*** Pavo has joined #openstack-kolla | 00:00 | |
sbezverk | any specific reason to run neutron 2.0.2 | 00:00 |
---|---|---|
sbezverk | ? | 00:00 |
kfox1111 | for now, everythings 2.0.2 in the gate. | 00:00 |
kfox1111 | 3.0.1 wasn't out at the time. | 00:01 |
sbezverk | I used 3.0.1 tags | 00:01 |
kfox1111 | we probably should get everything over with 2.0.2 just to keep it working, then upgrade the whole gate together. | 00:01 |
sbezverk | not sure, if entrypoint is possible with 3.0.1 tag, do not think anybody agrees pushing with 2.0.2 tag | 00:02 |
kfox1111 | we can keep entrypoint with a 3.0.1 tag. | 00:03 |
sbezverk | ok | 00:03 |
kfox1111 | its just the openstack services are 2.0.2 currently. | 00:03 |
kfox1111 | we're using the 3.0.1 kolla-toolbox in the gate, as the 2.0.2 one doesn't have the crudini/jq tools. | 00:04 |
sbezverk | what about "helm serve&" you are not using tiller in this case? | 00:04 |
kfox1111 | helm serve is the local repo. | 00:04 |
*** klindgren__ has joined #openstack-kolla | 00:05 | |
kfox1111 | shares packages out of ~/.helm/repos/local by default. | 00:05 |
jascott1 | tiller is release state machine | 00:05 |
jascott1 | term used loosely | 00:05 |
jascott1 | so hardcode KOLLA_CONFIG_STRATEGY? | 00:06 |
sbezverk | interesting | 00:06 |
kfox1111 | jascott1: yeah. we can reevaluate atht too at some point. would be easy to fix with a sed if we decide to do otherwise. | 00:07 |
*** ayoung has quit IRC | 00:07 | |
jascott1 | got it | 00:07 |
*** sdake_ has quit IRC | 00:07 | |
kfox1111 | sbezverk: helm repo's are basically just http servers that share out a dir full of helm tarballs. | 00:07 |
kfox1111 | helm repo add just adds a named url to point to it. | 00:07 |
*** williamcaban has joined #openstack-kolla | 00:08 | |
kfox1111 | and helm install <reponame>/package looks at the registered url for that package. fetches it and installs it. | 00:08 |
sbezverk | kfox1111: since sdake is still now show I +2 and +wf | 00:08 |
*** klindgren_ has quit IRC | 00:08 | |
kfox1111 | sbezverk: which thing? | 00:08 |
kfox1111 | neutron-server? | 00:08 |
sbezverk | yes | 00:08 |
kfox1111 | yes please. | 00:09 |
*** JRobinson_ is now known as JRobinson_afk | 00:09 | |
jascott1 | we could serve packages out of github pages. what were you guys thinking? | 00:09 |
openstackgerrit | Merged openstack/kolla-kubernetes: Helmize neutron-server https://review.openstack.org/404929 | 00:09 |
*** yingjun has joined #openstack-kolla | 00:09 | |
sbezverk | kfox1111: so for new deployments we will use common-deployment macro | 00:10 |
kfox1111 | jascott1: that would be interesting too. probably not in the main repo, but maybe we could do another one. | 00:10 |
kfox1111 | sbezverk: anything thats been using the other common-deployment lib we can do the same way. | 00:10 |
kfox1111 | should be able to just copy/tweak neutron-server to match the new service. | 00:11 |
*** yingjun has quit IRC | 00:11 | |
*** Pavo has quit IRC | 00:12 | |
*** williamcaban has quit IRC | 00:12 | |
*** TxGirlGeek has quit IRC | 00:12 | |
*** yingjun has joined #openstack-kolla | 00:13 | |
*** Pavo has joined #openstack-kolla | 00:13 | |
jascott1 | where is memcached_image_full set? | 00:13 |
jascott1 | in previous incarnations | 00:13 |
openstackgerrit | Joshua Harlow proposed openstack/kolla: Don't process/queue children not matched https://review.openstack.org/400467 | 00:14 |
*** lamt has quit IRC | 00:14 | |
sbezverk | do you guys have an account at docker hub? so we could dump entrypoint image there until kolla cores decides to push it into official kolla repo? | 00:14 |
jascott1 | we dont have a kolla memcached image right? | 00:14 |
kfox1111 | jascott1: prehelm, in kolla 's global vars. | 00:14 |
sbezverk | jascott1: we do | 00:14 |
kfox1111 | jascott1: with helm, we have to explicitly add image config bits in helm/all_values.yaml and put a little logic to build a sane default in the template itself. | 00:15 |
sbezverk | kfox1111: docker hub question ^^ | 00:15 |
jascott1 | should it be constructed from the pieces tho? host/repo/image:tag? | 00:15 |
jascott1 | instead of one liner memcached_full? | 00:15 |
openstackgerrit | Serguei Bezverkhi proposed openstack/kolla-kubernetes: Hel-ming openvswitch-vswitchd https://review.openstack.org/405107 | 00:16 |
kfox1111 | sbezverk: I may, but should double check with inc. | 00:16 |
kfox1111 | just for testing though, we can upload it to a different namespace we personally control. | 00:17 |
kfox1111 | I've used my kfox1111 user account to test things at times. | 00:17 |
kfox1111 | do you have a personal account? | 00:17 |
openstackgerrit | Serguei Bezverkhi proposed openstack/kolla-kubernetes: WIP Hel-ming Rabbitmq https://review.openstack.org/405422 | 00:17 |
sbezverk | kfox1111: nope I do not | 00:18 |
sbezverk | if it free? | 00:18 |
sbezverk | I meant is it free? | 00:18 |
openstackgerrit | Joshua Harlow proposed openstack/kolla: Allow for externally managed configuration https://review.openstack.org/332590 | 00:19 |
*** adrian_otto has joined #openstack-kolla | 00:20 | |
jascott1 | ok so should construct the image url in tpl like other examples or leave the full memcached_image_full in all_values.yaml? I see just the tag version for openswitch-ovsdb. follow that pattern? | 00:21 |
sbezverk | jascott1: yes | 00:23 |
jascott1 | alrighty | 00:23 |
kfox1111 | sbezverk: its free. | 00:24 |
kfox1111 | if you make a hub account, you get a space for free. | 00:24 |
kfox1111 | so, like, if you sign up for user sbezverk, | 00:25 |
kfox1111 | then you can: | 00:25 |
kfox1111 | docker login | 00:25 |
kfox1111 | enter your login details. | 00:25 |
kfox1111 | docker tag foo_image:whatever sbezverk/foo_image:whatever | 00:26 |
kfox1111 | docker push sbezverk/foo_image:whatever | 00:26 |
kfox1111 | then anywhere else can use sbezverk/foo_image:whatever directly. | 00:26 |
sbezverk | kfox1111: then at the gate I will just override registry info for that chart.. | 00:26 |
kfox1111 | yeah. can set image_full var to that. | 00:27 |
kfox1111 | and that one container will get redirected. | 00:27 |
sbezverk | I guess to pull image from that repo no authentication required | 00:28 |
kfox1111 | yup. its open. | 00:28 |
kfox1111 | authenticated repo's are where they make their money. | 00:28 |
sbezverk | cool, will give it a try | 00:28 |
*** ayoung has joined #openstack-kolla | 00:29 | |
*** duonghq_ is now known as duonghq | 00:29 | |
sbezverk | hm, I get this unauthorized: authentication required | 00:31 |
kfox1111 | on the push? | 00:33 |
kfox1111 | or login? | 00:33 |
*** srwilker has joined #openstack-kolla | 00:33 | |
openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Helmize openvswitch-agent and l3-agent https://review.openstack.org/404999 | 00:33 |
sbezverk | on the push | 00:33 |
sbezverk | I created repo there | 00:34 |
kfox1111 | and the login worked? | 00:34 |
kfox1111 | I see sbezverk/kolla-kubernetes | 00:35 |
portdirect | kfox1111/sbezverk: gate q for you if you have time - what args are fed into setup_gate.sh? https://github.com/openstack/kolla-kubernetes/blob/master/tools/setup_gate.sh#L57 | 00:35 |
kfox1111 | what tag are you trying to push? | 00:35 |
kfox1111 | portdirect: sec... let me dig up the code for you. | 00:35 |
portdirect | can wait | 00:35 |
sbezverk | docker push docker.io/sbezverk/kolla-kubernetes/centos-source-kubernetes-entrypoint:3.0.1 | 00:36 |
openstackgerrit | Joshua Harlow proposed openstack/kolla: Consistently use a exception vs sys.exit https://review.openstack.org/334690 | 00:36 |
kfox1111 | portdirect: https://github.com/openstack-infra/project-config/edit/master/jenkins/jobs/kolla-kubernetes.yaml | 00:37 |
jascott1 | eww got a new setup and dont see previous change id in subsequent commit -a | 00:37 |
kfox1111 | portdirect: and then the vars get permutated through: https://github.com/openstack-infra/project-config/blob/master/jenkins/jobs/projects.yaml#L6416 | 00:39 |
portdirect | kfox1111: awesome - cheers dude! | 00:40 |
kfox1111 | portdirect: no problem. the more of us that can grock the gate, the better. :) | 00:40 |
kfox1111 | sbezverk: oh. I see. | 00:40 |
kfox1111 | docker hub is ridgid in its naming. | 00:41 |
kfox1111 | you have your user name, your repo name, and then any tags. | 00:41 |
kfox1111 | no subdirs. | 00:41 |
kfox1111 | so would need: sbezverk/centos-source-kubernetes-entrypoint:3.0.1 | 00:42 |
sbezverk | oh ok, let me retag it | 00:42 |
sbezverk | :-( same thing | 00:43 |
kfox1111 | and you made the centos-source-kubernetes-entrypoint repo? | 00:43 |
sbezverk | no I made kolla-kubernetes repo | 00:43 |
kfox1111 | need to make one called centos-source-kubernetes-entrypoint, or whatever you want to call it. | 00:44 |
kfox1111 | the kolla in kolla/centos-source-foo is an org, but basically the same thing as a user. | 00:44 |
kfox1111 | the centos-source-foo is a repo, and then each push is tagged with a different tag for individual pulling. | 00:45 |
kfox1111 | the tags the stuff after the : | 00:45 |
*** JRobinson_afk is now known as JRobinson_ | 00:45 | |
*** awiddersheim has joined #openstack-kolla | 00:46 | |
kfox1111 | so each user has a list of repo's. each repo is a bucket of tagged containers (usually intended to be of the same kind of container, just diffreent releases of them) | 00:46 |
kfox1111 | similar to how github does things. | 00:46 |
kfox1111 | users have git repos. git repos have multiple tags in them. | 00:47 |
jascott1 | gerrit question: i dont see my change-id or commit msg from previous commit -a. if I commit -a anyway, I get a new change-id even if I paste the old in there | 00:47 |
jascott1 | will git review fix that up? | 00:47 |
kfox1111 | yeah. you should see the commit id. if not, somethings wrong and it iwll generate a new review. | 00:47 |
jascott1 | ya afraid of that | 00:47 |
*** tovin07 has joined #openstack-kolla | 00:47 | |
kfox1111 | if you paste it in as the last line though, it hsould do the right thing. | 00:47 |
jascott1 | docs say rerun git review -s, commit and gerrit will fix it up but I dont think thats working | 00:48 |
jascott1 | weird | 00:48 |
*** awiddersheim has quit IRC | 00:50 | |
Pavo | anyone here know anything about using custom configs with kolla-ansible? | 00:51 |
duonghq | morning guys | 00:51 |
portdirect | morning duonghq | 00:52 |
duonghq | kfox1111, around? | 00:52 |
kfox1111 | duonghq: yeah. whats up? | 00:53 |
*** unicell1 has quit IRC | 00:53 | |
duonghq | why you put test ceph_workflow in Helmize openvswitch-agent and l3-agent | 00:53 |
portdirect | just realised most of my problems are down to having a kolla-ansible running off a sshfs fuse mounts... that'll teach me for thinking I was being clever. Anyone know how to disable selinux context changing in ansible? | 00:56 |
*** unicell has joined #openstack-kolla | 00:56 | |
Pavo | portdirect http://docs.ansible.com/ansible/selinux_module.html | 00:57 |
kfox1111 | duonghq: its part of the gate jobs. | 00:57 |
kfox1111 | ensures the new helm stuff works. | 00:57 |
*** unicell has quit IRC | 00:57 | |
kfox1111 | the change there removes the jinja2 template and adds the helm based one. if the 'check experimental' works after that, then the helm template's good. | 00:58 |
*** unicell has joined #openstack-kolla | 00:58 | |
duonghq | kfox1111, I think you should add comment to commit message or put it in separate commit and base your helmize on it? | 00:58 |
*** jtriley has joined #openstack-kolla | 00:59 | |
kfox1111 | the commits should be atomic. we don't want to commit broken stuff. so doing it in one review proves the commit is good. | 00:59 |
kfox1111 | as for comment in the commit message. we could. but every commit would have the same extra message. | 01:00 |
portdirect | Pavo: I'm running in permissive already - ansible is doing the right thing and trying to change the coentext of files it's playing with - only you cant do that on a fuse mount... Think I'll just need to mirror my setup - where my code is on the boxes I'm running builds on and I edit the files via a remote mount - I was doing it the other wa as it let me PXE boot my build boxen and rebuild them without a 2 | 01:00 |
portdirect | nd thought.. | 01:00 |
duonghq | kfox1111, if you upload an ps, and other commit is based on its ps, all commit is green | 01:01 |
duonghq | agree with on you commits should be atomic | 01:01 |
duonghq | but 1 commits should do related works (in semantic,...) | 01:02 |
*** tovin07_ has joined #openstack-kolla | 01:02 | |
kfox1111 | duonghq: most openstack projects (if not more projects then that) like each commit to have a corrisponding test case, or updat the test case to continue to work. | 01:03 |
kfox1111 | it also helps the review process greatly. | 01:04 |
duonghq | kfox1111, ok, so can you put some comment on commit msg to reflect the reason you need ceph in your commit? | 01:04 |
kfox1111 | I have limited time to manually checkout a review, and test it. but if Icee the gate check passed, I'm pretty confident its safe. | 01:05 |
kfox1111 | I can start doing that. | 01:05 |
kfox1111 | for https://review.openstack.org/#/c/404999/, if I do, then it will force a recheck and take 30 more minutes. | 01:05 |
duonghq | kfox1111, nice | 01:06 |
jascott1 | cant seem to fix my gerrit :( | 01:06 |
kfox1111 | sbezverk: https://review.openstack.org/#/c/404999/ should be ready too. | 01:06 |
jascott1 | re cloned reinstalled got my checkout did commit ammend and still no existing change-id | 01:06 |
kfox1111 | jascott1: pastebin your commit message your trying to do? | 01:06 |
kfox1111 | and git review's installed? | 01:07 |
jascott1 | kfox1111 the problem is I am trying to ammend that memcached PS, but when I commit -a, the message doesnt have prev commit msg or change id like it should | 01:07 |
jascott1 | everything else seems to work, same system I submitted the original PS on | 01:08 |
jascott1 | the hook appears to be in .git/hooks idk wtf is going on | 01:09 |
duonghq | jascott1, I think should be commit --amend? | 01:09 |
kfox1111 | duonghq: know how to wrap a line while not making it contain spaces? | 01:09 |
kfox1111 | helms picky about that. | 01:10 |
*** zhubingbing has joined #openstack-kolla | 01:10 | |
jascott1 | -a is short for ammend | 01:10 |
jascott1 | --ammend | 01:10 |
kfox1111 | how did you pull your patch? | 01:10 |
sbezverk | jascott1: I usually do commit -a --amend | 01:10 |
jascott1 | from the download | 01:10 |
kfox1111 | got the link to the review handy? | 01:10 |
jascott1 | https://review.openstack.org/#/c/405677/ | 01:11 |
jascott1 | checkout with git fetch git://git.openstack.org/openstack/kolla-kubernetes refs/changes/77/405677/1 && git checkout FETCH_HEAD | 01:11 |
sbezverk | kfox1111: duonghq: I think we should leave that line as is, because that whole thing is an argument passed to helm. | 01:11 |
jascott1 | I think its just something w my local gerrit foobard | 01:11 |
jascott1 | thanks for the help, have to run pickup my daughter, bbiab | 01:12 |
kfox1111 | jascott1: that looks riht to me... | 01:12 |
sbezverk | still need to do commit -a --amend when you want to commit changes to workspace | 01:12 |
duonghq | jascott1, -a is add, not amend | 01:14 |
duonghq | kfox1111, sbezverk can we put in on env var? | 01:14 |
duonghq | and build this var before the command? | 01:14 |
duonghq | we have repeated args here | 01:15 |
sbezverk | kfox1111: your call, vars in theory should work | 01:15 |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Enable tempest test in gate https://review.openstack.org/402122 | 01:15 |
kfox1111 | sbezverk: yeah. that could work around it. | 01:16 |
duonghq | I think it make think more clear indeed, not just formating | 01:16 |
kfox1111 | maybe we make a common_vars env var, | 01:16 |
kfox1111 | then put in the stuff like the logger. | 01:16 |
kfox1111 | that doesn't shorten it up by much though. | 01:18 |
*** jrist has joined #openstack-kolla | 01:18 | |
duonghq | but we can build it value-by-value | 01:18 |
duonghq | it's also more convenient when we want to add more things to it | 01:18 |
kfox1111 | would have to rebuild it over and over. could work though... will take a stab. | 01:19 |
sbezverk | kfox1111: on the second thought, I think tha way now it is more clear, at least to me.. but it is up tpo you | 01:19 |
kfox1111 | sbezverk: yeah, I'm not very much a fan of 80 char limits. I don't run anything so limited as 80 char terms anywhere in ages. | 01:20 |
kfox1111 | this term window we're talking in is less then half my desktop wide and its over 100. | 01:21 |
v1k0d3n | sbezverk: hey guys...saw the ping | 01:21 |
v1k0d3n | what's up? | 01:21 |
v1k0d3n | portdirect: oh hey man | 01:21 |
v1k0d3n | what's up? | 01:21 |
kfox1111 | this is the only one thats so long though. can break it up as an exception. | 01:21 |
sbezverk | v1k0d3n: the helm issue when connection gets closed | 01:21 |
sbezverk | did you get any progress on its root cause? | 01:22 |
duonghq | kfox1111, terminal 80-char limited reasons it very out-dated. atm the reason it's make things easier for reader | 01:22 |
kfox1111 | actually, | 01:23 |
portdirect | sbezverk/v1k0d3n: https://github.com/kubernetes/helm/issues/1589 | 01:23 |
kfox1111 | this file is riddled with 80 char issues... | 01:23 |
kfox1111 | so I'd say it should be fixed on a ps of its own. fix them all, and then put up a gate check to prevent regressions. | 01:23 |
kfox1111 | or else we're going to be fighting it over and over. | 01:23 |
duonghq | sure, but more than 110-char is quite long,huh? | 01:24 |
duonghq | ah, ok, upcoming fix it is fine | 01:24 |
duonghq | should make thing work first | 01:24 |
duonghq | sorry | 01:24 |
*** dave-mccowan has joined #openstack-kolla | 01:24 | |
kfox1111 | duonghq: no, its good. we need some more eyes on reviews. its a good thing really. :) | 01:26 |
kfox1111 | thoughts on what we should put in description? | 01:27 |
kfox1111 | really, its a convention we should come up with. | 01:27 |
kfox1111 | or should we merge and have that discussion while we continue to merge other charts? | 01:28 |
duonghq | kfox1111, I commented on your ps, | 01:29 |
kfox1111 | k | 01:29 |
duonghq | I have meeting right now, so see you in about 1hr | 01:29 |
kfox1111 | kk | 01:30 |
