sounds good to me | 15:46 |
| sdake | sbezverk does that alleviate your concerns? | 15:46 |
| portdirect_ | so only files edited are variables.yaml, version in carts.yaml and copying in common_lib to approprate place? | 15:46 |
| sdake | portdirect_ | 15:46 |
| kfox1111 | portdirect_: yeah. | 15:46 |
| sdake | i want to be absolutely clear on one point | 15:46 |
| sdake | common_lib.yaml should not be modified by the prebuild step | 15:46 |
| sbezverk | sdake: if what portdirect_ just mentioned holds true then yes | 15:46 |
| portdirect_ | back to being a happy camper i think :) | 15:46 |
| sdake | I think if we want to have that conversation we can later | 15:46 |
| kfox1111 | sbezverk: right. common_lib is not modified. just copied in. | 15:46 |
| sbezverk | kfox1111: then 100% wfm | 15:47 |
| kfox1111 | sdake I mean. | 15:47 |
| kfox1111 | k. cool. | 15:47 |
| portdirect_ | yeah - this sounds good | 15:47 |
| sdake | sweet | 15:47 |
| sdake | victory declared :) | 15:47 |
| *** fguillot has joined #openstack-kolla | 15:47 | |
| kfox1111 | 98% there. :) | 15:47 |
| kfox1111 | there's one more thing to talk about, but minor, and can wait till later. | 15:47 |
| kfox1111 | gota head out. | 15:48 |
| sdake | ok well lets not mix that up with this work ;) | 15:48 |
| sdake | kfox1111 quick q | 15:48 |
| sdake | when will bild tool be ready | 15:48 |
| kfox1111 | sdake: it will come up in review. :) | 15:48 |
| sdake | sounds good | 15:48 |
| kfox1111 | I should be able to get things rekergiggered around this monring I hope. | 15:48 |
| kfox1111 | probably by noon. | 15:48 |
| sdake | thanks dude | 15:48 |
| sdake | you rock | 15:48 |
| portdirect_ | sweet, cheers man | 15:48 |
| *** sayantan_ has joined #openstack-kolla | 15:48 | |
| portdirect_ | sdake, when do you wanna get your vagrant stuff sorted? | 15:49 |
| kfox1111 | thanks all for the discussion. it was great to work through all the issues, and get everyhone happy. :) | 15:49 |
| kfox1111 | l8r | 15:49 |
| sdake | portdirect_ hrm... let me check calednar | 15:49 |
| *** jtriley has joined #openstack-kolla | 15:50 | |
| *** Jeffrey4l has quit IRC | 15:51 | |
| sdake | portdirect_ i am booked until about 3pm have a meeting in 15 mins but can cancel if you can get me sorted out in the next 1 hr | 15:51 |
| *** Jeffrey4l has joined #openstack-kolla | 15:51 | |
| sdake | the other meetings i cant cancel | 15:52 |
| portdirect_ | what time is where you are? | 15:52 |
| sdake | 8:52 am | 15:52 |
| portdirect_ | lets do after 3pm your time - I have a meeting in 30 that i cant cancel | 15:53 |
| sdake | cool | 15:53 |
| sdake | thanks :) | 15:53 |
| sdake | did duounghq get his env rolling? | 15:53 |
| portdirect_ | wont take an hour long - but I dont want to run out of time if anything weird crops up | 15:53 |
| sdake | ya i need to SSS anyway | 15:54 |
| *** sayantani01_ has joined #openstack-kolla | 15:54 | |
| *** portdirect_ has quit IRC | 15:54 | |
| *** portdirect_ has joined #openstack-kolla | 15:54 | |
| *** adrian_otto has joined #openstack-kolla | 15:55 | |
| sdake | kfox1111 thanks dude ;) ttyl | 15:55 |
| *** TxGirlGeek has quit IRC | 15:56 | |
| *** TxGirlGeek has joined #openstack-kolla | 15:56 | |
| rhallisey | sdake, yo | 15:58 |
| *** jascott1 has joined #openstack-kolla | 15:59 | |
| duonghq | sdake, my k8s cluster and helm is rolling | 15:59 |
| *** goldyfruit has joined #openstack-kolla | 15:59 | |
| duonghq | I'm diving in to kfox1111 and sbezverk PoC | 15:59 |
| duonghq | and your guys conversation | 15:59 |
| *** tovin07 has joined #openstack-kolla | 15:59 | |
| *** portdirect_ has quit IRC | 16:02 | |
| YefimG | Guys, I have installed kolla and kolla-ansible from master branch and it looks, like in both setup.cfg files package name is kolla, so they are uninstalling each other. Is it correct? I have changed kolla to kolla-ansible in it's setup.cfg and it went ok. | 16:02 |
| *** strigazi_AFK is now known as strigazi | 16:02 | |
| *** jascott1 has quit IRC | 16:03 | |
| *** duonghq has quit IRC | 16:09 | |
| kristian__ | sdake portdirect_ great news about my gpu bug :D, guys in libvirt have helped me and told me to move the ethernet adapter because its causing chaos in the iommu groups | 16:09 |
| sdake | kristian__ sweet :) | 16:10 |
| kristian__ | we will see | 16:10 |
| inc0 | kristian__, so not a version issue at the end of the day? | 16:10 |
| kristian__ | it looks like no | 16:10 |
| kristian__ | at first they thought there isnt a feature in the cpu but after investigation on current normal build of libvirt it looks like chaos in iommu groups | 16:11 |
| kristian__ | we will see | 16:11 |
| egonzalez90 | sean-k-mooney: know if the guys working on opendaylight still doing it? https://blueprints.launchpad.net/kolla/+spec/opendaylight-support | 16:13 |
| *** TxGirlGeek has quit IRC | 16:16 | |
| *** eaguilar has quit IRC | 16:17 | |
| mliima_ | need act here https://review.openstack.org/#/c/403689/ and https://review.openstack.org/#/c/401222/ | 16:18 |
| sean-k-mooney | egonzalez90: currently i dont think so. after i get the ovs-dpdk stuff done i might loop back to it but honestly i wont have time to look at it until pike | 16:19 |
| *** mdnadeem has quit IRC | 16:21 | |
| *** YefimG has quit IRC | 16:21 | |
| sean-k-mooney | egonzalez90: if you want to take a crack at it or know someone else who does feel free | 16:22 |
| openstackgerrit | Merged openstack/kolla-ansible: Move glance precheck into its own role https://review.openstack.org/399354 | 16:26 |
| kfox1111 | back. | 16:26 |
| *** huzhengchuan_ has quit IRC | 16:26 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: clean up kolla related files https://review.openstack.org/404330 | 16:29 |
| *** jtriley has quit IRC | 16:30 | |
| sbezverk | kfox1111: how do you want to deal with kind of PoC? | 16:30 |
| sbezverk | for what we discussed this morning | 16:31 |
| kfox1111 | sbezverk: I"m going to split up the https://review.openstack.org/#/c/396296/ into a few patches real quick, | 16:32 |
| sbezverk | kfox1111: for build/prebuild tool do you think you could put it together in one small PS without dependency on the gate etc? in this case we can start making templates and dealing with gates in parallel? | 16:32 |
| openstackgerrit | Merged openstack/kolla-ansible: Add solum ansible role https://review.openstack.org/402225 | 16:32 |
| kfox1111 | and tweak the build script into the prebuild script we talked about. | 16:33 |
| sbezverk | kfox1111: sounds good | 16:33 |
| kfox1111 | was going to pull out the helm gate support to a trunk level patch. | 16:33 |
| kfox1111 | so it can be reviewed independently. | 16:33 |
| kfox1111 | then do the poc on top. | 16:33 |
| kfox1111 | I also was thinking about it on the drive in, | 16:33 |
| kfox1111 | I think we can actually do the common lib as a pure helm package, | 16:33 |
| kfox1111 | rather then copy it in. | 16:34 |
| kfox1111 | due to the macro/include trick. | 16:34 |
| sbezverk | kfox1111: I think it will be impotrant to be able to merge build script asap | 16:34 |
| kfox1111 | sbezverk: yeah. I'm going to try and get something posted by noon. if not sooner. | 16:34 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: clean up kolla related files https://review.openstack.org/404330 | 16:34 |
| sbezverk | kfox1111: ok | 16:34 |
| *** pcaruana has quit IRC | 16:36 | |
| *** sbezverk has quit IRC | 16:37 | |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Initial Helm support in the gate https://review.openstack.org/404332 | 16:37 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Initial Helm support in the gate https://review.openstack.org/404332 | 16:39 |
| *** unicell1 has quit IRC | 16:40 | |
| *** tovin07 has quit IRC | 16:40 | |
| openstackgerrit | Mauricio Lima proposed openstack/kolla-ansible: Move manila precheck into its own role https://review.openstack.org/404338 | 16:42 |
| *** sbezverk has joined #openstack-kolla | 16:43 | |
| *** unicell has joined #openstack-kolla | 16:43 | |
| *** unicell has quit IRC | 16:43 | |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Initial Helm support https://review.openstack.org/396296 | 16:44 |
| *** unicell has joined #openstack-kolla | 16:44 | |
| *** bmace has joined #openstack-kolla | 16:46 | |
| openstackgerrit | Merged openstack/kolla: Update Percona repo to current release https://review.openstack.org/400889 | 16:46 |
| *** jtriley has joined #openstack-kolla | 16:47 | |
| *** jascott1 has joined #openstack-kolla | 16:47 | |
| openstackgerrit | Merged openstack/kolla: Add strongswan to neutron-vpn-agent source image https://review.openstack.org/404061 | 16:47 |
| *** rmart04 has quit IRC | 16:47 | |
| *** vhosakot has joined #openstack-kolla | 16:49 | |
| kfox1111 | oh... | 16:50 |
| kfox1111 | and doing a common lib package would let us break the lib up into multiple files. | 16:50 |
| kfox1111 | that would be much cleaner looking. | 16:50 |
| kfox1111 | hmm... | 16:51 |
| openstackgerrit | Merged openstack/kolla: Use kolla-ansible deploy_aio in gate https://review.openstack.org/401891 | 16:51 |
| kfox1111 | any helm experts awake? | 16:51 |
| kfox1111 | oh. I think it loads all files first, then processes macro's. | 16:52 |
| kfox1111 | so Ithink we're good. | 16:52 |
| v1k0d3n | hey guys have a question ssl related again (sorry already). | 16:53 |
| sbezverk | kfox1111: would it be acceptable for you to use the same folder structure as in my PS? it looks like people liked it since it is very easy to view and follow | 16:53 |
| v1k0d3n | i've generated certs, run deploy, post-deploy, added cert path to /etc/kolla/admin-openrc.sh, but get an error 'X509' object has no attribute '_x509'. has anyone seen this? | 16:54 |
| v1k0d3n | any ideas? | 16:54 |
| kfox1111 | sbezverk: thats the other thing I wanted to talk about. | 16:54 |
| *** portdirect_away is now known as pd_away | 16:54 | |
| kfox1111 | it assumes we know the exact layout of a super package, and that there wont be other types. | 16:55 |
| kfox1111 | I'm not sure we know either. :/ | 16:55 |
| *** sayantan_ has quit IRC | 16:55 | |
