| *** yangzb09 has joined #openstack-kolla | 00:06 | |
| *** sayantani01 has joined #openstack-kolla | 00:07 | |
| *** yangzb09 has quit IRC | 00:08 | |
| *** unicell has joined #openstack-kolla | 00:10 | |
| *** unicell is now known as Guest98378 | 00:11 | |
| *** sayantani01 has quit IRC | 00:11 | |
| *** chopmann has quit IRC | 00:17 | |
| *** Guest98378 has quit IRC | 00:17 | |
| openstackgerrit | Joseph Robinson proposed openstack/kolla: Build Deploy Guide Files and tox environment https://review.openstack.org/422931 | 00:20 |
|---|---|---|
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: WIP: Add a simple test to see if horizon is running https://review.openstack.org/422932 | 00:21 |
| *** schwicht has quit IRC | 00:23 | |
| *** schwicht has joined #openstack-kolla | 00:26 | |
| *** jtriley has joined #openstack-kolla | 00:28 | |
| *** schwicht has quit IRC | 00:29 | |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: WIP: test rebooting in the gate https://review.openstack.org/413193 | 00:29 |
| *** inc0 has joined #openstack-kolla | 00:29 | |
| kfox1111 | inc0: alive? | 00:30 |
| inc0 | indeed I am | 00:30 |
| *** yingjun has joined #openstack-kolla | 00:30 | |
| inc0 | barely, but here I am | 00:30 |
| kfox1111 | any chance we could get a ceph-mds and cephfs-fuse container still into ocata? | 00:31 |
| kfox1111 | might want to consume them in kolla-kubernetes, so would be nice to have in a stable release (soon) | 00:31 |
| *** salv-orl_ has quit IRC | 00:32 | |
| inc0 | kfox1111, so I'd be reluctant to this frankly | 00:32 |
| inc0 | we don't backport features | 00:32 |
| *** awiddersheim has joined #openstack-kolla | 00:33 | |
| kfox1111 | it woudlnt' be a backport. | 00:33 |
| inc0 | and we had standard of matching release of deployment tool with release of images | 00:33 |
| kfox1111 | it would be in ocata, and then we would depend on ocata for it. | 00:33 |
| inc0 | ahh | 00:33 |
| inc0 | for whatever reason I thought ocata was released already | 00:33 |
| kfox1111 | just saying, ocata will be stable soon. | 00:33 |
| inc0 | brain bad | 00:33 |
| kfox1111 | so would be nice to have befre ocata becomes stable. :) | 00:33 |
| duonghq | morning guys | 00:33 |
| inc0 | if you can crack code for it, sure | 00:33 |
| kfox1111 | k. guess its time to learn to add containers. | 00:34 |
| inc0 | I'll be happy to help | 00:34 |
| kfox1111 | cool. thanks. :) | 00:34 |
| kfox1111 | it should be just installing one package in each container really. | 00:34 |
| kfox1111 | ceph-mon in one, and cephfs-fuse in the other. | 00:34 |
| kfox1111 | sry, ceph-mds. | 00:34 |
| kfox1111 | ceph-mds works very similarlyl to ceph-mon. | 00:35 |
| kfox1111 | and the other one is just a fuse filesystem. so we can use it to mount things. | 00:35 |
| inc0 | hmm...do you want to use it for persistent volume? | 00:36 |
| kfox1111 | long term, yeah, the ability to do that. | 00:37 |
| inc0 | fine by me, it should be super straighforward patch | 00:37 |
| kfox1111 | it will take at least one more feature from k8s though, and i'm not sure it will make it into 1.6. :/ | 00:37 |
| kfox1111 | pushing hard for 1.7 though. if I have to code it myself. :) | 00:38 |
| inc0 | what's this feature? | 00:38 |
| *** adrian_otto has joined #openstack-kolla | 00:38 | |
| kfox1111 | docker shared mount namespaces. :/ | 00:38 |
| kfox1111 | still doesn't support them. :/ | 00:38 |
| inc0 | :S | 00:38 |
| inc0 | how does neutron work without it? | 00:38 |
| kfox1111 | once it has support, then we can run fuse filesystems in in containers managed by daemonsets. | 00:38 |
| kfox1111 | yes/no. | 00:38 |
| kfox1111 | it works, so long as you don't restart he container. :/ | 00:38 |
| inc0 | isnt this required for network namespace share? | 00:39 |
| inc0 | so it doesnt work.. | 00:39 |
| kfox1111 | fixing requires rebooting the node. :/ | 00:39 |
| inc0 | dafaq | 00:39 |
| kfox1111 | yeah, I've leaned on k8s hard as I can to get hte feature finished. | 00:39 |
| inc0 | you just saddened me greatly | 00:39 |
| kfox1111 | they tried and oversimplify the feature and then got spooked by it. :/ | 00:39 |
| kfox1111 | yeah, I'm really upset about it too. :/ | 00:40 |
| kfox1111 | could really use it for high energy physics work too. | 00:40 |
| inc0 | lol | 00:41 |
| inc0 | I need to ask you about your deployments sometime, seems like you're doing cool stuff | 00:42 |
| kfox1111 | they use a filesystem called cvmfs, which is fused based. | 00:42 |
| kfox1111 | they llike to break it in various versions though. :/ | 00:42 |
| kfox1111 | so containerizing it would be extremely helpful to stability. | 00:42 |
| kfox1111 | and managing it with k8s like everything else would also be extremely helpful. :/ | 00:42 |
| kfox1111 | sure. :) | 00:43 |
| kfox1111 | got a condor compute element running out of k8s now. does k8s orchestrated rulling updates nicely. :) | 00:43 |
| kfox1111 | working on shrinking that particular openstack and moving most of the nodes over to k8s. | 00:43 |
| kfox1111 | is ceph_mon_header not actually defined anywhere? | 00:44 |
| kfox1111 | just for overrides? | 00:44 |
| *** n0isyn0i1e has joined #openstack-kolla | 00:44 | |
| *** awiddersheim has left #openstack-kolla | 00:44 | |
| kfox1111 | oh... all the packages are just installed in ceph-base? | 00:45 |
| *** n0isyn0ise has quit IRC | 00:47 | |
| *** zhurong has joined #openstack-kolla | 00:48 | |
| kfox1111 | hmm... ceph-mds is already in ceph-base.... | 00:50 |
| inc0 | it's block | 00:50 |
| inc0 | we can create blocks for overrides | 00:50 |
| inc0 | just a label | 00:50 |
| kfox1111 | do you have to create them though for new packages, or are they just empty if undefined? | 00:51 |
| inc0 | empty if undefined | 00:51 |
| openstackgerrit | Zeyu Zhu proposed openstack/kolla: Modify the repo in multinode.rst file https://review.openstack.org/422943 | 00:52 |
| inc0 | you can inject stuff into them by redefining it in override | 00:52 |
| kfox1111 | k. | 00:52 |
| inc0 | if you add stuff inside block, it's default that can be overriden | 00:52 |
| kfox1111 | so, I just have to copy ceph-rgw to ceph-mds then, and s/rgw/mds/ and that one's done. | 00:52 |
| kfox1111 | and the same for fuse, and just add ceph-fuse package to ceph-base. | 00:53 |
| inc0 | sounds about right | 00:53 |
| kfox1111 | where do the ceph packages come from for ubuntu? | 00:54 |
| inc0 | you can install ceph-fuse in ceph-fuse container tho | 00:54 |
| kfox1111 | (so I can lookup the cephfs package name? | 00:54 |
| kfox1111 | ) | 00:54 |
| inc0 | I think cloud archive hosts it | 00:54 |
| *** bradjones has quit IRC | 00:55 | |
| *** tovin07 has joined #openstack-kolla | 00:57 | |
| *** schwicht has joined #openstack-kolla | 00:59 | |
| *** jascott1 has joined #openstack-kolla | 01:00 | |
| duonghq | kfox1111, ping | 01:01 |
| kfox1111 | duonghq: hi | 01:01 |
| duonghq | hi Kevin | 01:01 |
| duonghq | about the ceph situation, | 01:01 |
| duonghq | with Ansible deliverable, we can get partition at runtime and pass to container | 01:02 |
| duonghq | but does helm has any mechanism like this? | 01:02 |
| kfox1111 | currently, in kolla-kubernetes, we don't have a production ceph for running in k8s. | 01:03 |
| kfox1111 | so its expected the user would provide a ceph some other way. | 01:03 |
| *** cuongnv has joined #openstack-kolla | 01:03 | |
| duonghq | kfox1111, how about the ceph component microservice? | 01:03 |
| kfox1111 | the ceph server stuff in kolla-kubernetes is just intended to test out the ceph support in the packages. | 01:03 |
| kfox1111 | it will take a lot more work to make a production ready ceph on top of k8s. probably changes to k8s itself. :/ | 01:04 |
| *** jascott1 has quit IRC | 01:04 | |
| *** inc0 has quit IRC | 01:05 | |
| duonghq | so, what should I do with the ceph-mon, ceph-osd microservice atm | 01:05 |
| duonghq | I stuck with these | 01:05 |
| duonghq | I see that we still have ceph as a work item in helm-service bp | 01:05 |
| kfox1111 | I think we should convert the to helm, just so we can get rid of all the old python code for deployment. | 01:06 |
| kfox1111 | will simplify the code base greatly. | 01:06 |
| kfox1111 | the few ceph things are about the only thing left. | 01:07 |
| openstackgerrit | desheng proposed openstack/kolla: Modify punctuation https://review.openstack.org/422949 | 01:09 |
| openstackgerrit | Kevin Fox proposed openstack/kolla: Add ceph fs containers https://review.openstack.org/422950 | 01:12 |
| duonghq | kfox1111, but the helm stuff is just for testing, not for production? | 01:12 |
| kfox1111 | the ceph server stuff is just for testing. | 01:13 |
| duonghq | I still cannot figure out how to pass partition_num to ceph | 01:13 |
| kfox1111 | all the rest of the helm stuff is for production. | 01:13 |
| kfox1111 | through a helm value? | 01:13 |
| duonghq | ya, due to the value must be taken at runtime, on the nodes? | 01:14 |
| duonghq | It's not pre-determined | 01:14 |
| kfox1111 | yeah, but again, its only used for testing, so passing it through is ok. | 01:14 |
| kfox1111 | so for testing, we only launch it on the same node its being launched from, | 01:14 |
| kfox1111 | so we know its device properties. | 01:14 |
| duonghq | so, I can run some test to get the value, and hard-code to values? | 01:15 |
| kfox1111 | we can pass it through a helm value. | 01:16 |
| duonghq | In test code? | 01:16 |
| kfox1111 | so would be something like helm install kolla/test-ceph-osd --set id=0,device=... etc. | 01:16 |
| kfox1111 | from tests/bin/build_test_ceph.sh | 01:16 |
| *** yingjun has quit IRC | 01:16 | |
| duonghq | got it, thank you | 01:17 |
| kfox1111 | np. :) | 01:17 |
| duonghq | btw, anybody is working on horizon in helm-services bp? | 01:18 |
| duonghq | can I take the component, kfox1111? | 01:19 |
| duonghq | in the work items of bp I do not see anybody is working on that | 01:19 |
| *** adrian_otto has quit IRC | 01:20 | |
| kfox1111 | yeah, I think thats free to work on. :) | 01:21 |
| duonghq | roger | 01:21 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: WIP: Add a simple test to see if horizon is running https://review.openstack.org/422932 | 01:23 |
| *** chopmann has joined #openstack-kolla | 01:28 | |
| kfox1111 | sbezverk: I was talking to the horizon dev's. in order to do seamless upgrades at all with horzion we will need multiple instance support on that service too. | 01:29 |
| *** sp___ has quit IRC | 01:32 | |
| kfox1111 | I figured we would need it at least for major upgrades. | 01:34 |
| kfox1111 | but filenames can change on minor upgrades due to the way they do compression. | 01:35 |
| *** britthouser has quit IRC | 01:38 | |
| *** caowei has joined #openstack-kolla | 01:38 | |
| *** britthouser has joined #openstack-kolla | 01:39 | |
| *** sayantani01 has joined #openstack-kolla | 01:39 | |
| *** jascott1 has joined #openstack-kolla | 01:40 | |
| *** saneax-_-|AFK is now known as saneax | 01:43 | |
| *** sp___ has joined #openstack-kolla | 01:46 | |
| *** zhangkun_ has joined #openstack-kolla | 01:46 | |
| jascott1 | kfox1111 ping | 01:47 |
| *** zhurong has quit IRC | 01:49 | |
| *** zhurong has joined #openstack-kolla | 01:51 | |
| *** cuongnv has quit IRC | 01:52 | |
| *** tonanhngo has quit IRC | 01:52 | |
| kfox1111 | jascott1: ping | 01:54 |
| jascott1 | oh hey, how do we currently provide private instance of rabbit to neutron -server? | 01:55 |
| kfox1111 | we currently have no code exersising that option. | 01:56 |
| jascott1 | what I have done is add rabbit pieces to neutron-server requirements.yaml (which I had to create) | 01:56 |
| kfox1111 | right. | 01:56 |
| jascott1 | ok cool just checking in case I missed something. tahnks | 01:56 |
| jascott1 | thanks even | 01:56 |
| kfox1111 | np. sounds like the right aproach to me. | 01:56 |
| *** tonanhngo has joined #openstack-kolla | 02:01 | |
| openstackgerrit | Zeyu Zhu proposed openstack/kolla-kubernetes: Fix the issue in multi-node.rst https://review.openstack.org/421054 | 02:05 |
| *** cuongnv has joined #openstack-kolla | 02:06 | |
| *** tonanhngo has quit IRC | 02:06 | |
| *** chopmann has quit IRC | 02:07 | |
| *** schwicht has quit IRC | 02:07 | |
| *** zhubingbing has joined #openstack-kolla | 02:09 | |
| openstackgerrit | Merged openstack/kolla-ansible: Fix gate failure https://review.openstack.org/422734 | 02:14 |