sbezverk | portdirect: thanks but I do not see the error message, in the paste from kfox1111, it is clearly stating "Error: transport is closing" | 01:30 |
portdirect | ak sorry the error message posted in that bug is ther tiller side just as it is crashing | 01:31 |
portdirect | the paste from kfox shows the client side | 01:31 |
kfox1111 | sbezverk: thx | 01:32 |
portdirect | unless I'm getting the wrong end of the stick | 01:33 |
sbezverk | kfox1111: np, are you going to be around a bit more? | 01:33 |
kfox1111 | portdirect: yeah. waht I was seeing was the helm install side. | 01:33 |
kfox1111 | sbezverk: I can for a few if you need me. | 01:33 |
kfox1111 | whats up? | 01:33 |
sbezverk | kfox1111: no, I need to go but Jeffrey4l_ still not up | 01:34 |
sbezverk | if you see him, can you ask to push entrypoint image to docker? all my attempts to push my image failed :-( | 01:34 |
kfox1111 | k. | 01:35 |
sbezverk | thanks, good night folks.. | 01:36 |
kfox1111 | you tagged the file sbezverk/centos-source-kubernetes-entrypoint:3.0.1 ? | 01:36 |
kfox1111 | s/file/container/ | 01:36 |
portdirect | sbezverk: laters, have a good un | 01:36 |
kfox1111 | and then docker push sbezverk/centos-source-kubernetes-entrypoint:3.0.1 ? | 01:36 |
openstackgerrit | Merged openstack/kolla-kubernetes: Helmize openvswitch-agent and l3-agent https://review.openstack.org/404999 | 01:36 |
*** v1k0d3n has quit IRC | 01:38 | |
*** v1k0d3n has joined #openstack-kolla | 01:39 | |
*** unicell has quit IRC | 01:40 | |
*** unicell has joined #openstack-kolla | 01:40 | |
*** yingjun has quit IRC | 01:43 | |
openstackgerrit | Serguei Bezverkhi proposed openstack/kolla-kubernetes: Hel-ming openvswitch-vswitchd https://review.openstack.org/405107 | 01:44 |
*** v1k0d3n has quit IRC | 01:47 | |
*** v1k0d3n has joined #openstack-kolla | 01:47 | |
v1k0d3n | sorry sbezverk my connection got lost | 01:47 |
v1k0d3n | yeah, portdirect had the link. they're working on it. known issue (i think our responses nudged them along a bit). | 01:48 |
*** yingjun has joined #openstack-kolla | 01:49 | |
openstackgerrit | Serguei Bezverkhi proposed openstack/kolla-kubernetes: WIP Hel-ming Rabbitmq https://review.openstack.org/405422 | 01:51 |
Jeffrey4l_ | sbezverk, sup? | 01:53 |
Jeffrey4l_ | hmm. seem gone :( | 01:54 |
Jeffrey4l_ | kfox1111, sup | 01:54 |
*** gyx has joined #openstack-kolla | 01:58 | |
kfox1111 | Jeffrey4l_: he was hoping we could get a release of kubernetes-entrypoint posted to the kolla docker hub section. | 01:59 |
Jeffrey4l_ | kfox1111, i do not have the account. | 01:59 |
kfox1111 | ah. k. | 01:59 |
kfox1111 | I'll ask sdake then. | 02:00 |
Jeffrey4l_ | ok. | 02:00 |
kfox1111 | thanks | 02:00 |
openstackgerrit | Pete Birley proposed openstack/kolla-kubernetes: WIP: Add Development Setup Script for use with Kubernetes Helm https://review.openstack.org/405720 | 02:00 |
*** awiddersheim has joined #openstack-kolla | 02:02 | |
*** v1k0d3n has quit IRC | 02:05 | |
Pavo | how are you guys tonight? | 02:09 |
*** williamcaban has joined #openstack-kolla | 02:09 | |
Pavo | hows the kolla-k8s with helm coming along? | 02:10 |
duonghq | Pavo, we are moving to implement Helm microservice (upper layer of layer 3 in Ryan's spec) | 02:13 |
*** williamcaban has quit IRC | 02:13 | |
duonghq | hope that I can upload some ps when the week end | 02:14 |
zhubingbing | hello guys | 02:14 |
*** portdirect has quit IRC | 02:14 | |
*** portdirect has joined #openstack-kolla | 02:14 | |
duonghq | hi zhubingbing | 02:14 |
duonghq | have a nice week? | 02:15 |
Pavo | well found an issue today with kolla-ansible when I did a yum -y update | 02:15 |
zhubingbing | yeah | 02:15 |
Pavo | and needing some help with understanding how custom configs are done with kolla-ansible | 02:15 |
duonghq | Pavo, just ask | 02:15 |
*** adrian_otto has quit IRC | 02:16 | |
Pavo | so I understand that you have to have /etc/kolla/config/<< service name >>/<< config file >> | 02:17 |
Pavo | so would /etc/kolla/config/neutron/ml2_conf.ini work or does it have to be like /etc/kolla/config/neutron/neutron-server/ml2_conf.ini | 02:18 |
Pavo | and inside the custom config do you have to have all the settings or can you just have the settings you want to add that aren't default? | 02:19 |
*** jmccarthy has quit IRC | 02:19 | |
duonghq | Pavo, just your custom settings | 02:20 |
duonghq | and it will be merged to default one | 02:20 |
*** dave-mccowan has quit IRC | 02:20 | |
Pavo | then I guess I am doing them completely wrong then | 02:20 |
*** jmccarthy has joined #openstack-kolla | 02:20 | |
duonghq | (you can think it's layered as Docker image) | 02:20 |
Pavo | because I tried just adding my custom settings and thats all that got put into the ocnfigs on reconfigure | 02:21 |
*** pbourke_ has quit IRC | 02:21 | |
*** pbourke_ has joined #openstack-kolla | 02:22 | |
Pavo | so does it only have to be the parent service component or does it have to be the parent service component and the sub service component as the directory structure? | 02:22 |
Pavo | when making the config? | 02:22 |
Pavo | like /etc/kolla/config/neutron/ml2_conf.ini work or does it have to be like /etc/kolla/config/neutron/neutron-server/ml2_conf.ini | 02:22 |
*** unicell has quit IRC | 02:26 | |
openstackgerrit | jianyi proposed openstack/kolla-ansible: Enable senlin dashboard when enable_senlin is yes https://review.openstack.org/405078 | 02:26 |
awiddersheim | duonghq: this doesn't make much sense to me as is | 02:26 |
*** v1k0d3n has joined #openstack-kolla | 02:26 | |
awiddersheim | https://review.openstack.org/#/c/404092/ | 02:26 |
awiddersheim | but there is mention of a follow up | 02:26 |
awiddersheim | is there one yet? | 02:27 |
Pavo | anyone know the answer to that question? | 02:29 |
*** unicell has joined #openstack-kolla | 02:29 | |
awiddersheim | using heat as an example | 02:29 |
awiddersheim | https://github.com/openstack/kolla-ansible/blob/master/ansible/roles/heat/tasks/config.yml#L54-L57 | 02:29 |
awiddersheim | feel like `with_items` shouls really be `with_first_found` | 02:29 |
*** dave-mccowan has joined #openstack-kolla | 02:30 | |
awiddersheim | and remove the task that checks if `policy.json` exists and the conditional on the task itself | 02:30 |
*** v1k0d3n has quit IRC | 02:30 | |
duonghq | awiddersheim, not any follow up yet, | 02:30 |
Pavo | awiddersheim it looks like it would work to me | 02:30 |
duonghq | awiddersheim, can you fix that? | 02:31 |
awiddersheim | Pavo: it just doesn't make any sense as is | 02:31 |
awiddersheim | checking for this single file https://github.com/openstack/kolla-ansible/blob/master/ansible/roles/heat/tasks/config.yml#L47 | 02:31 |
awiddersheim | then later copying each of these | 02:32 |
Pavo | just the with_items section? | 02:32 |
awiddersheim | https://github.com/openstack/kolla-ansible/blob/master/ansible/roles/heat/tasks/config.yml#L54-L57 | 02:32 |
awiddersheim | only if that file exists | 02:32 |
awiddersheim | https://github.com/openstack/kolla-ansible/blob/master/ansible/roles/heat/tasks/config.yml#L54-L57 | 02:32 |
awiddersheim | duonghq: yes I can fix | 02:32 |
Pavo | yeah that kind of seems redundant to me | 02:32 |
awiddersheim | I am looking at some other problems | 02:32 |
*** v1k0d3n has joined #openstack-kolla | 02:32 | |
awiddersheim | with policy files | 02:32 |
awiddersheim | specifically around horizon | 02:32 |
duonghq | awiddersheim, I don't have much atm, I'm very appreciate if you can help me make more beautiful | 02:33 |
awiddersheim | will do | 02:33 |
duonghq | Pavo, iirc, it should be /etc/kolla/config/neutron/neutron-server.conf | 02:33 |
*** unicell has quit IRC | 02:33 | |
duonghq | Pavo, for your question | 02:33 |
Pavo | huh | 02:34 |
duonghq | awiddersheim, timing is tough for me atm, sorry guy | 02:34 |
Pavo | how does ml2 settings go into neutron-server.conf? | 02:34 |
Pavo | guess I'll just have to try both and see which one works | 02:35 |
*** unicell has joined #openstack-kolla | 02:35 | |
*** hfu has joined #openstack-kolla | 02:35 | |
*** kjelly has joined #openstack-kolla | 02:39 | |
*** prameswar has joined #openstack-kolla | 02:45 | |
*** srwilker has quit IRC | 02:47 | |
*** Jeffrey4l_ has quit IRC | 02:47 | |
*** yuanying_ has quit IRC | 02:48 | |
duonghq | Pavo, sorry, my fault | 02:49 |
duonghq | Pavo, around | 02:50 |
duonghq | Pavo, you can see it https://github.com/openstack/kolla-ansible/blob/master/ansible/roles/neutron/tasks/config.yml#L93 | 02:50 |
*** Jeffrey4l_ has joined #openstack-kolla | 02:50 | |
Pavo | ok so how does that only merge custom settings into default template? | 02:51 |
duonghq | Pavo, what do you mean? | 02:52 |
*** yuanying has joined #openstack-kolla | 02:52 | |
Pavo | seems like to me it would copy default template first, if there is a custom config it would over write the default template | 02:52 |
Pavo | second | 02:52 |
openstackgerrit | Zeyu Zhu proposed openstack/kolla: Ansible2.0 changes: ansible_ssh_user to ansible_user https://review.openstack.org/405752 | 02:54 |
duonghq | Pavo, it merges | 02:54 |
*** absubram has joined #openstack-kolla | 02:54 | |
duonghq | Pavo, your method: https://github.com/openstack/kolla-ansible/blob/master/ansible/action_plugins/merge_configs.py | 02:55 |
*** absubram_ has joined #openstack-kolla | 02:56 | |
Pavo | ok so "{{ node_custom_config }}/neutron/ml2_conf.ini" is the standard across the board and "{{ node_custom_config }}/neutron/{{ inventory_hostname }}/ml2_conf.ini" is for specific node correct? | 02:57 |
duonghq | yup | 02:58 |
*** coolsvap has joined #openstack-kolla | 02:58 | |
Pavo | ok I'll try adding only a few settings in my custom config tomorrow and see how it goes | 02:59 |
*** absubram has quit IRC | 02:59 | |
*** absubram_ is now known as absubram | 02:59 | |
Pavo | thanks for explaining duonghq | 02:59 |
duonghq | Pavo, my pleasure | 03:00 |
*** v1k0d3n has quit IRC | 03:08 | |
*** v1k0d3n has joined #openstack-kolla | 03:08 | |
*** v1k0d3n has quit IRC | 03:10 | |
*** williamcaban has joined #openstack-kolla | 03:10 | |
*** v1k0d3n has joined #openstack-kolla | 03:11 | |
*** prameswar has quit IRC | 03:13 | |
*** williamcaban has quit IRC | 03:15 | |
*** v1k0d3n has quit IRC | 03:18 | |
*** saneax-_-|AFK is now known as saneax | 03:27 | |
*** srwilker has joined #openstack-kolla | 03:28 | |
*** dave-mccowan has quit IRC | 03:28 | |
*** v1k0d3n has joined #openstack-kolla | 03:29 | |
*** unicell has quit IRC | 03:30 | |
*** elynn_ has joined #openstack-kolla | 03:30 | |
openstackgerrit | jianyi proposed openstack/kolla: Install senlin dashboard into horizon image https://review.openstack.org/405070 | 03:34 |
*** yuanying has quit IRC | 03:44 | |
*** prameswar has joined #openstack-kolla | 03:47 | |
*** DungHTD has joined #openstack-kolla | 03:49 | |
*** CBR09 has joined #openstack-kolla | 03:49 | |
*** CBR09 has left #openstack-kolla | 03:50 | |
*** v1k0d3n has quit IRC | 03:56 | |
*** srwilkers_ has joined #openstack-kolla | 03:57 | |
srwilkers_ | evening | 03:57 |
*** saneax is now known as saneax-_-|AFK | 03:58 | |
*** prameswar has quit IRC | 03:59 | |
openstackgerrit | Ethan Lynn proposed openstack/kolla-kubernetes: Use curl instead of wget in setup-kubectl.sh https://review.openstack.org/405785 | 03:59 |
*** Marcellin__ has quit IRC | 04:08 | |
duonghq | evening srwilkers_ | 04:11 |
*** williamcaban has joined #openstack-kolla | 04:12 | |
*** Pavo has quit IRC | 04:13 | |
*** v1k0d3n has joined #openstack-kolla | 04:17 | |
*** Pavo has joined #openstack-kolla | 04:18 | |
*** adrian_otto has joined #openstack-kolla | 04:20 | |
srwilkers_ | sup v1k0d3n | 04:21 |
*** williamcaban has quit IRC | 04:22 | |
openstackgerrit | Zeyu Zhu proposed openstack/kolla: Ansible2.0 changes: ansible_ssh_user to ansible_user. https://review.openstack.org/405752 | 04:22 |
openstackgerrit | Justin Scott proposed openstack/kolla-kubernetes: WIP Helm-ize memcached https://review.openstack.org/405677 | 04:30 |
jascott1 | wow what a total fscking idjut I am | 04:30 |
jascott1 | PSA: git commit -a != git commit --amend | 04:31 |
*** v1k0d3n has quit IRC | 04:33 | |
srwilkers_ | i feel your pain jascott1 | 04:33 |
srwilkers_ | jascott1: i too have made the same mistake | 04:33 |
jascott1 | just glad I took a break and came back and then saw the light :) | 04:35 |
jascott1 | was about to nuke my dev setup cause gerrit seemed irrevocably broken :| | 04:36 |
srwilkers_ | D: | 04:36 |
*** srwilkers_ has quit IRC | 04:36 | |
jascott1 | pilot error | 04:36 |
*** elynn_ has quit IRC | 04:36 | |
sbezverk | Jeffrey4l_: ping | 04:40 |
Jeffrey4l_ | sbezverk, pong | 04:41 |
*** mdnadeem has joined #openstack-kolla | 04:41 | |
sbezverk | Jeffrey4l_: current kolla master has one docker image which is needed for kolla-kubernetes gates. it is kubernetes-entrypoint. Is there any chance you could push it to docker hub with tag 3.0.1? | 04:42 |
Jeffrey4l_ | sbezverk, i do not have the account. you may need ask sdake or inc0 for this. | 04:43 |
sbezverk | Jeffrey4l_: got it thank you, I will try to get in touch with sdake.. | 04:44 |
Jeffrey4l_ | ok. | 04:44 |
*** JRobinson_ is now known as JRobinson_afk | 04:54 | |
*** v1k0d3n has joined #openstack-kolla | 04:55 | |
*** v1k0d3n has quit IRC | 05:10 | |
*** elynn has joined #openstack-kolla | 05:16 | |
*** williamcaban has joined #openstack-kolla | 05:18 | |
*** elynn has quit IRC | 05:21 | |
*** fragatina has joined #openstack-kolla | 05:21 | |
*** elynn has joined #openstack-kolla | 05:21 | |
*** williamcaban has quit IRC | 05:25 | |
*** fragatina has quit IRC | 05:33 | |
*** fragatina has joined #openstack-kolla | 05:33 | |
*** fragatina has quit IRC | 05:34 | |
*** srwilker has quit IRC | 05:37 | |
*** fragatina has joined #openstack-kolla | 05:38 | |
*** fragatina has joined #openstack-kolla | 05:38 | |
*** fragatina has joined #openstack-kolla | 05:39 | |
*** fragatina has quit IRC | 05:40 | |
*** fragatina has joined #openstack-kolla | 05:40 | |
*** JRobinson_afk is now known as JRobinson_ | 05:40 | |
*** fragatina has quit IRC | 05:42 | |
*** fragatina has joined #openstack-kolla | 05:43 | |
*** gyx has left #openstack-kolla | 05:47 | |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Use mariadb.cnf for mariadb custom file name https://review.openstack.org/399038 | 05:52 |
*** unicell has joined #openstack-kolla | 05:59 | |
openstackgerrit | zhubingbing proposed openstack/kolla: Install designate dashboard into horizon image https://review.openstack.org/405933 | 06:07 |
*** Jeffrey4l_ has quit IRC | 06:12 | |
*** Jeffrey4l has joined #openstack-kolla | 06:12 | |
*** Pavo has quit IRC | 06:13 | |
openstackgerrit | Zeyu Zhu proposed openstack/kolla: Ansible2.0 changes: ansible_ssh_user to ansible_user https://review.openstack.org/405752 | 06:13 |
*** prameswar has joined #openstack-kolla | 06:15 | |
*** Pavo has joined #openstack-kolla | 06:18 | |
*** adrian_otto has quit IRC | 06:19 | |
*** DungHTD has quit IRC | 06:21 | |
openstackgerrit | zhubingbing proposed openstack/kolla-ansible: Enable trove dashboard when enable_trove is yes https://review.openstack.org/405942 | 06:32 |
*** openstackgerrit has quit IRC | 06:33 | |
*** openstackgerrit has joined #openstack-kolla | 06:34 | |
openstackgerrit | Zeyu Zhu proposed openstack/kolla: Ansible2.0 changes: ansible_ssh_user to ansible_user https://review.openstack.org/405752 | 06:34 |
*** senk has joined #openstack-kolla | 06:41 | |
*** v1k0d3n has joined #openstack-kolla | 06:50 | |
*** v1k0d3n has quit IRC | 06:57 | |
openstackgerrit | Li Yingjun proposed openstack/kolla: Install searchlight-ui to horizon https://review.openstack.org/405956 | 06:57 |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Add qemu-img into glance api image https://review.openstack.org/405468 | 07:00 |
openstackgerrit | Li Yingjun proposed openstack/kolla: Install searchlight-ui to horizon https://review.openstack.org/405956 | 07:02 |
*** msimonin has joined #openstack-kolla | 07:02 | |
openstackgerrit | Li Yingjun proposed openstack/kolla-ansible: Enable searchlight ui when searchlight enabled https://review.openstack.org/405957 | 07:05 |
*** jrist has quit IRC | 07:14 | |
*** jrist has joined #openstack-kolla | 07:15 | |
*** ram__ has quit IRC | 07:18 | |
openstackgerrit | zhubingbing proposed openstack/kolla-ansible: Enable designate dashboard when enable_designate is yes https://review.openstack.org/405942 | 07:21 |
*** williamcaban has joined #openstack-kolla | 07:22 | |
*** DTadrzak has joined #openstack-kolla | 07:22 | |
openstackgerrit | Li Yingjun proposed openstack/kolla-ansible: Enable cloudkitty dashboard when enable_cloudkitty is yes https://review.openstack.org/405964 | 07:23 |
*** williamcaban has quit IRC | 07:26 | |
*** prameswar has quit IRC | 07:27 | |
*** mewald has joined #openstack-kolla | 07:29 | |
openstackgerrit | zhubingbing proposed openstack/kolla: add os_gnocchi_apache_log.lua in heka https://review.openstack.org/384599 | 07:29 |
openstackgerrit | zhubingbing proposed openstack/kolla: add aodh os_aodh_apache_log.lua in heka https://review.openstack.org/384602 | 07:31 |
*** saneax-_-|AFK is now known as saneax | 07:43 | |
*** absubram has quit IRC | 07:50 | |
openstackgerrit | Merged openstack/kolla: Install neutron-lbaas plugin in neutron-server container https://review.openstack.org/329487 | 07:51 |
*** mkoderer has joined #openstack-kolla | 07:56 | |
*** hfu has quit IRC | 07:56 | |
openstackgerrit | jianyi proposed openstack/kolla-ansible: Enable senlin dashboard when enable_senlin is yes https://review.openstack.org/405078 | 08:01 |
openstackgerrit | jianyi proposed openstack/kolla-ansible: Enable senlin dashboard when enable_senlin is yes https://review.openstack.org/405078 | 08:02 |
*** openstackgerrit has quit IRC | 08:03 | |
*** magicboiz has quit IRC | 08:04 | |
*** matrohon has joined #openstack-kolla | 08:12 | |
*** Pavo has quit IRC | 08:13 | |
*** Pavo has joined #openstack-kolla | 08:18 | |
*** Narayan has joined #openstack-kolla | 08:22 | |
*** williamcaban has joined #openstack-kolla | 08:23 | |
Narayan | I am using kolla to deploy openstack. how to restrict a package version on target machines using kolla. | 08:26 |