| kfox1111 | what about /helm/microservices/neutron-l3-agent, /helm/microservices/neutron-openvswitch-agent, etc? where each one is a chart? | 16:55 |
| kfox1111 | then for super packages, we can do /helm/service/neutron or something like that? | 16:56 |
| *** sayantan_ has joined #openstack-kolla | 16:56 | |
| kfox1111 | and use helm dependencies in the neutron package | 16:56 |
| sbezverk | kfox1111: what if by some reason we need a package per service as well? | 16:56 |
| sbezverk | example neutron package including all neutron components | 16:56 |
| kfox1111 | yeah. I think a helm dep will solve that? | 16:57 |
| kfox1111 | oh. | 16:57 |
| kfox1111 | I mean, requirements.yaml | 16:57 |
| kfox1111 | https://github.com/kubernetes/helm/blob/master/docs/charts.md | 16:57 |
| kfox1111 | "Managing Dependencies with requirements.yaml ? | 16:58 |
| jascott1 | heh | 16:58 |
| sbezverk | kfox1111: I think if we go with 3 level it will give better flexibility in terms of future packaging | 16:58 |
| jascott1 | good morning kolla peeps | 16:58 |
| *** Pavo has quit IRC | 16:58 | |
| sbezverk | brb | 16:58 |
| kfox1111 | sbezverk: I'm worried if we start nesting things then we will be rearanging it when we want to go to 3, or 4 levels. | 16:58 |
| kfox1111 | if we keep the microservice packages at a place all on their own, we can include them with requirements in the places they need. potentially including the same subpackage in multiple service packages if need be. | 16:59 |
| kfox1111 | rabbit for example. | 16:59 |
| *** eaguilar has joined #openstack-kolla | 16:59 | |
| jascott1 | using requirements? | 16:59 |
| kfox1111 | yeah. | 17:00 |
| jascott1 | each service would have its own rabbit if you use requirements like that right? | 17:00 |
| kfox1111 | yeah. | 17:00 |
| kfox1111 | which is something I want on my cloud. | 17:00 |
| kfox1111 | when a rabbit dies, less is affected that way. | 17:00 |
| *** TxGirlGeek has joined #openstack-kolla | 17:01 | |
| kfox1111 | a herd of rabbits. :) | 17:01 |
| kfox1111 | flock... | 17:01 |
| kfox1111 | whats the right term. | 17:01 |
| jascott1 | warren? | 17:01 |
| kfox1111 | "A group of rabbits is called a "herd", just like a group of cattle. A herd of rabbits lives in a warren." | 17:01 |
| *** adrian_otto has quit IRC | 17:01 | |
| kfox1111 | I guessed right the first time. :) | 17:01 |
| *** jtriley has quit IRC | 17:01 | |
| kfox1111 | but, this brings up an interesting point. | 17:02 |
| kfox1111 | where helm isn't very flexible. | 17:02 |
| *** egonzalez90 has quit IRC | 17:02 | |
| kfox1111 | what if you wanted to have a nova service package with an optional embeded rabbit? | 17:02 |
| kfox1111 | I don't think helm can handle that today. | 17:02 |
| kfox1111 | so you'd have to build two different service packages. one with rabbit and one without. :/ | 17:02 |
| inc0 | herd of rabbits? | 17:03 |
| inc0 | duh... | 17:03 |
| jascott1 | thats what sbezverk's plugin idea would handle | 17:03 |
| kfox1111 | yeah. that might work. | 17:03 |
| *** Pavo has joined #openstack-kolla | 17:03 | |
| *** eaguilar has quit IRC | 17:03 | |
| *** sdake_ has joined #openstack-kolla | 17:03 | |
| rhallisey | kfox1111, why have helm embed anything at all? | 17:04 |
| kfox1111 | rhallisey: some folks want that. (non operator use case) | 17:04 |
| sbezverk | kfox1111: but what forces them to bundle these together and not install one and then another? it is kind of artifial requirements | 17:05 |
| rhallisey | kfox1111, how about micro service helm charts with no deps | 17:05 |
| kfox1111 | sbezverk: I don't follow. | 17:05 |
| rhallisey | kfox1111, then charts that have the deps | 17:05 |
| kfox1111 | rhallisey: I'm thinking microservices helm charts with just one dep. kolla-common, where we put all of our common code. | 17:06 |
| sbezverk | kfox1111: covering base and adding reasonable flexibility (that works for your case) should be ok for now imho | 17:06 |
| *** sdake has quit IRC | 17:06 | |
| kfox1111 | sbezverk: yeah. | 17:06 |
| kfox1111 | sbezverk: just sayin,g I couild see a case for sure where you might want to have more then one implementation of "nova" service package. | 17:06 |
| *** eaguilar has joined #openstack-kolla | 17:06 | |
| kfox1111 | so putting the microservice inside the nova service package physically on the filesystem might not be a good fit. | 17:07 |
| kfox1111 | so the requirements.yaml thing might be the way to go with /helm/microservices | 17:07 |
| sbezverk | kfox1111: ok but now this scenario is not even handled by openstack | 17:08 |
| sbezverk | so we do not do any less than already is | 17:08 |
| kfox1111 | sbezverk: some operators deploy that way. | 17:08 |
| rhallisey | helm -> nova chart w/ deps -> nova micro service charts with no deps | 17:08 |
| kfox1111 | with rabbit per service. | 17:08 |
| *** jtriley has joined #openstack-kolla | 17:08 | |
| rhallisey | still complex tho | 17:08 |
| kfox1111 | we also do a secondary rabbit for ceilometer. | 17:08 |
| kfox1111 | rhallisey: in ha, everythings complex. :/ | 17:08 |
| rhallisey | kfox1111, ya but were talking about packaging here | 17:08 |
| jascott1 | dont we want HA rabbit clusters? | 17:09 |
| rhallisey | were packaing 2 models | 17:09 |
| kfox1111 | jascott1: for some things. for oslo notifications, I want ha. | 17:09 |
| kfox1111 | jascott1: for rpc things, maybe not. k8s restarting a container might be good enough, and simpler. | 17:09 |
| rhallisey | kfox1111, so in your rabbit scenario how would you add a rabbit per service | 17:09 |
| kfox1111 | helm install --name nova-rabbit kolla/rabbitmq | 17:10 |
| kfox1111 | helm install --name neutron-rabbit kolla/rabbitmq | 17:10 |
| kfox1111 | 2 rabbits. :) | 17:10 |
| sbezverk | kfox1111: for HA, it is not the only option to bundle right? you can still achieve HA with individual deployment no? | 17:10 |
| rhallisey | kfox1111, does nova-rabbit depend on nova? | 17:10 |
| rhallisey | or nova-* | 17:10 |
| kfox1111 | rhallisey: no. | 17:10 |
| kfox1111 | sbezverk: right. | 17:11 |
| jascott1 | kfox1111 but thats not using deps | 17:11 |
| kfox1111 | jascott1: right | 17:11 |
| jascott1 | ok | 17:11 |
| rhallisey | given that, I would define helm as a set of pods for a service to run | 17:11 |
| sbezverk | kfox1111: in near future, your build tool could evolve and package together components based on required components list | 17:11 |
| rhallisey | i.e. rpm | 17:11 |
| kfox1111 | sbezverk: I think helm can do that natively already. | 17:12 |
| kfox1111 | haven't tried it yet though. | 17:12 |
| kfox1111 | helm dependency update | 17:12 |
| sbezverk | kfox1111: me neither I want base working and start making temapltes for now ;-) | 17:12 |
| kfox1111 | will pull the subcharts. | 17:13 |
| kfox1111 | gona try poc'int that real quick. | 17:13 |
| *** adrian_otto has joined #openstack-kolla | 17:13 | |
| kfox1111 | hmm... | 17:16 |
| openstackgerrit | Alexis Rivera proposed openstack/kolla-ansible: renaming-kolla-build-command https://review.openstack.org/401005 | 17:16 |
| openstackgerrit | Alexis Rivera proposed openstack/kolla-ansible: renaming-kolla-build-command https://review.openstack.org/401005 | 17:19 |
| openstackgerrit | Eduardo Gonzalez proposed openstack/kolla-ansible: Tacker NFV Ansible support https://review.openstack.org/399179 | 17:19 |
| kfox1111 | arg.. helm definitely doesn't like symlinks. :/ | 17:20 |
| kfox1111 | a symlink would have worked beutifully here... | 17:20 |
| openstackgerrit | Alexis Rivera proposed openstack/kolla-ansible: renaming-kolla-build-command https://review.openstack.org/401005 | 17:20 |
| sbezverk | kfox1111: yeah portdirect mentioned that idea but it looks like helm does agree with him ;-) | 17:21 |
| inc0 | kfox1111, please...no symlinks | 17:21 |
| inc0 | it worked really really bad in kolla back in a day | 17:21 |
| kfox1111 | inc0: it just complicates things not using it in this case. | 17:22 |
| *** jtriley has quit IRC | 17:22 | |
| inc0 | yeah, just supporting this in repo and stuff is really hard | 17:23 |
| inc0 | also effectively closes lots of doors | 17:23 |
| kfox1111 | helm was built for making it easy to build a single package, with prebuilt dependencies. | 17:23 |
| kfox1111 | its not really tooled yet to easily build lots of packages. | 17:24 |
| inc0 | same as docker build really | 17:24 |
| inc0 | we had to write overarching tool (kolla_build) for it | 17:24 |
| kfox1111 | yeah. and am getting pushback for trying to go down that route. :/ | 17:25 |
| inc0 | maybe we could reuse similar pattern? | 17:25 |
| kfox1111 | I really want to use the same pattern. | 17:25 |
| *** zhubingbing has quit IRC | 17:25 | |
| inc0 | I mean each chart will be self-contained right? | 17:25 |
| inc0 | we will just do helm build in a loop | 17:25 |
| inc0 | how wrong am I with this statement? | 17:26 |
| kfox1111 | but deps have to be built before things that depend on them. | 17:26 |
| kfox1111 | so tree. :/ | 17:26 |
| *** matrohon has quit IRC | 17:26 | |
| kfox1111 | so, kolla-common, thjen microservice packages, then service packages. | 17:26 |
| *** vhosakot has quit IRC | 17:27 | |
| kfox1111 | the requirements stuff looks nice. | 17:27 |
| kfox1111 | but it looks like you have to have a working mirror running. | 17:27 |
| kfox1111 | so the build step would be: | 17:27 |
| kfox1111 | build kolla-common, copy to local repo. | 17:27 |
| kfox1111 | build microservices packages. copy to repo. | 17:27 |
| jascott1 | if we dont use helm's deps we dont need a repo | 17:28 |
| inc0 | thats how kolla-build works really | 17:28 |
| kfox1111 | build service packages. copy to repos. | 17:28 |
| inc0 | also true, what value do we get from helm native deps? | 17:28 |
| kfox1111 | jascott1: then we gota manually copy the build packages into the other subdirs. | 17:28 |
| pbourke | Jeffrey4l: would you expect your dynamic dashboard changes to work with reconfigure? | 17:28 |
| kfox1111 | 6 of one. | 17:29 |
| kfox1111 | do we use helm to do the copying, or python? | 17:29 |
| *** krtaylor has quit IRC | 17:29 | |
| kfox1111 | I thought we were "helm native" as much as possible. | 17:29 |