| openstackgerrit | lixing proposed openstack/kolla: Modify punctuation Bug-id: #1657933 https://review.openstack.org/422985 | 02:14 |
| *** prameswar has joined #openstack-kolla | 02:25 | |
| *** zhubingbing_ has joined #openstack-kolla | 02:26 | |
| openstackgerrit | Merged openstack/kolla-ansible: Add karbor ansible role https://review.openstack.org/402229 | 02:26 |
| *** zhubingbing has quit IRC | 02:27 | |
| openstackgerrit | lixing proposed openstack/kolla: Modify punctuation https://review.openstack.org/422985 | 02:32 |
| *** huikang has joined #openstack-kolla | 02:33 | |
| openstackgerrit | lixing proposed openstack/kolla: Modify punctuation https://review.openstack.org/422985 | 02:35 |
| openstackgerrit | Wei Cao proposed openstack/kolla-ansible: Add Karbor doc link in README.rst https://review.openstack.org/422995 | 02:36 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Add a simple test to see if horizon is running https://review.openstack.org/422932 | 02:37 |
| zhangqiankun | anybody can help me ,what the failure in https://review.openstack.org/#/c/407810/ | 02:43 |
| *** kbaegis has quit IRC | 02:44 | |
| zhangqiankun | Have a great god can help me :) | 02:51 |
| *** huikang has quit IRC | 02:51 | |
| *** huikang has joined #openstack-kolla | 02:52 | |
| *** masuberu has joined #openstack-kolla | 02:52 | |
| *** Masber_080 has quit IRC | 02:55 | |
| openstackgerrit | Merged openstack/kolla: Fix the gate failure https://review.openstack.org/422344 | 02:56 |
| *** huikang has quit IRC | 02:56 | |
| *** zhubingbing__ has joined #openstack-kolla | 02:59 | |
| *** zhubingbing_ has quit IRC | 02:59 | |
| *** huikang has joined #openstack-kolla | 03:04 | |
| *** huikang has quit IRC | 03:04 | |
| *** huikang has joined #openstack-kolla | 03:05 | |
| *** tovin07_ has joined #openstack-kolla | 03:06 | |
| *** huikang has quit IRC | 03:09 | |
| *** qiliang27 has quit IRC | 03:13 | |
| *** qiliang27 has joined #openstack-kolla | 03:13 | |
| openstackgerrit | Duong Ha-Quang proposed openstack/kolla-kubernetes: Horizon service package https://review.openstack.org/423016 | 03:18 |
| *** imcsk8 has joined #openstack-kolla | 03:20 | |
| openstackgerrit | zhubingbing proposed openstack/kolla: Fix designate dockerfile https://review.openstack.org/422366 | 03:21 |
| *** zhubingbing_ has joined #openstack-kolla | 03:23 | |
| *** zhubingbing__ has quit IRC | 03:23 | |
| *** saneax is now known as saneax-_-|AFK | 03:24 | |
| *** inc0 has joined #openstack-kolla | 03:25 | |
| inc0 | hey, have you read docker 1.13 change list? | 03:26 |
| inc0 | pretty neat | 03:26 |
| *** rhallisey has quit IRC | 03:26 | |
| tovin07 | yes | 03:26 |
| inc0 | I need to check out this squash flag | 03:27 |
| *** sdake has joined #openstack-kolla | 03:30 | |
| inc0 | yo sdake check out docker 1.13 change log | 03:30 |
| sdake | just woke up | 03:31 |
| sdake | and eating | 03:31 |
| sdake | ptfoed at like 330pm | 03:31 |
| sdake | got a link | 03:31 |
| inc0 | https://github.com/docker/docker/blob/master/CHANGELOG.md | 03:31 |
| sdake | s390 lol | 03:32 |
| inc0 | this squash feature might be extremally cool | 03:33 |
| portdirect | inc0: this it could be pretty good | 03:33 |
| portdirect | *think | 03:33 |
| sdake | ya i saw that in there | 03:34 |
| inc0 | yeah, it's experimental but worth looking at | 03:34 |
| sdake | evening portdirect | 03:34 |
| portdirect | inc0: these been other things around - Iand both me and sam have tried the in kolla | 03:34 |
| portdirect | an sve about 15-20% off the image size | 03:34 |
| portdirect | hey sdake :) | 03:35 |
| sdake | i see your back ;-) | 03:35 |
| * portdirect closes window | 03:35 | |
| sdake | my wife prepping to devconf takeoff on monday | 03:35 |
| portdirect | oh nice | 03:35 |
| portdirect | still need to do the pp for that | 03:35 |
| sdake | she is going early or late - not such which | 03:35 |
| sdake | gone for 8 days | 03:36 |
| sdake | home alone with the kids | 03:36 |
| * sdake whines | 03:36 | |
| * portdirect laughs manically | 03:36 | |
| sdake | did anyone fix the docker-engine gate problem in kolla-ansible? | 03:36 |
| duonghq | oh, have fun sdake | 03:36 |
| duonghq | sdake, iirc, yes | 03:36 |
| duonghq | Jeffrey or somebody pinned the docker-engine version | 03:37 |
| duonghq | and it's merged | 03:37 |
| inc0 | heh | 03:37 |
| inc0 | yeah | 03:37 |
| inc0 | new docker version | 03:37 |
| sdake | pinning is a bad long term solution :( | 03:37 |
| duonghq | sure, | 03:37 |
| sdake | does anyone know what actually causes the problem? | 03:37 |
| sdake | i wonder if this makes the devenv nonfunctional for kolla-kubernetes | 03:38 |
| *** zhurong has quit IRC | 03:39 | |
| duonghq | seem that something messed up with docker registry | 03:43 |
| openstackgerrit | Yao Lu proposed openstack/kolla: Change docker restart policy https://review.openstack.org/423026 | 03:44 |
| sbezverk | sdake: another crisis ;-) ?? | 03:48 |
| sdake | sbezverk could be | 03:48 |
| sdake | sbezverk are you available now for a bit of webex? | 03:48 |
| sdake | probably not since its 11pm there | 03:49 |
| sbezverk | sdake: I can irc chat for a bit by it is too late for live talk | 03:49 |
| sdake | ok makes sense sbezverk | 03:50 |
| kfox1111 | so is relying on an upstream that seems hell bent on breaking every release. ;) | 03:54 |
| kfox1111 | there's a reason I prefer redhat's docker most of the time. :) | 03:54 |
| duonghq | sdake, have you gotten dinner? | 03:54 |
| sdake | portdirerect RQ like, does the devenv setup ceph properly? | 03:55 |
| sdake | portdirect ^^ | 03:55 |
| sdake | duonghq ya I just ate | 03:55 |
| sdake | it was sort of breakfast i guess | 03:55 |
| sdake | although it was dinner :) | 03:55 |
| sdake | I pretty much hit the wall at 3pm today | 03:55 |
| sdake | its about 9pm now | 03:55 |
| portdirect | sdake - not kolla ceph :( | 03:55 |
| sdake | portdirect IDC | 03:55 |
| *** saneax-_-|AFK is now known as saneax | 03:55 | |
| portdirect | IDC? | 03:55 |
| sdake | portdirect what I want to know is if it does ceph at all | 03:55 |
| sdake | I Don't Care ;) | 03:56 |
| duonghq | sdake, not so good | 03:56 |
| portdirect | gotcha | 03:56 |
| sdake | duonghq meh it works for me :) | 03:56 |
| sbezverk | portdirect: hey man, how is your preso coming?? | 03:56 |
| portdirect | sbezverk: shut up :) | 03:56 |
| sdake | portdirect the goal here I have is to actually rewrite the QSG to be releevant to what we have | 03:56 |
| sbezverk | portdirect: ;) something was telling you would say that | 03:57 |
| sdake | portdirect (e.g. how do I get from QSG to fully deployed environment) | 03:57 |
| *** mdnadeem has joined #openstack-kolla | 03:57 | |
| sdake | portdirect i'm going to write those docs right now | 03:57 |
| sdake | as soon as i figure out how its done | 03:57 |
| portdirect | sbezverk: I'll storm off slamming doors in a munite :) | 03:57 |
| sdake | portdirect what i need to know is, ceph_workflow* requires ceph | 03:57 |
| sdake | sbezverk don't run portdirect off plz :) | 03:57 |
| sdake | portdirect will ceph_workflow_* workflows work with the devenv? | 03:58 |
| portdirect | sdake: to get kolla ceph running outside of gat is noth somthing ive managed to do really | 03:58 |
| portdirect | 2 sec | 03:58 |
| zhangkun_ | Can anyone help review? https://review.openstack.org/#/c/422377/ | 03:58 |
| zhangkun_ | thx. | 03:58 |
| kfox1111 | I don't know if anyone has tried since we converted to helm packages. | 03:59 |
| sdake | i guess a lecture here wouldn't help :) | 03:59 |
| portdirect | last time was pre that: https://review.openstack.org/#/c/405720/9/tests/bin/build_dev_ceph.sh | 03:59 |
| portdirect | and that was only single node via major hackery | 03:59 |
| sdake | ok, what exactly can I document and stand up in thenext 8-10 hours | 03:59 |
| kfox1111 | sdake: active development, recently refactered, blablabla. :) | 03:59 |
| sdake | kfox1111 that doesn't work for devs, devs need a way to do development | 04:00 |
| kfox1111 | I plan on deploying it on real hardware soon. | 04:00 |
| kfox1111 | so I can take notes. | 04:00 |
| sdake | kfox1111 i need it now not soon :) | 04:00 |
| portdirect | kfox1111: is the best here | 04:00 |
| sdake | kfox1111 so - lets assume I'll do the documenting | 04:00 |
| sdake | can anyone throw me bones on what is missing wrt the devenv and ceph? | 04:00 |
| kfox1111 | sdake: k. | 04:00 |
| portdirect | i'm just using this I'm afraid | 04:00 |
| sdake | which portdirect | 04:01 |
| portdirect | https://github.com/portdirect/ceph-helm | 04:01 |
| kfox1111 | would be interested in hearing how well that will work out. | 04:01 |
| portdirect | and i may be quicker getting that working with kolla, than kolla working out of the gate :/ | 04:01 |
| kfox1111 | I was just planning on using ceph-deploy. | 04:02 |
| kfox1111 | its been rock solid for us for years. | 04:02 |
| portdirect | though again kfox1111 is in a much better postion to give an idea of t/f | 04:02 |
| sdake | portdirect will that ceph-helm run on single disk? | 04:02 |
| portdirect | no | 04:02 |
| portdirect | needs 3 nodes | 04:02 |
| sdake | 3 disks or 3 nodes? | 04:03 |
| sdake | or 1 disk * 3 nodes | 04:03 |
| portdirect | nodes the way the chart is written | 04:03 |
| portdirect | is uses bind mounts | 04:03 |
| kfox1111 | yeah ceph can run single node, but they strongly recommend against it. | 04:03 |
| sdake | yes I know it runs single node | 04:04 |
| portdirect | so discs are not really relivant | 04:04 |
| kfox1111 | I had problems deploying ceph in gate with just one disk though. | 04:05 |
| kfox1111 | became much easier when I made at least 2 osd's. | 04:05 |
| kfox1111 | the way it was configured, it should have worked. but for some reason, just didn't. | 04:05 |
| portdirect | kfox, how eay would it be to update gate to run outside gate? | 04:05 |
| kfox1111 | earlier versions did just work in that config. | 04:05 |
| portdirect | *easy | 04:05 |
| sdake | portdirect that is called third party gating | 04:05 |
| *** PramodJ has quit IRC | 04:05 | |
| sdake | portdirect third party gating cannot be voting | 04:05 |
| kfox1111 | with or without ceph? | 04:05 |
| portdirect | with ceph | 04:06 |
| sdake | portdirect i've never set up third party gating, however, it is probably relatively easy | 04:06 |
| portdirect | is that not what you kinda need though sdake ? | 04:06 |
| portdirect | single node poc? | 04:06 |
| kfox1111 | sorry. didn't word that carefully. | 04:06 |
| sdake | portdirect what I need is -> | 04:06 |
| kfox1111 | with internal or external ceph? | 04:06 |
| sdake | on my development node/nodes, deploy kolla-kubernetes | 04:06 |
| sdake | Ive got 3 nodes to work with | 04:06 |
| sdake | portdirect i'd like to use the devenv we got | 04:07 |
| sdake | portdirect wondering how to stitch your ceph-helm chart and the devenv we got together to get up and running by running ceph_workflow_* | 04:07 |
| kfox1111 | sdake: I broke out ceph_workflow so I could share it with minikube. | 04:08 |
| sdake | do I need 3 baremetal nodes? - I've got that to work with | 04:08 |
| kfox1111 | haven't gotten back round to trying it though, but the intention was to be able to share it. | 04:08 |
| sdake | kfox1111 I really want ot use the devenv because it is simple to setup | 04:08 |
| sdake | kfox1111 you probably have never used the devnev (vagrant) environemnt | 04:08 |
| sdake | kfox1111 its dead simple to setup and practically invisible once launched | 04:08 |
| kfox1111 | yeah. I tried vagrent again a few days ago. | 04:09 |
| portdirect | sdake - once its up and running -t shoudl be just like external ceph i think | 04:09 |
| kfox1111 | made me mad again. | 04:09 |
| kfox1111 | they don't support kvm well at all. :/ | 04:09 |
| sdake | kfox1111 mind expanding on what problems you had | 04:09 |
| portdirect | kfox1111: how hard is is to ge going with external ceph? | 04:09 |
| portdirect | kfox1111: ubuntu/kvm? | 04:09 |
| kfox1111 | portdirect: pretty easy I think. the only thing undefined is the exact steps to bring in the config/keys. | 04:09 |
| kfox1111 | theres stuff in the gate to do that, but needs a little generaliazation or documentation. | 04:10 |
| portdirect | could you helm sdake with that? | 04:10 |
| kfox1111 | portdirect: centos7/kvm. | 04:10 |