*** williamcaban has quit IRC | 08:28 | |
*** magicboiz has joined #openstack-kolla | 08:30 | |
*** Serlex has joined #openstack-kolla | 08:31 | |
*** openstackgerrit has joined #openstack-kolla | 08:34 | |
openstackgerrit | Jack Ning proposed openstack/kolla-ansible: Deploy openstack with specified version https://review.openstack.org/405992 | 08:34 |
*** egonzalez90 has joined #openstack-kolla | 08:35 | |
*** unicell1 has joined #openstack-kolla | 08:42 | |
*** unicell has quit IRC | 08:42 | |
*** jascott1_ has joined #openstack-kolla | 08:45 | |
*** jascott1 has quit IRC | 08:46 | |
*** JRobinson_ has quit IRC | 08:47 | |
*** magicboiz has quit IRC | 08:49 | |
openstackgerrit | jianyi proposed openstack/kolla: Install senlin dashboard into horizon image https://review.openstack.org/405070 | 08:51 |
*** shardy has joined #openstack-kolla | 08:58 | |
*** shardy has quit IRC | 08:58 | |
*** shardy has joined #openstack-kolla | 08:59 | |
*** zhubingbing has quit IRC | 09:02 | |
*** openstackgerrit has quit IRC | 09:03 | |
*** openstackgerrit has joined #openstack-kolla | 09:05 | |
openstackgerrit | jianyi proposed openstack/kolla: Install senlin dashboard into horizon image https://review.openstack.org/405070 | 09:05 |
openstackgerrit | Jack Ning proposed openstack/kolla-ansible: Added docker image tag parameter in kolla-ansible https://review.openstack.org/405992 | 09:06 |
openstackgerrit | jianyi proposed openstack/kolla: Install senlin dashboard into horizon image https://review.openstack.org/405070 | 09:08 |
*** sdake has joined #openstack-kolla | 09:11 | |
*** matrohon has quit IRC | 09:14 | |
*** hieulq has joined #openstack-kolla | 09:15 | |
*** zhugaoxiao has joined #openstack-kolla | 09:16 | |
duonghq | sdake, ping | 09:21 |
*** msimonin has quit IRC | 09:21 | |
*** msimonin has joined #openstack-kolla | 09:23 | |
*** williamcaban has joined #openstack-kolla | 09:24 | |
*** Serlex has quit IRC | 09:25 | |
sdake | morning peeps | 09:26 |
sdake | on crazy schedule | 09:27 |
sdake | chilren wake me up at 3am daily | 09:27 |
openstackgerrit | Jack Ning proposed openstack/kolla-ansible: Fix can't generate heka-custom.toml on multi-nodes https://review.openstack.org/406016 | 09:27 |
sdake | sup duonghq | 09:27 |
sdake | you think they would get the hint abut the closed door | 09:27 |
sdake | duonghq my brain is not warmed up so I hope your quesiton is easy | 09:28 |
jascott1_ | heheh | 09:28 |
sdake | jascott1_ i see your on a crazy schedule too :) | 09:28 |
jascott1_ | bout to crash... dang physical office i have to goto in the morn | 09:28 |
*** williamcaban has quit IRC | 09:28 | |
duonghq | seem that everytime I ping you I ask something, :P, just a simple question, do we need mechanism to gather all customization block and put in some place? (w/ some docs) | 09:29 |
jascott1_ | ive going to sleep to videos about learning Rust.. I go right out :) | 09:29 |
sdake | duonghq that is a question for inc0 | 09:29 |
sdake | or possibly pbourke_ could answer | 09:29 |
sdake | Jeffrey4l ping | 09:29 |
Jeffrey4l | sdake, pong | 09:30 |
jascott1_ | sdake, all have a great consciousness cycle! | 09:30 |
*** jascott1_ is now known as jascott1_zZz | 09:31 | |
duonghq | sdake, ok I'll ask when I see them | 09:33 |
sdake | jascott1_zZz groan | 09:33 |
duonghq | anw, what kind of hint about the closed door which you was asked? | 09:33 |
sdake | duonghq oh my door to my bedroom, you think the kids would ge tthe hint "DND" | 09:34 |
*** elynn has quit IRC | 09:34 | |
*** matrohon has joined #openstack-kolla | 09:34 | |
duonghq | oh | 09:34 |
*** bjolo has quit IRC | 09:36 | |
zhangshuai | http://paste.openstack.org/show/591220/ , pls have a look , 192.168.98.254 is my host, 192.68.98.79 is the docker about karbor-api service, is it right about all IP addr? | 09:37 |
openstackgerrit | Jack Ning proposed openstack/kolla-ansible: Fix can't generate heka-custom.toml on multi-nodes https://review.openstack.org/406016 | 09:39 |
pbourke_ | duonghq: hi what was the question? | 09:49 |
*** zhugaoxiao has quit IRC | 09:54 | |
*** jascott1_zZz has quit IRC | 09:54 | |
*** tovin07 has quit IRC | 09:58 | |
*** zhangshuai has quit IRC | 09:58 | |
*** zhangshuai has joined #openstack-kolla | 10:00 | |
duonghq | pbourke_, around? | 10:00 |
pbourke_ | morning duonghq | 10:00 |
duonghq | morning pbourke_ | 10:00 |
duonghq | my question is: do we need mechanism to gather all customization block and put in some place? (w/ some docs) | 10:00 |
openstackgerrit | Li Yingjun proposed openstack/kolla: Install searchlight-ui to horizon https://review.openstack.org/405956 | 10:01 |
*** mkoderer has quit IRC | 10:03 | |
openstackgerrit | Li Yingjun proposed openstack/kolla-ansible: Enable searchlight ui when searchlight enabled https://review.openstack.org/405957 | 10:04 |
duonghq | pbourke_, iirc, you live in Ireland? | 10:04 |
pbourke_ | duonghq: I do :) | 10:04 |
pbourke_ | duonghq: wrt gathering blocks, we could do that I guess | 10:05 |
*** yingjun has quit IRC | 10:05 | |
pbourke_ | I often find if I want to customise a specific piece of functionality I just look in the Dockerfile | 10:05 |
pbourke_ | if there's a block there, great, if not I look at if I need to add one | 10:05 |
duonghq | It's hard for pure ops? | 10:06 |
pbourke_ | so personally I would check there before docs which may be out of date anyway | 10:06 |
pbourke_ | if you wanted to write something like that I would not say no | 10:06 |
duonghq | it's better if we have some doc generation mechanism for keep it up-to-date? | 10:06 |
*** strigazi_AFK is now known as strigazi | 10:06 | |
pbourke_ | it would be nice sure | 10:06 |
duonghq | so I'll jot down a bp for note about the idea | 10:07 |
*** ram_ has joined #openstack-kolla | 10:07 | |
duonghq | thank you pbourke_ | 10:07 |
*** Pavo has quit IRC | 10:13 | |
duonghq | pbourke_, FYI: https://blueprints.launchpad.net/kolla/+spec/extract-customisation-block | 10:15 |
pbourke_ | duonghq: thanks, marked as approved | 10:17 |
*** matrohon has quit IRC | 10:18 | |
*** Pavo has joined #openstack-kolla | 10:18 | |
duonghq | pbourke_, thanks, I think it's suitable for o-3? with medium priority? | 10:19 |
duonghq | or low for this cycle? | 10:20 |
*** portdirect is now known as portdirect_away_ | 10:20 | |
duonghq | do not many work here, imo | 10:20 |
*** tovin07_ has quit IRC | 10:21 | |
*** williamcaban has joined #openstack-kolla | 10:25 | |
*** rstarmer has joined #openstack-kolla | 10:25 | |
*** slagle has quit IRC | 10:28 | |
*** williamcaban has quit IRC | 10:29 | |
*** slagle has joined #openstack-kolla | 10:29 | |
pbourke_ | i see it as a nice to have | 10:31 |
*** prameswar has joined #openstack-kolla | 10:31 | |
pbourke_ | wishlist priority | 10:31 |
duonghq | pbourke_, the lowest priority is low | 10:32 |
pbourke_ | that's fine | 10:33 |
duonghq | thanks | 10:33 |
pbourke_ | whatever you like :) | 10:33 |
*** pbourke_ is now known as pbourke | 10:33 | |
*** Serlex has joined #openstack-kolla | 10:33 | |
*** jascott1_zZz has joined #openstack-kolla | 10:55 | |
*** magicboiz has joined #openstack-kolla | 10:57 | |
duonghq | bye guys | 10:57 |
*** duonghq has quit IRC | 10:57 | |
*** jascott1_zZz has quit IRC | 10:59 | |
*** hieulq has quit IRC | 11:04 | |
*** mliima has joined #openstack-kolla | 11:09 | |
SamYaple | yea. i bet pbourke _would_ know what the lowest priority was | 11:20 |
*** matrohon has joined #openstack-kolla | 11:20 | |
*** harbor has joined #openstack-kolla | 11:23 | |
*** harbor is now known as pd_away | 11:23 | |
*** pd_away is now known as pordirect | 11:24 | |
pordirect | mornin' all :) | 11:24 |
*** msimonin has quit IRC | 11:24 | |
jmccarthy | Morning ! | 11:25 |
*** Bico_Fino has joined #openstack-kolla | 11:25 | |
*** williamcaban has joined #openstack-kolla | 11:26 | |
pordirect | mariso: with difficulty yes - but the project is undergoing a major rework currently - we hope that by about the 20th the user experience will be getting toward *AWESOME* but currently I would describe it as anything but | 11:27 |
sdake | sup jmccarthy portdirect_away_ | 11:27 |
sdake | hmm confused by that - wonder if my client lost messages | 11:28 |
sdake | wierd, my tab completion isn't working for you portdirect | 11:29 |
*** msimonin has joined #openstack-kolla | 11:29 | |
rstarmer | has anyone been testing the lbaas/fwaas functionality? Seems that the neutron server container is getting created but is missing plugins to enable. | 11:29 |
sdake | yet my client says your connected - and it would typically highlight your name | 11:29 |
sdake | anyway if yourr there yo :) | 11:29 |
sdake | rstarmer not much | 11:29 |
sdake | rstarmer i see reviews going by making it happen now | 11:29 |
sdake | rstarmer i haven't reviewed them | 11:30 |
mliima | morning guys | 11:30 |
rstarmer | seems like the config aspect is "ok", but neutron server blows up "can't find plugin" | 11:30 |
*** Bico_Fino has quit IRC | 11:30 | |
pbourke | rstarmer: yes | 11:30 |
sdake | rstarmer this implies to me that in whatevery ou got - it doesn't got it :) | 11:30 |
pbourke | rstarmer: should work | 11:30 |
*** williamcaban has quit IRC | 11:30 | |
rstarmer | sdake: yeah... :( | 11:30 |
sdake | pbourke right works in master | 11:30 |
pbourke | yes | 11:30 |
sdake | pbourke but not in anything rstarmer is using | 11:30 |
rstarmer | pbourke: was using stable/newton | 11:30 |
pbourke | rstarmer: probably some backports needed | 11:31 |
sdake | ya definately not in newton | 11:31 |
rstarmer | or rather, trying stable/newton | 11:31 |
pbourke | i'll take a look | 11:31 |
sdake | and we dont backport features | 11:31 |
pbourke | actually | 11:31 |
pbourke | yes was about to say :) | 11:31 |
sdake | so we would have to evaluate if that is a feature or a defect | 11:31 |
rstarmer | is it a feature, or a bug? | 11:31 |
pbourke | you could still enable it via template-overrides in newton | 11:31 |
sdake | i'm not really sure | 11:31 |
pbourke | feature | 11:31 |
sdake | i could se eit has a bug too | 11:31 |
*** pordirect has quit IRC | 11:31 | |
rstarmer | well, regardless, if there's a way to override, that'd be fine by me, I just need it to work... | 11:32 |
pbourke | rstarmer: http://docs.openstack.org/developer/kolla/image-building.html#plugin-functionality | 11:32 |
sdake | but whoever makign the work i think makes that call :) | 11:32 |
*** harbor has joined #openstack-kolla | 11:32 | |
*** harbor is now known as pordirect | 11:32 | |
sdake | pbourke yo uthink that iwll work for newton? | 11:32 |
*** pordirect is now known as portdirect_ | 11:32 | |
pbourke | sdake: yes | 11:32 |
sdake | as a workaround? | 11:32 |
pbourke | all we're doing in ocata is making it easier | 11:33 |
portdirect_ | sup sdake | 11:33 |
sdake | sweet! | 11:33 |
*** msimonin has quit IRC | 11:33 | |
sdake | i didn't know that | 11:33 |
sdake | i am working really hard on other things paul atm - having a tough time staying on top of kolla-ansible reviews | 11:33 |
pbourke | also a nice thing Jeffrey4l has done for ocata is to make the dashboard enable/disable dynamically | 11:33 |
sdake | but trying to help where i can | 11:33 |
rstarmer | cool, on all accounts. | 11:34 |
pbourke | sdake: cant stay on it all :) | 11:34 |
sdake | pbourke right thats why we are a team :) | 11:34 |
rstarmer | I'll see if I can find where the lbaas plugin comes from (though i thought it was a default part of the neutron service, clearly not any longer), and see if I can get the override to work. | 11:34 |
*** bjolo has joined #openstack-kolla | 11:34 | |
rstarmer | will report back later :) | 11:34 |
rstarmer | but now, sleep is needed. | 11:34 |
rstarmer | thanks for the pointer(s) | 11:35 |
sdake | rstarmer dude i was there 10 hrs ago and got 10 hrs :) enjoy :) | 11:35 |
rstarmer | sdake: Well, 3am seems to have become my normal EOD, so this isn't too bad... | 11:35 |
*** Bico_Fino has joined #openstack-kolla | 11:37 | |
*** portdirect_ is now known as portdirect | 11:38 | |
sdake | ya i had a weirdly long day yesterday | 11:38 |
sdake | it sort of ended at 6pm | 11:38 |
*** coolsvap has quit IRC | 11:39 | |
sdake | rstarmer this is actually the start of my day | 11:39 |
sdake | any night ;) | 11:40 |
sdake | anywayway night I mean ;) | 11:40 |
sdake | goodnight | 11:40 |
rstarmer | gn :) | 11:40 |
*** rstarmer has quit IRC | 11:40 | |
portdirect | sdake: hows your selinux? I'm trying to get all like super optimal and as usual this is kicking in: https://xkcd.com/1319/, I'm trying keep my source on my local workstation and run builds etc on a remote server via a mount (currently sshfs) - works fine apart from ansible wanting to get the selinux context right - so running kolla genconfig fails? Know of any way round that? | 11:44 |
*** eaguilar has joined #openstack-kolla | 11:44 | |
*** Jeffrey4l has quit IRC | 11:44 | |
sdake | portdirect i dont run selinux | 11:45 |
sdake | portdirect kolla works in the gate | 11:45 |
sdake | with selinux | 11:45 |
sdake | completely enabled | 11:45 |
sdake | the difficulty is completely setting tags and whatnot on yoru system | 11:45 |
sdake | especially if you use tools like su | 11:45 |
sdake | everything has to go thorugh sudo i think | 11:45 |
sdake | or if you runas the root user for example | 11:45 |
sdake | that breaks the sysstem | 11:46 |
sdake | i am not sure about su | 11:46 |
sdake | but definately root - breaks the system | 11:46 |
sdake | and i dont know how to fix it | 11:46 |
sdake | i know it can be fixed with tools | 11:46 |
sdake | i dont totally get how its fixed on your host if it gets busted | 11:46 |
sdake | if you did a gate type install 9totally fresh) it would work - because that iswhat the gate does | 11:47 |
*** jtriley has quit IRC | 11:48 | |
sdake | getting the contexts right nad keeping them that wya are not trivial things | 11:48 |
portdirect | yeah - the gat type stuff has been working fine with it in enforcing (pretty impressive, otb) but it's the lack of selinux context on fuse mounts thats causing me headakes | 11:48 |
portdirect | I supect the answer is as simple as, turn it off, and get back on with life :) | 11:49 |
sdake | i think its probably easy for an operator that has bought into the selinux model to keep track of | 11:49 |
sdake | but if your not an operator that has bought into selinux, then - well its not really a good diea;) | 11:49 |
sdake | portdirect yup thats what I'd do :) | 11:49 |
sdake | sorry last sentence poorlty worded if your an operator that has not bought into se linux | 11:50 |
sdake | has NOT bought sorry :) | 11:50 |
portdirect | lol - I never got a copy of Dan Walshes infamous SeLinux coloring book, but I've been a convert the last few years now | 11:51 |
sdake | ya - i cant run it in dev | 11:52 |
sdake | it gets in the way | 11:52 |
sdake | in prod, id rpobablly ru nit | 11:52 |
sdake | so we make iit work | 11:52 |
sdake | but its hard to dev with it on | 11:52 |
portdirect | ok: brb, just rebooting to make my systems less secure | 11:55 |
sdake | you can do it without a reboot | 11:55 |
sdake | sudo setenforce 0 | 11:55 |
*** jascott1_zZz has joined #openstack-kolla | 11:56 | |
portdirect | still getting errors in permissive - I thought that should fix it as well | 11:56 |
sdake | my main workstation rusn selinux tho :) | 11:56 |
sdake | gov could prob get in to it | 11:57 |
sdake | but i think its unlikely anyone else could | 11:57 |
*** ccesario has quit IRC | 11:57 | |
sdake | unless they were professionals | 11:57 |
*** Jeffrey4l has joined #openstack-kolla | 11:58 | |
sdake | i have machines I trust ,and I have machines i dont trust | 11:58 |
sdake | but the machinesi dont trust are used by others | 11:58 |
portdirect | sdake: summary at bottom of page 2: https://www.usenix.org/system/files/1401_08-12_mickens.pdf | 11:58 |
*** palexster has quit IRC | 11:58 | |
*** jascott1_zZz has quit IRC | 12:00 | |
sdake | portdirect reading article | 12:00 |
sdake | i knwo mickens | 12:00 |
sdake | from long ago | 12:00 |
sdake | article looks interesting so far | 12:01 |
berendt | does it makes sense to use something like serverspec for our CI? | 12:01 |
sdake | sort of knwo of him | 12:01 |
sdake | i was in a security focused area very eary in my university :) | 12:01 |
berendt | Can somebody please merge https://review.openstack.org/#/c/404822/ | 12:01 |
sdake | i was exploring security - rather then fixing it | 12:01 |
portdirect | yeah, your work at sun was interesting | 12:02 |
sdake | looking back it was the wrong appproach | 12:02 |
sdake | wow how hte hell doy ou knwo about htat port? | 12:02 |
sdake | i think fixing it is better then exploring it | 12:03 |
sdake | but i didn't know nay better ;) | 12:03 |
sdake | i was interested in it | 12:03 |
sdake | and nobody around to guide me :) | 12:03 |
portdirect | drinking brandy in barcelona | 12:03 |
*** palexster has joined #openstack-kolla | 12:03 | |
sdake | that musthave been some good brandy | 12:03 |
sdake | i dont remember that ;) | 12:04 |
sdake | anyway reading article | 12:04 |
*** Narayan has quit IRC | 12:05 | |
*** rhallisey has joined #openstack-kolla | 12:05 | |
*** eaguilar has quit IRC | 12:09 | |
*** ccesario has joined #openstack-kolla | 12:10 | |
*** portdirect has quit IRC | 12:11 | |
*** shardy is now known as shardy_lunch | 12:12 | |
sdake | wow just got to the MOSSAD threat model | 12:12 |
sdake | he is right on there | 12:12 |
sdake | yup i dont really work on security as a researcher | 12:12 |
*** shardy_lunch is now known as shardy | 12:12 | |
sdake | its not something I do now | 12:12 |
sdake | Its something I did in the past | 12:13 |
*** Pavo has quit IRC | 12:13 | |
sdake | i don't know how to remember passwords | 12:13 |
sdake | its too damn hard | 12:13 |
*** eaguilar has joined #openstack-kolla | 12:13 | |
sdake | i agree the pssword problem needs to be solved | 12:13 |
berendt | sdake: have you tried diceware? | 12:13 |
sdake | i use lastpass unfortunately | 12:14 |
sdake | its a tradeoff for me | 12:14 |
berendt | diceware is pretty cool to generate passwords that a human can remember | 12:14 |
sdake | trust that lastpass can run a fortress | 12:14 |
*** shardy is now known as shardy_lunch | 12:14 | |
berendt | http://world.std.com/~reinhold/diceware.html | 12:15 |
sdake | the problem is that the passwords need to be different on every ddamn website on teh planet | 12:15 |