| jascott1 | inc0 helm deps will download your sub charts and install them with the values you have for them. but thats it and really doesnt handle ordering of those deps iirc | 17:30 |
| sbezverk | kfox1111: copy is the part of your built script | 17:30 |
| sbezverk | so it is pre-helm step | 17:30 |
| *** vhosakot has joined #openstack-kolla | 17:30 | |
| mliima_ | pbourke, https://review.openstack.org/#/c/401222/ | 17:30 |
| kfox1111 | sbezverk: so, I think we can do common lib as a kolla-common package, | 17:30 |
| mliima_ | can you review? | 17:30 |
| mliima_ | :) | 17:30 |
| kfox1111 | and break up the lib into multiple files for easier maintenence/readability. | 17:30 |
| mliima_ | https://review.openstack.org/#/c/403689/ | 17:30 |
| kfox1111 | but then we probably don't want to copy the whole kolla-common subpackage into each microservice on prebuild? | 17:31 |
| kfox1111 | rather, go helm native and do a | 17:31 |
| kfox1111 | helm dependency update; helm package . ? | 17:31 |
| *** diogogmt has joined #openstack-kolla | 17:31 | |
| sbezverk | kfox1111: just to be sure I follow you idea kolla-common is kind of global chart/template with bunch of smaller templates per macro? | 17:31 |
| kfox1111 | sbezverk: exactly. | 17:32 |
| mliima_ | thanks pbourke | 17:32 |
| kfox1111 | then its pure helm. | 17:32 |
| sbezverk | kfox1111: I think that what we agreed, copy common into microservice | 17:32 |
| kfox1111 | sbezverk: yeah. but as I was copying more and more stuff into common_lib just now, it was starting to get unwealdy. looking at alternative ideas. | 17:33 |
| jascott1 | i dont see how deps will work until we get sbezverk's plugin to skip a dep. if you embed rabbit in a service then how do I use that service with my existing rabbit cluster for example? | 17:33 |
| sbezverk | so when microservice gets packaged/deployed common code will be used by helm rendering tool | 17:33 |
| sbezverk | jascott1: we kind of agree to go with microservice level | 17:34 |
| kfox1111 | yeah. and I think a helm dep will do the same. but helm native. | 17:34 |
| sbezverk | and then prebuilt tool will add dependencies in terms of variable/values to the microservice chart | 17:34 |
| kfox1111 | so no prebuild touching templates/* at all. just charts/values | 17:34 |
| kfox1111 | so, even more helm native. | 17:35 |
| sbezverk | kfox1111: right | 17:35 |
| sbezverk | jascott1: it seems we can achive even without plugin | 17:35 |
| jascott1 | so flat charts with external tooling, im down with that | 17:36 |
| sbezverk | later when plugon will become available, current templates still can be used for going funcy with this plugin as they are completely helm native | 17:36 |
| *** eaguilar has quit IRC | 17:37 | |
| kfox1111 | heh. now there's a helm bug... | 17:38 |
| sbezverk | kfox1111: so back to file structure :-) | 17:38 |
| kfox1111 | helm dep update | 17:38 |
| kfox1111 | ...Unable to get an update from the "local" chart repository (http://127.0.0.1:8879/charts): | 17:38 |
| kfox1111 | echo $? | 17:38 |
| kfox1111 | 0 | 17:38 |
| kfox1111 | :/ | 17:38 |
| kfox1111 | broken. | 17:38 |
| jascott1 | ya the name has to match exactly | 17:38 |
| jascott1 | 127.0.0.1 != localhost | 17:38 |
| *** unicell has quit IRC | 17:38 | |
| *** zhubingbing has joined #openstack-kolla | 17:38 | |
| kfox1111 | no, the brokenness is if it cant download something important, it still returns success. :/ | 17:38 |
| jascott1 | oh ha | 17:39 |
| kfox1111 | meeting. bbiab | 17:40 |
| openstackgerrit | Merged openstack/kolla-ansible: Remove fragments of docker files https://review.openstack.org/401222 | 17:43 |
| *** matrohon has joined #openstack-kolla | 17:43 | |
| *** TxGirlGeek has quit IRC | 17:48 | |
| sbezverk | kfox1111: I read through requirements.yaml doc and it seems it can be used for the use case you mentioned like nova+rabbit, by specifying rabbit in this file helm is supposed to add it at the rendering time.. | 17:50 |
| *** msimonin has quit IRC | 17:51 | |
| *** TxGirlGeek has joined #openstack-kolla | 17:52 | |
| kfox1111 | sbezverk: yeah, it can be done by changing the charts, but not by using a prebuilt package. :/ | 17:52 |
| kfox1111 | one of my goals is to make it so that users don't normally have to build charts. just use prebuilt packages. | 17:52 |
| kfox1111 | I hardly ever build an rpm from source. | 17:52 |
| *** adrian_otto has quit IRC | 17:53 | |
| jascott1 | "but when I do, I do it in PRD" | 17:53 |
| jascott1 | kfox1111 are you saying you cannot install a dep without a package? | 17:53 |
| *** strigazi is now known as strigazi_AFK | 17:54 | |
| kfox1111 | jascott1: no, I'm saying if you want to make add/remove rabbit from nova service package, you gota edit a file and rebuild the package from source. | 17:54 |
| kfox1111 | rather then be able to do helm install nova --set no_rabbit=true | 17:54 |
| jascott1 | gotcha. cause reqs wants to pull the package | 17:55 |
| sbezverk | kfox1111: well this limitation is not clear from reading the doc, but I guess you tried and it did not work | 17:55 |
| kfox1111 | it would be nice if there were optional deps. :/ | 17:55 |
| kfox1111 | or optional instantiation really... | 17:56 |
| kfox1111 | maybe we can do somethign with a macro though, in the rabbit package. | 17:56 |
| kfox1111 | make an override that skips launching the template if a var is set. | 17:56 |
| jascott1 | but if you have that chart in subfolder it still pulls? | 17:56 |
| kfox1111 | yeah. all subcharts always get instantiated. | 17:56 |
| *** zhubingbing has quit IRC | 17:56 | |
| jascott1 | i thought it was using my local subchart | 17:57 |
| jascott1 | hmm | 17:57 |
| kfox1111 | oh. I didn't understand the question. | 17:57 |
| kfox1111 | if you have a subchart already in there, I think it uses that. | 17:57 |
| kfox1111 | meting next. bbiab. | 17:58 |
| *** harbor has joined #openstack-kolla | 17:59 | |
| *** unicell has joined #openstack-kolla | 18:00 | |
| *** pd_away has quit IRC | 18:01 | |
| *** TxGirlGeek has quit IRC | 18:02 | |
| *** unicell has quit IRC | 18:05 | |
| *** jgriffith_away is now known as jgriffith | 18:05 | |
| pbourke | Jeffrey4l: any chance you can double check the lbaas dashboard is working for you? | 18:05 |
| pbourke | Jeffrey4l: have it enabled and can see the enabled files but its not showing up | 18:06 |
| pbourke | (in horizon) | 18:06 |
| *** kristian__ has quit IRC | 18:06 | |
| *** JoseMello has joined #openstack-kolla | 18:07 | |
| *** TxGirlGeek has joined #openstack-kolla | 18:08 | |
| *** gfidente is now known as gfidente|afk | 18:08 | |
| sdake_ | jascott1 is intel stlil running management by objectives? (MBO?) | 18:09 |
| sdake_ | jascott1 surprised intel still running product requirements documents (PRD) | 18:10 |
| *** mkoderer has quit IRC | 18:13 | |
| *** tonanhngo has quit IRC | 18:14 | |
| inc0 | sdake_, I'm from intel and I have no idea what are you talknig about | 18:15 |
| sdake_ | [10:53:21] <jascott1>"but when I do, I do it in PRD" | 18:16 |
| sdake_ | maybe PRD has some other magic tla inside intc now :) | 18:17 |
| *** papacz has quit IRC | 18:23 | |
| *** TxGirlGeek has quit IRC | 18:24 | |
| *** athomas has quit IRC | 18:25 | |
| *** inc0 has quit IRC | 18:26 | |
| *** sayantan_ has quit IRC | 18:30 | |
| jascott1 | sorry I meant PROD | 18:31 |
| jascott1 | but yes MBOs are still a thing at Intel | 18:31 |
| jascott1 | but not so much in our org afaik.. idk new to the org | 18:31 |
| sdake_ | jascott1 oh got it :) | 18:32 |
| jascott1 | http://imgur.com/gallery/y7Hm9 | 18:32 |
| sdake_ | lol | 18:33 |
| *** shardy has quit IRC | 18:34 | |
| jascott1 | we have a lot of agile in the software group | 18:34 |
| jascott1 | no idea what the chip makers use | 18:35 |
| sdake_ | copy exactly | 18:35 |
| sdake_ | Wendy Vittori led a session on "Whats wrong with strategy" when i was a young tike out of school | 18:36 |
| sdake_ | it fundamentally changed how i viewed the world | 18:36 |
| sdake_ | too bad it wasn't recorded :( | 18:36 |
| sdake_ | it took a few years to sink in | 18:37 |
| sdake_ | maybe 5 ;) | 18:37 |
| sdake_ | i think it was geared at the old dogs like you and me jascott1 :) | 18:37 |
| *** eaguilar has joined #openstack-kolla | 18:38 | |
| jascott1 | woof | 18:38 |
| sdake_ | that old idiom old dogs can't be taught new tricks doesn't apply to humans :) | 18:39 |
| *** hvlad has quit IRC | 18:39 | |
| *** jgriffith is now known as jgriffith_away | 18:40 | |
| jrich523 | lol, i kinda feel like it applies more to human than any other creature | 18:40 |
| jascott1 | no doubt | 18:40 |
| *** inc0 has joined #openstack-kolla | 18:41 | |
| *** kristian__ has joined #openstack-kolla | 18:42 | |
| jascott1 | mammals rule! | 18:44 |
| *** unicell1 has joined #openstack-kolla | 18:44 | |
| inc0 | reptilians rule, you just don't know it | 18:46 |
| *** krtaylor has joined #openstack-kolla | 18:46 | |
| jascott1 | lol... would agree reptilian part of our brains seems to call the shots too often | 18:49 |
| *** lrensing has quit IRC | 18:52 | |
| *** lrensing has joined #openstack-kolla | 18:53 | |
| *** athomas has joined #openstack-kolla | 18:55 | |
| *** Pavo has quit IRC | 18:58 | |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Move aodh precheck into its own role https://review.openstack.org/404389 | 19:02 |
| *** Pavo has joined #openstack-kolla | 19:03 | |
| sean-k-mooney | hi. has anyone used manila? | 19:03 |
| *** athomas has quit IRC | 19:03 | |
| *** kristian__ has quit IRC | 19:04 | |
| *** eaguilar has quit IRC | 19:05 | |
| mliima_ | \o | 19:09 |
| mliima_ | sean-k-mooney, | 19:09 |
| sean-k-mooney | hi | 19:10 |
| mliima_ | hi | 19:10 |
| mliima_ | any problem with manila? | 19:10 |
| mliima_ | I used and i can try to help you | 19:11 |
| sean-k-mooney | mliima_: am yes basically i was trying to create a share on stable/newton but it fails. | 19:12 |
| mliima_ | hmm | 19:12 |
| sean-k-mooney | on the manila-share container i am seeing the follwoing message ServiceInstanceException: SSH connection has not been established to 10.254.0.7 in 300s. Giving up. | 19:12 |
| mliima_ | nova list --all-tenants | 19:13 |
| mliima_ | please | 19:13 |
| mliima_ | are you using generic driver? | 19:13 |