| * sdake documenting | 04:10 | |
| kfox1111 | yeah. but probably more tomorrow then today. | 04:10 |
| sdake | here is the deal a whole lot of people are blocked because they dont know how to get kolla off the ground | 04:10 |
| sdake | they read the docs and are like "thes edocs are 6 months old" | 04:10 |
| portdirect | sdake: I'm in that boat as well :( | 04:10 |
| sdake | we should just remove them if they are irrelevant | 04:11 |
| portdirect | +2 | 04:11 |
| sdake | ok fellas i'm not a dummy - I just need some general hints :) | 04:11 |
| kfox1111 | +1 | 04:11 |
| sdake | so - setup devenv, that works | 04:11 |
| portdirect | yup | 04:11 |
| sdake | build a ceph-helm chart, that works (?) | 04:11 |
| portdirect | yup | 04:11 |
| sdake | then do what, deploy the ceph-helm chart *where8 | 04:12 |
| sdake | on bare metal, or in the devenv? | 04:12 |
| portdirect | either | 04:12 |
| kfox1111 | either. | 04:12 |
| portdirect | id go for dev env | 04:12 |
| sdake | cool | 04:12 |
| sdake | i'll do devenv | 04:12 |
| sdake | and that will give me what, external ceph? | 04:12 |
| portdirect | throw 8gb of ram at each node and 4 cores | 04:12 |
| portdirect | yes | 04:12 |
| portdirect | external ceph running on same nodes... | 04:13 |
| sdake | i've got 32gb of ram in my box - so that hsoudl work :) | 04:13 |
| sdake | why is external ceph needed, is that just so openstack works properly? | 04:13 |
| kfox1111 | should be plenty. | 04:13 |
| kfox1111 | I did a full deploy in under 8 in minikube. | 04:13 |
| sdake | or is it a functional requirement for the devenv? | 04:13 |
| kfox1111 | including 2 vm's. :) | 04:14 |
| portdirect | not a requirement for deve env | 04:14 |
| sdake | portdirect I thought k8s required ceph? | 04:14 |
| portdirect | no | 04:14 |
| sdake | how does k8s do persistent storage? | 04:15 |
| portdirect | its very unual to run ceph in k8s | 04:15 |
| portdirect | rememebr it usually runs *in* a cloud | 04:15 |
| portdirect | so cinder | 04:15 |
| kfox1111 | that will change I think. :) | 04:15 |
| sdake | k8s uses cinder for persistent storage? | 04:15 |
| sdake | focus :) | 04:16 |
| portdirect | it can | 04:16 |
| portdirect | or gce | 04:16 |
| portdirect | or aws | 04:16 |
| sdake | ok.. | 04:16 |
| sdake | i don't want to know what it can do | 04:16 |
| sdake | I want to know what we recommend for a devenv :) | 04:16 |
| sdake | if we were dictating a devenv to nwe contribs | 04:16 |
| sdake | what would it look like | 04:16 |
| kfox1111 | if your doing multinode devenv, I'd still recommend ceph. | 04:17 |
| kfox1111 | if all in one, you can get away without it. | 04:17 |
| sdake | kfox1111 the devenv is mutinode | 04:17 |
| sdake | kfox1111 (via virtualization) | 04:17 |
| portdirect | sdake - I have not maanged to get kolla running outside of gate since december | 04:17 |
| sdake | portdirect that is bad bad news | 04:17 |
| *** adrian_otto has joined #openstack-kolla | 04:17 | |
| portdirect | i dont think anyone is running it outside of gate | 04:17 |
| sdake | portdirect i know your talented and have probably tried everything | 04:17 |
| sdake | portdirect yes, I think your right | 04:17 |
| sdake | we need ot fix that | 04:17 |
| sdake | portdirect where are you stuck there? | 04:18 |
| portdirect | kfox1111: are you? | 04:18 |
| portdirect | config gen | 04:18 |
| kfox1111 | portdirect: I havent' tried outside the gate in a while. but like I said, in a week or two I'm palnning on deploying it on our test harware. | 04:18 |
| sdake | portdirect which config gen, the secrets for ceph? | 04:18 |
| kfox1111 | ah. genconfig's kind of a pain. still. :/ | 04:18 |
| kfox1111 | as that's one of the last holdouts for kollakube. | 04:19 |
| kfox1111 | and depneds on kolla-ansible. | 04:19 |
| sdake | i understand | 04:19 |
| kfox1111 | and that install process is a little rough. | 04:19 |
| sdake | so, starting at the bottom | 04:19 |
| sdake | we have ceph | 04:19 |
| portdirect | its not really possible to customise the config | 04:19 |
| sdake | portdirect customize which config? | 04:19 |
| kfox1111 | you can customize most of it the same way as kolla-ansible. | 04:19 |
| portdirect | the configmaps that kolla-k8s uses (i just dont know how) | 04:20 |
| kfox1111 | you configure /etc/kolla/globals.yaml | 04:20 |
| kfox1111 | run kolla-ansible genconfig to build the configmap source material | 04:20 |
| kfox1111 | then kollakube res create configmap .... | 04:20 |
| kfox1111 | which is in ceph_workflow | 04:21 |
| kfox1111 | it requres a bit of config options to be added to /etc/kolla/globals.yaml, that is in kolla-kuberntes/tests/conf/ceph-all-in-one/kolla-config | 04:21 |
| kfox1111 | stuff like swithcing the configmap generation to orcestration k8s and a few other bits. | 04:22 |
| sdake | kfox1111 I was planning to follow ceph_workflow.sh | 04:22 |
| kfox1111 | (yes, I agree that stuff is defintely suboptimal at the moment.) | 04:23 |
| sdake | kfox1111 the issue I am strugglign with currently is ceph_workflow I assume depends on some kind of ceph somewhere | 04:23 |
| sdake | I don't know which ceph it depends on, and i don't know where it depends on it | 04:23 |
| kfox1111 | yeah. ceph is setup before ceph_workflow kicks off. | 04:23 |
| kfox1111 | setup_gate.sh is the whole setup. | 04:23 |
| kfox1111 | which includes, setting up loopback devices for ceph | 04:23 |
| kfox1111 | installing kolla-ansible so genconfig will work, | 04:24 |
| kfox1111 | copying a bit of default config into /etc/kolla/globals.yaml | 04:24 |
| kfox1111 | setting up k8s | 04:24 |
| kfox1111 | setting up a test ceph inside k8s, | 04:24 |
| kfox1111 | pulling credentials out of ceph and uploading them so kolla-kubernetes can use them. | 04:24 |
| kfox1111 | and then launching kolla-kubernetes. | 04:25 |
| kfox1111 | ceph_workflow.sh is the last step there. | 04:25 |
| sdake | got it | 04:25 |
| sdake | so the pulling the creds out of ceph | 04:25 |
| sdake | i think that is where the integraiton point is between helm-ceph and ceph_workflow.sh for the devenv we would dictate if we were to do that | 04:25 |
| sdake | got a link to where that is done currently? | 04:25 |
| kfox1111 | see tools/setup_simple_ceph_users.sh | 04:25 |
| kfox1111 | and tools/setup-ceph-secrets.sh | 04:26 |
| kfox1111 | and tools/setup_rbd_volumes.sh | 04:27 |
| sdake | portdirect could you confirm my assertion above? I understand the rest of kfox1111 's statements, just not sure on the integration between helm-ceph and ceph_workflows.h | 04:27 |
| kfox1111 | those take the existing ceph and set it up for kolla-kube to use. | 04:27 |
| sdake | portdirect i had for some reason thought the devenv setup ceph during vagrant up | 04:27 |
| portdirect | no | 04:27 |
| kfox1111 | tests/bin/basic_tests.sh sets up the test ceph from scratch. | 04:27 |
| sdake | ok fellas- this weekend i am going to be reworking the docs significantlly | 04:28 |
| portdirect | it just installs the required packages on the host to make use of ceph once it is set up | 04:28 |
| kfox1111 | those three files above, can be tweked to talk to ceph-helm instead. | 04:28 |
| sdake | would appreciate fast merges on monday - even if they are not perfect gramatically :) | 04:28 |
| kfox1111 | sdake: sounds great. :) | 04:28 |
| portdirect | sounds great to me as well :) | 04:28 |
| kfox1111 | I can help answer questions. just ping me | 04:28 |
| sdake | sounds good kfox1111 | 04:28 |
| sdake | portdirect - in the context switching there - I saw this: | 04:29 |
| sdake | [21:28:08] <portdirect>it just installs the required packages on the host to make use of ceph once it is set up | 04:29 |
| sdake | would you mind expanding on that, that didn't totally parse for me | 04:29 |
| sdake | it being what? | 04:29 |
| portdirect | yeah - you need to have rbd etc intalled on the host | 04:30 |
| portdirect | 2 sec | 04:30 |
| sdake | you mean outside of the devenv? | 04:30 |
| portdirect | no: https://github.com/att-comdev/halcyon-kubernetes/blob/1fb6b919b27985731a87da01bfa5df82f4845f82/kube-deploy/roles/upgrade-os/tasks/centos.yml#L40 | 04:31 |
| zhangkun_ | Can anyone help review? https://review.openstack.org/#/c/422377/ thx. | 04:31 |
| portdirect | and https://github.com/att-comdev/halcyon-kubernetes/blob/e4547655105f46fff4dcfa1cdb101fbaf1664a3e/kube-deploy/roles/kube-prep/tasks/patch-kube.yml#L17 | 04:31 |
| portdirect | and https://github.com/att-comdev/halcyon-kubernetes/blob/e4547655105f46fff4dcfa1cdb101fbaf1664a3e/kube-deploy/roles/kube-prep/tasks/prep-host-dns.yml#L54 | 04:32 |
| portdirect | thats all it does | 04:32 |
| portdirect | which is required for k8s to make use of ceph as its not a typical use case | 04:32 |
| kfox1111 | kolla-kubernetes has an alternate way of doing that support, | 04:32 |
| kfox1111 | that doesn't require installing stuff on the host. | 04:33 |
| kfox1111 | the ceph-rbd daemonset. | 04:33 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Fix the kolla_toolbox module do not work on docker engine 1.10 https://review.openstack.org/422781 | 04:33 |
| portdirect | yeah - kfox you are the only person who really understands how that works | 04:33 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Optimize reconfigure action for nova https://review.openstack.org/407416 | 04:33 |
| kfox1111 | portdirect: if you have your ceph setup to work with kolla-kubernetes, | 04:33 |
| kfox1111 | it "just works(TM)" | 04:34 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla: install murano dashboard package in horizon https://review.openstack.org/395957 | 04:34 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Detect openstack_release variable when it auto https://review.openstack.org/422324 | 04:34 |
| portdirect | kfox1111: but we should have kolla-k8s ceph working with kolla-k8s :) | 04:34 |
| kfox1111 | I dont think kolla-kubernetes should provide ceph personally. | 04:34 |
| kfox1111 | ceph-rbd is just to help not have to install it on the k8s hosts. | 04:35 |
| portdirect | its a bit funny that the easyes way to get it running outside of the gate is to use a side project of mine :/ | 04:35 |
| kfox1111 | we usualy deploy or cephs on dedicated storage bricks. | 04:35 |
| *** sdake_ has joined #openstack-kolla | 04:35 | |
| sdake_ | portdirect was switching locations | 04:36 |
| sdake_ | portdirect I clicked through those 3 links | 04:36 |
| kfox1111 | to be perfectly honest, we would be a lot further along with that goal, if it wasn't for taking a long time to migrate to helm. | 04:36 |
| kfox1111 | I think it was worth the effort, but it did set things back quite a bit. | 04:37 |
| kfox1111 | the docs worked pretty well prehelm. | 04:37 |
| sdake_ | set some things back, other things forward | 04:37 |
| kfox1111 | true. | 04:37 |
| sdake_ | what got set back were the docs | 04:37 |
| kfox1111 | yeah. it basically threw out the docs. | 04:37 |
| sdake_ | what got set forward is it seems to me our community is growing rapidly now | 04:37 |
| sdake_ | vs was fairly stagnet | 04:37 |
| kfox1111 | and config's still in transition, | 04:38 |
| sdake_ | in the end, the docs are easy to fix | 04:38 |
| kfox1111 | so its not as easy as it shoudl be currently. :/ | 04:38 |
| sdake_ | the community aspect hard to fix - helm fixed that :) | 04:38 |
| kfox1111 | that is very true. | 04:38 |
| sdake_ | i dont care how *hard* it is, just that is is documented :) | 04:38 |
| sdake_ | and i'll do that | 04:38 |
| kfox1111 | yeah. thanks for that. :) | 04:38 |
| sdake_ | portdirect if you ahven't unplugged out of being maddened - could use a couple moer pointers on the ceph charty ou produced | 04:38 |
| sdake_ | portdirect i see the devenv installs ceph | 04:38 |
| sdake_ | portdirect (in the firstl ink) | 04:39 |
| portdirect | yeah - thats just to pull in rbd on the host | 04:39 |
| *** sdake has quit IRC | 04:39 | |
| sdake_ | how is that useful for the helm chart you wrote? | 04:39 |
| sdake_ | or is it just noise? | 04:39 |
| portdirect | it doesnt do anything eles with it | 04:39 |
| sdake_ | got it | 04:39 |
| sdake_ | so its essentially not helpful :) | 04:39 |
| kfox1111 | k8s needs rbd to mount volumes. | 04:39 |
| portdirect | you need it to *mount* ceph volumes with k8s | 04:39 |
| kfox1111 | so mariadb/rabbitmq need it. | 04:40 |
| kfox1111 | to back the pv's. | 04:40 |