sdake | i got over 500 passwords | 12:15 |
sdake | possibly more | 12:15 |
sdake | more then i can possibly remember | 12:15 |
berendt | reduce your number of accounts :) | 12:15 |
sdake | i'd love to | 12:15 |
sdake | consilidation is a solution | 12:15 |
sdake | i dont know how ot really do that | 12:16 |
sdake | this guy told me his gmail got broken into | 12:16 |
sdake | lastpass model is that each website needs a different SPRNGed generated password | 12:17 |
sdake | anyway finishign article :) | 12:17 |
sdake | maybe i'll learn something useful :) | 12:17 |
*** Pavo has joined #openstack-kolla | 12:17 | |
*** jascott1_zZz has joined #openstack-kolla | 12:19 | |
sdake | i do have to say not everything is in lastpass | 12:20 |
*** portdirect has joined #openstack-kolla | 12:22 | |
*** jascott1_zZz has quit IRC | 12:23 | |
*** eaguilar_ has joined #openstack-kolla | 12:29 | |
sdake | cognitive overhead | 12:29 |
sdake | cool term | 12:29 |
sdake | first time i've seen it used | 12:29 |
*** eaguilar has quit IRC | 12:30 | |
*** fragatina has quit IRC | 12:32 | |
*** fragatina has joined #openstack-kolla | 12:32 | |
*** duonghq has joined #openstack-kolla | 12:33 | |
sdake | ya there is cognitive overhead with selinux as a dev - i think ops deal with it | 12:35 |
sdake | if they want to :) | 12:36 |
*** Bico_Fino has quit IRC | 12:36 | |
*** mewald has left #openstack-kolla | 12:36 | |
portdirect | yeah - it's not overload (just), but boy does it make you have to take a step back and think | 12:37 |
sdake | the threat he presents for gmail applies to websites quite well | 12:38 |
sdake | or any type of web service | 12:38 |
sdake | personal machines are under another threat | 12:38 |
sdake | and thatis that they contain information you may not want others to have | 12:38 |
sdake | especially since I let other people use my gear ;) | 12:39 |
sdake | it would be easy for "organized crime' for example to obtian information on personal machines | 12:41 |
sdake | another problem is other websites do get password dumped | 12:42 |
sdake | happens all the time | 12:42 |
sdake | its really disruptive to my life, because my accounts get reset | 12:43 |
sdake | for web services - his article is completely fantastic | 12:43 |
sdake | it leaves unadressed other htings, but he says "the one thing I want for christmas" | 12:43 |
*** senk has quit IRC | 12:51 | |
*** eaguilar has joined #openstack-kolla | 12:53 | |
*** eaguilar has quit IRC | 12:55 | |
*** hvlad has joined #openstack-kolla | 12:55 | |
*** eaguilar_ has quit IRC | 12:55 | |
*** jrist has quit IRC | 12:59 | |
*** sdake has quit IRC | 13:05 | |
*** eaguilar has joined #openstack-kolla | 13:06 | |
ram_ | Hi. I just followed http://docs.openstack.org/developer/kolla/quickstart.html to deploy kolla . I opened horizon. But default what username and password will be used to login Horizon. | 13:08 |
duonghq | ram_, did you run init_runonce and post-deploy? | 13:08 |
ram_ | duonghq: I ran post-deploy . and when I ran sudo kolla/tools/init-runonce it was giving command not found like that. | 13:12 |
ram_ | duonghq: Hi. sudo: kolla/tools/init-runonce: command not found | 13:12 |
duonghq | can you cd in to kolla/tools? | 13:13 |
*** msimonin has joined #openstack-kolla | 13:13 | |
openstackgerrit | Merged openstack/kolla-ansible: Enable cloudkitty ui when enable_cloudkitty is yes https://review.openstack.org/403520 | 13:15 |
*** mdnadeem has quit IRC | 13:15 | |
openstackgerrit | Merged openstack/kolla-ansible: Enable trove dashboard when enable_trove is yes https://review.openstack.org/405090 | 13:15 |
ram_ | duonghq: Yes. Still it is giving sudo: init-runonce: command not found | 13:16 |
duonghq | sudo ./init-runonce | 13:16 |
duonghq | comment not found? | 13:17 |
openstackgerrit | Merged openstack/kolla-ansible: Add missing pull task for neutron-vpnaas-agent https://review.openstack.org/401240 | 13:17 |
duonghq | (no colon) | 13:17 |
openstackgerrit | Merged openstack/kolla-ansible: Fix barbican upgrade permission issue https://review.openstack.org/404595 | 13:17 |
*** lamt has joined #openstack-kolla | 13:17 | |
*** DTadrzak has quit IRC | 13:18 | |
ram_ | duonghq: I ran as vedams@kolla1:~/kolla/tools$ sudo ./init-runonce. output: sudo: ./init-runonce: command not found | 13:18 |
duonghq | ram_, moment | 13:20 |
duonghq | ram_, | 13:20 |
duonghq | you should run in kolla-ansible | 13:20 |
duonghq | I guess that you use master? | 13:20 |
duonghq | ram_, document is not updated | 13:21 |
*** dave-mccowan has joined #openstack-kolla | 13:21 | |
ram_ | duonghq: Yes. It is master | 13:21 |
duonghq | ram_, it's in kolla-ansible | 13:22 |
ram_ | duonghq: where koola-ansible located? | 13:23 |
*** senk has joined #openstack-kolla | 13:23 | |
duonghq | ram_, it's in its own repository | 13:24 |
duonghq | ram_, http://git.openstack.org/cgit/openstack/kolla-ansible/ | 13:24 |
duonghq | here | 13:24 |
ram_ | duonghq: Actually, I cloned git clone https://git.openstack.org/openstack/kolla | 13:26 |
*** williamcaban has joined #openstack-kolla | 13:27 | |
ram_ | duonghq: I followed http://docs.openstack.org/developer/kolla/quickstart.html. Is there any new doc apart from this. | 13:30 |
*** krtaylor has quit IRC | 13:31 | |
*** magicboiz has quit IRC | 13:31 | |
*** williamcaban has quit IRC | 13:31 | |
*** dave-mcc_ has joined #openstack-kolla | 13:32 | |
*** portdirect is now known as portdirect_away | 13:34 | |
ram_ | duonghq: My another concern is : I enabled magnum service in globals.yml before running $sudo kolla-ansible deploy. But that service not deployed . why? do you know any idea? | 13:34 |
duonghq | ram_, the doc is outdated | 13:35 |
*** dave-mccowan has quit IRC | 13:35 | |
*** portdirect_away has quit IRC | 13:35 | |
duonghq | ram_, I haven't tested it yet, sorry | 13:35 |
*** v1k0d3n has joined #openstack-kolla | 13:36 | |
ram_ | duonghq: So what do you suggest me now. Actually, I want to prepare a nice POC on Kolla. | 13:36 |
ram_ | dounghq: Could you please help me in this. I am new to kolla. | 13:36 |
*** portdirect_away has joined #openstack-kolla | 13:37 | |
duonghq | ram_, so, just get the basic setting run first? | 13:38 |
*** portdirect_away has quit IRC | 13:38 | |
*** saneax is now known as saneax-_-|AFK | 13:38 | |
duonghq | we do not have much test against magnum yet, so we will find our path later? | 13:38 |
ram_ | duonghq: basic testing means? can you please clear me | 13:38 |
duonghq | ram_, I mean basic compute kit: keystone, nova, neutron, glance, cinder | 13:39 |
*** portdirect_away has joined #openstack-kolla | 13:39 | |
*** dmsimard is now known as dmsimard|pto | 13:39 | |
ram_ | duonghq: OK. Those services are automatically deployed right without enabling in global.yml? | 13:40 |
duonghq | yup | 13:40 |
*** lbeliveau has quit IRC | 13:41 | |
ram_ | duonghq: I deployed now right. Cinder service is not there. http://paste.openstack.org/show/591239/ | 13:41 |
*** mgiles has joined #openstack-kolla | 13:42 | |
*** goldyfruit has joined #openstack-kolla | 13:43 | |
ram_ | duonghq: The document what I used is outdated right. Can please provide me working document for test basic kolla functionality. | 13:43 |
duonghq | ram_, We're writing it :) | 13:43 |
duonghq | ram_, wait me about 5min, I need restart my box | 13:43 |
*** duonghq has quit IRC | 13:44 | |
ram_ | duonghq: OK. NP. restart it. | 13:44 |
*** duonghq has joined #openstack-kolla | 13:45 | |
duonghq | ram_, I'm back | 13:45 |
*** lbeliveau has joined #openstack-kolla | 13:46 | |
ram_ | duonghq: Hi. My intension is to know as much about openstack container projects. Just started with Kolla first. Can you please give me a suggetion what to proceed and what not to proceed for Kolla. | 13:48 |
duonghq | we have some use case in production | 13:49 |
duonghq | so, basically, it can run at least all core projects in openstack | 13:50 |
duonghq | at least with projects I listed above | 13:50 |
duonghq | I don't sure if I satisfied your question | 13:50 |
ram_ | duonghq: It has run all core openstack projects like : Keystone, glance, nova, cinder and neutron right. I want to have practical knowledge. Means I need to deploy kolla with the mentioned openstack projects. What I have to do now for that? | 13:52 |
*** zhubingbing has joined #openstack-kolla | 13:53 | |
duonghq | ram_, you can build docker image using kolla project and deploy using kolla-ansible project | 13:56 |
duonghq | other steps are same as the guide | 13:56 |
*** senk has quit IRC | 13:56 | |
ram_ | duonghq: OK. I understood the guide. But doing deployment failed. Because that is outdated as you mentioned. When that new document will be ready for operator usage? | 13:58 |
duonghq | not sure, if I can arrange my time, it can be done in this weekend | 13:59 |
*** fguillot has joined #openstack-kolla | 13:59 | |
ram_ | duonghq: Hmmm. I have a question . In globals.yml , I made #enable_cinder: "yes" but that service not deploy now using kolla-ansible? Is there any specific reason? | 14:01 |
duonghq | ram_, you should uncomment this line | 14:02 |
ram_ | duonghq: Oh. Whatever we want to deploy , we will have to 1)uncomment 2)made that service value "no" to "yes" right. | 14:03 |
duonghq | yup | 14:06 |
ram_ | duonghq: And In globals.yml, I didn't see options like : enable_nova, enable_glance, enable_neutron, enable_keystone options. But we have enable_cinder option. This mean nova,neutron,glance,and keystone will deploy automatically. They are default services to deploy by kolla-ansible right? | 14:06 |
duonghq | ram_, it get default config: enable | 14:07 |
ram_ | In globals.yml , default config: enable option is not there. | 14:08 |
ram_ | duonghq: don't mind. Have small small questions. | 14:09 |
duonghq | I do not say that, but nova, glance, neutron, keystone are core services and are enabled by default. | 14:09 |
duonghq | anw, just let it rol | 14:10 |
duonghq | roll | 14:10 |
*** Pavo has quit IRC | 14:13 | |
*** hfu has joined #openstack-kolla | 14:14 | |
ram_ | duonghq: OH. As you said, When I ran $./init-runonce it will initialize an environment with a glance image and neutron networks right. What This initialization provide us? And using what credentials I can open horizon UI from browser? | 14:14 |
duonghq | ram_, you can get from /etc/kolla/admin-openrc | 14:15 |
*** eaguilar has quit IRC | 14:16 | |
*** athomas has joined #openstack-kolla | 14:16 | |
*** Pavo has joined #openstack-kolla | 14:17 | |
*** eaguilar has joined #openstack-kolla | 14:19 | |
ram_ | duonghq: Hi. $./init-runonce is not done correctly for me right. Then what are the consequences I will face? | 14:21 |
duonghq | ram_, can you give me it's output | 14:22 |
*** yingjun has joined #openstack-kolla | 14:22 | |
openstackgerrit | Li Yingjun proposed openstack/kolla-ansible: Enable searchlight ui when searchlight enabled https://review.openstack.org/405957 | 14:22 |
*** athomas has quit IRC | 14:24 | |
*** athomas has joined #openstack-kolla | 14:25 | |
ram_ | duonghq: Actually, under kolla/tools there is not init-runonce script. Can we generate manually? Due to that When I ran $sudo ./init-runonce it was showing http://paste.openstack.org/show/591243/ | 14:26 |
duonghq | ram_, it's in kolla-ansible repo, not kolla one | 14:27 |
*** williamcaban has joined #openstack-kolla | 14:28 | |
*** shardy_lunch is now known as shardy | 14:29 | |
duonghq | ram_, you can read this file, it's self explain | 14:29 |
openstackgerrit | Merged openstack/kolla: Corrected Punctuation in quickstart.rst https://review.openstack.org/404907 | 14:31 |
ram_ | duonghq: OK. Fine. That kolla-ansible repo not cloned. Giving error: http://paste.openstack.org/show/591247/ | 14:31 |
duonghq | ram_, tried this one: https://github.com/openstack/kolla-ansible | 14:32 |
*** yingjun has quit IRC | 14:32 | |
*** yingjun has joined #openstack-kolla | 14:32 | |
*** williamcaban has quit IRC | 14:32 | |
*** jtriley has joined #openstack-kolla | 14:33 | |
ram_ | duonghq: Now Can I do the entire setup based on old doc? | 14:34 |
*** williamcaban has joined #openstack-kolla | 14:35 | |
duonghq | ram_, yeah, just notice that the ansible bit is in kolla-ansible, which mean almost commands (except image building) is done in kolla-ansible repo | 14:35 |
*** jheroux has joined #openstack-kolla | 14:35 | |
*** hfu has quit IRC | 14:36 | |
*** yingjun has quit IRC | 14:36 | |
*** yingjun has joined #openstack-kolla | 14:36 | |
*** eaguilar has quit IRC | 14:38 | |
ram_ | duonghq: instead of kolla-build what I need to run for building openstack service docker images? | 14:39 |
duonghq | I mean you should build the image using kolla repo | 14:39 |
duonghq | in tools/build.py | 14:39 |
duonghq | or if you installed kolla using pip, you can do kolla-build | 14:39 |
*** jascott1_zZz has joined #openstack-kolla | 14:41 | |
*** jtriley has quit IRC | 14:41 | |
*** eaguilar has joined #openstack-kolla | 14:44 | |
ram_ | duonghq: You mean, I installed kolla using pip. So I can run kolla-build. I have question. We need to to clone both kolla repo and kolla ansible repo right. | 14:44 |
duonghq | yup | 14:45 |
duonghq | I mean if you install kolla from source using pip, you'll get kolla-ansible | 14:45 |
openstackgerrit | Li Yingjun proposed openstack/kolla: Install searchlight-ui to horizon https://review.openstack.org/405956 | 14:45 |
*** jascott1_zZz has quit IRC | 14:45 | |
ram_ | So here kolla-ansible repo will be used only for initialize the environment by glance image and neutron networks only right. | 14:46 |
*** senk has joined #openstack-kolla | 14:47 | |
duonghq | ram_, the kolla-ansible contain the ansible bit | 14:47 |
ram_ | duonghq; I installed kolla as $sudo pip install kolla. | 14:47 |
duonghq | which mean all ansible playbook | 14:47 |
duonghq | the Ansible playbook take care deploy openstack (in form of docker image) to cluster | 14:48 |
duonghq | the docker image is built from kolla repo | 14:48 |
duonghq | from dockerfile | 14:48 |
*** absubram has joined #openstack-kolla | 14:48 | |
*** absubram_ has joined #openstack-kolla | 14:50 | |
*** lrensing has joined #openstack-kolla | 14:53 | |
*** absubram has quit IRC | 14:53 | |
*** absubram_ is now known as absubram | 14:53 | |
ram_ | duonghq: That is OK. First I ran $sudo pip install -U ansible then I ran $sudo pip install kolla and next I cloned kolla repo. Then kolla-ansible, kolla-build and kolla-genpwd utilities came. | 14:54 |
*** inc0 has joined #openstack-kolla | 14:55 | |
inc0 | o/ | 14:55 |
duonghq | evening inc0 | 14:55 |
inc0 | morning duonghq ;) | 14:55 |
*** yingjun has quit IRC | 14:56 | |
sbezverk | inc0: morning inc0 need your helm urgently | 14:57 |
*** yingjun has joined #openstack-kolla | 14:57 | |
sbezverk | inc0: for kube gate we need entrypoint image | 14:57 |
*** senk has quit IRC | 14:58 | |
sbezverk | inc0: I know it is in the master but since we do not build images at the gate we pull them from docker hub | 14:58 |
ram_ | duonghq: But from kolla repo we didn't find init-runonce. So for that only we have to clone kolla-ansible repo. And then we can initilize the environment right? | 14:58 |
duonghq | seem that Kollish really really need update the docs :) | 14:58 |
openstackgerrit | Merged openstack/kolla-ansible: Disable notifications for cinder when no consumer https://review.openstack.org/398669 | 14:58 |
*** dave-mcc_ has quit IRC | 14:58 | |
sbezverk | inc0: could you please push kubernetes-entrypoint image to kolla repo on docker hub with tag 3.0.1 | 14:58 |
*** lrensing has quit IRC | 14:58 | |
sbezverk | ? | 14:58 |
duonghq | ram_, if we choose to clone kolla from git, we need to clone kolla and kolla-ansible | 15:00 |
*** lrensing has joined #openstack-kolla | 15:00 | |
duonghq | kolla contains dockefile and related stuff for building the Docker image | 15:00 |
duonghq | kolla-ansible contains Ansible playbook and related ultilities for deploy previous images to cluster | 15:00 |
*** yingjun has quit IRC | 15:01 | |
*** jtriley has joined #openstack-kolla | 15:02 | |
*** krtaylor has joined #openstack-kolla | 15:02 | |
ram_ | duonghq: Hmmm. That good to know for me. But I saw globals.yml , passwords.yml files from kolla-ansible repo also. But in kolla-ansible we can't find build.py under kolla-ansible/tools | 15:03 |
ram_ | But we need to edit globals.yml of kolla repo only right. | 15:04 |
*** eaguilar has quit IRC | 15:04 | |
duonghq | I prefer kolla-ansible one | 15:05 |
*** yingjun has joined #openstack-kolla | 15:07 | |
*** eaguilar has joined #openstack-kolla | 15:07 | |
*** DaveTurner has quit IRC | 15:09 | |
ram_ | duonghq: You mean , I know need to edit globals.yml of kolla repo right. But if I run kolla-genpwd where it will edit 1)passwords.yml of kolla repo (or) passwords.yml file of kolla-ansible file. | 15:09 |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Do not mount /var/lib/nova/mnt unless nfs cinder backend is enabled https://review.openstack.org/406168 | 15:10 |
duonghq | ram_, finally, globals.yml, passwords.yml and kolla-build.conf will go to /etc/kolla/ | 15:11 |
openstackgerrit | Li Yingjun proposed openstack/kolla-ansible: Enable searchlight ui when searchlight enabled https://review.openstack.org/405957 | 15:11 |
*** dave-mccowan has joined #openstack-kolla | 15:11 | |
duonghq | kolla-genpwd knows it should find passwords.yml in /etc/kolla/ | 15:11 |
duonghq | same for globals.yml | 15:11 |
ram_ | duonghq: OK. Thanks. For all i one kolla deployment we no need to create docker registry right? | 15:17 |
duonghq | if you use AIO config and just for test, yes | 15:17 |
*** inc0 has quit IRC | 15:17 | |
duonghq | if not: no | 15:17 |
ram_ | duonghq: If we are not creating docker registry, Locally it will store docker images for AIO? | 15:19 |
duonghq | just in this node, but other nodes cannot get the images | 15:19 |
ram_ | duonghq: When we need to generate and edit the kolla-build.conf. Actually, for my current setup I didn't generated kolla-build.cong file? | 15:21 |
*** inc0 has joined #openstack-kolla | 15:23 | |
duonghq | ram_, when you want to customise this | 15:23 |
duonghq | fox ex: when you need using local docker registry | 15:23 |
duonghq | or customise the images | 15:24 |
inc0 | sbezverk, back, just got it | 15:24 |
inc0 | ubuntu source is pushing | 15:24 |