| *** matrohon has quit IRC | 19:14 | |
| sean-k-mooney | yep i am. one sec i deleted the vms because i also could not delete the shares from error state | 19:14 |
| *** tonanhngo has joined #openstack-kolla | 19:14 | |
| *** inc0 has quit IRC | 19:15 | |
| mliima_ | sean-k-mooney, which image are you using to build an instance? | 19:15 |
| sean-k-mooney | i was using the one in the docs following https://github.com/openstack/kolla/blob/master/doc/manila-guide.rst | 19:16 |
| mliima_ | ok | 19:16 |
| *** athomas has joined #openstack-kolla | 19:17 | |
| *** inc0 has joined #openstack-kolla | 19:18 | |
| mliima_ | try to create a share again | 19:18 |
| mliima_ | and run nova list --all-tenants | 19:19 |
| mliima_ | :) | 19:19 |
| sean-k-mooney | http://paste.openstack.org/show/590871/ | 19:20 |
| mliima_ | sean-k-mooney, | 19:21 |
| mliima_ | run cat /etc/kolla/config/manila-share.conf | 19:21 |
| mliima_ | "BadRequest: Flavor 100 could not be found." | 19:21 |
| sean-k-mooney | ya i have fixed that | 19:22 |
| *** jtriley has joined #openstack-kolla | 19:22 | |
| kfox1111 | back. | 19:22 |
| sean-k-mooney | i updated the flavoer which cased the id to chagne but i deleted and recreated it | 19:22 |
| sean-k-mooney | mliima_: do i have to create a /etc/kolla/config/manila-share.conf is if i create the flavor with id 100 | 19:23 |
| mliima_ | no | 19:23 |
| sean-k-mooney | http://paste.openstack.org/show/590872/ | 19:24 |
| sean-k-mooney | that is my manila conf | 19:24 |
| sean-k-mooney | kolla@silpixa00385225:~$ openstack flavor list | 19:24 |
| sean-k-mooney | +--------------------------------------+-----------------------+-------+------+-----------+-------+-----------+ | 19:24 |
| sean-k-mooney | | ID | Name | RAM | Disk | Ephemeral | VCPUs | Is Public | | 19:24 |
| sean-k-mooney | +--------------------------------------+-----------------------+-------+------+-----------+-------+-----------+ | 19:24 |
| sean-k-mooney | | 100 | manila-service-flavor | 128 | 0 | 0 | 1 | True | | 19:24 |
| sean-k-mooney | | 1132d6d1-0a12-4f62-9ad4-7aae93d21109 | all-in-one | 14336 | 50 | 0 | 10 | True | | 19:24 |
| sean-k-mooney | | 616b1e9a-4419-414a-8c46-8e737a4bcc39 | medium | 4096 | 20 | 0 | 4 | True | | 19:24 |
| sean-k-mooney | | 62d0625d-ba94-4679-8619-42735db9d8fb | large | 8192 | 40 | 0 | 8 | True | | 19:24 |
| sean-k-mooney | | 795a70b9-cd96-40a3-90c5-c3b114bb8d4c | cirros | 64 | 1 | 0 | 1 | True | | 19:24 |
| sean-k-mooney | | 9403c6a1-ff39-406c-9915-d3fa19896742 | smoke-test | 4096 | 10 | 0 | 2 | True | | 19:24 |
| sean-k-mooney | | f25d0593-037e-4a47-8b69-dc5154110bee | small | 1024 | 10 | 0 | 1 | True | | 19:24 |
| sean-k-mooney | +--------------------------------------+-----------------------+-------+------+-----------+-------+-----------+ | 19:24 |
| mliima_ | ok, its right sean-k-mooney | 19:24 |
| mliima_ | docker restart manila_share | 19:25 |
| mliima_ | and try to create another share again | 19:25 |
| sean-k-mooney | am sure i can do that one sec | 19:25 |
| *** jtriley has quit IRC | 19:26 | |
| *** eaguilar has joined #openstack-kolla | 19:27 | |
| *** sayantan_ has joined #openstack-kolla | 19:28 | |
| sdake_ | sup kfox1111 | 19:28 |
| kfox1111 | sdake_: hi | 19:29 |
| sean-k-mooney | mliima_: it has a different error "ServiceInstanceUnavailable: Service instance is not available." | 19:29 |
| kfox1111 | sdake_: so, had an idea for an even more helm native thing, | 19:29 |
| kfox1111 | it needs more of a build step. | 19:29 |
| kfox1111 | kind of a hard call. | 19:29 |
| sean-k-mooney | mliima_: the nova instance is there though | 19:29 |
| kfox1111 | but the more I pla ywith it, the more I think we will want a build step anyway. as helm doesn't have a good way to build multiple packages anyway. | 19:29 |
| kfox1111 | need a command to build all microservices for example. | 19:30 |
| sdake_ | ok - so i think my default respones here would be lets tackle what we agreed to - and sort it out in the gerrit reiew :) | 19:30 |
| mliima_ | manila list sean-k-mooney | 19:30 |
| sdake_ | and then sort out that idea next | 19:30 |
| kfox1111 | hmm... | 19:30 |
| kfox1111 | k. | 19:30 |
| mliima_ | manila try to build an instance at the same time try to connect using ssh | 19:30 |
| sean-k-mooney | kolla@silpixa00385225:~$ manila list | 19:30 |
| sean-k-mooney | +--------------------------------------+-------------+------+-------------+----------+-----------+--------------------+---------------------------------+-------------------+ | 19:30 |
| sean-k-mooney | | ID | Name | Size | Share Proto | Status | Is Public | Share Type Name | Host | Availability Zone | | 19:31 |
| sean-k-mooney | +--------------------------------------+-------------+------+-------------+----------+-----------+--------------------+---------------------------------+-------------------+ | 19:31 |
| sean-k-mooney | | 0ed0aa91-4f97-4bc6-8081-cd5744c2a00a | demo-share1 | 200 | NFS | deleting | True | default_share_type | silpixa00385226@generic#GENERIC | nova | | 19:31 |
| sean-k-mooney | | 84730297-a9dc-4247-a61d-d22270b355e6 | demo-share1 | 200 | NFS | creating | True | default_share_type | silpixa00385226@generic#GENERIC | nova | | 19:31 |
| *** TxGirlGeek has joined #openstack-kolla | 19:31 | |
| sean-k-mooney | +--------------------------------------+-------------+------+-------------+----------+-----------+--------------------+---------------------------------+-------------------+ | 19:31 |
| sdake_ | sean-k-mooney any chance you can use a paste service ;) | 19:31 |
| mliima_ | manila try to build an instance at the same time try to connect using ssh until timeout | 19:31 |
| sean-k-mooney | sdake_: ya sorry | 19:31 |
| sdake_ | sean-k-mooney itso k i do it too | 19:31 |
| sdake_ | although usually 1-3 short lines | 19:31 |
| mliima_ | so,this is normal | 19:31 |
| sdake_ | thats really hard to read in irc anyway :) | 19:31 |
| mliima_ | just wait | 19:31 |
| mliima_ | :) | 19:31 |
| sean-k-mooney | mliima_: roughly how long should it take? | 19:32 |
| *** fguillot has quit IRC | 19:33 | |
| mliima_ | It depends | 19:33 |
| mliima_ | 2 minutes | 19:34 |
| *** fguillot has joined #openstack-kolla | 19:34 | |
| mliima_ | but this depends on your resources sean-k-mooney | 19:34 |
| mliima_ | run manila list | 19:34 |
| sean-k-mooney | it just went into error state. | 19:34 |
| mliima_ | oh god | 19:34 |
| mliima_ | let me see manila_share logs | 19:35 |
| sean-k-mooney | mliima_: so far all shares i have create end in error state | 19:35 |
| mliima_ | let me see manila_share logs sean-k-mooney | 19:36 |
| sean-k-mooney | http://paste.openstack.org/show/590877/ | 19:36 |
| sean-k-mooney | same issue ServiceInstanceException: SSH connection has not been established to 10.254.0.8 in 300s. Giving up. | 19:36 |
| *** harbor_ has joined #openstack-kolla | 19:37 | |
| *** kristian__ has joined #openstack-kolla | 19:37 | |
| sdake_ | sup harbie | 19:37 |
| sdake_ | harbor | 19:37 |
| sdake_ | harbor_ :) | 19:37 |
| mliima_ | sean-k-mooney, neutron net-list | 19:37 |
| harbor_ | wait for the transformation - just shedding my lizard skin | 19:37 |
| *** harbor_ is now known as portdirect | 19:38 | |
| mliima_ | manila share-network-list | 19:38 |
| portdirect | hey sdake_ :) | 19:38 |
| mliima_ | and which command are you using to create a share? | 19:38 |
| sdake_ | portdirect have meetings for next couple hours then should be good to go | 19:39 |
| sdake_ | on env setup | 19:39 |
| sdake_ | need to get rolling :) | 19:39 |
| sean-k-mooney | mliima_: ill put it in the paste | 19:39 |
| mliima_ | ok | 19:40 |
| sdake_ | portdirect i'm on your schedule after that :) | 19:40 |
| sean-k-mooney | mliima_: http://paste.openstack.org/show/590880/ | 19:40 |
| portdirect | sdake_: I'm reverting to full mad scientist :/ getting a centos and ubuntu box set up right now so will be able to play in all the sandpits :) | 19:40 |
| sdake_ | tarpits ? :) | 19:40 |
| *** kristian__ has quit IRC | 19:41 | |
| mliima_ | sean-k-mooney, can you ping 10.254.0.8 ? | 19:41 |
| *** kristian__ has joined #openstack-kolla | 19:42 | |
| mliima_ | from your host? | 19:42 |
| sean-k-mooney | no that network would conflict with my lab network | 19:42 |
| sean-k-mooney | 10.254.0.8 is on the private tenant network created by manila | 19:42 |
| sean-k-mooney | mliima_: i could try from the network namespace of the dhcp server | 19:43 |
| alanmeadows | kfox1111: interesting, I can't reproduce the "inconsistency" with a sub chart define in v2.0.0 | 19:43 |
| alanmeadows | I wonder if this was an rc2 thing | 19:43 |
| mliima_ | you need to run this command from a vm that can ping manila-instance | 19:45 |
| mliima_ | sean-k-mooney, | 19:45 |
| kfox1111 | alanmeadows: which do you mean. defining in a subchart, then using in a main template? | 19:45 |
| kfox1111 | defining a var I mean. | 19:45 |
| alanmeadows | right | 19:45 |
| kfox1111 | maybe their hashing has just gotten a bit more consistant, but will still have issues. | 19:45 |
| kfox1111 | we should see if we can find a bug for it. | 19:45 |
| kfox1111 | if its fixed, thats great. | 19:46 |
| kfox1111 | if its just more consistatnly randomly working, thats bad. :/ | 19:46 |
| alanmeadows | maybe the stars are aligning for this particular inclusion ordering | 19:46 |
| sdake_ | kfox1111 :) | 19:46 |
| alanmeadows | not sure if its because im in a different template combo | 19:46 |
| kfox1111 | I had one that worked 10 tims in a row, then broke on the 11th. :/ | 19:46 |
| alanmeadows | or if it was rc2 only | 19:46 |
| sdake_ | kfox1111 welcome to my world ;) | 19:46 |
| mliima_ | it is clear that this is a problem in the network configuration, "SSH connection has not been established to 10.254.0.8 in 300s" | 19:47 |
| alanmeadows | same... but now consistent I don't know what to believe anymore ;-) | 19:47 |
| mliima_ | I have to go now, but tomorrow I can keep trying to help you sean-k-mooney | 19:48 |
| kfox1111 | sdake_: yeah. got bit by that so often, I'm really leary of it. if I ever see it, then I don't trust it to workf or a long time, or someone showing me the bug fix notes. :) | 19:48 |