| sdake_ | got it, so without pvs, there is no persistent storage | 04:40 |
| portdirect | yup | 04:40 |
| sdake_ | and without ceph installed in the devenv, thee is no way to mount the rbds which provide the pvs | 04:40 |
| sdake_ | or something to that effect :) | 04:40 |
| portdirect | pretty much | 04:41 |
| kfox1111 | right. | 04:41 |
| kfox1111 | or, | 04:41 |
| * sdake_ just woke up - brain booting | 04:41 | |
| kfox1111 | you can install ceph-rbd microservice. | 04:41 |
| sdake_ | portdirect how does your helm package help with that problem? | 04:41 |
| kfox1111 | it will setup a wrapper on the host that calls into the kolla container to do the rbd commands. | 04:41 |
| sdake_ | kfox1111 which ceph_rbd microservice? | 04:41 |
| kfox1111 | so ceph is only installed really in the container. | 04:41 |
| kfox1111 | helm/microservice/ceph-rbd-daemonset/ | 04:42 |
| portdirect | my helm package is for running a ceph storage cluster, you can use it with halcyon to use cep in a hyperconverged (buzzword) way | 04:42 |
| *** huikang has joined #openstack-kolla | 04:42 | |
| kfox1111 | we ran a cloud hyperconverged for a while. it ended badly. :/ | 04:43 |
| sdake_ | ok presented with two paths here | 04:43 |
| sdake_ | wondering which one is faster in terms of documenting | 04:43 |
| *** huikang_ has joined #openstack-kolla | 04:43 | |
| sdake_ | use helm/mciroservice/ceph-rbd-daemonset | 04:43 |
| sdake_ | or portdirect 's helm-ceph package | 04:43 |
| sdake_ | recommendations? | 04:43 |
| portdirect | kfox will know best | 04:43 |
| sdake_ | portdirect i'd like your opinion as well :) | 04:44 |
| portdirect | I would use mine - but thats only cause om not familair with the thing kfox built | 04:44 |
| kfox1111 | helm-ceph package is for all of ceph deployment? or just rbd on the host? | 04:44 |
| sdake_ | portdirect you understand the dev env in detail | 04:44 |
| portdirect | all of ceph | 04:44 |
| portdirect | kfox have you looked at it? | 04:44 |
| sdake_ | kfox1111 it took me 3 minutes to understand | 04:44 |
| kfox1111 | for all of ceph, I'd recommend either ceph-deploy or portdirec'ts ceph packages. | 04:44 |
| sdake_ | kfox1111 can you llook at repo - even though its super late? | 04:44 |
| portdirect | https://github.com/portdirect/ceph-helm#quckstart | 04:45 |
| kfox1111 | I'd use ceph-rbd just so that you don't have to install anything on the host. | 04:45 |
| kfox1111 | though I guess if you are using ceph deploy, that will happen anyway. | 04:45 |
| sdake_ | kfox1111 i am actually using virtual machines | 04:46 |
| kfox1111 | at least in a 3 node dev env. | 04:46 |
| sdake_ | kfox1111 so I am not installing anything on the host | 04:46 |
| portdirect | kfox1111: that deosnt help for dev | 04:46 |
| kfox1111 | sdake_: sorryu, we have multip[le definitions of "host" here. | 04:46 |
| portdirect | we dont need good ceph, we just need working ceph | 04:46 |
| kfox1111 | host being the vm's os's. | 04:46 |
| sdake_ | portdirect *5 | 04:46 |
| kfox1111 | true. | 04:46 |
| sdake_ | kfox1111 got it | 04:47 |
| sdake_ | kfox1111 I am not writing an operator guide, I am writing a dev guide | 04:47 |
| *** huikang has quit IRC | 04:47 | |
| sdake_ | kfox1111 operator guide comes later | 04:47 |
| kfox1111 | sdake_: part of the question though, is do devs want an env that sort of resembles a prod one? | 04:47 |
| kfox1111 | otherwise, issues arise with compatability. | 04:47 |
| portdirect | kfox1111: they often want to work on a laptop | 04:47 |
| sdake_ | kfox1111 right - and at pressent, we have a devenv that sets up 4 virtual machines | 04:47 |
| kfox1111 | laptop, right. me too. | 04:48 |
| sdake_ | kfox1111 not buy 4 bare metal machines | 04:48 |
| kfox1111 | sure. | 04:48 |
| kfox1111 | the vm's are a replacement physical ones. | 04:48 |
| kfox1111 | but do you deploy stuff in the vm's that look like what one would deploy on physical onesm, | 04:48 |
| sdake_ | ignoring the memory problem with laptops, lets assume the contrib has 32 gb of ram available in their server they aredoing work on | 04:48 |
| kfox1111 | or do you do different things, as its "just for dev"? | 04:48 |
| sdake_ | kfox1111 i think thati si an open question - i'd like to unify that at some point, but i'd like first to unblock dev | 04:49 |
| sdake_ | I've been blocked on dev for months because of $$DAYJOB meetings | 04:49 |
| kfox1111 | sdake_: its relevent right now though. | 04:49 |
| portdirect | kfox1111: lets go for MVP | 04:49 |
| kfox1111 | if you want something that looks like what someone would dpeploy on real hardware, I'd use ceph-deploy. | 04:49 |
| kfox1111 | if you want something just to get something up and running only for devs, I'd use helm-ceph. | 04:49 |
| portdirect | so why do we even have ceph charts then?! | 04:50 |
| sdake_ | kfox1111 cool - we can unify that later | 04:50 |
| kfox1111 | portdirect: initially, it was to support minikube, as it doesn't work with ceph deploy. | 04:50 |
| sdake_ | kfox1111 could you expand on that point please | 04:50 |
| kfox1111 | I never went back and came up with ceph-deploy instructions yet. | 04:50 |
| kfox1111 | minikube is a very quick/easy to deploy k8s in a vm. | 04:51 |
| kfox1111 | it runs a very stripped down os, so basically only can run containers. | 04:51 |
| kfox1111 | ceph-deploy installs rpms on the os. so can't install ceph on minikube's vm. | 04:51 |
| portdirect | I use it a lot, but this is kinda changing the target | 04:51 |
| kfox1111 | so I wrote a minimal set of test ceph charts to deploy something that woudl work in minikube. | 04:51 |
| openstackgerrit | zhubingbing proposed openstack/kolla: Fix designate dockerfile https://review.openstack.org/422366 | 04:52 |
| sdake_ | portdirect mind expanding - didnt' quite parse that | 04:52 |
| * kfox1111 shrugs. feel free to use whatever ceph deployment tool is easiest. | 04:52 | |
| sdake_ | kfox1111 - the reason i asked is I was curios - not passing judgement :) | 04:52 |
| kfox1111 | I just know how to use/debug ceph-deploy. so can help with that. | 04:52 |
| kfox1111 | sdake_: yeah. not offence taken. | 04:53 |
| portdirect | we have a k8s cluster, we want to be able to deploy kolla-k8s on it, is this possible using kolla-k8s deliverables? the current suggestion is no? | 04:53 |
| portdirect | but we use kolla-k8s ceph in the gate? | 04:53 |
| portdirect | so it must be yes | 04:53 |
| kfox1111 | depends on if you scope ceph in or not. | 04:53 |
| portdirect | this is where i get mentally blocked | 04:53 |
| kfox1111 | with external ceph, yes. | 04:54 |
| kfox1111 | with internal ceph, there be draggons. | 04:54 |
| sdake_ | internal ceph causes danger because why? | 04:54 |
| kfox1111 | the ceph server in kolla-kubernetes was intended only for testing. | 04:54 |
| kfox1111 | its not robust enough to be used for anything else. | 04:54 |
| portdirect | developers are only testing... they are running this on laptops and desktops not datacenters | 04:54 |
| portdirect | anyway - 4:55am gonna call it a night :) | 04:55 |
| portdirect | laters | 04:55 |
| * sdake_ sadfaces | 04:55 | |
| kfox1111 | it can be used for that, yheah. | 04:55 |
| sdake_ | night portdirect | 04:55 |
| kfox1111 | so there are 3 possibilites then. | 04:55 |
| *** portdirect is now known as portdirect_away | 04:55 | |
| *** sdake_ is now known as sdake | 04:55 | |
| kfox1111 | ceph-deploy, helm-ceph, and the test ceph in kolla-kubernetes. | 04:55 |
| kfox1111 | I only really created the kolla-kubernetes test ceph because of minikube restrictions. | 04:56 |
| kfox1111 | the dev environment does not have the restrictions that required the cration of the kolla-kubernetes test ceph. | 04:57 |
| sdake | kfox1111 cool - so which distro does minikube run? | 04:57 |
| *** krtaylor has joined #openstack-kolla | 04:57 | |
| kfox1111 | in fact, I'd like to switch the gate to ceph-deploy at some point. just hasn't been a priority. | 04:57 |
| kfox1111 | sdake: a stripped down one. havent' been able to identify it. | 04:57 |
| kfox1111 | no package manager I coudl find. | 04:57 |
| sdake | read only /var? | 04:58 |
| sdake | and read only /usr? | 04:58 |
| kfox1111 | its a modified boot2docker vm I think. | 04:58 |
| sdake | can minikube be made to run centos? | 04:58 |
| kfox1111 | no. it was intended just to be an easy way to run one command and get a working k8s. | 04:59 |
| kfox1111 | uses virtualbox by defauilt, | 04:59 |
| kfox1111 | so works on windows/mac/linux with very little effort. | 04:59 |
| kfox1111 | statically linked go binary so curl and go. | 05:00 |
| sdake | kfox1111 cool thanks - thats no good for a real deployment :) | 05:00 |
| kfox1111 | http://boot2docker.io/ says its based on tiny core linux. | 05:00 |
| kfox1111 | yeah. its just a dev tool. | 05:00 |
| kfox1111 | lets random dev spin up k8s really easy and do some stuff. then tare it all down just as easily. | 05:01 |
| sdake | kfox1111 mind parsing this and integrating it into your brain and spitting out what comes up:) | 05:01 |
| sdake | https://github.com/portdirect/ceph-helm#namespace-activation | 05:01 |
| kfox1111 | yeah. thats similar to the ceph setup scripts I pointed you at before. | 05:01 |
| sdake | kfox1111 which one in particular | 05:02 |
| sdake | kfox1111 I assume your headed to bed soon - so I'll dissect it this evening | 05:02 |
| sdake | I'm basically starting my work day early :) | 05:02 |
| kfox1111 | yeah. I gota head out very soon. | 05:02 |
| *** adrian_otto has quit IRC | 05:02 | |
| sdake | cool - which script in particular would be the parallel to what portdirect_away has there? | 05:02 |
| kfox1111 | really, I think you should try and parse through the setup_gate.sh script. | 05:03 |
| kfox1111 | it does everythying from a fresh installed vm all the way up. | 05:03 |
| sdake | i have 4 vms not 1 | 05:03 |
| sdake | and it come sfrom teh dev env | 05:03 |
| kfox1111 | tools/setup-ceph-secrets.sh is the closest I think. | 05:03 |
| sdake | i will parse through setup_gate.sh | 05:03 |
| kfox1111 | there isn't much differenced between 4 nodes and 1 in k8s. | 05:03 |
| sdake | id like to integrate the code in setup_gate.sh and children | 05:03 |
| sdake | and what we have in the devenv documentation | 05:03 |
| sdake | into "now my devenv is seutp, whats next" | 05:04 |
| kfox1111 | setup_gate.sh has a 2node setup, which should be basically the same for 4node. | 05:04 |
| kfox1111 | with just 2 more compute nodes. | 05:04 |
| sdake | what other storage can be used as a backing store for kubernetes? | 05:04 |
| sdake | iscsi/nfs/etc? | 05:05 |
| kfox1111 | yes. | 05:05 |
| sdake | is there a list somewhere? | 05:05 |
| kfox1111 | basically setup_gate.sh is the distilled knowlege of everything I know about deploying. :) | 05:05 |
| kfox1111 | so if I'm not around, thats the place to look. :) | 05:05 |
| kfox1111 | yeah. | 05:05 |
| sdake | kfox1111 i understnd that totally | 05:05 |
| sdake | i am going to examine that deeply | 05:05 |
| sdake | kfox1111 here is the deal - today people use the gate to test kolla-kubernetes | 05:06 |
| kfox1111 | sdake: https://kubernetes.io/docs/user-guide/volumes/ | 05:06 |
| sdake | rather then doing dev on their personal boxes | 05:06 |
| sdake | kfox1111 thats a huge problem - need people to be able to dev on their own boxes | 05:06 |
| sdake | waiting on the gate is slow | 05:06 |
| kfox1111 | I get that. | 05:06 |
| sdake | i am here to fix it :) | 05:06 |
| kfox1111 | thats why I started breaking upt he gate into more reusable chunks. | 05:06 |
| kfox1111 | though never tried it after breaking it up. | 05:06 |
| kfox1111 | so not sure just how reusable the chunks are. | 05:06 |
| sdake | kfox1111 thanks for link on the storage types | 05:07 |
| kfox1111 | if you find thins need some rearanging to be more reusable in the docs, lets fix it. | 05:07 |
| kfox1111 | gota head out. | 05:07 |
| sdake | kfox1111 thanks dude - goodnight :) | 05:07 |
| kfox1111 | good night. :) | 05:07 |
| *** sacharya_ has quit IRC | 05:10 | |
| *** sacharya has joined #openstack-kolla | 05:11 | |
| *** inc0 has quit IRC | 05:15 | |
| *** msimonin1 has left #openstack-kolla | 05:19 | |
| openstackgerrit | Steven Dake proposed openstack/kolla-kubernetes: add a cd to halcyon-vagrant-kubernetes https://review.openstack.org/423039 | 05:22 |
| *** imcsk8_ has joined #openstack-kolla | 05:30 | |