ram_ | duonghq: Thank you very much for your patience in giving answers. I got good information from you. | 15:27 |
duonghq | ram_, my pleasure | 15:27 |
openstackgerrit | prameswar proposed openstack/kolla: QSG and bifrost improved https://review.openstack.org/406177 | 15:28 |
duonghq | prameswar, ping | 15:29 |
prameswar | hi | 15:30 |
prameswar | http://docs.openstack.org/developer/kolla/bifrost.html guide improved | 15:30 |
duonghq | nice | 15:30 |
duonghq | prameswar, can you do me a favor if you have time: update the QSG of kolla to reflect our repo split? | 15:31 |
duonghq | due to you are improving docs | 15:31 |
prameswar | ok | 15:32 |
prameswar | i am doing | 15:32 |
duonghq | prameswar, thank you very much | 15:32 |
*** openstackgerrit has quit IRC | 15:33 | |
*** openstackgerrit has joined #openstack-kolla | 15:33 | |
openstackgerrit | Merged openstack/kolla-kubernetes: Kolla-Kubernetes Vagrant development environment documentation https://review.openstack.org/404493 | 15:33 |
*** jascott1_zZz has joined #openstack-kolla | 15:42 | |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Do not mount /var/lib/nova/mnt unless nfs cinder backend is enabled https://review.openstack.org/406168 | 15:43 |
*** jascott1_zZz has quit IRC | 15:46 | |
*** eaguilar has quit IRC | 15:48 | |
*** inc0 has quit IRC | 15:53 | |
*** sayantan_ has joined #openstack-kolla | 15:54 | |
*** inc0 has joined #openstack-kolla | 15:54 | |
*** srwilkers_ has joined #openstack-kolla | 15:57 | |
srwilkers_ | good morning everyone | 15:57 |
duonghq | srwilkers_, nice to see you | 15:59 |
srwilkers_ | you too duonghq :) | 15:59 |
*** egonzalez90 has quit IRC | 15:59 | |
*** sayantan_ has quit IRC | 16:02 | |
*** jascott1_zZz has joined #openstack-kolla | 16:03 | |
*** sayantan_ has joined #openstack-kolla | 16:03 | |
*** absubram has quit IRC | 16:05 | |
*** jascott1_zZz has quit IRC | 16:08 | |
openstackgerrit | prameswar proposed openstack/kolla: QSG and bifrost improved https://review.openstack.org/406177 | 16:11 |
openstackgerrit | prameswar proposed openstack/kolla: QSG and bifrost improved https://review.openstack.org/406177 | 16:12 |
*** Pavo has quit IRC | 16:13 | |
*** yingjun has quit IRC | 16:14 | |
*** yingjun has joined #openstack-kolla | 16:15 | |
prameswar | Hi duonghq, done . can you please review now ? | 16:15 |
duonghq | prameswar, I just post a review, | 16:15 |
*** jascott1_zZz has joined #openstack-kolla | 16:16 | |
openstackgerrit | Eduardo Gonzalez proposed openstack/kolla: Fix barbican upgrade permission issue https://review.openstack.org/406202 | 16:16 |
*** david-lyle_ has joined #openstack-kolla | 16:17 | |
*** Pavo has joined #openstack-kolla | 16:18 | |
openstackgerrit | Eduardo Gonzalez proposed openstack/kolla: Fix aodh role https://review.openstack.org/406203 | 16:18 |
openstackgerrit | prameswar proposed openstack/kolla: QSG and bifrost improved https://review.openstack.org/406177 | 16:18 |
*** david-lyle has quit IRC | 16:19 | |
*** jascott1_zZz has quit IRC | 16:20 | |
openstackgerrit | Eduardo Gonzalez proposed openstack/kolla-ansible: Add custom policies in service.json https://review.openstack.org/399712 | 16:23 |
*** inc0 has quit IRC | 16:23 | |
*** inc0 has joined #openstack-kolla | 16:23 | |
*** adrian_otto has joined #openstack-kolla | 16:25 | |
duonghq | prameswar, ping | 16:30 |
duonghq | actually, the repo split make documentation need more details, | 16:31 |
kfox1111 | morning. | 16:31 |
duonghq | I'm writing some comments but we still need more opinion on it | 16:31 |
*** jistr is now known as jistr|biab | 16:33 | |
duonghq | sorry, just realize that this ps is about Kolla QSG | 16:33 |
srwilkers_ | morning kfox1111 | 16:34 |
duonghq | prameswar, what is your tz, can I take care the docs when you sleep? I'm in tz +7 | 16:35 |
prameswar | hi | 16:35 |
kfox1111 | srwilkers_: morning | 16:35 |
*** sayantan_ has quit IRC | 16:36 | |
*** zhubingbing has quit IRC | 16:36 | |
prameswar | shall i remove split part from my patch and someone will add new patch for this | 16:36 |
*** Serlex has quit IRC | 16:36 | |
*** sayantan_ has joined #openstack-kolla | 16:36 | |
sbezverk | kfox1111: morning, we got the entrypoint pushed to docker hib, thanks to inc0 so testing it | 16:37 |
duonghq | prameswar, sure, but you should put a note in commit message | 16:37 |
prameswar | duonghq my tz is IST . i am available for next 2 hour | 16:37 |
prameswar | will you add split part ? | 16:38 |
kfox1111 | sbezverk: morning. cool. :) | 16:38 |
duonghq | prameswar, not sure, hope that I can write these things in tomorrow evening (UTC+7) | 16:38 |
*** saneax-_-|AFK is now known as saneax | 16:39 | |
prameswar | duonghq , ok then i am removing spit part from QSG | 16:39 |
duonghq | prameswar, okay, | 16:40 |
duonghq | goodnight prameswar | 16:40 |
duonghq | night all | 16:40 |
*** vhosakot has joined #openstack-kolla | 16:41 | |
kfox1111 | oh... | 16:41 |
kfox1111 | you know what... | 16:41 |
kfox1111 | I think there is a slight gate breakage. | 16:41 |
kfox1111 | kind of. | 16:41 |
kfox1111 | sbezverk: we're not specifying distro to helm in the gate. | 16:41 |
kfox1111 | so the helm stuff in the ubuntu job is running centos containers right now... | 16:42 |
*** Jeffrey4l has quit IRC | 16:42 | |
openstackgerrit | prameswar proposed openstack/kolla: QSG and bifrost improved https://review.openstack.org/406177 | 16:43 |
prameswar | goodnight duonghq | 16:44 |
*** duonghq has quit IRC | 16:44 | |
*** sayantan_ has quit IRC | 16:45 | |
*** bmace has quit IRC | 16:47 | |
*** bmace has joined #openstack-kolla | 16:47 | |
*** matrohon has quit IRC | 16:48 | |
*** sdake has joined #openstack-kolla | 16:48 | |
sdake | 2nd morning again peeps :) | 16:49 |
sbezverk | kfox1111: it seems it is picking up centos/binary | 16:49 |
sbezverk | sdake: morning | 16:49 |
*** strigazi is now known as strigazi_AFK | 16:50 | |
kfox1111 | sbezverk: its using the default in all_values. I'm whiping up a quick ifx. | 16:50 |
kfox1111 | fix | 16:50 |
srwilkers_ | yo sdake | 16:50 |
*** sdake_ has joined #openstack-kolla | 16:51 | |
*** yingjun has quit IRC | 16:51 | |
openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Fix distro in gate https://review.openstack.org/406223 | 16:51 |
*** yingjun has joined #openstack-kolla | 16:52 | |
openstackgerrit | Paul Bourke (pbourke) proposed openstack/kolla-ansible: Add missing per service Keystone config augments https://review.openstack.org/406224 | 16:52 |
sdake_ | sbezverk we dont push to dockerhub until a release is tagged | 16:52 |
sdake_ | i dont wnat to push to dockerhu at all | 16:52 |
sdake_ | however several other cores want it | 16:52 |
kfox1111 | sdake_: it really simplifies testing and kicking the tires. | 16:53 |
sbezverk | sdake_: I agree with them, for some cases this flexibility is good.. | 16:53 |
kfox1111 | no 30+ minute build step. | 16:53 |
sdake_ | i get the value kfox1111 | 16:53 |
sdake_ | kfox1111 have heard the arguemnts about why its a good idea | 16:53 |
sdake_ | kfox1111 there are also reasons why its a bad idea | 16:53 |
sbezverk | sdake_: when inc0's internal registry is up, then it will not be that usefull, but for now it is. | 16:54 |
kfox1111 | yeah. it shuldn't be relied apon without some kind of security update mechanism. | 16:54 |
*** sdake has quit IRC | 16:54 | |
sdake_ | kfox1111 there is a whole process for that for dockerhub | 16:54 |
sdake_ | its long and involved | 16:54 |
kfox1111 | but thats generally true of the docker hub in general. its a sespool of stuff not getting updates in general. | 16:54 |
sdake_ | if the core team actually did that, and we used that, I'd be for it :) | 16:54 |
sdake_ | there are some type of certified images | 16:55 |
sdake_ | i looked into it | 16:55 |
kfox1111 | sdake_: do you have a pointer to docs to do that? | 16:55 |
kfox1111 | I'm curious. | 16:55 |
sdake_ | yup moment | 16:55 |
*** saneax is now known as saneax-_-|AFK | 16:55 | |
*** yingjun has quit IRC | 16:56 | |
sdake_ | https://docs.docker.com/docker-hub/official_repos/ | 16:56 |
sdake_ | keep in mind the cor eteam would have to maintian 180 of those dockerfiles - probably separately | 16:56 |
sdake_ | and build them and get them through a docker review process | 16:56 |
sdake_ | the first step in my mind to getting there is through VMT | 16:57 |
sdake_ | we can't get VMT done | 16:57 |
sdake_ | i spent alot of time on it during my term as ptl | 16:57 |
sdake_ | I'm not sure why it didn't get done | 16:57 |
sdake_ | I couldnt' do it alone | 16:57 |
sdake_ | need the security team's help and the core team's help | 16:57 |
sdake_ | the think that concerns me about official images is the docker best practices | 16:58 |
sdake_ | for the most part we do that | 16:58 |
sdake_ | on our generated dockerfiles | 16:58 |
sdake_ | they start out in j2 format | 16:58 |
sdake_ | also we have a layered himage approach to building | 16:59 |
sdake_ | wihch is not really atypical but different | 16:59 |
sdake_ | most softwwre in the ofifical repo is 1 image | 16:59 |
sdake_ | kolla is 180... | 16:59 |
kfox1111 | oh... they are doing like one build repo to rule them all. | 17:00 |
sdake_ | you got it | 17:00 |
sdake_ | recommend reading link | 17:01 |
kfox1111 | I don't think that will scale will with our number of containers. | 17:01 |
sdake_ | it looks straightfoward | 17:01 |
kfox1111 | well. | 17:01 |
sdake_ | but seems challenging in practice | 17:01 |
sdake_ | vmt looked straightforward too :) | 17:02 |
srwilkers_ | yeah, can definitely see the amount of work required to get that going for kolla | 17:02 |
kfox1111 | I don't think we nessisarily need to go that far. we just need to ensure we have some process to watch for updates, and push updates. | 17:02 |
srwilkers_ | this is definitely a good read though, thanks sdake_ | 17:02 |
sdake_ | kfox1111 there is a middleground yoru right | 17:03 |
sdake_ | kfox1111 we can't really maintain a distro of openstack - because that really isn't our job :) | 17:03 |
sdake_ | to do that job.. | 17:04 |
sdake_ | reqiures more people then we have | 17:04 |
kfox1111 | yes/no. we're not maintaining a distro. rdo/ubuntu is. | 17:04 |
kfox1111 | we're just building containers based on those distro's. | 17:04 |
sdake_ | rdo is a pristine implementation of openstack | 17:04 |
sdake_ | it is not maintained for security updates | 17:04 |
sbezverk | kfox1111: so the entrypoint worked for centos, but then on other gate jobs glance went south :-( | 17:05 |
kfox1111 | hmm... I guess I'm not so worried about the openstack packages itself. | 17:05 |
kfox1111 | more, the underlying distro packages we bundle. | 17:05 |
kfox1111 | ie, openssl for example. | 17:05 |
kfox1111 | those do get security updates. | 17:05 |
kfox1111 | fairly frequently. | 17:05 |
sdake_ | right thats part of centos | 17:05 |
sdake_ | i am not really sure what security updates centos gets | 17:06 |
sdake_ | i dont know the answer to tha tq | 17:06 |
sdake_ | i can look :) | 17:06 |
kfox1111 | thats something automatable. look at container in hub. do a docker run kolla/foo yum upgrade -y | 17:06 |
kfox1111 | check to see if it did anything. if so, time to build a new container and push it. | 17:06 |
kfox1111 | same for ubuntu, just apt. | 17:06 |
kfox1111 | that alone would go a long long way. | 17:07 |
sdake_ | hadn't thought of that approach | 17:07 |
kfox1111 | its something that could go in a gate periodic job I think. | 17:07 |
sdake_ | i haven't been supportive of docker hub model at all | 17:07 |
sdake_ | so i haven't been thinking how to make it happen | 17:08 |
sdake_ | i assume people that want that will make it hapepn :) | 17:08 |
kfox1111 | we wouldn't be making security updates per'se, but would just be making sure we kept up with the distro's updates. | 17:08 |
kfox1111 | yeah, thats me. :) might make it happen eventually, if no one else does first. | 17:09 |
sdake_ | wfm - i can add any core reviewer to the docker hub account | 17:09 |
sdake_ | coolsvap was the person pushing images most recently i think | 17:09 |
sdake_ | but i am not really sure | 17:09 |
sbezverk | sdake_: can you add kfox1111 and myself please, I could not move forward whole days yesterday because of missing image.. | 17:10 |
*** tonanhngo has joined #openstack-kolla | 17:10 | |
sdake_ | sorry about that - ha dto sleep :) | 17:14 |
sdake_ | ok well the real problem here is the gate is not building image | 17:15 |
sdake_ | s | 17:15 |
sdake_ | but i get you dont want to blcok on solving that problem | 17:15 |
*** sayantan_ has joined #openstack-kolla | 17:15 | |
sdake_ | are you registered on dockerhub | 17:15 |
*** sayantan_ has quit IRC | 17:15 | |
sbezverk | sdake_: yes | 17:15 |
sdake_ | let me see if i can find ya | 17:15 |
openstackgerrit | Paul Bourke (pbourke) proposed openstack/kolla: Install magnum dashboard into the horizon image https://review.openstack.org/405350 | 17:16 |
*** sayantan_ has joined #openstack-kolla | 17:17 | |
*** sayantan_ has quit IRC | 17:18 | |
kfox1111 | we need a periodic gate job that builds docker images and caches them for the other gate jobs. | 17:18 |
*** sayantan_ has joined #openstack-kolla | 17:18 | |
kfox1111 | we don't want to break kolla-kube's gates just becase something slips through in kolla's containers. | 17:19 |
kfox1111 | the non direct coupling there would allow us to contianue to develop on kolla-kube while kolla gets fixed. | 17:19 |
sdake_ | kfox1111 your already registered dude :) | 17:19 |
sbezverk | kfox1111: I think inc0 is working on it | 17:19 |
sdake_ | kfox1111 that is what cross gating solves | 17:20 |
kfox1111 | I also have been testing with 2.0.x containers, and would like to continue to do so. will give us some good infrastructure to test upgrades. | 17:20 |
sbezverk | kfox1111: as a part of internal registry idea | 17:20 |
sdake_ | kfox1111 you asked me to register you once before so you could push - i did | 17:20 |
kfox1111 | sdake_: ah. I couldn't remember if I had access or not. and I wanted to double check if it was ok to push entrypoint before I tried. | 17:20 |
sdake_ | pushing a 2.0.2 version of entrypoint is wierd | 17:21 |
sbezverk | sdake_ can you add me then, since kfox1111 memory sometimes suffers ;-) | 17:21 |
kfox1111 | we won't need a 2.0.2 version. | 17:21 |
kfox1111 | just mitaka containers for the actual openstack stuff. | 17:22 |
sdake_ | so current process is we build from tags | 17:22 |
sdake_ | thats what the core team sort of does | 17:22 |
openstackgerrit | prameswar proposed openstack/kolla: QSG and bifrost improved https://review.openstack.org/406177 | 17:22 |
sdake_ | so when 2.0.1 is git tagged a build is done | 17:22 |
sdake_ | or 3.0.1 | 17:23 |
sdake_ | or whatever | 17:23 |
sdake_ | if you want to really kick the idea around automating the dockerhub wfm - recommend ml discussion on the topic | 17:23 |
*** Bico_Fino has joined #openstack-kolla | 17:23 | |
sdake_ | see who is up for actually doing the job | 17:23 |
kfox1111 | sdake_: yeah. will do at some point soon. wana get a little bit more through the helm stuff first. | 17:24 |
kfox1111 | I want to do the same for the helm packages too. | 17:24 |
sdake_ | ya it doesn'treally seem like a priority to me :) | 17:24 |
Bico_Fino | Hello, I just did a redeploy of Kolla(removed all containers and images). | 17:24 |
kfox1111 | prebuild all the helm packages into a helm repo so that someone can just download helm, point it at the kolla repo, and go. | 17:24 |
kfox1111 | nothing kolla specific to build/download. | 17:24 |
Bico_Fino | But when I logged again on Horizon interface, all stuff was there(all flavors, network, etc). | 17:25 |
Bico_Fino | Wasn’t suppose to be a new installation since I deleted all containers/images, or I’m missing something inside /etc/kolla? | 17:25 |
kfox1111 | sdake_: the way I look at it, I'm either going to have to come up with the ci infrastructure at my own site to keep the containers up to date, | 17:26 |
prameswar | pbourke, done :) can you please review again ? | 17:26 |
kfox1111 | or contribute it back so that everyone benifits. | 17:26 |
prameswar | https://review.openstack.org/406177 | 17:26 |
kfox1111 | so I kind of prefer to contribute it. :) | 17:26 |
sdake_ | kfox1111 yup - its interesting long term - i think the issue is there is nobody to do the work of maintaining it :) | 17:27 |
sdake_ | but not really my problem to solve ;) | 17:27 |
sdake_ | the gate being blocked- my problem to solve | 17:27 |
*** jascott1 has joined #openstack-kolla | 17:27 | |
sdake_ | so what do you guys need there | 17:27 |
sdake_ | sbezverk i added you | 17:27 |
sdake_ | just permissions in dockerhub? | 17:28 |
sbezverk | sdake_: thank you.. | 17:28 |
sdake_ | i'd prefer we focus our efforts on having a generallly better ci | 17:29 |
sbezverk | kfox1111: you need to push ubuntu-binary-haproxy to the hub | 17:29 |
sdake_ | ci of kolla kolla-ansible is suboptimal | 17:29 |
kfox1111 | yeah. | 17:29 |
sdake_ | why does that need to be pushed | 17:29 |
kfox1111 | sbezverk: there should be one | 17:29 |
sbezverk | kfox1111: it did not find it | 17:29 |
sbezverk | http://paste.openstack.org/show/591269/ | 17:30 |
sdake_ | deivey was looking at doing a better job maintaining hte dockerhub securitywise | 17:30 |
sdake_ | kfox1111 might talk to him when he is about | 17:30 |
sbezverk | needs tag 3.0.1 | 17:31 |
kfox1111 | https://hub.docker.com/r/kolla/ubuntu-binary-haproxy/tags/ | 17:31 |
kfox1111 | yup. someone forgot to build/push that container. :/ | 17:31 |
kfox1111 | sdake_: k. sounds good. | 17:32 |
*** fragatina has quit IRC | 17:32 | |
kfox1111 | sbezverk: I'll revert to 2.0.2 for now. gate previous was using that anyway. | 17:32 |
sbezverk | kfox1111: sure, nothing new I suspect was added between 2.0.2 and 3.0.1 to haproxy | 17:33 |
openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Fix distro in gate https://review.openstack.org/406223 | 17:33 |
*** fragatina has joined #openstack-kolla | 17:34 | |
openstackgerrit | Merged openstack/kolla: Run docker registry in gate https://review.openstack.org/404405 | 17:34 |
*** unicell1 has quit IRC | 17:36 | |
*** inc0 has quit IRC | 17:36 | |
*** inc0 has joined #openstack-kolla | 17:36 | |
kfox1111 | wow.. thats a lot of kolla patches in zuul. | 17:38 |