| *** portdirect has quit IRC | 19:48 | |
| *** portdirect has joined #openstack-kolla | 19:48 | |
| sean-k-mooney | mliima_: i guess i will have to read up on manila then as i dont understand how manila instance withc is connected to a different neutron network is ment to be pinable form the private network in the kolla docs example | 19:48 |
| sean-k-mooney | mliima_: thanks for your help | 19:49 |
| *** portdirect has quit IRC | 19:49 | |
| mliima_ | perhaps this may be a bug sean-k-mooney | 19:49 |
| sean-k-mooney | perhaps ill try to do a littl more diging thanks | 19:50 |
| mliima_ | or I may have done some step that is not in guide | 19:50 |
| mliima_ | bye | 19:51 |
| *** mliima_ has quit IRC | 19:51 | |
| openstackgerrit | Michal Jastrzebski (inc0) proposed openstack/kolla: Run docker registry in gate https://review.openstack.org/404405 | 19:55 |
| openstackgerrit | caoyuan proposed openstack/kolla: Replaces uuid.uuid4 with uuidutils.generate_uuid() https://review.openstack.org/404406 | 19:56 |
| *** jgriffith_away is now known as jgriffith | 19:56 | |
| *** portdirect has joined #openstack-kolla | 19:56 | |
| *** msimonin has joined #openstack-kolla | 19:59 | |
| *** senk has joined #openstack-kolla | 20:00 | |
| *** unicell1 has quit IRC | 20:01 | |
| *** jascott1 has quit IRC | 20:01 | |
| *** sayanta__ has joined #openstack-kolla | 20:03 | |
| *** sayantan_ has quit IRC | 20:05 | |
| *** jascott1 has joined #openstack-kolla | 20:05 | |
| *** senk has quit IRC | 20:06 | |
| *** msimonin has quit IRC | 20:06 | |
| *** unicell1 has joined #openstack-kolla | 20:09 | |
| *** msimonin has joined #openstack-kolla | 20:09 | |
| kfox1111 | Error: YAML parse error on neutron-openvswitch-agent/templates/l3_agent_daemonset.yaml: error converting YAML to JSON: yaml: line 77: mapping values are not allowed in this context | 20:11 |
| kfox1111 | arg... really not helpful helm... | 20:12 |
| kristian__ | sdake_ what do you think this means and a possible fix | 20:12 |
| kristian__ | vfio: error, group 1 is not viable | 20:12 |
| kfox1111 | helm needsa good way to dump out a macro expanded template, before processing. :/ | 20:17 |
| alanmeadows | kfox1111: apparently a newer version of Go in 2.0.0 build (vs rc2) | 20:19 |
| alanmeadows | and (I quote) "that could have had an impact... but my money's on luck." | 20:19 |
| alanmeadows | not sure if like me you were using 2.0.0rc2 | 20:19 |
| alanmeadows | as we are in production | 20:19 |
| kfox1111 | i may still be using 2.0.0rc2 on this box. | 20:20 |
| kfox1111 | I think we should assume luck then and just ckeep using the include trick. | 20:20 |
| kfox1111 | I saw some place in the documentation that said it shouldn't work either. | 20:20 |
| kfox1111 | cant remember exactly where though. | 20:20 |
| inc0 | "assume luck" makes me worried...should I be worried? | 20:21 |
| alanmeadows | haha thats what I said ;-) | 20:21 |
| alanmeadows | The expert helm developer response | 20:21 |
| alanmeadows | leaves me no more confident then before I arrived | 20:21 |
| sbezverk | kfox1111: can you push what you have so far, so we could give it a spin in different environment to see if this issue is a test bed specific? | 20:22 |
| inc0 | how fun based our very architecture on not-released stuff | 20:22 |
| kfox1111 | sbezverk: I don't have a reporducer for that behavior. just something I tried a couple weeks ago to see what would work and what not. | 20:22 |
| kristian__ | inc0 sdake_ portdirect: it booted with GPU :D :D :D gonna check from the host if there is gpu attached in the vm | 20:23 |
| inc0 | good news | 20:23 |
| inc0 | :) | 20:23 |
| kristian__ | yeaaah | 20:24 |
| kristian__ | its in the host | 20:24 |
| kristian__ | finallllyyyyy | 20:24 |
| kristian__ | 00:05.0 VGA compatible controller: NVIDIA Corporation GM204 [GeForce GTX 970] (rev a1) | 20:24 |
| kristian__ | right from the output | 20:24 |
| sbezverk | kfox1111: just one more thing, would it not be a bit more managable still to have all microservice under the same service folder. Example rabbit: I have 5 separate charts, if these all are under microservices folder then missing one would be very easy, but if the are under rabbit/ then it becomes more visible, what do you think? | 20:24 |
| kristian__ | going to fix cinder and I can go and make the gaming server :D | 20:25 |
| inc0 | kristian__, which game? | 20:26 |
| kristian__ | watch dogs 2 | 20:26 |
| portdirect | kristian__: fingers crossed dude :D | 20:26 |
| kfox1111 | sbezverk: I think it will be highly unlikely someone would be building every one manually, risking missing one. | 20:26 |
| inc0 | I couldnt handle watch dogs 1 | 20:26 |
| kfox1111 | there will just be too many to build that way. | 20:26 |
| kristian__ | and others I think | 20:26 |
| kristian__ | thanks | 20:26 |
| kristian__ | thank you for all of your help | 20:26 |
| inc0 | not very captivating | 20:26 |
| kfox1111 | helm bilding takes like under a second. | 20:26 |
| inc0 | but good luck:) | 20:27 |
| kfox1111 | so a build_microservices.py should be real quick. | 20:27 |
| kristian__ | Im so hyped about that game | 20:27 |
| kfox1111 | we could do microservices/neutron/neutron-openvswitch-agent I guess, but more typing. | 20:27 |
| kfox1111 | I guess I don't have muhc of a preference there. | 20:27 |
| *** jtriley has joined #openstack-kolla | 20:28 | |
| kristian__ | also I want it to be on openstack because first I want to learn it and second I want to use it as a gaming server for new games | 20:28 |
| portdirect | kristian__: could you close that bug you opened? It's really usefull to get the info of how you managed to fix the issue incase someone else hits the same thing :) | 20:28 |
| kristian__ | yes I will | 20:28 |
| portdirect | nice one | 20:29 |
| kristian__ | give me some time so I can quickly fix my cinder problem | 20:29 |
| sdake_ | meh who cares on the directory layout | 20:31 |
| kfox1111 | ok. this seems to work about right... | 20:32 |
| kfox1111 | a few more min and should be able to post something. | 20:32 |
| alanmeadows | just fyi, according to helm devs, defines in subchart, accessed in parents is a use case that is supposed to work, so if it does not, its a bug | 20:32 |
| alanmeadows | I like confirmation I'm not using unintended magic ;-) | 20:33 |
| kfox1111 | really? | 20:33 |
| kfox1111 | very cool. | 20:33 |
| kristian__ | portdirect: where can I close the bug? | 20:33 |
| kfox1111 | alanmeadows: and thats with $foo style vars? | 20:34 |
| kfox1111 | fyi, I am using beta2 on my test box. | 20:34 |
| alanmeadows | thats with {{ define }} | 20:34 |
| alanmeadows | and if you use a --values=foo.yaml during install, its suppose to merge with the charts defaults | 20:35 |
| kfox1111 | define is for macro's touhg. | 20:35 |
| kfox1111 | not variables? | 20:35 |
| alanmeadows | it is, one mans macro could be another man's variable ;-) | 20:35 |
| kfox1111 | alanmeadows: well, yeah. we're using them to tranfer vars. | 20:35 |
| alanmeadows | but just thinking out loud | 20:35 |
| kfox1111 | but not explicitly {{ $foo = "some value in child template }} | 20:35 |
| kfox1111 | {{ $foo }} in parent | 20:36 |
| alanmeadows | I actually haven't tried that, given there is some pre "global evaluation" in order to accomplish the defines | 20:36 |
| alanmeadows | I wonder. | 20:36 |
| alanmeadows | the context would be messed up | 20:36 |
| alanmeadows | you'd have to Child.$foo | 20:36 |
| alanmeadows | but anyways | 20:37 |
| alanmeadows | given two things | 20:37 |
| openstackgerrit | Michal Jastrzebski (inc0) proposed openstack/kolla: Run docker registry in gate https://review.openstack.org/404405 | 20:37 |
| alanmeadows | 1) you can depend on child defines bubbling up for macros, and 2) you can merge in a top level set of common values with --values=cross-chart-values.yaml. | 20:37 |
| alanmeadows | if those were to reliably hold true | 20:37 |
| alanmeadows | no precopy step | 20:37 |
| sdake_ | portdirect i've got a direct dependency on your work and kevins work - so as soon as your ready to go, can you help me get my dev env setup :) | 20:38 |
| alanmeadows | and you can refer to common values without redefining them everywhere or relying on globals | 20:38 |
| alanmeadows | it just requires the operator to always install charts with --values=... | 20:39 |
| sdake_ | without those two things, i can't do mariadb which is critical path | 20:39 |
| *** diogogmt has quit IRC | 20:40 | |
| kristian__ | sdake_ can you help me close the bug? | 20:40 |
| sdake_ | kristian__ link? | 20:40 |
| sdake_ | was the bug user error? | 20:40 |
| kristian__ | https://bugs.launchpad.net/nova/+bug/1642419 | 20:40 |
| openstack | Launchpad bug 1642419 in kolla-ansible "GPU Passthrough isn't working" [Low,Confirmed] | 20:40 |
| kristian__ | kinda yes | 20:40 |
| kristian__ | its written there at the end | 20:41 |
| *** msimonin has quit IRC | 20:41 | |
| sdake_ | kristian__ cool - i dont have access to close nova bugs | 20:41 |
| sdake_ | so the nova bug triage team will have to handle that | 20:41 |
| kristian__ | ok going there now | 20:41 |
| *** diogogmt has joined #openstack-kolla | 20:42 | |
| sdake_ | portdirect ? :) | 20:42 |
| sdake_ | paying 1 AUD for documentation on a development environment | 20:42 |
| portdirect | portdirect, studiously ignores sdake_ for 10 mins :/ | 20:43 |
| portdirect | hopfully 5 | 20:43 |
| sdake_ | portdirect roger | 20:43 |
| openstackgerrit | Michal Jastrzebski (inc0) proposed openstack/kolla: Run docker registry in gate https://review.openstack.org/404405 | 20:45 |
| kfox1111 | alanmeadows: I prototyped making the whole thing just a kolla-common subpackage to the mircroservices. it works. | 20:47 |
| kfox1111 | but I we agreed apon something else, so I'll suggest that as a followon patch. | 20:48 |
| kfox1111 | Its much more clean I think, but has more steps/requirements of the build system. | 20:48 |
| kfox1111 | as helm is't very good at building multilple packages. | 20:48 |
| alanmeadows | makes sense to me | 20:50 |
| kfox1111 | really close now. :/ | 20:51 |
| kfox1111 | ./helm_build_microservices.py ~/.helm/repository/local/ | 20:51 |