| *** imcsk8 has quit IRC | 05:33 | |
| *** mcnanci has quit IRC | 05:35 | |
| openstackgerrit | Merged openstack/kolla: Fix aodh dockerfile https://review.openstack.org/421645 | 05:36 |
| *** mcnanci has joined #openstack-kolla | 05:43 | |
| *** sp___ has quit IRC | 05:45 | |
| *** zhurong has joined #openstack-kolla | 05:49 | |
| openstackgerrit | Steven Dake proposed openstack/kolla-kubernetes: Install kubectl on the host OS in the development environment docs https://review.openstack.org/423046 | 05:50 |
| openstackgerrit | Merged openstack/kolla: Fix gnocchi dockerfile https://review.openstack.org/421648 | 06:10 |
| *** tonanhngo has joined #openstack-kolla | 06:12 | |
| *** tonanhngo has quit IRC | 06:13 | |
| openstackgerrit | Steven Dake proposed openstack/kolla-kubernetes: Install kubectl and helm on the host OS in docs https://review.openstack.org/423046 | 06:13 |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Unify haproxy prechecks condition order https://review.openstack.org/415510 | 06:15 |
| openstackgerrit | Steven Dake proposed openstack/kolla-kubernetes: add a cd to halcyon-vagrant-kubernetes https://review.openstack.org/423039 | 06:19 |
| openstackgerrit | Steven Dake proposed openstack/kolla-kubernetes: Install kubectl and helm on the host OS in docs https://review.openstack.org/423046 | 06:19 |
| openstackgerrit | Steven Dake proposed openstack/kolla-kubernetes: Add an overview section to the documentation https://review.openstack.org/423067 | 06:19 |
| *** absubram has joined #openstack-kolla | 06:19 | |
| * sdake groans at the state of the documentation across Kolla | 06:19 | |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Unify haproxy prechecks condition order https://review.openstack.org/415510 | 06:22 |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Unify haproxy prechecks condition order https://review.openstack.org/415510 | 06:25 |
| *** huikang_ has quit IRC | 06:25 | |
| *** qiliang27 has quit IRC | 06:30 | |
| *** sayantani01 has quit IRC | 06:31 | |
| *** qiliang27 has joined #openstack-kolla | 06:32 | |
| openstackgerrit | zhubingbing proposed openstack/kolla: Fix designate dockerfile https://review.openstack.org/422366 | 06:36 |
| sdake | duonghq you about? | 06:37 |
| *** satyar has joined #openstack-kolla | 06:44 | |
| openstackgerrit | zhubingbing proposed openstack/kolla: remove /var/log/apache on keystone dockerfile https://review.openstack.org/423082 | 06:46 |
| duonghq | sdake, afternoon | 06:47 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Optimize reconfiguration for tacker https://review.openstack.org/422036 | 06:53 |
| duonghq | sdake, still cannot figure out why docker 1.13 destroy the gate | 06:54 |
| sdake | duonghq Jeffrey4l_ has fixed it as I understand by pinning to an older version of docker | 06:54 |
| duonghq | sdake, yup, but as I told, we should not realize on this workaround | 06:55 |
| sdake | duonghq quick quesiton, when you do kolla-kubernetes devlopment, do you use your personal box | 06:55 |
| Jeffrey4l_ | duonghq, i have no idea too. so just pin the docker in the gate and try to fix it later. | 06:55 |
| sdake | duonghq ya the workaround is not good | 06:55 |
| sdake | Jeffrey4l_ i think one thing worth trying is turning off selinux | 06:55 |
| Jeffrey4l_ | sdake, tried. not work. | 06:55 |
| sdake | Jeffrey4l_ or alternately check the audit logs | 06:55 |
| sdake | Jeffrey4l_ ok -worth a shot ;) | 06:55 |
| Jeffrey4l_ | sdake, here https://review.openstack.org/422786 | 06:55 |
| duonghq | sdake, currently, yes, my personal PC, but once the vagrant is up and running, it's very risky to turn off the box due to Internet connection can go worse anytime | 06:56 |
| sdake | duonghq cool - cn you ell me how you get tiller server running | 06:56 |
| sdake | duonghq I am getting an image pull error in the dev env | 06:56 |
| sdake | rather helm server running | 06:56 |
| sdake | duonghq i am on a mission to document the entire "dev process" for kolla-kubernetes | 06:57 |
| sdake | duonghq to help bring others up to speed | 06:57 |
| sdake | duonghq we have a good document already in the dev env | 06:57 |
| duonghq | sdake, I saw, | 06:57 |
| *** duonghq has left #openstack-kolla | 06:57 | |
| *** duonghq has joined #openstack-kolla | 06:57 | |
| sdake | duonghq coud use some assistancein the form of training :) | 06:57 |
| duonghq | I simply do vagrant destroy and vagrant up, until the ansible do not failed anything | 06:58 |
| sdake | duonghq at present, tiller is on an image bckoff loop | 06:58 |
| sdake | duonghq do you ahve any custom config, or running helm 2.1.0 as is defined in the halycon repo? | 06:58 |
| duonghq | after tired with vagrant destroy and up, I run the Ansbile directly | 06:58 |
| sdake | duonghq I am using helm 2.1.3 client | 06:58 |
| * duonghq use k8s 1.5.1 w/ helm 2.1.3 both for server and client | 06:58 | |
| jascott1 | sdake I have seen that when I install halycon vagrant without internet access (proxy set not on vpn etc) | 06:59 |
| sdake | duonghq the dev env sets up helm 2.1.0 by default | 06:59 |
| sdake | jascott1 i am on a vpn-free network | 06:59 |
| *** portdirect_away is now known as portdirect | 06:59 | |
| portdirect | centos was busted last week | 06:59 |
| sdake | jascott1 all other services are pulled correctly but tiller is not | 06:59 |
| openstackgerrit | zhubingbing proposed openstack/kolla: Fix manila-base dockerfile format https://review.openstack.org/423089 | 07:00 |
| portdirect | pulling from gcr.io | 07:00 |
| sdake | portdirect i pulled a fresh repo an hour ago | 07:00 |
| duonghq | sdake, I custom both config.rb and halcyon kubernetes | 07:00 |
| sdake | duonghq please past econfig.rb | 07:00 |
| sdake | portdirect the last time i tried the dev env (dec) it was working well with helm | 07:01 |
| portdirect | :( yeah - they had that as a priority fix so it should be in | 07:01 |
| portdirect | it took me down for a day | 07:01 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Fix ironic inspector https://review.openstack.org/420122 | 07:01 |
| duonghq | sdake, http://paste.openstack.org/show/595733/ here | 07:02 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Fix ironic inspector https://review.openstack.org/380774 | 07:02 |
| sdake | portdirect mind taking a look at my system? | 07:02 |
| portdirect | np | 07:02 |
| sdake | portdirect or are you sleeping :) | 07:02 |
| portdirect | lol - just change the fuel cell | 07:03 |
| *** zhurong has quit IRC | 07:03 | |
| duonghq | sdake, and the all.yaml of halcyon-kubernetes http://paste.openstack.org/show/595734/ | 07:03 |
| *** JeroenBo has joined #openstack-kolla | 07:06 | |
| *** absubram has quit IRC | 07:08 | |
| sdake | duonghq thanks :) | 07:09 |
| *** salv-orlando has joined #openstack-kolla | 07:09 | |
| *** Jeffrey4l_ has quit IRC | 07:10 | |
| openstackgerrit | qiankun.zhang proposed openstack/kolla: Install tacker-horizon into horizon image https://review.openstack.org/407810 | 07:11 |
| duonghq | sdake, I re-run halcyon-kubernetes (the ansible one), many time, and it successes magically | 07:12 |
| jascott1 | thats ansible isnt it | 07:12 |
| sdake | jascott1 your up late too i see :) | 07:13 |
| jascott1 | aw yeah | 07:13 |
| *** salv-orlando has quit IRC | 07:14 | |
| jascott1 | usually doesnt feel late to me until about 1 | 07:14 |
| sdake | i slept in the middle of the day so i'll be awake until cob tomorrow :( | 07:14 |
| jascott1 | yeah I had to take a lil nap myself | ) | 07:14 |
| sdake | duonghq magic success is not good :( | 07:15 |
| sdake | jascott1 my nap was more like 5 or 6 hrs | 07:15 |
| jascott1 | haha nice. power coma | 07:15 |
| sdake | jascott1 although my emploeyr doesn't care when i work | 07:15 |
| sdake | jascott1 i dont work on a clock, ratheri work on projects | 07:15 |
| jascott1 | yeah thats great | 07:15 |
| jascott1 | we get a lot of freedom but its still centers around a more or less 9-5 life | 07:16 |
| jascott1 | i envy you wrt to scheduling | 07:16 |
| duonghq | sdake, I have meeting for about 20mins, cya | 07:17 |
| jascott1 | i hate having to make myself 'go to sleep' for no good reason (other than maintaining employment optics) when im in the zone | 07:17 |
| sdake | thanks again duonghq | 07:17 |
| duonghq | sdake, my pleasure | 07:17 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Add zun ansible role https://review.openstack.org/416179 | 07:17 |
| sdake | jascott1 portdirect if you fellas get a chance can you review the last 3 patches that hit gerrit related to docs and kolla-kubernetes deliverable? | 07:25 |
| *** zhubingbing_ has quit IRC | 07:29 | |
| jascott1 | looking good, agree with portdirect and will take another look on next revs | 07:37 |
| * duonghq back | 07:41 | |
| *** pcaruana has joined #openstack-kolla | 07:43 | |
| duonghq | portdirect, can you review this: https://review.openstack.org/#/c/423016/1 | 07:43 |
| *** afranc has joined #openstack-kolla | 07:44 | |
| *** bmace has quit IRC | 07:46 | |
| *** harbie_ has joined #openstack-kolla | 07:48 | |
| *** Aju has quit IRC | 07:51 | |
| *** dgonzalez has quit IRC | 07:51 | |
| *** harbie has quit IRC | 07:51 | |
| *** bmace_ has joined #openstack-kolla | 07:51 | |
| *** dgonzalez has joined #openstack-kolla | 07:51 | |
| *** harbie_ is now known as harbie | 07:51 | |
| *** harbie has quit IRC | 07:51 | |
| *** harbie has joined #openstack-kolla | 07:51 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Add block for dumb-init installation https://review.openstack.org/422249 | 07:53 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Code coverage improvement for kolla https://review.openstack.org/422444 | 07:55 |
| *** matrohon has joined #openstack-kolla | 07:57 | |
| portdirect | duonghq: looking now | 07:57 |
| jascott1 | sdake i tried to make the one about obliterating your os less scary | 07:58 |
| duonghq | thank portdirect | 07:58 |
| portdirect | duonghq: is is easy for you to make another ps? | 07:59 |
| duonghq | portdirect, sorry but do not get your idea | 07:59 |
| portdirect | can you edit that patchset easily? | 08:00 |
| portdirect | I have a change i would like but its not major | 08:00 |
| duonghq | sure | 08:00 |
| portdirect | sweet :) | 08:01 |
| portdirect | would be best to install horizon last | 08:01 |
| portdirect | as we dont have any test for it - and it should be really straightforward | 08:02 |
| portdirect | other than that looks great :) | 08:02 |
| duonghq | portdirect, sure, I thought about this before, I push horizon first just for alphabetical order :) | 08:02 |
| duonghq | moment | 08:02 |
| openstackgerrit | Duong Ha-Quang proposed openstack/kolla-kubernetes: Horizon service package https://review.openstack.org/423016 | 08:04 |
| portdirect | jascott1: you work with any of the sdi crew? | 08:04 |
| openstackgerrit | Steven Dake proposed openstack/kolla-kubernetes: add a cd to halcyon-vagrant-kubernetes https://review.openstack.org/423039 | 08:05 |
| openstackgerrit | Steven Dake proposed openstack/kolla-kubernetes: Install kubectl and helm on the host OS in docs https://review.openstack.org/423046 | 08:05 |
| jascott1 | i do not but have met a few | 08:05 |
| jascott1 | inc0 knows some of them peeps | 08:05 |
| jascott1 | ^portdirect | 08:06 |
| portdirect | nice - I'll need to hit him up - would like to find out more about snap | 08:06 |
| jascott1 | ya i have looked at it but havent tried it | 08:07 |
| portdirect | thoughts? of literally just looked | 08:07 |
| sdake | portdirect do you think pinning to docker 0.12.z will fix all the tagging things you did there? | 08:08 |
| portdirect | yeah should do | 08:08 |
| sdake | weave and calico | 08:09 |
| sdake | I noticed they are in all.yaml | 08:09 |
| jascott1 | seems great, it seems their is some interest and ive read some good impressions | 08:09 |
| sdake | aren't they competing things? | 08:09 |
| portdirect | though I'm kinda temped to "say screw you centos" and go upstream.. | 08:09 |
| sdake | its probably a docker problem not a centos problem | 08:09 |
| portdirect | we cover all the major cni plugins | 08:09 |
| portdirect | no this one is deffo centos | 08:10 |
| sdake | do we knwo what the root cause is? | 08:10 |
| portdirect | cant fint the bug report, but it's theirs | 08:10 |
| sdake | we are not using docker from centos are we? | 08:10 |