*** gmmaha has quit IRC | 17:38 | |
kfox1111 | what that about? | 17:38 |
srwilkers_ | :x | 17:39 |
*** fragatina has quit IRC | 17:39 | |
*** msimonin has quit IRC | 17:39 | |
kfox1111 | oh. might have been merge-checks. | 17:39 |
*** jascott1 has quit IRC | 17:39 | |
*** athomas has quit IRC | 17:39 | |
*** jascott1 has joined #openstack-kolla | 17:39 | |
kfox1111 | meeting time. bbiab. | 17:40 |
*** gmmaha has joined #openstack-kolla | 17:40 | |
*** jascott1 has quit IRC | 17:40 | |
*** jascott1 has joined #openstack-kolla | 17:41 | |
*** sdake_ has quit IRC | 17:42 | |
*** unicell has joined #openstack-kolla | 17:43 | |
*** jascott1 has quit IRC | 17:44 | |
*** javcasalc has joined #openstack-kolla | 17:48 | |
*** jistr|biab is now known as jistr | 17:53 | |
*** jascott1 has joined #openstack-kolla | 17:54 | |
*** javcasalc has quit IRC | 17:54 | |
*** jascott1 has quit IRC | 17:54 | |
*** jascott1 has joined #openstack-kolla | 17:55 | |
*** adrian_otto1 has joined #openstack-kolla | 17:56 | |
*** adrian_otto has quit IRC | 17:58 | |
*** javcasalc has joined #openstack-kolla | 18:02 | |
*** fragatina has joined #openstack-kolla | 18:04 | |
*** sdake has joined #openstack-kolla | 18:04 | |
*** unicell has quit IRC | 18:07 | |
*** david-lyle_ is now known as david-lyle | 18:08 | |
*** Pavo has quit IRC | 18:13 | |
*** shardy has quit IRC | 18:13 | |
*** javcasalc has quit IRC | 18:14 | |
*** matrohon has joined #openstack-kolla | 18:15 | |
*** Pavo has joined #openstack-kolla | 18:18 | |
*** prameswar has quit IRC | 18:20 | |
*** senk has joined #openstack-kolla | 18:20 | |
jascott1 | kfox1111 what sup w this version "3.0.0-1"? | 18:21 |
*** Bico_Fino has quit IRC | 18:21 | |
kfox1111 | in the chart? | 18:22 |
*** inc0 has quit IRC | 18:22 | |
jascott1 | ooooh | 18:22 |
jascott1 | just figured it out | 18:22 |
jascott1 | it was in the helm install wasnt thinking clearly.. must be Friday | 18:23 |
kfox1111 | gonna have to work through that stuff at some point in more detail. but trying to mirror best practices by other package managers. have a revision number on the package. | 18:23 |
jascott1 | sounds legit | 18:23 |
*** magicboiz has joined #openstack-kolla | 18:23 | |
kfox1111 | no worries. | 18:23 |
*** Pavo_ has joined #openstack-kolla | 18:26 | |
Pavo_ | evening everyone | 18:26 |
Pavo_ | question | 18:26 |
Pavo_ | how do you guys enable the magnum horizon plugin in the current kolla-ansible? | 18:27 |
*** fragatina has quit IRC | 18:27 | |
*** fragatina has joined #openstack-kolla | 18:27 | |
sbezverk | kfox1111: gate fix failed again on ubuntu tool box, I just pushed 3.0.1 | 18:28 |
openstackgerrit | Merged openstack/kolla: Don't process/queue children not matched https://review.openstack.org/400467 | 18:29 |
kfox1111 | sbezverk: there wasn't a 3.0.1 toolbox either? | 18:30 |
kfox1111 | weird. | 18:30 |
kfox1111 | k | 18:30 |
sbezverk | nope :-( | 18:31 |
kfox1111 | been working some other things this morning, so haven't been paying too close attention yet. | 18:31 |
kfox1111 | thatnks for looking after it. | 18:31 |
sbezverk | no worries just do check exp again, if I see something missing I will rebuilt it and push | 18:31 |
kfox1111 | k. cool. thx. | 18:32 |
sbezverk | kfox1111: gate is not stable today :-( sometime it works sometimes it doesn't for vswitchd so I am just playing pin-pong with the gate | 18:32 |
sdake | portdirect_away - NICE JOB ON DOCS | 18:36 |
*** Bico_Fino has joined #openstack-kolla | 18:39 | |
srwilkers_ | sdake, +1 | 18:41 |
sdake | srwilkers_ ya its super impressive | 18:42 |
sdake | needs to be tested ithink | 18:42 |
sdake | i'm not sure whwere the status of the docs testing is :) | 18:42 |
*** senk has quit IRC | 18:42 | |
*** unicell has joined #openstack-kolla | 18:42 | |
sdake | i have n early dev env, based upon portdirect's fork, i'll run through the instructions and validate em | 18:43 |
srwilkers_ | definitey. can confirm its working though -- using it currently | 18:43 |
srwilkers_ | but im biased, so dont take my word for it | 18:43 |
*** jheroux has quit IRC | 18:44 | |
sdake | srwilkers_ cool | 18:44 |
sdake | nah that sounds good | 18:44 |
sdake | i think i want to get on the standard dev env | 18:44 |
sdake | rather then a fork of an earlier peice of work | 18:45 |
*** eaguilar has joined #openstack-kolla | 18:45 | |
sdake | so dont need to validate it if you have | 18:45 |
openstackgerrit | Justin Scott proposed openstack/kolla-kubernetes: WIP Helm-ize memcached https://review.openstack.org/405677 | 18:45 |
*** jascott1_ has joined #openstack-kolla | 18:45 | |
*** jascott1 has quit IRC | 18:46 | |
srwilkers_ | sdake, question. im sure its been brought up before, but i cant recall. have you looked at the helm work in our repo? | 18:46 |
sdake | i have | 18:46 |
sdake | i have also looked at others | 18:47 |
kfox1111 | sbezverk: yeah, I htink the docker hubs been a litlte unstable the last few days. :/ | 18:50 |
*** diogogmt has joined #openstack-kolla | 18:50 | |
kfox1111 | sbezverk: 2016-12-02 18:34:12.629052 | Tag 3.0.1 not found in repository docker.io/kolla/ubuntu-binary-openvswitch-db-server | 18:51 |
sdake | yup | 18:51 |
kfox1111 | did any of the 3.0 stuff for ubuntu get posted? | 18:51 |
sdake | shouldn't be used for gating :) | 18:51 |
sdake | anything unreliable ideally shouldn't be used for gating - because it blocks gating | 18:52 |
*** hvlad has quit IRC | 18:52 | |
kfox1111 | sbezverk: all the gates are improper there. | 18:52 |
kfox1111 | if we switch to building containers, | 18:52 |
kfox1111 | then we just push the external pulling to rpms/debs instead. | 18:53 |
sdake | those are also unreliable ;) | 18:53 |
kfox1111 | I'd rather pull a known good set of containers, rather then a unknown set of rpms. | 18:53 |
kfox1111 | for testing against. | 18:53 |
sdake | rpms are gated in rdo | 18:53 |
kfox1111 | sdake: yeah, but we don't just build against pure rdo rpms. | 18:54 |
sdake | not sure about ubuntu | 18:54 |
kfox1111 | its all the other crap too. | 18:54 |
kfox1111 | openssl, glibc, etc | 18:54 |
sdake | yup there is some pulls from non repos | 18:54 |
sdake | rdo is gated, openssl is gated, glibc is gated | 18:54 |
sdake | we have some things in our build from repo pulls which are not gated | 18:55 |
sdake | not sure how to fix that | 18:55 |
sdake | it needs to be fixed | 18:55 |
sdake | we do things like curl -L etc | 18:55 |
kfox1111 | yeah. but not gated by integration test. | 18:55 |
sdake | thats gotta go | 18:55 |
sdake | rdo is totally gated by an integration testing | 18:55 |
sdake | dmsimard|pto does that work | 18:55 |
sdake | upstream of us | 18:55 |
*** schwicht has joined #openstack-kolla | 18:56 | |
sdake | thats the integration testing | 18:56 |
kfox1111 | we fix it by having a job build the containers from packages periodically. run against kolla-ansible trunk gates and kolla-kubernetes trunk gates. | 18:56 |
kfox1111 | if all tests pass, those containers are cached as the newest tested set. | 18:56 |
sdake | cross repo gating is better | 18:56 |
kfox1111 | then the kolla-ansible/ kolla-kubernetes review gates use those cached containers. they are konwn good. | 18:56 |
kfox1111 | we cant cross gate against glibc/rdo/mariadb/etc. | 18:56 |
kfox1111 | thats what the periodic job though trys to acomplish. at least do sanity checks and prevent breakage from leaking into unrelated places. | 18:57 |
sdake | i understand what your saying with caching | 18:57 |
sbezverk | kfox1111: can you check what is wrong with gate? when I delploy vswitchd in my test bed it works fine, but in the gate it shows bunch of containers, not related to vswitchd getting stuck in Init state | 18:57 |
kfox1111 | have the link handy? | 18:58 |
*** eaguilar_ has joined #openstack-kolla | 18:58 | |
sdake | its a hard problem to solve | 18:58 |
sbezverk | http://logs.openstack.org/07/405107/6/experimental/gate-kolla-kubernetes-deploy-centos-binary-ceph-multi-nv/ff2f144/ | 18:58 |
kfox1111 | sdake: agreed. | 18:58 |
sbezverk | kfox1111: funny thing that it deployed fine once.. | 18:59 |
sdake | cross repo gating makes it fairly easy - since rdo is pretty well gated, and kolla containers will be gated on kolla-ansible ccross-repo gates | 18:59 |
sdake | we build a web of gating at the top | 18:59 |
sdake | and depend on gating below | 18:59 |
*** eaguilar has quit IRC | 18:59 | |
sdake | or depend on our upstream gating is a better way of putting it | 19:00 |
sdake | i wouldn't totally get wrapped around the axle on gating | 19:01 |
sdake | its tken kolla 2 years to even get a reasonably functional gate, and it is suboptimal | 19:01 |
*** eaguilar_ has quit IRC | 19:01 | |
kfox1111 | sdake: yeah. we should cross repo gate kolla changes against kolla-ansible and kolla-kubernetes. | 19:01 |
sdake | and by reasonably functional, I mean reliable | 19:01 |
sdake | kfox1111 jeffrey4l is working on the kolla-ansible part | 19:01 |
kfox1111 | I don't think it would be too hard from the kolla-kubernetes side. | 19:01 |
sdake | waiting to see wht jefrey4l comes up with before i crosss repo gate kolla-kubernetes | 19:02 |
sdake | its helpful if all our gates are similar in design | 19:02 |
kfox1111 | would just be checking out kolla-kubernetes trunk, then running setup_gate.sh with the right args I think. | 19:02 |
*** Bico_Fino has quit IRC | 19:03 | |
sdake | ya jeffrey4l last i saw was jammed up about how to get the correct version of kolla-ansible (master) into the kolla gate to do the job | 19:03 |
sdake | there is a tool zuul-cloner which doess this job perfectly | 19:03 |
sdake | i dont know how to use it unfortunately :( | 19:03 |
kfox1111 | sdake: yeah. see the kolla-kubernetes gate. it uses it. | 19:03 |
sdake | yup i suggested jeffrey4l reach out to you | 19:04 |
kfox1111 | dug it out of the infra folks. :) | 19:04 |
*** adrian_otto1 has quit IRC | 19:04 | |
kfox1111 | sbezverk: oh. one thing to consider. | 19:05 |
sdake | if you come up with some new gating technique we can roll with that :) | 19:05 |
*** eaguilar has joined #openstack-kolla | 19:05 | |
sdake | inc0 also working on it | 19:05 |
sdake | i'd ask that poeple work together on the gating so its consistent :) | 19:05 |
sdake | thats hrad to sort out | 19:05 |
kfox1111 | sbezverk: in the before helm time, all the config was done in setup_config.sh, then the kollakube res create pod ... stuff used those presetup vars. | 19:05 |
kfox1111 | sbezverk: with helm though, the vars have to be passed at helm launch time. | 19:05 |
*** sayantan_ has quit IRC | 19:05 | |
*** adrian_otto has joined #openstack-kolla | 19:05 | |
kfox1111 | so I'm guessing for the openvswitch one, some env stuff needs overridden. | 19:06 |
kfox1111 | sdake: +1 | 19:06 |
sbezverk | kfox1111: not sure becase 1: other pods. not directly related also getting stuck on Init, 2: vswitchd only comput, gets stuck the "network" mode works fine.. | 19:07 |
openstackgerrit | Justin Scott proposed openstack/kolla-kubernetes: WIP Helm-ize memcached https://review.openstack.org/405677 | 19:08 |
*** matrohon has quit IRC | 19:09 | |
sdake | i always had a dream of making the gate for kolla-ansible for example super robust | 19:10 |
sdake | that never happened | 19:10 |
sdake | and also do all integration testing necessary | 19:11 |
sdake | inc0 is on it, maybe he will have more success :) | 19:11 |
*** mliima has quit IRC | 19:11 | |
sdake | pbourke and jeffrey4l and now inc0 and i and you know how to work on the gate | 19:12 |
sdake | not sure nayone else does | 19:12 |
sdake | coolsvap possibly | 19:12 |
kfox1111 | sdake: yeah. I think having a gate for our site is going to be important, | 19:12 |
kfox1111 | and having it upstream rather then in our infra makes sense to me, so I'm ok trying to push as much as I can into making the kolla gate test useful things. | 19:12 |
sdake | nice | 19:12 |
sdake | its onerous expecting all the core reviewerst o understand exactly how the gate works | 19:13 |
sdake | or to make that a requirement for cr | 19:13 |
kfox1111 | semi agree. | 19:14 |
sbezverk | kfox1111: do you collect kubelet's logs from the gate? | 19:14 |
kfox1111 | at first, I agree. | 19:14 |
kfox1111 | long term, lots of folks should understand it. | 19:14 |
sdake | you remember the pain you went through to learn the gate | 19:14 |
sdake | if thats a requirement for cr, we wouldn't have any crs ;) | 19:14 |
kfox1111 | sdake: yeah. and the hit by a bus scenerio's real. | 19:14 |
srwilkers_ | :) | 19:14 |
kfox1111 | it literally happened in our group. :/ | 19:14 |
sdake | yup agree | 19:15 |
srwilkers_ | ive been looking at them a bit each night | 19:15 |
sdake | airbus does happen | 19:15 |
srwilkers_ | has actually helped quite a bit | 19:15 |
kfox1111 | drunk driver and one of our team members was a bicleist. not a good mix. :( | 19:15 |
sdake | bummer | 19:15 |
srwilkers_ | damn kfox1111, that sucks | 19:15 |
sdake | that happened to one of robyn's good friends | 19:15 |
sdake | who maintained yum | 19:15 |
sdake | same shit | 19:15 |
Pavo_ | how do you guys enable the magnum horizon plugin in the current kolla-ansible? | 19:15 |
sdake | drunk driver, bicyclist, etc | 19:15 |
kfox1111 | yeah. | 19:15 |
sdake | uber is a thing and its cheap | 19:15 |
kfox1111 | worst thing about it was it was a repeat offender. :/ | 19:16 |
sdake | same shit repeat offender | 19:16 |
kfox1111 | maddening. :/ | 19:16 |
sdake | i dont rent cars on the road | 19:16 |
sdake | why bother - uber gets me around | 19:17 |
srwilkers_ | yep | 19:17 |
sdake | at home i dont drink :) | 19:17 |
sdake | although i take uber if i do at dinner or something | 19:17 |
sdake | which is rare | 19:17 |
sdake | Pavo_ pbourke is your fella there | 19:18 |
sdake | Pavo_ i believe its documented but i dont recall | 19:18 |
Pavo_ | ok thanks sdake | 19:18 |
sdake | a similar question came up yesterday | 19:19 |
sdake | or the day before | 19:19 |
sdake | i dont recall teh answer | 19:19 |
kfox1111 | sbezverk: have alook at tests/conf/ceph-all-in-one/kolla_kubernetes_config. | 19:19 |
sdake | its in the logs (click the logs in the topic) | 19:19 |
kfox1111 | looks like the settings are a bit different. | 19:19 |
kfox1111 | ovs_add_port is false | 19:19 |
*** msimonin has joined #openstack-kolla | 19:19 | |
kfox1111 | that could easily cause some bad things. | 19:20 |
kfox1111 | did on my boxes. | 19:20 |
sbezverk | kfox1111: I see your point, but it was always like this | 19:22 |
sbezverk | kfox1111: git blame shows, that no was original value | 19:22 |
*** kbyrne has quit IRC | 19:23 | |
kfox1111 | the override's been in the gate since its begining. | 19:23 |
kfox1111 | that particular setting is very dangerious when true, | 19:23 |
*** portdirect_away has quit IRC | 19:23 | |
kfox1111 | as any slight misconfiguration will pull the whole node off the network. | 19:23 |
kfox1111 | happened to me multiple times until I figured out what was going on. :/ | 19:24 |
kfox1111 | so, for helm, I might even say we should consider defaulting it to false. just to prevent new users from shooting themselves accidently. | 19:24 |
sbezverk | kfox1111: according to vswitchd pod log, it was going to add port | 19:25 |
sbezverk | check line 20: http://logs.openstack.org/07/405107/6/experimental/gate-kolla-kubernetes-deploy-centos-binary-ceph-multi-nv/ff2f144/logs/pods/kolla-openvswitch-vswitchd-compute-g2mdw.txt | 19:26 |
kfox1111 | yeah. and its probably trying to add the wrong one. | 19:26 |
*** portdirect_away has joined #openstack-kolla | 19:26 | |
sbezverk | do you mean wrong interface? | 19:26 |
kfox1111 | but the gate's designed to add the port elsewhere in the gate. so can safely be skipped in the pod. | 19:26 |
*** kbyrne has joined #openstack-kolla | 19:26 | |
kfox1111 | yeah. wrong interface. | 19:26 |
*** portdirect_away is now known as portdirect | 19:26 | |
*** adrian_otto has quit IRC | 19:27 | |
*** ccesario has quit IRC | 19:27 | |
srwilkers_ | hey portdirect, hows it going? | 19:27 |
sbezverk | kfox1111: I am still not convinced, because the actual pod is not even running | 19:27 |
portdirect | kfoxx1111/srwilkers_: dont think either of you are crazy (yet) :) | 19:27 |
srwilkers_ | :) | 19:27 |
*** msimonin has quit IRC | 19:27 | |
kfox1111 | portdirect: hehe. then you haven't seen just how crazy I can be yet. ;) | 19:28 |
sbezverk | I mean even Init pod, so it does not get to the point of adding anything, to me it looks kube problem.. kubelet has no logs for creating the pod at all | 19:28 |
*** inc0 has joined #openstack-kolla | 19:29 | |
*** Pavo_ has quit IRC | 19:29 | |
inc0 | I'm back | 19:30 |
portdirect | i thought as much kfox1111, though we should prob do this in the comments on gerrit - just wanted to run a few things past you as it would be quicker here. | 19:30 |
kfox1111 | sbezverk: there could be some other issues too. haven't had time to dig too deeply into it yet. but I bet thats at least one of the problems. | 19:31 |
kfox1111 | portdirect: sounds good | 19:31 |
portdirect | re getting setup_dev.sh in a container - sounds like a great idea - I was playing around with it as well, but thought it might be too extreme as well | 19:31 |
portdirect | (too many as wells..) | 19:31 |
jascott1_ | well well well | 19:31 |
jascott1_ | portdirect so you started setup_dev? | 19:32 |
kfox1111 | portdirect: so, https://github.com/kfox1111/kolla-kubernetes-toolbox | 19:32 |
kfox1111 | made that for the minikube docs origionally. its a bit bitrotten by now, but could be a good starting place. | 19:33 |
portdirect | great - I wondered where that came from | 19:33 |
kfox1111 | was planning on contributing it to the kolla repo, but other things kept poping up. | 19:33 |
*** serverascode has quit IRC | 19:34 | |
*** NobodyCam has quit IRC | 19:34 | |
*** sayantani01_ has quit IRC | 19:34 | |
*** NachoDuck has quit IRC | 19:34 | |
*** berendt has quit IRC | 19:34 | |
portdirect | I think it would make sense in the tools dir of kolla-k8s? | 19:34 |
portdirect | my plan was to take sdakes idea of a devstack type env - based on the gate (and kolla-toolbox) | 19:35 |
sdake | portdirect dude nice job on the docs | 19:35 |
sdake | portdirect dev env setup looks fairly straightforward now | 19:35 |
sdake | ;) | 19:35 |