| kfox1111 | Error: open /home/kfox2/.helm/repository/local/index.yaml: no such file or directory | 20:51 |
| kfox1111 | not sure why it should care.. hmm.... | 20:51 |
| *** msimonin has joined #openstack-kolla | 20:54 | |
| kfox1111 | ok. got it to work... | 20:54 |
| portdirect | sdake ping | 20:54 |
| sdake_ | wound me portdirect | 20:54 |
| portdirect | I taught i saw a wappit | 20:55 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Initial Helm support https://review.openstack.org/396296 | 20:56 |
| kfox1111 | ok. | 20:56 |
| sdake_ | cool lets get to it | 20:56 |
| kfox1111 | first pass at a poc with the discussed layout. | 20:56 |
| sdake_ | cool lets get to it was at portdirect :) | 20:56 |
| kfox1111 | got a meeting to go to now though. | 20:56 |
| portdirect | I've thrown up an updated gist of the setup I'm using: https://gist.github.com/intlabs/9ce71a4a6b93b4e2e90faec500ef70d0 | 20:56 |
| kfox1111 | sdake_: alanmeadows sbezverk please have a look at that review. | 20:56 |
| kfox1111 | bbl. | 20:57 |
| *** YefimG has joined #openstack-kolla | 20:57 | |
| *** eaguilar has quit IRC | 20:57 | |
| *** matrohon has joined #openstack-kolla | 20:57 | |
| sdake_ | kfox1111 will do | 20:57 |
| portdirect | where were we with your setup before? - we had go you up with vagrant but is was not happy? | 20:57 |
| sdake_ | right ipv6 | 20:57 |
| sdake_ | was imploding | 20:57 |
| sdake_ | portdirect is that README.md your original work? | 20:58 |
| portdirect | ok - that was prob my bad and docker weirdness | 20:58 |
| portdirect | yes | 20:58 |
| sdake_ | cool looks like i owe you one AUD | 20:58 |
| sdake_ | can you get those docs in a review | 20:58 |
| portdirect | np | 20:58 |
| *** Pavo has quit IRC | 20:58 | |
| portdirect | before I do that lets get you running then I'll fire them straigt in | 20:59 |
| sdake_ | sounds good | 21:00 |
| sdake_ | let me run through them | 21:00 |
| sdake_ | need to switch networks to make that happen | 21:00 |
| *** adrian_otto has joined #openstack-kolla | 21:00 | |
| *** kristian__ has quit IRC | 21:00 | |
| *** kristian__ has joined #openstack-kolla | 21:01 | |
| *** JoseMello has quit IRC | 21:03 | |
| sdake_ | portdirect let me have a quick look at kevin's review - its failing the gate, to unblock me i need to unblock him :) | 21:03 |
| *** Pavo has joined #openstack-kolla | 21:03 | |
| *** TxGirlGeek has quit IRC | 21:05 | |
| srwilkers_ | sdake_, looks like pep8 errors | 21:05 |
| srwilkers_ | some unused imports and lines too long | 21:06 |
| *** adrian_otto1 has joined #openstack-kolla | 21:06 | |
| *** TxGirlGeek has joined #openstack-kolla | 21:06 | |
| *** dave-mccowan has quit IRC | 21:06 | |
| *** adrian_otto has quit IRC | 21:07 | |
| *** fguillot has quit IRC | 21:07 | |
| *** gfidente|afk has quit IRC | 21:07 | |
| portdirect | np - kubeadm throwing a hissyfit too :/ | 21:08 |
| *** inc0 has quit IRC | 21:09 | |
| *** jtriley has quit IRC | 21:09 | |
| mgiles | Whew - Finally got my kubernetes dev setup working (after many failed attempts - been working on it all day) | 21:12 |
| sdake_ | mgiles ya its rough :) | 21:12 |
| sdake_ | mgiles still working on it :) | 21:13 |
| mgiles | portdirect Those instructions look like a good start but probably won't work behind a proxy/firewall. There were a bunch of extra steps for me to work around that. | 21:13 |
| sdake_ | srwilkers_ what helps when you find the gate error is to add a comment to the review with a link to the gate failure location | 21:13 |
| portdirect | mgiles: yeah - I'm not needing a proxy - but had to openup quite a bit on firewalld for my fedora boxen | 21:14 |
| sdake_ | that helps the author understand what needs fixing in the python code exposed by the linter | 21:14 |
| srwilkers_ | still doing my review sdake_, got a snippet pasted inline already | 21:14 |
| sdake_ | srwilkers_ sweet | 21:14 |
| srwilkers_ | been making it a point to get more familiar with the gates and whats inside -- havent done that much in the past | 21:14 |
| *** jtriley has joined #openstack-kolla | 21:16 | |
| mgiles | portdirect Are you going to add this to the docs? Would be useful for others starting out. If you do, we can then edit with instructions to accommodate a proxy | 21:17 |
| sdake_ | bb-10 need to collect child | 21:17 |
| *** sdake_ has quit IRC | 21:17 | |
| portdirect | mgiles: I'll get a pr in soon - if you could add any addtional steps you needed then I'll happly put them in | 21:17 |
| *** msimonin has quit IRC | 21:18 | |
| portdirect | mgiles: Would also be great to cover mac/ubuntu vagrant setups as well | 21:18 |
| mgiles | I did this on ubuntu. my steps were slightly different, but substantially similar. Haven't tried it on my Mac yet, but can try that sometime if it would help. | 21:19 |
| *** TxGirlGeek has quit IRC | 21:19 | |
| jascott1 | I did on mac yesterday without problems | 21:20 |
| *** TxGirlGeek has joined #openstack-kolla | 21:20 | |
| srwilkers_ | yeah, halcyon should work mostly out of the box with the vagrant set up, as long as youve got everything listed in config.rb | 21:21 |
| srwilkers_ | this worked fine for me portdirect | 21:21 |
| *** TxGirlGeek has quit IRC | 21:22 | |
| portdirect | srwilkers_/jascott1: great :) | 21:22 |
| portdirect | I'll aslo tidy up the ansible mod's I've made and get them over into your repo is poss - seems messy/confusing to work from a fork? | 21:23 |
| srwilkers_ | helm repo update | 21:24 |
| srwilkers_ | oops | 21:24 |
| portdirect | lol | 21:24 |
| *** krtaylor has quit IRC | 21:26 | |
| srwilkers_ | gotta head home -- be back in an hour or so | 21:28 |
| *** inc0 has joined #openstack-kolla | 21:29 | |
| *** sdake has joined #openstack-kolla | 21:31 | |
| sdake | srwilkers i can show you if you like | 21:32 |
| *** adrian_otto1 has quit IRC | 21:32 | |
| srwilkers_ | yeah sdake, thatd be awesome | 21:32 |
| *** msimonin has joined #openstack-kolla | 21:32 | |
| sdake | srwilkers htis is the top level file: http://logs.openstack.org/96/396296/29/check/gate-kolla-kubernetes-pep8-ubuntu-xenial/8ab0b44/console.html | 21:32 |
| sdake | srwilkers to figure out where it went wrong, hunt to the correct location and click | 21:33 |
| sbezverk | srwilkers_: you made the same mistake :-) as I did, we need to use helm_prebuilt.py | 21:33 |
| srwilkers_ | sbezverk, whoops | 21:33 |
| *** adrian_otto has joined #openstack-kolla | 21:33 | |
| sdake | srwilkers http://logs.openstack.org/96/396296/29/check/gate-kolla-kubernetes-pep8-ubuntu-xenial/8ab0b44/console.html#_2016-11-29_20_57_27_323284 | 21:33 |
| *** goldyfruit has quit IRC | 21:34 | |
| sdake | srwilkers to get the corerct time stamp in there just click on the timestamp | 21:34 |
| sdake | srwilkers then paste that into the log | 21:34 |
| sdake | srwilkers its fantastic when people analyze gate failures for others | 21:34 |
| sdake | it makes the commit process go much faster | 21:34 |
| sdake | you paste the url into the log that is :) | 21:34 |
| *** Jeffrey4l has quit IRC | 21:35 | |
| sdake | to get to that top level page, click the gate job that failed | 21:35 |
| sdake | pep8 = linters | 21:35 |
| sdake | there was an effort to rename pep8 to linters but it was overwhelming for the infra team to tackle so they left it as pep8 | 21:35 |
| sdake | ok portdirect lemme read ure docs and try to reproduce :) | 21:36 |
| *** rhallisey has quit IRC | 21:37 | |
| openstackgerrit | Serguei Bezverkhi proposed openstack/kolla-kubernetes: WIP adding openvswitch-ovsdb microservice https://review.openstack.org/403959 | 21:37 |
| sbezverk | sdake: just pushed ps based on new agreement | 21:38 |
| *** sdake_ has joined #openstack-kolla | 21:38 | |
| sbezverk | it is missing values.yaml as I could not generate one using kfox1111 prebuilt script | 21:39 |
| *** sdake has quit IRC | 21:41 | |
| *** YefimG has quit IRC | 21:42 | |
| *** jtriley has quit IRC | 21:42 | |
| *** diogogmt has quit IRC | 21:46 | |
| *** srwilkers_ has quit IRC | 21:46 | |
| *** lrensing has quit IRC | 21:47 | |
| *** Jeffrey4l has joined #openstack-kolla | 21:47 | |
| *** sdake has joined #openstack-kolla | 21:47 | |
| sdake | ok portdirect - now kubernetes ;) | 21:47 |
| sdake | portdirect i'll report back my findings - should take about 10 mins | 21:47 |
| sdake | portdirect good to have people test out the docs :) | 21:48 |
| portdirect | nice - my vagrant is busted :/ | 21:48 |
| sdake | doesn't sound inspiring :) | 21:48 |
| portdirect | all the tests in that readme passed? | 21:48 |
| sdake | i haven' tstarted yet | 21:48 |
| sdake | getting there | 21:48 |
| sdake | - starting now | 21:48 |
| *** sdake_ has quit IRC | 21:51 | |
| *** matrohon has quit IRC | 21:54 | |
| *** matrohon has joined #openstack-kolla | 21:54 | |
| openstackgerrit | Michal Jastrzebski (inc0) proposed openstack/kolla: WIP: packing and uploading registry https://review.openstack.org/404435 | 21:57 |
| *** TxGirlGeek has joined #openstack-kolla | 21:57 | |
| *** JRobinson__ has joined #openstack-kolla | 22:03 | |
| *** jrist has quit IRC | 22:08 | |
| *** TxGirlGeek has quit IRC | 22:09 | |
| *** diogogmt has joined #openstack-kolla | 22:12 | |
| *** Serlex has left #openstack-kolla | 22:13 | |
| *** athomas has quit IRC | 22:17 | |
| *** inc0 has quit IRC | 22:20 | |
| mgiles | portdirect regarding the instructions: I had problems with the vagrant plugin install within the Vagrantfile. | 22:21 |
| mgiles | I had to manually install the plugins and then comment these lines out of the Vagrantfile. | 22:21 |
| mgiles | Had the problem on a couple of different systems so not sure why others aren't hitting it. Maybe a different version of vagrant? | 22:21 |
| portdirect | think thats an issue - I'm not that familiar with vagrant :( what version are you using? | 22:22 |
| *** matrohon has quit IRC | 22:22 | |
| mgiles | 1.9.0 | 22:22 |
| portdirect | my f24 comp has 1.8.1 whihc seems to work fine - but 1.7.x was giving me problems on centos | 22:23 |
| mgiles | seemed to cause a recursive vagrant call | 22:23 |
| portdirect | what platform you using/virt backend? | 22:23 |
| *** TxGirlGeek has joined #openstack-kolla | 22:23 | |