| portdirect | halcyon does | 08:10 |
| sdake | i see | 08:10 |
| portdirect | as thats what k8s recommends | 08:10 |
| *** salv-orlando has joined #openstack-kolla | 08:11 | |
| sdake | rather then docker-engine? | 08:11 |
| portdirect | though the dont hav very strong feelings on the matter | 08:11 |
| portdirect | *they dont | 08:11 |
| portdirect | not really | 08:11 |
| sdake | kolla-ansible recommends docker-engine | 08:11 |
| portdirect | the acutaly prefer older dockers | 08:11 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Load murano dashboard dynamic https://review.openstack.org/402156 | 08:11 |
| sdake | ya, 1.10 is what i typically run :) | 08:11 |
| sdake | since it doesn't have all the bloat | 08:11 |
| portdirect | 0.14 still on one of my boxes | 08:12 |
| sdake | rather 0.10 i mean | 08:12 |
| portdirect | dude you need to upgrade more :) | 08:12 |
| sdake | i got enuf upgrades going on in my life :) | 08:12 |
| sdake | [sdake@minime-03 halcyon-vagrant-kubernetes]$ docker --version | 08:13 |
| sdake | Docker version 1.12.6, build 78d1802 | 08:13 |
| portdirect | 0.14 is like 2.5 years old | 08:13 |
| sdake | that is docker-engine | 08:13 |
| sdake | ya i mean 1.10 :) | 08:13 |
| sdake | -ETOOMANYNUMBERS | 08:13 |
| satyar | hi sdake | 08:14 |
| jascott1 | anyone got a good map of compute kit interdependencies? working on subchart enabling/disabling via values | 08:14 |
| sdake | jascott1 in my head have a map :) | 08:14 |
| *** shardy has joined #openstack-kolla | 08:14 | |
| jascott1 | do you by chance also have a data jack or floppy drive in there? | 08:15 |
| *** n0isyn0i1e has quit IRC | 08:15 | |
| jascott1 | ya didnt know if there is some well known openstack website that shows this stuff, i have it all long hand and its not that complicated but prior art would be great | 08:16 |
| portdirect | is compute-kit a real term? | 08:16 |
| portdirect | its a good one :) | 08:17 |
| *** n0isyn0ise has joined #openstack-kolla | 08:17 | |
| jascott1 | its a legit kolla term | 08:17 |
| portdirect | that works for me :) | 08:17 |
| portdirect | keystone, glance, nova, neutron? | 08:17 |
| portdirect | cinder? | 08:17 |
| jascott1 | sure! | 08:18 |
| openstackgerrit | Steven Dake proposed openstack/kolla-kubernetes: add a cd to halcyon-vagrant-kubernetes https://review.openstack.org/423039 | 08:20 |
| openstackgerrit | Steven Dake proposed openstack/kolla-kubernetes: Install kubectl and helm on the host OS in docs https://review.openstack.org/423046 | 08:20 |
| openstackgerrit | Steven Dake proposed openstack/kolla-kubernetes: Add an overview section to the documentation https://review.openstack.org/423067 | 08:20 |
| sdake | jascott1 welcome to the matrix :) | 08:20 |
| sdake | portdirect yes compute kit is a legit openstack term | 08:20 |
| portdirect | what does it cover? | 08:20 |
| portdirect | the ones i meantiond i assume | 08:21 |
| sdake | https://governance.openstack.org/tc/reference/tags/starter-kit_compute.html | 08:21 |
| portdirect | nice | 08:21 |
| sdake | portdirect how do I go about delting this stable/mysql chart | 08:25 |
| sdake | nm figured it out | 08:26 |
| *** shardy is now known as shardy_afk | 08:26 | |
| *** klindgren has quit IRC | 08:27 | |
| *** klindgren has joined #openstack-kolla | 08:27 | |
| *** dasTor has quit IRC | 08:27 | |
| *** dasTor has joined #openstack-kolla | 08:28 | |
| openstackgerrit | zhang kun proposed openstack/kolla-ansible: Fix Bootstraping the mongodb replication set https://review.openstack.org/422377 | 08:30 |
| *** tovin07 has quit IRC | 08:30 | |
| *** tovin07 has joined #openstack-kolla | 08:30 | |
| openstackgerrit | Steven Dake proposed openstack/kolla-kubernetes: Clean up mysql installed during test of helm https://review.openstack.org/423120 | 08:30 |
| openstackgerrit | Surya Prakash Singh proposed openstack/kolla: Code coverage improvement for kolla https://review.openstack.org/422444 | 08:31 |
| *** harmw has quit IRC | 08:31 | |
| *** harmw has joined #openstack-kolla | 08:32 | |
| sdake | ok - so the development environment works, with a pin on docker - if portdirect can sort that out | 08:33 |
| ScoopIcicle | Pursuing this route will enable us to differentiate a predictability; this is why escalations strengthen the Chief Business Planning Officer. Controlling should prioritize hyper-hybrid, bi-face, value chains, while environments challenge us to hunt the business down. | 08:34 |
| sdake | ScoopIcicle should be called "things that shadowman wouldn't say" | 08:35 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Fix docker restart policy issue on docker engine 1.13 https://review.openstack.org/423122 | 08:35 |
| sdake | portdirect how would I change th eCNI plugin? | 08:36 |
| portdirect | i find him strangely motivational | 08:37 |
| portdirect | you cant easily once cluster is up | 08:37 |
| portdirect | but | 08:37 |
| sdake | how about prior | 08:37 |
| portdirect | this is the list of supported ones https://github.com/att-comdev/halcyon-kubernetes/tree/master/kube-deploy/roles/kube-sdn/tasks | 08:38 |
| portdirect | and hes where you do it: https://github.com/att-comdev/halcyon-kubernetes/blob/master/kube-deploy/group_vars/all.yml#L34 | 08:39 |
| *** egonzalez has joined #openstack-kolla | 08:42 | |
| sdake | sup egonzalez | 08:43 |
| sdake | portdirect ok so dev environment is up, only 3 nodes came up (one had an error during initialization iirc so that is to be expected) | 08:46 |
| sdake | and I think the dev env docs are in good shape now with my new commits | 08:47 |
| sdake | rahter git reviews | 08:47 |
| sdake | tested and working with whtever hackery you did on my box ;-) | 08:47 |
| portdirect | stuxnet | 08:47 |
| sdake | which i think was pulling a speicla tiller version and changing from weave to calico | 08:47 |
| sdake | should be document those changes in the docs? | 08:47 |
| portdirect | not a special tiller, just pulled from a different domain | 08:48 |
| sdake | now - the next set of problems | 08:48 |
| portdirect | //logout | 08:48 |
| sdake | do I absolutely need ceph for kolla-kubernetes to run? | 08:48 |
| * sdake grins | 08:48 | |
| portdirect | right | 08:49 |
| portdirect | i dont think you do | 08:49 |
| jascott1 | ScoopIcicle: you are too serious! need more silly jibberish | 08:49 |
| portdirect | but have absolutely no idea if im right, or how to run without | 08:49 |
| portdirect | jascott1: is ScoopIcicle yours? | 08:50 |
| duonghq | portdirect, can you review my ps (again): https://review.openstack.org/#/c/423016/ | 08:50 |
| jascott1 | nope | 08:50 |
| sdake | why do we have wait for pods in the service script | 08:51 |
| sdake | shoudln't entrypoint manage that? | 08:51 |
| portdirect | kfox likes em | 08:52 |
| portdirect | it does also help pace the gate | 08:52 |
| *** shardy_afk is now known as shardy | 08:52 | |
| sdake | i get why in the microservices | 08:52 |
| jascott1 | sdake where you at in the code? we had gate before all entrypoints maybe? | 08:52 |
| sdake | ya - i'm just wondering if they can safely be removed | 08:52 |
| sdake | jascott1 ^^ | 08:53 |
| portdirect | i think they probably can | 08:53 |
| portdirect | but we would need one at the end | 08:53 |
| portdirect | before running functional tests | 08:53 |
| *** bjolo has joined #openstack-kolla | 08:53 | |
| bjolo | morning | 08:53 |
| duonghq | morning bjolo | 08:53 |
| bjolo | hi duonghq | 08:54 |
| bjolo | how are ya? | 08:54 |
| duonghq | totally fine | 08:54 |
| portdirect | hey bjolo, I'm just frying some butter in old chip oil, you want any? | 08:54 |
| openstackgerrit | Yao Lu proposed openstack/kolla: Change docker restart policy https://review.openstack.org/423026 | 08:55 |
| bjolo | sounds yummy | 08:55 |
| bjolo | :) | 08:55 |
| portdirect | ;) | 08:55 |
| bjolo | whos going to the PTG? | 08:55 |
| openstackgerrit | Eduardo Gonzalez proposed openstack/kolla-ansible: Optimize reconfiguration for tacker https://review.openstack.org/422036 | 08:55 |
| portdirect | not sure, holding out hope for good news today | 08:55 |
| duonghq | the TSP result is not sent yet? | 08:55 |
| jascott1 | portdirect haha me too | 08:56 |
| jascott1 | but from upper mgmt | 08:56 |
| jascott1 | ya its T-1month right | 08:56 |
| *** salv-orlando has quit IRC | 08:56 | |
| portdirect | bjolo: you going? | 08:56 |
| bjolo | looks like it | 08:56 |
| duonghq | nice | 08:57 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Fix the kolla_toolbox module do not work on docker engine 1.10 https://review.openstack.org/422781 | 08:57 |
| portdirect | nice :) your site up and running yet? | 08:57 |
| portdirect | or still in dev? | 08:57 |
| bjolo | final final betatesting | 08:57 |
| bjolo | fricking neutron issues | 08:57 |
| bjolo | enabling MTU 9000 everywhere in the configs, but even so the linux bridges comes up with mtu 1500 | 08:58 |
| portdirect | :( | 08:58 |
| portdirect | least that not *too* bad | 08:58 |
| bjolo | nova also has one issue. VMs dont autostart when compute node reboots | 08:59 |
| portdirect | I've had to tune mtu in the source before | 08:59 |
| sdake | bjolo good to hear your close :) | 08:59 |
| *** matrohon has quit IRC | 08:59 | |
| sdake | bjolo that is the way nova behaves I believe | 08:59 |
| bjolo | sdake, still up huh! | 08:59 |
| bjolo | well there is a config flag for it, but does not seem to take it | 08:59 |
| sdake | bjolo nah I slept at 3pm, woke up 8 or 9pm | 08:59 |
| jascott1 | speaking of, have a good one folks! | 08:59 |
| sdake | bjolo so I guess I had a compressed awake schedule on thursday :) | 09:00 |
| sdake | later jascott1 | 09:00 |
| jascott1 | portdirect, all good luck on PTG stuff | 09:00 |
| sdake | jascott1 ya ptg is feb 24 iirc | 09:00 |
| *** jascott1 has quit IRC | 09:00 | |
| portdirect | cheers bro :) | 09:01 |
| portdirect | hope to meet you and duonghq there :) | 09:01 |
| *** prameswar has quit IRC | 09:01 | |
| duonghq | portdirect, /me too | 09:01 |
| *** openstackgerrit has quit IRC | 09:02 | |
| bjolo | anyone worked with kolla tempest? i.e. how to actually run the tests? | 09:05 |
| egonzalez | also waiting for TSP, first round was denied | 09:05 |
| duonghq | egonzalez, did you receive explicit denial email? | 09:08 |
| duonghq | iirc, it'was just postponed | 09:08 |
| egonzalez | just saying that we will be in touch mid january | 09:09 |
| duonghq | ya | 09:10 |
| egonzalez | berendt: receive his tickets at the end of DEC | 09:10 |
| egonzalez | lucky guy ;) | 09:10 |
| duonghq | seems that not many people as lucky as berendt | 09:10 |
| *** openstackgerrit has joined #openstack-kolla | 09:11 | |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Optimize reconfigure action for gnocchi https://review.openstack.org/423133 | 09:11 |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Optimize reconfigure action for rabbitmq https://review.openstack.org/422705 | 09:12 |
| ScoopIcicle | There is no alternative to pillars. | 09:13 |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Optimize reconfiguration for heat https://review.openstack.org/421184 | 09:14 |
| bjolo | one would figure that tempest verify-config should not stacktrace at least :) | 09:16 |
| *** Serlex has joined #openstack-kolla | 09:16 | |
| *** gfidente has joined #openstack-kolla | 09:21 | |
| *** gfidente has quit IRC | 09:21 | |
| *** gfidente has joined #openstack-kolla | 09:21 | |
| *** tovin07 has left #openstack-kolla | 09:21 | |
| *** dgonzalez_ has joined #openstack-kolla | 09:23 | |
| *** msimonin has joined #openstack-kolla | 09:23 | |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Update the upgrade.yml for trove https://review.openstack.org/423142 | 09:25 |
| *** dgonzalez has quit IRC | 09:27 | |
| *** dgonzalez_ is now known as dgonzalez | 09:27 | |
| openstackgerrit | Merged openstack/kolla: Modify punctuation https://review.openstack.org/422949 | 09:28 |
| *** jascott1 has joined #openstack-kolla | 09:33 | |
| *** matrohon has joined #openstack-kolla | 09:44 | |
| sdake | well i've hit the wall yet again | 09:48 |
| sdake | see ya all later :) | 09:48 |
| *** zhubingbing has joined #openstack-kolla | 09:48 | |
| duonghq | sdake, night | 09:49 |
| *** sdake has quit IRC | 09:52 | |
| *** sdake has joined #openstack-kolla | 09:53 | |
| *** sacharya has quit IRC | 09:57 | |
| *** sacharya has joined #openstack-kolla | 09:58 | |