sdake | now i wont have to walk a million people through it :) | 19:35 |
portdirect | if its good enough to acheive that then it's ok :) | 19:36 |
portdirect | thanks dude | 19:36 |
sbezverk | kfox1111: I will change in all_values to not plug inteface | 19:36 |
sbezverk | and run test again | 19:36 |
portdirect | jascott1_: https://review.openstack.org/#/c/405720/ | 19:37 |
kfox1111 | sbezverk: ok. cool. | 19:37 |
openstackgerrit | Serguei Bezverkhi proposed openstack/kolla-kubernetes: Hel-ming openvswitch-vswitchd https://review.openstack.org/405107 | 19:37 |
kfox1111 | sbezverk: if it still fails, I'll dig in more then. | 19:37 |
*** sayantan_ has joined #openstack-kolla | 19:37 | |
kfox1111 | portdirect: I think we've been putting all the containers in the kolla repo so far. | 19:37 |
*** sayantan_ has quit IRC | 19:37 | |
*** sayantan_ has joined #openstack-kolla | 19:38 | |
sdake | brb need to run an errand | 19:38 |
portdirect | kfox1111: yeah - though I was wondering if this was a 'special' case where we shouldn't actually build it, so that bitrot never becomes an issue - as it would just be a developer tool? | 19:38 |
*** NachoDuck has joined #openstack-kolla | 19:40 | |
kfox1111 | portdirect: I think we need to keep it up to date, and come up with infra that helps with that. | 19:41 |
*** inc0 has quit IRC | 19:41 | |
kfox1111 | we mwant to make it as easy as possible to spin up new devs. | 19:41 |
kfox1111 | good docs, and easy to setup build envs really go a long way therre. | 19:42 |
kfox1111 | thats what I was atempting to do with the minikube stuff origionally. | 19:42 |
kfox1111 | then got sidetracked as I was spending too much time testing rather then having a gate do the testing. so had to stop and work on that. | 19:42 |
*** magicboiz has quit IRC | 19:43 | |
portdirect | yeah - your minikube stuff was great | 19:43 |
sbezverk | kfox1111: I think you might be right. since first vswitchd always runs successfully, it plugs eth1, which could potentially screw up the whole cluster and could explain strange impact on a non-related pods | 19:43 |
kfox1111 | yeah. the openvswitch stuff, and neutron stuff can break things in really weird looking ways. once those are done, it should be muhc easier. :/ | 19:44 |
portdirect | though limited by minikube itself - but that ux was pretty close to what id like us to have - the only real difference being that I'd like to keep (at least as an option) the kolla-kube code outside of the container and bind mounted in - so that a developer gets perisitance of that | 19:45 |
kfox1111 | portdirect: thanks. was hoping it would be useful. | 19:45 |
kfox1111 | portdirect: +1 for a bind mount option. would like to bundle it too though. | 19:46 |
kfox1111 | so if all you want to do is kick the tires, it just works. if you want to use it for further dev, then bindmounting works well. | 19:46 |
sbezverk | kfox1111: need to chat with you about pv/pvc/svc/job charts since in rabbitmq I have all these. but it can wait for later if you are busy with other things | 19:46 |
kfox1111 | sbezverk: yeah. k. sounds good. | 19:47 |
portdirect | kfox1111: sounds good - right I'm gonna head home and then have a crack at that after dinner, will you be about in a couple of hours? (may need some assistance with multinode ceph) | 19:48 |
kfox1111 | portdirect: yeah. should be here. | 19:49 |
*** adrian_otto has joined #openstack-kolla | 19:49 | |
*** unicell has quit IRC | 19:49 | |
*** serverascode has joined #openstack-kolla | 19:50 | |
portdirect | kfox1111: nice, laters :) | 19:50 |
*** portdirect has quit IRC | 19:50 | |
*** adrian_otto has quit IRC | 19:54 | |
*** jrich523 has quit IRC | 19:56 | |
*** sdake has quit IRC | 20:04 | |
*** sdake has joined #openstack-kolla | 20:05 | |
*** rstarmer has joined #openstack-kolla | 20:08 | |
rstarmer | hello | 20:08 |
*** adrian_otto has joined #openstack-kolla | 20:09 | |
rstarmer | I'm trying to get LBaaS and VPNaaS functions running in an ubuntu-source based deployment in 3.0.0 | 20:09 |
rstarmer | following some suggestions from last night, I've been looking at the plugin customization process, and it appears that I _should_ be able to just add plugin specific blocks to /etc/kolla/kolla-build.conf | 20:10 |
*** jrich523 has joined #openstack-kolla | 20:10 | |
rstarmer | one for [neutron-server-plugin-lbaas] and one for [neutron-server-plugin-vpnaas] | 20:11 |
*** ipsecguy_ has joined #openstack-kolla | 20:12 | |
*** adrian_otto has quit IRC | 20:12 | |
rstarmer | but this doesn't seem to work. When I try to re-build the neutron containers, I get 'ERROR:kolla.image.build:Unhandled error when running BuildTask(neutron-server)' | 20:12 |
rstarmer | source['reference'].replace('/', '-')) | 20:12 |
rstarmer | AttributeError: 'NoneType' object has no attribute 'replace' | 20:12 |
*** adrian_otto has joined #openstack-kolla | 20:13 | |
*** Pavo has quit IRC | 20:13 | |
*** msimonin has joined #openstack-kolla | 20:13 | |
*** ipsecguy has quit IRC | 20:15 | |
*** NobodyCam has joined #openstack-kolla | 20:16 | |
*** Pavo has joined #openstack-kolla | 20:18 | |
*** msimonin has quit IRC | 20:20 | |
*** msimonin has joined #openstack-kolla | 20:24 | |
sbezverk | kfox1111: it got better but still some other failure on glance even before any openvswitch pods get started, please check.https://review.openstack.org/#/c/405107/ | 20:25 |
*** berendt has joined #openstack-kolla | 20:27 | |
*** sayantani01_ has joined #openstack-kolla | 20:28 | |
sdake | yo | 20:29 |
sdake | back | 20:29 |
sdake | rstarmer Pavo can you share notes on the plugins ? :) | 20:29 |
sdake | rstarmer i see you have an error | 20:30 |
sdake | pavo was looking for a plugin for magnum | 20:30 |
sdake | how to make it work that is | 20:30 |
sdake | rstarmer do you have a line number on that error? | 20:31 |
sdake | rather backtrace | 20:31 |
rstarmer | ERROR:kolla.image.build:Unhandled error when running BuildTask(neutron-server) | 20:31 |
rstarmer | Traceback (most recent call last): | 20:31 |
rstarmer | File "/usr/local/lib/python2.7/dist-packages/kolla/image/build.py", line 485, in run | 20:31 |
rstarmer | task.run() | 20:31 |
rstarmer | File "/usr/local/lib/python2.7/dist-packages/kolla/image/build.py", line 271, in run | 20:31 |
rstarmer | self.builder(self.image) | 20:31 |
rstarmer | File "/usr/local/lib/python2.7/dist-packages/kolla/image/build.py", line 396, in builder | 20:31 |
rstarmer | archive_path = self.process_source(image, plugin) | 20:31 |
rstarmer | File "/usr/local/lib/python2.7/dist-packages/kolla/image/build.py", line 317, in process_source | 20:31 |
rstarmer | source['reference'].replace('/', '-')) | 20:31 |
rstarmer | AttributeError: 'NoneType' object has no attribute 'replace' | 20:31 |
sdake | paste service pls in future ;) | 20:31 |
sdake | which version are you using, newton? | 20:32 |
sdake | or mitaka | 20:32 |
rstarmer | oops, sorry | 20:32 |
sdake | its ok | 20:32 |
rstarmer | newton | 20:32 |
sdake | ok give me a min to read the code | 20:32 |
sdake | see if anything pops out | 20:32 |
rstarmer | ok | 20:32 |
sdake | no software should backtrace | 20:32 |
sdake | no matter the input | 20:32 |
rstarmer | I agree, and it'd be nice if it gave you an error in a fashion that was useful for a non-developer, but hey, the tool works better than any other for this kind of thing :) | 20:33 |
sdake | the reality is if the software is backtracing, it is impossible to give a useful error | 20:33 |
sdake | for non-developers to understand | 20:33 |
sdake | hence why no software should backtrace ;) | 20:34 |
sdake | rstarmer while i look at the code, please file a defect | 20:34 |
sdake | i'll mark it criticl | 20:34 |
sdake | rstarmer did you set reference in in kolla-build.conf? | 20:36 |
sdake | can you paste your kolla-build.conf if possible? | 20:36 |
sdake | attributeerror nonetype means source.reference is unset | 20:37 |
sdake | unset | 20:37 |
sdake | nonetype is the default type | 20:37 |
*** unicell has joined #openstack-kolla | 20:38 | |
*** portdirect_away_ has quit IRC | 20:39 | |
rstarmer | yes, I set a reference | 20:39 |
rstarmer | http://pastebin.com/hjWSCbcV | 20:41 |
*** sayantan_ has quit IRC | 20:44 | |
srwilkers_ | reference is spelled wrong there rstarmer | 20:44 |
srwilkers_ | s/refernce/reference | 20:44 |
rstarmer | ... ugh. | 20:44 |
rstarmer | ok, re-trying. | 20:44 |
*** sayantan_ has joined #openstack-kolla | 20:45 | |
rstarmer | thanks for catching my typo... | 20:45 |
kfox1111 | sbezverk: thats weird. | 20:45 |
srwilkers_ | it happens man :) | 20:45 |
kfox1111 | very rare that something works on multinode but not single node. | 20:45 |
kfox1111 | still feels like thats a misconfiguration to me. | 20:45 |
kfox1111 | lets see... | 20:45 |
kfox1111 | oh... even weirder. | 20:46 |
kfox1111 | the external ovs one failed too. | 20:46 |
rstarmer | sdake: yep, that fixed it! | 20:46 |
rstarmer | now to see if I can actually get the service to work :) | 20:47 |
srwilkers_ | rstarmer, awesome. glad that took care of it. cheers | 20:47 |
sdake | rstarmer good to hear | 21:03 |
sdake | srwilkers_ typos for me = type fast with lots of mistakes! | 21:03 |
vhosakot | he is no ordinary rstarmer... he is "QOS QOS baby" rstarmer ;) | 21:03 |
sdake | code needs better error checking | 21:04 |
sdake | all code that is ;-) | 21:04 |
rstarmer | vhosakot: OMG, now that's going to follow me around? :) I just wish I'd gotten the music right to start, my backup dancers were primed and ready :) | 21:04 |
vhosakot | rstarmer: the talk was great :) | 21:05 |
rstarmer | better error responses, like "can't find 'reference' parameter in kolla-build.conf" would be much more useful than 'AttributeError: 'NoneType' object has no attribute 'replace'' | 21:07 |
rstarmer | vhosakot: thanks :) | 21:08 |
*** fguillot has quit IRC | 21:13 | |
kfox1111 | sbezverk: I just poked the gate fix review again just to see if there was gate funnyness. it is passing like always. so something is funny about your review. just having a hard time finding it. it seems like its an actual breakage though. or else super bad luck. | 21:18 |
*** eaguilar has quit IRC | 21:18 | |
kfox1111 | next thing to test I think is 2.0.2 | 21:18 |
*** jascott1_ has quit IRC | 21:19 | |
*** jascott1 has joined #openstack-kolla | 21:20 | |
*** jascott1 has quit IRC | 21:20 | |
openstackgerrit | sayantani proposed openstack/kolla: Add new guide for single-node deployment https://review.openstack.org/404993 | 21:21 |
*** jascott1 has joined #openstack-kolla | 21:21 | |
*** portdirect_away_ has joined #openstack-kolla | 21:23 | |
*** portdirect_away_ is now known as portdirect | 21:23 | |
*** jascott1 has quit IRC | 21:25 | |
rstarmer | well, I'm not out of the woods yet... Now the neutron bootstrap container fails to start. | 21:30 |
*** turvey has joined #openstack-kolla | 21:30 | |
*** eaguilar has joined #openstack-kolla | 21:33 | |
rstarmer | http://pastebin.com/fNTizEKt | 21:33 |
openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Fix distro in gate https://review.openstack.org/406223 | 21:34 |
*** sayantan_ has quit IRC | 21:39 | |
*** eaguilar has quit IRC | 21:39 | |
*** jascott1 has joined #openstack-kolla | 21:40 | |
*** sayantan_ has joined #openstack-kolla | 21:40 | |
*** sdake has quit IRC | 21:43 | |
*** absubram has joined #openstack-kolla | 21:46 | |
*** sayantan_ has quit IRC | 21:47 | |
*** sayantan_ has joined #openstack-kolla | 21:51 | |
*** schwicht has quit IRC | 22:00 | |
*** jtriley has quit IRC | 22:04 | |
*** sdake has joined #openstack-kolla | 22:04 | |
*** sdake_ has joined #openstack-kolla | 22:06 | |
*** sdake__ has joined #openstack-kolla | 22:07 | |
*** sdake has quit IRC | 22:07 | |
*** schwicht has joined #openstack-kolla | 22:09 | |
*** mgiles has quit IRC | 22:09 | |
*** sdake_ has quit IRC | 22:10 | |
*** sdake has joined #openstack-kolla | 22:10 | |
*** sdake has quit IRC | 22:10 | |
*** sdake has joined #openstack-kolla | 22:10 | |
*** msimonin has quit IRC | 22:11 | |
*** sdake has quit IRC | 22:11 | |
kfox1111 | sbezverk: https://review.openstack.org/#/c/406223/ should be ready | 22:12 |
*** Pavo has quit IRC | 22:13 | |
openstackgerrit | Serguei Bezverkhi proposed openstack/kolla-kubernetes: Hel-ming openvswitch-vswitchd https://review.openstack.org/405107 | 22:14 |
*** sdake__ has quit IRC | 22:14 | |
sbezverk | kfox1111: done, for vswitch I changed tag for now and extra && | 22:16 |
*** sayantan_ has quit IRC | 22:16 | |
kfox1111 | k. am curious if that'll fix it. | 22:16 |
*** sayantan_ has joined #openstack-kolla | 22:16 | |
openstackgerrit | Merged openstack/kolla-kubernetes: Fix distro in gate https://review.openstack.org/406223 | 22:17 |
*** Pavo has joined #openstack-kolla | 22:17 | |
*** sayantan_ has quit IRC | 22:25 | |
*** lrensing has quit IRC | 22:29 | |
*** inc0 has joined #openstack-kolla | 22:31 | |
inc0 | howdy | 22:32 |
*** schwicht has quit IRC | 22:32 | |
wirehead_ | meow. | 22:33 |
kfox1111 | hi | 22:33 |
rhallisey | meow meow | 22:33 |
wirehead_ | I haven’t touched Barcelona photos yet, inc0. | 22:33 |
jascott1 | whatup inc0 ready for the big trip? | 22:33 |
inc0 | almost | 22:33 |
wirehead_ | inc0: You do know that this is the town you are moving to, right? http://www.kgw.com/news/local/portland-womans-stolen-car-returned-with-note-gas-money-and-a-crazy-story/342692897 | 22:34 |
inc0 | yeah I've seen that one | 22:34 |
wirehead_ | OK. | 22:34 |
inc0 | someone makes a mistake and tries to be nice, it makes news | 22:34 |
*** inc0 has quit IRC | 22:36 | |
*** inc0 has joined #openstack-kolla | 22:37 | |
wirehead_ | Well, there’s a category of news that’s “Portland stories" | 22:39 |
wirehead_ | It’s either less capitalistic or less comunistic than “San Francisco stories" | 22:40 |
wirehead_ | And very different from “Florida stories" | 22:40 |
SamYaple | whats teh status of k8s kolla? | 22:42 |
SamYaple | neutron working? instance launched? | 22:42 |
kfox1111 | SamYaple: we can launch a whole compute kit with it. | 22:42 |
SamYaple | kfox1111: neutron as well? | 22:43 |
kfox1111 | including launching vm's, writing to a cinder vollume on one vm, moving it to another, and reading it back, | 22:43 |
kfox1111 | and neutron networkaccess, including security groups and floating ips. | 22:43 |
srwilkers_ | hey inc0 | 22:44 |
SamYaple | kfox1111: how are you solving the network namespace issues with neutron? | 22:44 |
*** vhosakot_ has joined #openstack-kolla | 22:44 | |
kfox1111 | for now? praying that l3agent doesn't die, and leaning on k8s to support shared. | 22:44 |
kfox1111 | :/ | 22:44 |
SamYaple | esh. ok | 22:44 |
SamYaple | might i suggest reverting to thick containers for a bit | 22:45 |
srwilkers_ | oh hey rhallisey, hows it going? | 22:45 |
SamYaple | thats what i had t oswitch kolla-ansible too when shared support was flaking for kolla | 22:45 |
SamYaple | thats how it was originally because i could make thin contaienrs work | 22:45 |
kfox1111 | I'd hand patch k8s and support a temporary fork before doing that. | 22:45 |
SamYaple | no one is going to use your patched k8s :P | 22:45 |
kfox1111 | there's a patch already in flight. | 22:46 |
SamYaple | ok. well once that lands it sounds promising | 22:46 |
kfox1111 | yeah. I don't think its far out. | 22:46 |
*** vhosakot has quit IRC | 22:46 | |
rhallisey | srwilkers, hey! good thanks :) | 22:46 |
SamYaple | are you keeping l3,dhcp,metadata-agent in the same "pod"? | 22:46 |
SamYaple | so they stay on the same host? | 22:46 |
kfox1111 | no. daemonsets. | 22:46 |
SamYaple | i dont know much about k8s | 22:47 |
SamYaple | they stay on the same host though, right? | 22:47 |
kfox1111 | a daemonset is a object type that lands a single pod on each host that matches the appropriate label. | 22:47 |
SamYaple | hmmm so they might not all land on the same host? | 22:48 |
kfox1111 | if you use the same selector you will get the desired outcome. | 22:48 |
SamYaple | this shouldnt be in the users control, being on the same host is a hard requirement though | 22:49 |
SamYaple | is there no way t oenfore that? | 22:49 |
kfox1111 | for example, label your compute nodes kolla_compute=true, then make your l3 and openvswitch daemsonets target kolla_compute=true. | 22:49 |
SamYaple | doesnt that make it prone to user error? from lack of knowlege or just a config mistake? | 22:50 |
kfox1111 | We're doing things multilevel. base building blocks like openstack does. flexible and unopinionated. | 22:50 |
wirehead_ | Well, consider all of the possibilities for config errors with regular OpenStack. | 22:50 |
kfox1111 | orchestration level can become more opinionanted, and try and save the operator from themselves. | 22:50 |
kfox1111 | (which I plan not to use for a while... that kind of "saving" usually bites me) | 22:51 |
SamYaple | i mean thats not opinionated, it doesnt work wihtout those things in place | 22:51 |
SamYaple | or worse, in the case of neutron, it _kinda_ works | 22:51 |
kfox1111 | openstack's built the same way. you can misconfigure it a million ways. | 22:52 |
kfox1111 | but thats important too, when you do crazy stuf like I do with running vxlans over ip over ib. ;) | 22:52 |
SamYaple | but... lol this is a deployment project. its here to fix that problem | 22:52 |
SamYaple | thats the whole point | 22:52 |
kfox1111 | yes/no. | 22:52 |
kfox1111 | with a layered aproach, both goals can be met. | 22:53 |
SamYaple | ok. well just fyi, youre in a world of troubleshooting pain without enforcing some of this stuff | 22:53 |
SamYaple | it does _NOT_ fail well without having the componets it expects | 22:53 |
kfox1111 | the upper layer provides the ease of deployment/safety net kinds of things. | 22:53 |
kfox1111 | the lower level provides building blocks that you can asemble however needed. | 22:53 |
*** schwicht has joined #openstack-kolla | 22:54 | |
kfox1111 | SamYaple: its really not any different then openstack's compoennts already behave. | 22:54 |
kfox1111 | if you run neutron-openvswitch-agent without openvswitch, you get badness. | 22:54 |
kfox1111 | whether or not your running it in containers or not. | 22:54 |
SamYaple | kfox1111: im aware, but thats why openstack is complicated to install and we have a bunch of deployment projects ANYWAY | 22:54 |