| mgiles | saw it on ubuntu and then again on my mac | 22:23 |
| mgiles | same problem | 22:23 |
| portdirect | vbox virt? | 22:24 |
| mgiles | yes | 22:24 |
| portdirect | but your env works now with the changes you have made? | 22:24 |
| mgiles | yes, it did. I'm retrying following your instructions though. | 22:25 |
| *** jtriley has joined #openstack-kolla | 22:25 | |
| *** TxGirlGeek has quit IRC | 22:25 | |
| portdirect | great - I'm gonna grab a bite to eat then I'll get them up on gerrit so we can work properly on them | 22:27 |
| mgiles | sounds good. I may be offline by then, but will post comments in the morning. | 22:27 |
| *** msimonin has quit IRC | 22:28 | |
| *** msimonin1 has joined #openstack-kolla | 22:28 | |
| *** msimonin1 has quit IRC | 22:28 | |
| portdirect | sdake: (if your about - you need vagrant 1.8.1 (i've tested it) on centos for this to work - if your running the 1.7.x that I had in the gist then you may have issues with the workers getting added to k8s: uninstall the vagrant rpm and 'rm -rf ~/.vagrant.d' and start over - sorry dude.. | 22:29 |
| *** jtriley has quit IRC | 22:29 | |
| portdirect | sdake: though if you've not touched vagrant since we set it up the other day then i think you should be ok | 22:30 |
| portdirect | mgiles: cheers | 22:30 |
| *** chas has joined #openstack-kolla | 22:30 | |
| *** srwilkers_ has joined #openstack-kolla | 22:31 | |
| *** adrian_otto has quit IRC | 22:31 | |
| mgiles | cheers portdirect | 22:31 |
| *** chas has quit IRC | 22:34 | |
| sdake | portdirect i started over ;) | 22:39 |
| *** TxGirlGeek has joined #openstack-kolla | 22:39 | |
| sdake | portdirect hang tight - seem to have some progress | 22:39 |
| *** jheroux has quit IRC | 22:45 | |
| *** goldyfruit has joined #openstack-kolla | 22:51 | |
| sdake | portdirect got rolling :) | 22:51 |
| sdake | portdirect had errors with running outside vagrant | 22:52 |
| sdake | portdirect had to ssh into the vagrant node to get any work done :) | 22:52 |
| portdirect | ok - so almost home and dry | 22:52 |
| portdirect | what errors you getting? | 22:53 |
| sdake | moment let me log out and i'll tell you | 22:53 |
| sdake | note i was able to get ovsdb-server running by labeling the node properly | 22:53 |
| sdake | which is in the rest of the docs | 22:54 |
| sdake | or ovsdbsomething or naother ;) | 22:54 |
| sdake | like we were doing yesterday | 22:54 |
| sdake | i do see this: openvswitch-ovsdb-compute-rkrkj 0/1 Init:0/1 0 14m | 22:55 |
| sdake | stuck in init state | 22:55 |
| sdake | is that normal? | 22:55 |
| sdake | so two problems ;) | 22:55 |
| sdake | portdirect note i am using vagrant 1.7.z from the magic place you mentioned yesterday | 22:55 |
| portdirect | ok - lets look at them one at time - you pick whihc one first :) | 22:55 |
| sdake | lets look at outsideth e vagrant box first | 22:56 |
| portdirect | oh - ok didnt work for me on a fesh node - but good to know it's working for you :) | 22:56 |
| sbezverk | sdake: not it is probably missing something | 22:56 |
| *** slagle has quit IRC | 22:57 | |
| sbezverk | the easiest way to find out what is to run kubectl describe pod openvswitch-ovsdb-compute-rkrkj -n kolla | 22:57 |
| *** unicell1 has quit IRC | 22:57 | |
| sdake | [root@minime-03 halcyon-vagrant-kubernetes]# kubectl get pods | 22:58 |
| sdake | The connection to the server localhost:8080 was refused - did you specify the right host or port? | 22:58 |
| *** Pavo has quit IRC | 22:58 | |
| sdake | sbezverk lets tackle that one next | 22:58 |
| portdirect | ok - did you run './get-k8s-creds.sh' ? | 22:58 |
| sdake | i ran . ./get-k8s-creds.sh | 22:59 |
| sdake | and ./get-k8s-creds.sh | 22:59 |
| sdake | but let me log out of root context | 22:59 |
| sdake | and try just ./get-k8s-creds.sh | 22:59 |
| sdake | portdirect [root@minime-03 halcyon-vagrant-kubernetes]# ./get-k8s-creds.sh | 23:01 |
| sdake | Connection to 192.168.121.62 closed. | 23:01 |
| jascott1 | sdake you have to label a node kolla_compute=true also for sbezverk's PS | 23:01 |
| sdake | jascott1 yup i got that :) | 23:01 |
| jascott1 | kewl | 23:01 |
| portdirect | try a kubectl command again | 23:01 |
| *** diogogmt has quit IRC | 23:01 | |
| portdirect | should work this time | 23:02 |
| sdake | ok it works | 23:02 |
| *** Pavo has joined #openstack-kolla | 23:02 | |
| sdake | so - i think kubernetes was still doing its setup | 23:02 |
| portdirect | I'll add better feedback to that script: https://raw.githubusercontent.com/portdirect/halcyon-vagrant-kubernetes/master/get-k8s-creds.sh | 23:02 |
| Pavo | evening | 23:02 |
| sdake | sup pavo | 23:03 |
| Pavo | gotta question | 23:03 |
| srwilkers_ | yo pavo | 23:03 |
| portdirect | could you also run kubectl get nodes? just to satisfy my paranoia that they are all checking in? | 23:03 |
| Pavo | how hard would it be to use ESXi VMWare as a hypervisor using kolla-ansible? | 23:03 |
| sdake | portdirect 4 nodes in ready state | 23:04 |
| sdake | pavo no idea | 23:04 |
| jascott1 | thats a nova question right | 23:04 |
| Pavo | was just wondering if would be possible | 23:04 |
| sdake | openstack config = black magic to me :) | 23:04 |
| sdake | not an operator | 23:04 |
| Pavo | lol | 23:04 |
| sdake | not clear to me how we orchestrate a system using helm | 23:05 |
| sdake | pod labeling etc | 23:05 |
| sdake | oh well - problem for later ;) | 23:06 |
| srwilkers_ | baby steps ;) | 23:06 |
| sdake | thats right | 23:06 |
| portdirect | sdake: wanna have a look at that pod now? | 23:07 |
| sdake | portdirect which one | 23:07 |
| sdake | mariadb? | 23:07 |
| portdirect | the one stuck in init | 23:07 |
| sdake | oh right | 23:07 |
| sdake | sorry - mortal form grows weary :) | 23:07 |
| sdake | need sustinance | 23:08 |
| sdake | so lets rock on that pod stuck :) | 23:08 |
| portdirect | np, sbezverk you around too? | 23:08 |
| sbezverk | sdake: kubectl describe pod openvswitch-ovsdb-compute-rkrkj -n kolla | 23:08 |
| sdake | http://paste.fedoraproject.org/493523/48046094/ -> http://paste.fedoraproject.org/493523/48046094 | 23:09 |
| sdake | failedmount | 23:09 |
| sbezverk | sdake: you need to create one config map | 23:09 |
| sdake | ok - why? | 23:09 |
| sdake | for the external config maps? | 23:10 |
| sbezverk | which is used for a workaround we have for dns and pods with net=host | 23:10 |
| sdake | pods with net=host = dataplane? | 23:10 |
| sbezverk | sdake: It is long type I would prefer to exaplain verbally | 23:10 |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Move barbican precheck into its own role https://review.openstack.org/404465 | 23:10 |
| portdirect | i can quickly try | 23:10 |
| sbezverk | but let's do it differently | 23:11 |
| sbezverk | you can disable workaround in values.yaml | 23:11 |
| portdirect | as I have worked round this on that k8s setup sdake is using | 23:11 |
| sbezverk | and then pod will not require this config map | 23:11 |
| portdirect | so it's not needed anymore | 23:11 |
| sdake | ok shall i delete this helm install then? | 23:11 |
| portdirect | (I'll shut up for a bit) | 23:12 |
| sbezverk | sdake: change values.yaml for enable_resolve_conf_net_host_workaround: false | 23:12 |
| sbezverk | sdake: yes helm ls | 23:12 |
| sbezverk | then helm delete {what you see in previous output} | 23:12 |
| sbezverk | portdirect: what workaround you found for this? | 23:13 |
| sdake | sbezverk yup i know ls nd delete ;) | 23:14 |
| portdirect | sbezverk: https://github.com/portdirect/halcyon-kubernetes/blob/kolla_dev/kube-deploy/roles/kube-prep/templates/resolv.conf.j2 | 23:14 |
| sdake | is only my second rodeo :) | 23:14 |
| portdirect | sbezverk: and I set up a route to the k8s dns service ip range from the hosts | 23:14 |
| sbezverk | portdirect: great, will work too. | 23:15 |
| sbezverk | sdake: when you change that values please run helm install again | 23:16 |
| kfox1111 | back. | 23:16 |
| portdirect | sbezverk: this is required for the kubelet to be able to mount RBD devices and a few other things as well | 23:16 |
| *** yuanying has joined #openstack-kolla | 23:16 | |
| sbezverk | kfox1111: can you check the error I saw when try to run helm_prebuild? | 23:16 |
| sdake | running | 23:17 |
| kfox1111 | sbezverk: yeah. just starting to look at the reviews again. | 23:17 |
| sdake | sup kfox1111 | 23:17 |
| sdake | ok ready for m0ar! | 23:17 |
| sdake | kfox1111 any chance ou can bet your review into shape | 23:17 |
| sdake | kfox1111 i think it looks pretty close as is | 23:17 |
| portdirect | I'm gonna grab that food - back in 30 | 23:18 |
| sbezverk | kfox1111: thanks, I reformated my openvswitch ps for new format, just waiting to get values.yaml generation | 23:18 |
| sdake | kfox1111 know your tired and whatnot :) | 23:18 |
| sdake | bet/beat | 23:18 |
| kfox1111 | sbezverk: I didn't check in the results of tools/helm_prebuild.py as I didn't want to complicate the initial review. | 23:18 |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Move gnocchi precheck into its own role https://review.openstack.org/404468 | 23:18 |
| kfox1111 | did you try that first? | 23:18 |
| sbezverk | kfox1111: Yes I ran helm_prebuild and it gave me some weird error | 23:19 |
| kfox1111 | hmm... | 23:19 |
| kfox1111 | maybe I missed checking in a file... | 23:19 |
| kfox1111 | I'll do a fresh checkout of it to test. | 23:19 |
| sdake | kfox1111 might check the review comments sa well | 23:20 |
| sdake | not passing pep8 gate etc | 23:20 |
| sdake | ok lets leave kfox1111 to it :) | 23:20 |
| kfox1111 | yeah. was just lookingn for general idea of it was the right aproach. can polish once thats ok. :) | 23:21 |
| sdake | kfox1111 there is only one question in the review from me about api related thing | 23:21 |
| sdake | rest looks good to me - minus the not working part ;) | 23:21 |
| kfox1111 | so, fresh checkout, tools/helm_prebuild.py worked for me. | 23:22 |
| kfox1111 | and tools/helm_build_microservices.py ~/.helm/repository/local/ too. | 23:22 |
| sdake | hmm my review was rushed i don't recall seeing two tools | 23:22 |
| sdake | (had to pick up kids) | 23:22 |