| *** hfu has quit IRC | 10:00 | |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Optimize reconfiguration for kibana https://review.openstack.org/423158 | 10:00 |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Optimize reconfiguration for kibana https://review.openstack.org/423158 | 10:02 |
| *** athomas has joined #openstack-kolla | 10:10 | |
| *** zhubingbing has quit IRC | 10:12 | |
| *** zhubingbing has joined #openstack-kolla | 10:13 | |
| *** ingmar23 has joined #openstack-kolla | 10:13 | |
| *** sambetts|afk is now known as sambetts | 10:13 | |
| openstackgerrit | Surya Prakash Singh proposed openstack/kolla: Code coverage improvement for kolla https://review.openstack.org/422444 | 10:14 |
| *** ingmar23 has quit IRC | 10:14 | |
| *** ingmar23 has joined #openstack-kolla | 10:15 | |
| *** satyar has quit IRC | 10:19 | |
| *** duonghq has quit IRC | 10:20 | |
| *** zhubingbing has quit IRC | 10:20 | |
| *** sacharya has quit IRC | 10:23 | |
| *** sacharya has joined #openstack-kolla | 10:24 | |
| *** cuongnv has quit IRC | 10:26 | |
| *** saneax is now known as saneax-_-|AFK | 10:31 | |
| *** prameswar has joined #openstack-kolla | 10:31 | |
| *** tovin07_ has quit IRC | 10:36 | |
| openstackgerrit | Wei Cao proposed openstack/kolla: Add dragonflow container https://review.openstack.org/418193 | 10:40 |
| *** Jeffrey4l has joined #openstack-kolla | 10:40 | |
| Jeffrey4l | hi egonzalez this bp is implemented. | 10:42 |
| Jeffrey4l | https://blueprints.launchpad.net/kolla-ansible/+spec/octavia-ansible-role | 10:42 |
| Jeffrey4l | but could u approve it? egonzalez | 10:42 |
| *** jascott1_ has joined #openstack-kolla | 10:43 | |
| egonzalez | done, thanks | 10:43 |
| *** jascott1_ has quit IRC | 10:48 | |
| *** ipsecguy_ has quit IRC | 10:51 | |
| *** zhangkun_ has quit IRC | 10:52 | |
| *** msimonin has left #openstack-kolla | 10:57 | |
| Jeffrey4l | thanks. | 11:01 |
| *** salv-orlando has joined #openstack-kolla | 11:06 | |
| *** ScoopIcicle has quit IRC | 11:07 | |
| *** salv-orlando has quit IRC | 11:24 | |
| *** ipsecguy has joined #openstack-kolla | 11:32 | |
| openstackgerrit | Wei Cao proposed openstack/kolla: Add dragonflow container https://review.openstack.org/418193 | 11:32 |
| openstackgerrit | Mauricio Lima proposed openstack/kolla-ansible: Unify haproxy prechecks condition order https://review.openstack.org/415510 | 11:32 |
| openstackgerrit | Surya Prakash Singh proposed openstack/kolla: Code coverage improvement for kolla https://review.openstack.org/422444 | 11:40 |
| mliima | morning guys | 11:46 |
| bjolo | hi mliima | 11:54 |
| *** pcaruana has quit IRC | 11:59 | |
| openstackgerrit | Christian Berendt proposed openstack/kolla-ansible: Remove Ubuntu 14.04 specific precheck https://review.openstack.org/423231 | 12:00 |
| *** hfu has joined #openstack-kolla | 12:01 | |
| *** caowei has quit IRC | 12:02 | |
| *** saneax-_-|AFK is now known as saneax | 12:02 | |
| *** msimonin1 has joined #openstack-kolla | 12:03 | |
| openstackgerrit | Eduardo Gonzalez proposed openstack/kolla-ansible: Remove hardcoded designate backend https://review.openstack.org/423233 | 12:03 |
| *** jtriley has quit IRC | 12:04 | |
| *** hfu has quit IRC | 12:05 | |
| *** pcaruana has joined #openstack-kolla | 12:06 | |
| sdake | morning folks | 12:14 |
| egonzalez | morning sdake | 12:14 |
| *** bjolo has quit IRC | 12:14 | |
| *** bjolo_ has joined #openstack-kolla | 12:14 | |
| *** sdake has quit IRC | 12:18 | |
| *** sdake_ has joined #openstack-kolla | 12:19 | |
| openstackgerrit | Sajauddin proposed openstack/kolla: Implements: blueprint kolla on power server https://review.openstack.org/423239 | 12:24 |
| openstackgerrit | Sajauddin proposed openstack/kolla: Implements: blueprint kolla on power server https://review.openstack.org/423239 | 12:29 |
| *** rhallisey has joined #openstack-kolla | 12:30 | |
| *** shardy is now known as shardy_lunch | 12:36 | |
| openstackgerrit | Merged openstack/kolla-kubernetes: Remove support for py33 https://review.openstack.org/422342 | 12:37 |
| *** slagle_ is now known as slagle | 12:37 | |
| *** mattmceuen has joined #openstack-kolla | 12:40 | |
| openstackgerrit | Eduardo Gonzalez proposed openstack/kolla: Remove empty spaces dind dockerfile https://review.openstack.org/423249 | 12:43 |
| *** hfu has joined #openstack-kolla | 12:50 | |
| *** mgiles has joined #openstack-kolla | 12:51 | |
| openstackgerrit | Mark Giles proposed openstack/kolla-kubernetes: Nova-compute service deployment https://review.openstack.org/422688 | 12:51 |
| *** JoseMello has joined #openstack-kolla | 12:52 | |
| *** hfu has quit IRC | 12:53 | |
| *** satyar has joined #openstack-kolla | 12:54 | |
| *** hfu has joined #openstack-kolla | 12:56 | |
| *** hfu has quit IRC | 12:58 | |
| *** hfu has joined #openstack-kolla | 13:00 | |
| *** Jeffrey4l_ has joined #openstack-kolla | 13:00 | |
| *** hfu has quit IRC | 13:01 | |
| *** hfu has joined #openstack-kolla | 13:02 | |
| *** Jeffrey4l has quit IRC | 13:04 | |
| *** hfu has quit IRC | 13:04 | |
| *** athomas has quit IRC | 13:10 | |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: WIP: Add a simple test to see if horizon is running https://review.openstack.org/422932 | 13:14 |
| *** caowei has joined #openstack-kolla | 13:14 | |
| *** sdake has joined #openstack-kolla | 13:14 | |
| *** shardy_lunch is now known as shardy | 13:14 | |
| *** athomas has joined #openstack-kolla | 13:16 | |
| *** sdake_ has quit IRC | 13:17 | |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: WIP: test rebooting in the gate https://review.openstack.org/413193 | 13:17 |
| openstackgerrit | Mauricio Lima proposed openstack/kolla: introduces OVS with DPDK containers https://review.openstack.org/342354 | 13:19 |
| *** JeroenBo has quit IRC | 13:21 | |
| *** hfu has joined #openstack-kolla | 13:21 | |
| mliima | morning sdake | 13:22 |
| *** jheroux has joined #openstack-kolla | 13:25 | |
| *** kbaegis has joined #openstack-kolla | 13:25 | |
| *** sayantani01 has joined #openstack-kolla | 13:26 | |
| *** mdnadeem has quit IRC | 13:27 | |
| *** shardy has quit IRC | 13:27 | |
| *** shardy has joined #openstack-kolla | 13:27 | |
| openstackgerrit | Wei Cao proposed openstack/kolla: Add dragonflow container https://review.openstack.org/418193 | 13:29 |
| *** zhubingbing has joined #openstack-kolla | 13:29 | |
| sdake | portdirect ping | 13:30 |
| sdake | portdirect what is the tiller URI | 13:31 |
| *** hfu has quit IRC | 13:32 | |
| kfox1111 | morning | 13:33 |
| kfox1111 | tiller uri? | 13:33 |
| kfox1111 | it doesn't really have one. | 13:33 |
| *** v1k0d3n has joined #openstack-kolla | 13:34 | |
| *** zhubingbing has quit IRC | 13:36 | |
| *** zhubingbing has joined #openstack-kolla | 13:37 | |
| *** bradjones has joined #openstack-kolla | 13:38 | |
| *** bradjones has quit IRC | 13:38 | |
| *** bradjones has joined #openstack-kolla | 13:38 | |
| openstackgerrit | Francisco Edigleison Barbosa proposed openstack/kolla: Add how configure backend to cinder HNAS iSCSI driver guide https://review.openstack.org/422157 | 13:41 |
| *** schwicht has joined #openstack-kolla | 13:41 | |
| *** zhubingbing has quit IRC | 13:42 | |
| *** ipsecguy has quit IRC | 13:44 | |
| *** ipsecguy has joined #openstack-kolla | 13:45 | |
| *** ipsecguy has quit IRC | 13:46 | |
| *** ipsecguy has joined #openstack-kolla | 13:47 | |
| openstackgerrit | Merged openstack/kolla-ansible: Detect openstack_release variable when it auto https://review.openstack.org/422324 | 13:47 |
| Serlex | Anyone try halcyon-kolla-kubernetes environment? | 13:50 |
| *** schwicht has quit IRC | 13:50 | |
| openstackgerrit | Merged openstack/kolla: remove /var/log/apache on keystone dockerfile https://review.openstack.org/423082 | 13:52 |
| *** sp___ has joined #openstack-kolla | 13:54 | |
| *** salv-orlando has joined #openstack-kolla | 13:57 | |
| *** zhubingbing has joined #openstack-kolla | 13:58 | |
| *** zhubingbing has quit IRC | 13:59 | |
| *** skramaja has quit IRC | 14:02 | |
| *** dgonzalez has quit IRC | 14:06 | |
| *** hfu has joined #openstack-kolla | 14:06 | |
| *** sdake_ has joined #openstack-kolla | 14:07 | |
| *** dave-mccowan has joined #openstack-kolla | 14:07 | |
| *** zhubingbing has joined #openstack-kolla | 14:07 | |
| *** schwicht has joined #openstack-kolla | 14:09 | |
| *** sdake has quit IRC | 14:09 | |
| *** mattmceuen has quit IRC | 14:11 | |
| *** dgonzalez has joined #openstack-kolla | 14:11 | |
| *** fguillot has joined #openstack-kolla | 14:14 | |
| *** sayantani01 has quit IRC | 14:14 | |
| *** sayantani01 has joined #openstack-kolla | 14:15 | |
| *** saneax is now known as saneax-_-|AFK | 14:18 | |
| openstackgerrit | Francisco Edigleison Barbosa proposed openstack/kolla: Add how configure backend to cinder HNAS iSCSI driver guide https://review.openstack.org/422157 | 14:18 |
| zhubingbing | hello guys | 14:19 |
| kfox1111 | hi | 14:19 |
| mliima | hi zhubingbing | 14:19 |
| zhubingbing | hi guys | 14:19 |
| zhubingbing | sup <mliima> | 14:20 |
| zhubingbing | sup kfox1111 | 14:20 |
| zhubingbing | ；） | 14:20 |
| sdake_ | kfox1111 kfox1111 ping | 14:20 |
| sdake_ | two kfox pings :) | 14:20 |
| sdake_ | kfox1111 standing up kolla-kubernetes atm | 14:20 |
| sdake_ | kfox1111 to fix the docs | 14:20 |
| kfox1111 | hey. | 14:20 |
| kfox1111 | nice. | 14:20 |
| kfox1111 | how goes it? | 14:21 |
| sdake_ | kfox1111 wondering if you could help me with what the /etc/kolla/globals.yml should look like | 14:21 |
| sdake_ | i see a kolla-kubernetes.yml in etc dir | 14:21 |
| sdake_ | kfox1111 got dev env up and running althoug htiller is busted because the devenv uses 1.13 of docker-engine | 14:21 |
| kfox1111 | cat kolla_kubernetes/tests/conf/ceph-all-in-one/kolla_config >> /etc/kolla/globals.yml | 14:21 |
| kfox1111 | oh. does 1.13 not work with k8s? | 14:23 |
| kfox1111 | haven't tried it yet. | 14:23 |
| *** sacharya has quit IRC | 14:23 | |
| *** sacharya_ has joined #openstack-kolla | 14:23 | |
| *** zhubingbing_ has joined #openstack-kolla | 14:24 | |
| *** sacharya_ has quit IRC | 14:24 | |
| *** sacharya has joined #openstack-kolla | 14:25 | |
| *** zhubingbing has quit IRC | 14:25 | |
| sbezverk | morning | 14:26 |
| kfox1111 | morning | 14:26 |
| *** sacharya has quit IRC | 14:30 | |
| *** hfu has quit IRC | 14:30 | |
| *** kbaegis has quit IRC | 14:34 | |
| *** kbaegis has joined #openstack-kolla | 14:35 | |
| *** kbaegis has quit IRC | 14:38 | |
| sbezverk | kfox1111: qq we always have a single configmap per service right? regardless of number of microservices under the service umbrella | 14:42 |
| *** absubram has joined #openstack-kolla | 14:42 | |
| kfox1111 | no, I think nova has multiple configmaps in its single service. | 14:43 |
| kfox1111 | I'm dusting off the horizon rolling upgrade patch set and I think it will have multiple confimaps too. | 14:43 |
| sbezverk | kfox1111: ah right :( | 14:43 |
| *** bjolo_ has quit IRC | 14:43 | |
| sbezverk | kfox1111: it makes life of an operator more difficult as it needs to maintin multiple configmaps needed for service | 14:44 |
| sdake_ | kfox1111 dissecting the setup_gate.sh | 14:44 |
| sdake_ | is this mandatory: | 14:44 |
| sdake_ | TOOLBOX=$(kollakube tmpl bootstrap neutron-create-db -o json | jq -r '.spec.template.spec.containers[0].image') | 14:44 |
| sdake_ | sudo docker pull $TOOLBOX > /dev/null | 14:44 |
| sdake_ | timeout 240s tools/setup-resolv-conf.sh | 14:44 |
| kfox1111 | sbezverk: yeah. | 14:44 |
| kfox1111 | sdake_: its just a test to ensure that the dns pods in k8s has successfully launched and is working before progress continues. | 14:45 |
| sdake_ | the docker pull part? | 14:45 |
| kfox1111 | the pull just ensures the toolbox is downloaded before staring the 240s countdown. | 14:46 |
| kfox1111 | makes it more likely to succeed in time. | 14:46 |
| kfox1111 | can probably skip it. | 14:46 |
| *** schwicht has quit IRC | 14:47 | |
| *** schwicht has joined #openstack-kolla | 14:50 | |
| *** breitz has quit IRC | 14:50 | |
| sbezverk | kfox1111: there has to be an association in place to the progam can select right config for right microservice. when you want to use two configmaps for reconfiguration, I need to have naming convention neutron-l3-0.4.0.-1-X wheere X is configuation revision, would it work for you? | 14:50 |