openstackgerrit | Joshua Harlow proposed openstack/kolla: Add ability to output the things built/failed/not build https://review.openstack.org/395273 | 22:54 |
SamYaple | this isnt just another openstack project, its is openstack dpeloyment | 22:55 |
*** dave-mccowan has quit IRC | 22:55 | |
kfox1111 | SamYaple: lots of projects use kolla's containers, but not the deployment tools. | 22:55 |
kfox1111 | the same can be said about kolla-kubernetes going forward. | 22:55 |
kfox1111 | theyu can use the base k8s object building blocks, without the orchestration components. | 22:55 |
kfox1111 | idaelly everyone would collaborate on all parts of the stack. | 22:56 |
kfox1111 | but that seems impossible, at least for now. | 22:56 |
kfox1111 | at least with the layered aproach, folks stil have room to contribute to the parts they care about. | 22:56 |
openstackgerrit | Joshua Harlow proposed openstack/kolla: Add ability to output the things built/failed/not build https://review.openstack.org/395273 | 22:56 |
kfox1111 | and the whole project is better for it. | 22:56 |
SamYaple | kfox1111: its entirely possible i dont understand kolla-k8s. but from what youre describing it sounds like youre focusing it more on the dev side of openstack than the OPS side of openstack | 22:56 |
SamYaple | just like the rest of openstack is focused on dev rather than ops | 22:56 |
kfox1111 | hehe. quite the oposite. :) | 22:56 |
kfox1111 | I'm an op wanting to use this on a production system in the very neer future, | 22:57 |
jascott1 | kfox1111 should helm install chart fail if name is not supplied? | 22:57 |
jascott1 | i would think not but idk | 22:57 |
kfox1111 | and am terrified of an orchestration system "deciding" my fate. | 22:57 |
jascott1 | im talking about our desired implementation | 22:57 |
SamYaple | there are very few people that know that l3,dhcp,metadata must run on the same host, and why | 22:57 |
kfox1111 | jascott1: it won't. it is like docker. it will just make up a random one. | 22:57 |
kfox1111 | SamYaple: yup. and by default, the settings have them co'exist due to shared default labels. | 22:58 |
kfox1111 | if you purposfully change the default, then its on you to debug. :) | 22:58 |
jascott1 | im templating in name as per comments on memcached, with name its memcached-<name> otherwise it would be memcached- | 22:58 |
kfox1111 | oh... hmm.. | 22:58 |
sbezverk | kfox1111: it looks like we hit it again: "2016-12-02 22:37:48.108753 | Error: transport is closing" | 22:58 |
jascott1 | can just do logic to make it memcache (without -) if name wasnt set | 22:59 |
openstackgerrit | Joshua Harlow proposed openstack/kolla: Don't process/queue children not matched https://review.openstack.org/406389 | 22:59 |
kfox1111 | jascott1: I was pretty sure there is a chart var to get taht. | 22:59 |
SamYaple | kfox1111: oh. thats totally ok then. the default is to deploy them together on the same host | 22:59 |
SamYaple | kfox1111: kolla-ansible does the same thing. you can seperate them | 22:59 |
openstackgerrit | Serguei Bezverkhi proposed openstack/kolla-kubernetes: Hel-ming openvswitch-vswitchd https://review.openstack.org/405107 | 22:59 |
SamYaple | kfox1111: its just a bad idea | 22:59 |
SamYaple | kfox1111: if there is a default enforment of this then thats perfectly fine | 22:59 |
jascott1 | kfox1111 so better to default to name of chart than hardcode memcached? | 23:00 |
SamYaple | kfox1111: the way you made it sound was there was nowhere in the project that linked these things into a "working" config | 23:00 |
kfox1111 | jascott1: https://github.com/kubernetes/helm/blob/master/docs/charts.md | 23:00 |
kfox1111 | jascott1: predefined values | 23:00 |
*** srwilkers_ is now known as srwilkers_away | 23:01 | |
jascott1 | so use memcached-<name> and if name was not set, use Charts.name | 23:01 |
SamYaple | so everything should be working for kolla-k8s , except metadata and container restarts for neutron | 23:01 |
kfox1111 | jascott1: I think, just use the value {{ .Values.Release.Name | quote }} | 23:02 |
kfox1111 | should always be set I think. | 23:02 |
kfox1111 | SamYaple: yeah. | 23:02 |
kfox1111 | and that should be fixed pretty quickly I hope. | 23:02 |
SamYaple | kfox1111: multinode/multiple services? | 23:03 |
kfox1111 | SamYaple: works. and gate tested. :) | 23:03 |
jascott1 | i dont see that in the other templates but if thats the way we want to go im fine | 23:03 |
SamYaple | kfox1111: nice! | 23:03 |
SamYaple | wait multinode is gate tested? | 23:03 |
kfox1111 | jascott1: yeah, I think we want to do the others at some point too. but the others largely don't need multiinstance. | 23:03 |
kfox1111 | memcache/rabbit/mariadb do. | 23:03 |
jascott1 | good point | 23:03 |
kfox1111 | SamYaple: yes. | 23:03 |
jascott1 | so if name is speciied take the whole name not just append to memcached- | 23:03 |
jascott1 | if thats the case, got it | 23:04 |
kfox1111 | SamYaple: with ceph suport, k8s spread across both nodes, one labled compute and the other controller, and pods scheduled as appropriate. | 23:04 |
kfox1111 | jascott1: yeah. I think that shoudl work. then thhe user can set it towhatever they want. | 23:05 |
jascott1 | cool | 23:05 |
kfox1111 | shoudl always be unique then too. | 23:05 |
openstackgerrit | Joshua Harlow proposed openstack/kolla: Not much need to cache good, bad, unmatched https://review.openstack.org/406391 | 23:07 |
*** vhosakot_ has quit IRC | 23:08 | |
kfox1111 | SamYaple: http://logs.openstack.org/07/405107/8/experimental/gate-kolla-kubernetes-deploy-centos-binary-ceph-multi-nv/eff8258/ have a look if your curious. :) | 23:09 |
sbezverk | kfox1111: the version of gate where you added extra barrier for helm, was it merged already? if not it seems it does not solve ever case.. | 23:09 |
*** inc0 has quit IRC | 23:09 | |
sbezverk | ever/every/ | 23:09 |
kfox1111 | sbezverk: If Iremember correctly, the extra barior didn't actually help and I backed it out. as there's another barior before helm's used. | 23:12 |
kfox1111 | they did ask me to put in some logging thougt to see if we can atch it doing something. give me a few and I'll see if Ican see anything. | 23:12 |
kfox1111 | yeah... it failed at the ls. | 23:13 |
sbezverk | kfox1111: intersting that I have never seen in in baremetal setup | 23:14 |
SamYaple | kfox1111: nice! | 23:14 |
SamYaple | kfox1111: you can actually run ceph without having host pid namespace in k8s | 23:15 |
SamYaple | because you ahve pods | 23:15 |
SamYaple | you can share the pid namespace with the ceph-osd and no need for host | 23:15 |
kfox1111 | SamYaple: some tings yes, some no. if you use net=host, you need to use pid namespace too. | 23:15 |
kfox1111 | or else you get collection reconnect storms. | 23:16 |
kfox1111 | it all works, just much noisier then it should be. | 23:16 |
*** dave-mccowan has joined #openstack-kolla | 23:16 | |
sbezverk | SamYaple: sending ceph traffic over kube network overlay will impact seriously performace.. | 23:16 |
kfox1111 | the servers keep track of client connection by hostname/pid that the client tells them. if it sees a connection from the same hostname/pid telling its a "new" client, it evicts the old one. | 23:17 |
SamYaple | i dont think youve caught what i was saying sbezverk or kfox1111 | 23:17 |
kfox1111 | fully containerized net=host without pid namespaces ends up having multiple isolated clients contact the same servier with the same hostname/pid and they shoot eachother. | 23:17 |
SamYaple | with kolla-ansible you _must_ run pid=host to make multiple ceph-osd on the same host work | 23:18 |
kfox1111 | ture. | 23:18 |
SamYaple | with k8s pod support (with shared pid namespsaces) you dont have to do that | 23:18 |
SamYaple | this has nothing to do with net=host | 23:18 |
kfox1111 | yoru talking osd's not clients. | 23:18 |
kfox1111 | right. | 23:18 |
kfox1111 | ceph servers in kolla-k8s is for testing, not for production. | 23:19 |
kfox1111 | k8s doesn't yet have a very good model for dealing with ceph. | 23:19 |
kfox1111 | (servers) | 23:19 |
*** eaguilar has joined #openstack-kolla | 23:19 | |
jascott1 | kfox1111 if we use release name it wont be unique but thats ok? | 23:19 |
kfox1111 | jascott1: release names are garanteed to be unique I believe. | 23:19 |
SamYaple | kfox1111: hmm. wouldnt it be the same model as dealing with compute? this container must run here kind of thing? | 23:19 |
jascott1 | oh i see so calling helm install memcached twice will get you two releases. thanks | 23:20 |
kfox1111 | SamYaple: yeah, but how do you model a pool of nodes with multiple osd's? possibly a disjoint where some nodes have more disks then others? | 23:20 |
kfox1111 | jascott1: right. | 23:20 |
SamYaple | kfox1111: oh no clue. i dont know k8s. in ansible i check the host for number of osds and launch the appropriate contaienrs | 23:21 |
kfox1111 | you could do it with a series of daemonsets. but that feels clunky. | 23:21 |
SamYaple | kfox1111: so i get your point | 23:21 |
kfox1111 | yo ucould do it by launching individual pods bound to nodes, but that too is clunky. :/ | 23:21 |
kfox1111 | all of the clouds I manage are external ceph managed by ceph-deploy anyway, and that's working great, so not too much pressure on making ceph servers work yet other then to test tthe gate. | 23:22 |
SamYaple | how are you handling persistent mariadb and/or rabbitmq data in k8s? | 23:22 |
kfox1111 | ceph rbd backed. | 23:22 |
kfox1111 | k8s pv/pvc pair today. | 23:22 |
kfox1111 | once k8s 1.5 lands, maybe just dynamic pvc. | 23:22 |
SamYaple | oh. so ceph is a hard requirement right now | 23:22 |
SamYaple | ok | 23:23 |
*** lamt has quit IRC | 23:23 | |
kfox1111 | well, yes/no. | 23:23 |
SamYaple | does it work without ceph? | 23:23 |
kfox1111 | you can create pv's with any backing driver. | 23:23 |
kfox1111 | so, gce backed volumes if you wanted. | 23:23 |
kfox1111 | I just don't think much other then ceph right now is very practical. | 23:23 |
sbezverk | SamYaple: on a signle node you can use host fs | 23:24 |
kfox1111 | that works too, if you don't need any kind of failover. | 23:24 |
SamYaple | right, but for the project ceph is a hard requirement right now? | 23:24 |
SamYaple | other than AIO, which im not counting | 23:24 |
kfox1111 | SamYaple: we're not providing stuff other then ceph, but because we're using native k8s and building block phlosophy, | 23:24 |
kfox1111 | should be able to use anything k8s has a driver for. | 23:25 |
kfox1111 | so, say you have a gluster system or an an expensive ha nfs server. should in theory work. | 23:25 |
kfox1111 | we just havent tested it. but k8s has. | 23:25 |
SamYaple | one thing I will say kfox1111, i started building kolla-ansible the same way. and it was, early on. look at it now | 23:25 |
kfox1111 | and we're just consuming the k8s api there. | 23:25 |
kfox1111 | yup. hoping kolla-kubernetes will have the same growth. :) | 23:26 |
SamYaple | the whole "you can drop anything into place" hasn't lasted like id have hoped in kolla-ansible | 23:26 |
SamYaple | i wasnt saying it like that | 23:26 |
kfox1111 | ah. | 23:27 |
kfox1111 | well, I need a system thats more flexable then any general deployoment tool that I've run across so far. | 23:27 |
SamYaple | as bugs pop up (human error), the "fix" for those bugs because removing this flexible piece | 23:27 |
kfox1111 | so, so far as my contribuations, they will be at the lower levels where the building blocks are. | 23:27 |
SamYaple | sure i get that. this is why i wanted to seperate docker from ansible for so long | 23:28 |
kfox1111 | otherwise, if it becomes more opinionated, I can't continue to use it. | 23:28 |
SamYaple | and YAY its finally done | 23:28 |
kfox1111 | yeah. :) | 23:28 |
SamYaple | so should worst happen, its an easy fork | 23:28 |
kfox1111 | yeah. | 23:28 |
SamYaple | cool. well kfox1111 thanks for the info | 23:29 |
SamYaple | do you know when well know more about neutron and shared namespaces in k8s? | 23:29 |
*** goldyfruit has quit IRC | 23:29 | |
kfox1111 | SamYaple: any time. :) | 23:31 |
kfox1111 | SamYaple: don't know... I really really need it for another project too. | 23:31 |
kfox1111 | fuse filesystems in containers with k8s daemonsets are way easier to manage then with just the host. | 23:32 |
kfox1111 | SamYaple: https://github.com/kubernetes/kubernetes/pull/31504 is one of the issues. | 23:35 |
kfox1111 | kidn of a stopgap while the issue is discussed more. | 23:35 |
kfox1111 | SamYaple: https://github.com/kubernetes/kubernetes/pull/20698 is another option. | 23:36 |
*** schwicht has quit IRC | 23:36 | |
kfox1111 | I kind of prefer the latter, but whatever. I don't really care so long as its finally solved. | 23:36 |
sbezverk | kfox1111: ready to go: https://review.openstack.org/405107 | 23:37 |
SamYaple | should be sweet | 23:37 |
SamYaple | hey sbezverk and kfox1111 , might i suggest using openvswitch dpdk? its going to save you alot of headaches here | 23:37 |
SamYaple | no kernel support needed | 23:38 |
SamYaple | ive been using it at the house for about 6 months, worth it | 23:38 |
kfox1111 | SamYaple: can the kolla containers support it? we're just using whats there. | 23:38 |
SamYaple | its all userspace | 23:38 |
SamYaple | well its mostly userspace | 23:38 |
kfox1111 | if not, we need someone to add support to the containers. | 23:38 |
SamYaple | oh you mean _currently_ | 23:38 |
kfox1111 | sbezverk: nice. | 23:39 |
SamYaple | no it cant, im working on a patch and a few other openstack things | 23:39 |
SamYaple | for kolla that is | 23:39 |
sbezverk | SamYaple: dpdk is on to do list.. for now the main target is compute kit.. | 23:39 |
kfox1111 | SamYaple: awesome. once its in, it should be pretty easy to enable? | 23:40 |
kfox1111 | if I remember, its justly just about hardware acceleration, right? | 23:40 |
SamYaple | no its about not switching contexts | 23:40 |
SamYaple | all the network processign is done in userspace | 23:40 |
SamYaple | faster iteration, more testing, more stable, less..... kernel | 23:41 |
kfox1111 | oh. I thought thats the layer I saw some network vendors hooking in direct vxlan hardware acceleration. | 23:41 |
SamYaple | well that exists in the traditional ovs | 23:41 |
openstackgerrit | Merged openstack/kolla-kubernetes: Hel-ming openvswitch-vswitchd https://review.openstack.org/405107 | 23:41 |
sbezverk | kfox1111: we use dpdk a lot in cisco nfvi things | 23:41 |
kfox1111 | hmm. ok. | 23:41 |
kfox1111 | does it and up just being a config option in the end? | 23:42 |
kfox1111 | (just wondering how much kolla-kubernetes has to care) | 23:42 |
sbezverk | kfox1111: it would be a differnet image for dpdk support I think | 23:43 |
kfox1111 | oh. it is a build time option? | 23:43 |
kfox1111 | so we'd just have, say a openvswitch container built with it, and choose the alternate container? | 23:43 |
sbezverk | kfox1111: pretty much | 23:44 |
kfox1111 | k. soulnds easy to support on our end. :) | 23:44 |
SamYaple | yep | 23:45 |
sbezverk | kfox1111: I think plugging and defining bridge is a little differnet but from a client perspective it looks the same just faster | 23:45 |
SamYaple | thats teh whole deal | 23:45 |
sbezverk | SamYaple: Have you seen or played with vpp yet? | 23:46 |
SamYaple | sbezverk: neither | 23:46 |
sbezverk | SamYaple: it is even faster alternative to ovs/dpdk | 23:47 |
SamYaple | is it ovs/vpp? or a different thing entirely | 23:47 |
kfox1111 | cool. | 23:47 |
SamYaple | because im using OVS dpdk outside of openstack and then hooking openstack onto it | 23:47 |
sbezverk | https://fd.io/technology | 23:47 |
sbezverk | a month or so ago there was announcement on neutron mailer it is called networking-vpp | 23:48 |
jascott1 | anyone seen %!s(<nil>)-logging as logging-configmap name? | 23:49 |
SamYaple | interesting | 23:49 |
kfox1111 | sbezverk: interesting. seems like a tradeoff between bandwidth and latency though. | 23:50 |
kfox1111 | jascott1: yeah. need a var defined in all_values. | 23:50 |
SamYaple | kfox1111: they do talk about that, where latency is stable because of a few mechanics | 23:50 |
SamYaple | kfox1111: but it was alot of "averge latency" talk | 23:50 |
kfox1111 | SamYaple: there's average, and there's low. ;) | 23:51 |
SamYaple | :) | 23:51 |
kfox1111 | same with real time operating systems. | 23:51 |
SamYaple | i meant averaged* | 23:51 |
sbezverk | kfox1111: why do you say that? if you have sufficient horse power it can push 40/100g | 23:51 |
kfox1111 | they are predictable. they become predictable by running things really slow. :) | 23:51 |
SamYaple | kfox1111: i sent you a pm with a question when you get a chance | 23:51 |
kfox1111 | sbezverk: it functions by batching. batching incurs latency. | 23:51 |
sbezverk | kfox1111: dpdk always does that | 23:52 |
*** javcasalc has joined #openstack-kolla | 23:52 | |
SamYaple | yea sbezverk it looks comparable but younger if im reading it right | 23:52 |
sbezverk | no matter which top level software uses it.. the good thing it uses pull mode | 23:52 |
SamYaple | its interesting for sure | 23:52 |
sbezverk | SamYaple: there was an article comparing performance of ovs/dpdk and vpp/dpdk and vpp showed more stable results under the stress | 23:53 |
jascott1 | kfox1111 funny when I add that key i dont see yaml from --dry-run anymore | 23:54 |
jascott1 | will look into it | 23:54 |
kfox1111 | jascott1: weird. | 23:54 |
jascott1 | let me try diff chart | 23:55 |
jascott1 | same thing for neutron server (i didnt change it since its already set with the logging configmap ) | 23:56 |
SamYaple | sbezverk: its added to my list :) thanks | 23:56 |
jascott1 | tools/helm_prebuild.py && tools/helm_build_microservices.py ~/.helm/repository/local && helm install kolla/neutron-server --dry-run | 23:56 |
jascott1 | that will seriously suck for debbuging unless its just me doing something wrong | 23:57 |
sbezverk | SamYaple: if you have question, let me know I have direct access to developers working on this beast.. | 23:58 |
*** klindgren_ has joined #openstack-kolla | 23:58 | |
SamYaple | sbezverk: awesome. good to know | 23:59 |
sbezverk | jascott1: it works on my side --dry-run --debug | 23:59 |
jascott1 | ok thanks | 23:59 |
*** rhallisey has quit IRC | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!