| sbezverk | hm, not sure what it is then :-( | 23:23 |
| sbezverk | kfox1111: to create a new microservice do I need to add anything in all_values.yaml? | 23:23 |
| sdake | srwilkers_ you went all out in the review :) | 23:24 |
| kfox1111 | helm install kolla/neutron-openvswitch-agent --dry-run --debug --set selector_key=kfoxwashere | grep kfox | 23:24 |
| kfox1111 | kfoxwashere: "true" | 23:24 |
| kfox1111 | overrides seem to work too. | 23:24 |
| sdake | srwilkers_ typically its just good enough to point to the gate failure not mark each indivdiual line | 23:24 |
| sdake | srwilkers_ but up to you :) | 23:24 |
| *** sayanta__ has quit IRC | 23:24 | |
| kfox1111 | sbezverk: yeah. it looks for a key based on the service name. | 23:24 |
| sdake | srwilkers_ maybe you can make some new workflows for us :) | 23:24 |
| kfox1111 | we could make that optional too. | 23:24 |
| srwilkers_ | whoops | 23:24 |
| kfox1111 | oh. maybe thats whats failing for you maybe. | 23:25 |
| sdake | srwilkers_ no woops about it | 23:25 |
| sbezverk | kfox1111: ahh that was missing then | 23:25 |
| kfox1111 | if there are other directories there. | 23:25 |
| *** unicell has joined #openstack-kolla | 23:25 | |
| sdake | portdirect so i think your directions are good to go - i've tested em | 23:25 |
| sbezverk | so before running tool I need to add section into all_values.yaml | 23:25 |
| sdake | portdirect lets get em in the docs in some way - they need a bit of tidying but we can iterate on em | 23:25 |
| *** Pavo has quit IRC | 23:25 | |
| srwilkers_ | sdake, pardon the ignorance but what workflows? | 23:26 |
| sdake | srwilkers_ the gerrit review work you did - is the openstaack workflow | 23:26 |
| kfox1111 | sbezverk: yeah. do you want me to make that optional? | 23:26 |
| sdake | srwilkers_ followed by two +2 reviews | 23:26 |
| kfox1111 | probably fine I guess. | 23:26 |
| sdake | or -1s with comments fixed | 23:26 |
| sdake | some projects are harder then others | 23:27 |
| sdake | kolla is pretty loose with the reviews | 23:27 |
| sdake | getting something into neutron for example - mission impossible ;) | 23:27 |
| srwilkers_ | ah, i got you. wasnt sure what you meant by new ones :) | 23:27 |
| srwilkers_ | no kidding | 23:27 |
| sbezverk | kfox1111: please because even if I added empty section it does not like it either | 23:27 |
| sdake | an hp manager once told me they bring donuts into the office every time a nova patch merges they produced ;) | 23:28 |
| *** TxGirlGeek has quit IRC | 23:28 | |
| sdake | i was like "are you serious?" | 23:28 |
| sbezverk | so it really want to see something inside. having copy just common part should be mandatory | 23:28 |
| sdake | he was serious. ;) | 23:28 |
| openstackgerrit | caoyuan proposed openstack/kolla: Replaces uuid.uuid4 with uuidutils.generate_uuid() https://review.openstack.org/404406 | 23:28 |
| srwilkers_ | must be depressing getting donuts 2-3 times a year | 23:28 |
| sdake | srwilkers_ wow | 23:28 |
| sdake | srwilkers_ :) | 23:28 |
| srwilkers_ | implied sarcasm of course | 23:28 |
| srwilkers_ | ;) | 23:28 |
| kfox1111 | sbezverk: k. will do in a few min. | 23:29 |
| sdake | fortunately there are no donut parties for kolla | 23:30 |
| sdake | still laughing about that one srwilkers_ :) | 23:32 |
| *** vhosakot has quit IRC | 23:32 | |
| srwilkers_ | ;) | 23:32 |
| sdake | laughed so much my daughter wanted me to explain it to her | 23:33 |
| srwilkers_ | oh man | 23:33 |
| kfox1111 | sdake: show her this: https://www.youtube.com/watch?v=4ht22ReBjno :) | 23:35 |
| kfox1111 | "waaaa??" | 23:35 |
| kfox1111 | :) | 23:35 |
| *** Pavo has joined #openstack-kolla | 23:35 | |
| jascott1 | k8s for kids! | 23:36 |
| srwilkers_ | yes! | 23:36 |
| jascott1 | "do you wish you had your own HA Minecraft cluster?" | 23:36 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Initial Helm support https://review.openstack.org/396296 | 23:36 |
| Pavo | so if I deploy on say 3 nodes and a few weeks later I want to add another 3 nodes, do I have to do a destroy to add the new nodes? | 23:37 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Initial Helm support in the gate https://review.openstack.org/404332 | 23:38 |
| kfox1111 | Pavo: shouldn't have to | 23:38 |
| Pavo | reconfigure? | 23:38 |
| jascott1 | sdake all are we agreeing to use entrypoint for init container for microservice deployment yamls? | 23:38 |
| kfox1111 | jascott1: optionally. | 23:38 |
| jascott1 | pavo are you talking about adding k8s nodes? | 23:39 |
| sdake | jascott1 its in the spec | 23:39 |
| Pavo | no no not yet, I am told kolla-ansible currently | 23:39 |
| *** mgiles has quit IRC | 23:39 | |
| jascott1 | ah | 23:39 |
| kfox1111 | jascott1: don't really need my pods having service accounts. just entrypoint maybe for blocking until unix sockets come up. | 23:39 |
| jascott1 | yes! | 23:39 |
| Pavo | so only have to use kolla-ansible reconfigure after I updated my inventory file correct? | 23:40 |
| sdake | jascott1 in an older review that probably flew by you I asked kfox1111 where was init-container, and he suggested we tackle that in a followon review once we get the structure in place | 23:40 |
| kfox1111 | +1. | 23:40 |
| sdake | rather init-container with entrypoint | 23:40 |
| jascott1 | cool so I am working on glance charts based on sbezverk's patch | 23:40 |
| kfox1111 | lets do jinja2 -> gotl/helm in one ps, then add entrypoint as a new feature ps. | 23:41 |
| sdake | i think portdirect is too :) | 23:41 |
| jascott1 | wondering how to handle glances db dependency | 23:41 |
| sdake | so the way we handle conflicts on the work is via launchpad | 23:41 |
| sdake | that way we dont both do the same thing | 23:41 |
| kfox1111 | jascott1: I think this is another case where we might want to use helm deps. | 23:41 |
| portdirect | nope - keystone/ceph | 23:41 |
| sdake | portdirect ooops thought you mentioned you wer edoing glance | 23:41 |
| portdirect | np :) | 23:42 |
| sdake | hard to keep it all straight- hence the launchpad :) | 23:42 |
| kfox1111 | have the kolla-common kind of thing where we put skeleton jobs, like the keystone-endpoint creation job, | 23:42 |
| jascott1 | why dont we just make (for instance) glance-db at same level as glance-api and glance-registry? then its optional and maybe use values file to control if it gets installed | 23:42 |
| kfox1111 | and then make a glance-create-entrypoint job package that includes it and calls it with the right params. | 23:43 |
| kfox1111 | jascott1: I think your up a level at service packages. | 23:44 |
| jascott1 | in service there is micro1 micro2 which share the same db | 23:44 |
| jascott1 | where to put db? | 23:44 |
| jascott1 | (chart) | 23:45 |
| *** inc0 has joined #openstack-kolla | 23:45 | |
| kfox1111 | jascott1: which review are you working off of? | 23:46 |
| jascott1 | https://review.openstack.org/#/c/403959/ | 23:46 |
| *** lamt has quit IRC | 23:46 | |
| jascott1 | basically taking the glance k8s stuff from kolla-k8s and converting to gotpl following that lead | 23:47 |
| kfox1111 | jascott1: yeah. there's kind of two parallel prototypes going right now... | 23:47 |
| kfox1111 | https://review.openstack.org/#/c/396296/ is the other one. | 23:47 |
| jascott1 | right | 23:47 |
| jascott1 | we talked a lot yesterday about a more flat chart structure and not using requirements | 23:47 |
| kfox1111 | the layout is different. | 23:47 |
| kfox1111 | I think requirements might solve some issues. | 23:49 |
| kfox1111 | for example, if we make a microservie rabbitmq, then the glance service package can include a rabbit of its own for rpc. | 23:49 |
| kfox1111 | nova can too. | 23:49 |
| jascott1 | yes but what is advantage of using helms requirements? | 23:50 |
| sbezverk | guys with this new approach, do NOT forget to re-run helm_prebuild whenever you change all_values | 23:50 |
| jascott1 | i can install a named rabbit and tell my service about it | 23:50 |
| sbezverk | :-) I was scratching my head why I changed to ignore workaround but it still looking for workaround config map | 23:50 |
| kfox1111 | jascott1: exactly that. being able to have a microservice instantiated in more then one service package as one. | 23:50 |
| jascott1 | sorry i might be behind, you were proposing dynamically manipulating requirements file tho right? | 23:51 |
| sbezverk | kfox1111: do you think you can print out nicely the values you copy to microservice? | 23:51 |
| kfox1111 | jascott1: thinking about it. if we keep versions in a single file, it would be pretty easy to fill in requirements with the right versions I think. | 23:51 |
| kfox1111 | sbezverk: yeah. its just a yaml file. | 23:52 |
| sbezverk | since they are stored in a not very nice format in microservice directory | 23:52 |
| kfox1111 | sbezverk: I was debating that a bit... we could generate it nicer, but then it would be more tempting for users to edit it directly then too. | 23:52 |
| sbezverk | kfox1111: I got it, I thought maybe if you print them out nicely it will give a way to visually inspect them one more time | 23:53 |
| sbezverk | but it is up to you | 23:53 |
| kfox1111 | oh... but... the way I'm writing it, I might be able to put a comment up at tthe top saying its generated.... | 23:53 |
| kfox1111 | then pretty print it. | 23:53 |
| kfox1111 | let me try that... | 23:53 |
| sbezverk | kfox1111: so I was able install chart template using helm_prebuild tool and then helm install I am very very happy !!!! | 23:54 |
| kfox1111 | sbezverk: awesome. :) | 23:54 |
| *** goldyfruit has quit IRC | 23:55 | |
| sbezverk | kfox1111: thank you! | 23:55 |
| kfox1111 | np. :) | 23:57 |
| *** Pavo has quit IRC | 23:57 | |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Initial Helm support https://review.openstack.org/396296 | 23:58 |
| kfox1111 | sbezverk: try that one for values exported. | 23:59 |
| kfox1111 | should be prettier. | 23:59 |
| kfox1111 | and disclaimered. :) | 23:59 |
| sbezverk | kfox1111: one question though. do we really want oto check in this generated file into git? example now it has some my test bed specific settings | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!