| *** lrensing has joined #openstack-kolla | 14:50 | |
| *** lamt has joined #openstack-kolla | 14:51 | |
| *** breitz has joined #openstack-kolla | 14:51 | |
| *** sayantani01 has quit IRC | 14:51 | |
| *** lrensing has quit IRC | 14:54 | |
| kfox1111 | yeah. versioning will be important. | 14:55 |
| kfox1111 | was kind of hoping we could delay until the ptg to discuss, as it will take a while to figure out the right thing to do. | 14:55 |
| *** jtriley has joined #openstack-kolla | 14:55 | |
| kfox1111 | but I'm starting to run into it too. :/ | 14:55 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: WIP: Support rolling upgrades of horizon https://review.openstack.org/333996 | 14:56 |
| kfox1111 | maybe we can start to discuss it though. | 14:57 |
| kfox1111 | so, neutron-l3-0.4.0.-1-X would work. | 14:58 |
| kfox1111 | though for versioning of the package, I'm thinking maybe we do neutron-l3-agent-3.0.2-0.4.0-1 | 14:58 |
| kfox1111 | so the configmap woudl end up being neutron-l3-agent-3.0.2-0.4.0-1-X. which might be a bit much. | 14:59 |
| sbezverk | kfox1111: ok if we agree that configmap will ALWAYS be named after microservice then it will make thing easier | 14:59 |
| kfox1111 | another option would be to just do neutron-l3-X | 14:59 |
| sbezverk | kfox1111: I like latter | 14:59 |
| kfox1111 | do we have cases where a configmaps shared between objects? | 14:59 |
| kfox1111 | yeah. I kind of like neutron-l3-X better too. | 15:00 |
| kfox1111 | it doesn't let you reset the number as often, but not sure that really matters. | 15:00 |
| sbezverk | kfox1111: ony between job that bootstrap process | 15:00 |
| sbezverk | and the actual process | 15:00 |
| kfox1111 | k. | 15:01 |
| *** JeroenBo has joined #openstack-kolla | 15:01 | |
| kfox1111 | so, maybe we prototype neutron-l3-X for now and see if we run into any issues? | 15:01 |
| sbezverk | kfox1111: since I have already everything for memcached, I would prefer continue using it | 15:07 |
| sbezverk | it is more simple and less dependent service | 15:08 |
| kfox1111 | using neutron-l3-0.4.0.-1-X ? | 15:08 |
| sbezverk | no no, that was just an example | 15:08 |
| kfox1111 | oh. | 15:08 |
| sbezverk | memcached-XXX | 15:08 |
| kfox1111 | yeah, I meant just the -X thing rather then the 0.4.0-1-x thing. | 15:08 |
| sbezverk | memcached-001 etc | 15:08 |
| kfox1111 | yeah. +1. | 15:08 |
| kfox1111 | <configmapname>-<version> | 15:09 |
| sbezverk | kfox1111: sounds good I really want to see helm upgrade works with the same chart and dofferent configmap | 15:10 |
| kfox1111 | me too. :) | 15:10 |
| kfox1111 | If it works, I'll probably convert the high energy physics k8s objects I hae to helm. :) | 15:10 |
| kfox1111 | cause being able to make a config change, then do a rolling upgrade to deploy there is really critical. | 15:11 |
| *** jemcevoy has quit IRC | 15:13 | |
| *** david-lyle has joined #openstack-kolla | 15:13 | |
| *** JeroenBo has quit IRC | 15:15 | |
| *** kristian__ has joined #openstack-kolla | 15:16 | |
| *** TxGirlGeek has joined #openstack-kolla | 15:23 | |
| *** david-lyle has quit IRC | 15:25 | |
| *** absubram has quit IRC | 15:25 | |
| *** david-lyle has joined #openstack-kolla | 15:25 | |
| *** masuberu has quit IRC | 15:28 | |
| *** sacharya has joined #openstack-kolla | 15:35 | |
| sambetts | Hey, I'm trying to setup kolla-kubernetes, and I'm looking at the setup_gate.sh script, I know I need a ceph backing store, and I see some scripts like build_test_ceph.sh, is this the script I need to run to setup ceph or is there something else, and does it run ceph in a container or on the k8s host?? | 15:37 |
| openstackgerrit | caoyuan proposed openstack/kolla-ansible: Optimize reconfiguration for senlin https://review.openstack.org/423336 | 15:37 |
| *** nrado has joined #openstack-kolla | 15:37 | |
| *** mgoddard has joined #openstack-kolla | 15:39 | |
| sbezverk | sambetts: scritps you are looking at were desinged for the gate testinh, so everything run in containers, even ceph.. | 15:39 |
| sdake_ | sbezverk i thought ceph was external? | 15:40 |
| *** david-lyle has quit IRC | 15:42 | |
| *** ingmar23 has quit IRC | 15:43 | |
| sbezverk | sdake_: at the gate everything runs as a container | 15:44 |
| *** salv-orlando has quit IRC | 15:44 | |
| sdake_ | sbezverk cool - kfox1111 indicated that isn't ideal | 15:45 |
| sbezverk | sdake_: sure it was just for the gate where we do not have too much of luxury | 15:46 |
| sdake_ | sbezverk cool so documenting how to do deploy without the gate | 15:47 |
| sdake_ | wondering can I just run build_test_ceph.sh? | 15:47 |
| sdake_ | kfox1111 ^^ | 15:47 |
| *** shardy has quit IRC | 15:48 | |
| *** shardy has joined #openstack-kolla | 15:49 | |
| sdake_ | is setup_gate_lopback.sh run on the vm or in a vm on a vm? | 15:50 |
| *** adrian_otto has joined #openstack-kolla | 15:51 | |
| *** masuberu has joined #openstack-kolla | 15:53 | |
| sdake_ | kfox1111 sbezverk ^^ | 15:53 |
| sdake_ | writing docs guys , need some love :) | 15:53 |
| sbezverk | sdake_: kfox1111 would be better resource for ceph. but I do not think running build_test_ceph is siutable for non gate env.. | 15:55 |
| sdake_ | sbezverk i get its not suitable; I want to document how to setup ceph and haven't hte faintest idea how | 15:56 |
| erhudy | can i get another +2 on https://review.openstack.org/#/c/422249/ to push it over the finish line | 15:58 |
| *** egonzalez has quit IRC | 15:58 | |
| *** jascott1_ has joined #openstack-kolla | 15:58 | |
| *** sp___ has quit IRC | 16:01 | |
| *** sayantani01 has joined #openstack-kolla | 16:02 | |
| *** pcaruana has quit IRC | 16:03 | |
| *** jascott1_ has quit IRC | 16:03 | |
| sdake_ | sbezverk kfox1111 is there a diagnostic command more advanced then kubectl get pod rabbitmq-0? | 16:05 |
| openstackgerrit | zhubingbing proposed openstack/kolla-ansible: Use port variable replace 5000 and 35357 https://review.openstack.org/423348 | 16:08 |
| *** hieulq has quit IRC | 16:09 | |
| mgiles | sdake_ depending on what you're looking for kubectl describe pod rabbitmq-0 or kubectl logs rabbitmq-0 may give more detail | 16:10 |
| sdake_ | mgiles thanks | 16:12 |
| sdake_ | mgiles i'm on a mission to make a proepr quickstart guide for kolla-kubernetes | 16:13 |
| sdake_ | mgiles curious whaty our approach to ceph was | 16:13 |
| sdake_ | mgiles or if instead your using the gate for testing | 16:13 |
| sdake_ | mgiles morning btw :) | 16:13 |
| mgiles | good morning | 16:13 |
| mgiles | I've tried to replicate the ceph setup the gate uses. It's not ideal, but I don't know anything about ceph so haven't figured out a better way | 16:14 |
| sdake_ | define tried | 16:14 |
| sdake_ | did you write a script? | 16:14 |
| sdake_ | mgiles i want to do exacty the same thing and document it | 16:14 |
| openstackgerrit | sean mooney proposed openstack/kolla-ansible: This change allow setting kernel commandline args https://review.openstack.org/408876 | 16:14 |
| openstackgerrit | sean mooney proposed openstack/kolla-ansible: extends ansible roles to deploy ovsdpdk https://review.openstack.org/408872 | 16:14 |
| openstackgerrit | sean mooney proposed openstack/kolla-ansible: adds ovs-dpdkctl tool https://review.openstack.org/408871 | 16:14 |
| mgiles | it's a bit of a long answer - give me a minute ;) | 16:14 |
| sdake_ | mgiles devs r e blocked on deving because of ceph | 16:14 |
| mliima | hey sdake_, can you review it https://review.openstack.org/#/c/417094/ :) | 16:15 |
| mgiles | need to look back at my notes | 16:15 |
| *** hieulq_ has joined #openstack-kolla | 16:15 | |
| *** hieulq has joined #openstack-kolla | 16:15 | |
| *** sdake_ has quit IRC | 16:15 | |
| sbezverk | sdake_: what exactly you want to check with rabbitmq-0? | 16:16 |
| mgiles | after the vagrant env is up, I ssh'd into kube2 (vagrant ssh kube2) and run the contents of the setup_gate_loopback.sh script there | 16:16 |
| *** sdake has joined #openstack-kolla | 16:16 | |
| mgiles | The problem is that in that env, the loopback devices are different device numbers than expected by some of the later commands, so I had to tweak things after that | 16:17 |
| mgiles | Also it's necessary to have kube2 be the only node labelled as a control node so that all the things that need ceph are there | 16:17 |
| mgiles | kind of hacky all around, but it's gotten things working for me | 16:17 |
| sdake | mgiles taking all suggesetions :) | 16:18 |
| sdake | mgiles are you talking the halcyon env? | 16:18 |
| mgiles | if you pm me your email address, I can send you more specifics. Too long for irc | 16:18 |
| sdake | stdake@csico.com | 16:18 |
| sdake | thanks mgiles :) | 16:19 |
| *** satyar_ has joined #openstack-kolla | 16:19 | |
| *** satyar has quit IRC | 16:20 | |
| mgiles | also bmace_ has been working on some scripts to make this dev env setup smoother so you should check with him when he's online (might be a bit early for him yet) | 16:20 |
| kfox1111 | back. | 16:21 |
| *** zhubingbing_ has quit IRC | 16:22 | |
| kfox1111 | sdake: if you setup the two loopback devices prior to running build_test-ceph.sh, it might just work. | 16:23 |
| *** l4yerffeJ has joined #openstack-kolla | 16:23 | |
| *** l4yerffeJ has left #openstack-kolla | 16:24 | |
| *** l4yerffeJ has joined #openstack-kolla | 16:24 | |
| kfox1111 | oh. probably need some bits from setup_config.sh too. | 16:24 |
| kfox1111 | as it tweaks ceph to be ok with 1 node, 2 osd's. | 16:24 |
| *** inc0 has joined #openstack-kolla | 16:24 | |
| sdake | kfox1111 yup - working on writing a "deploy" script | 16:25 |
| sdake | that works outside of the gate | 16:25 |
| inc0 | good morning | 16:25 |
| sdake | sup inc0 | 16:25 |
| kfox1111 | sdake: cool. | 16:25 |
| sdake | kfox1111 any hints you could provide for that would be helpful | 16:25 |
| sdake | I have dug into setup_gate.sh quite a bit today and last night | 16:25 |
| sdake | as well as ceph_workflow.sh | 16:25 |
| kfox1111 | setup_gate does a complete setup. most of it is deployment related. a little bit is gate related. | 16:25 |
| sdake | i understand - i'e studied it for about 8 hrs | 16:26 |
| kfox1111 | cool. :) | 16:26 |
| sdake | the helm deployments end in a 1/1 deployed state | 16:26 |
| sdake | however it takes forever and the docker containers themselves keep restarting | 16:26 |
| kfox1111 | yeah. | 16:26 |
| sdake | while some work | 16:26 |
| kfox1111 | hmm... so almost there. | 16:26 |
| sdake | right - i think the missing component is ceph | 16:27 |
| kfox1111 | yeah. if ceph isn't there, it will do that. | 16:27 |
| kfox1111 | as mariadb/rabbit depend on ceph, | 16:27 |
| sdake | ok well that h elps :) | 16:27 |
| kfox1111 | so everything else won't work. | 16:27 |
| kfox1111 | ceph's the very bottom of the layer cake of storage. | 16:27 |
| kfox1111 | so, I think your close. | 16:28 |
| sdake | got it and understand it (or some other storage provider for kubernetes) is mandatory | 16:28 |
| kfox1111 | yeah. | 16:28 |
| *** dave-mccowan has quit IRC | 16:29 | |
| *** salv-orlando has joined #openstack-kolla | 16:31 | |
| *** nrado has quit IRC | 16:32 | |
| *** sudipto has joined #openstack-kolla | 16:32 | |
| *** sudipto_ has joined #openstack-kolla | 16:32 | |
| *** hieulq_ has quit IRC | 16:32 | |
| sdake | mgiles ya know, I should have asked you to mail that to the mailing list for others | 16:37 |
| sdake | rather then eveyrone singlethread on me | 16:37 |
| sdake | mgiles would you be willing to do that? | 16:37 |
| sdake | mgiles i see this often in every project | 16:38 |
| sdake | there are 3 aspects | 16:38 |
| sdake | code, docs, and gating | 16:38 |
| sdake | a project does really well at 2 | 16:38 |
| sdake | and not so well at 1 | 16:39 |
| sdake | kolla-kubernetes hasn't done a great job of keeping the docs up to date | 16:39 |
| sdake | if we could get a minimal quickstart guide written, that would help everyone | 16:39 |
| sdake | and I'm willing to write it or parts of it | 16:39 |
| mgiles | sdake agreed about the docs. Glad you're tackling some of it | 16:40 |
| mgiles | sdake I think ideally any set of instrucitons should go into docs rather the ml. Not everybody follows the ml. | 16:40 |
| sdake | mgiles right - i just want to get your notes distributed | 16:41 |
| mgiles | sdake ...and I keep hoping there's a more elegant solution than what I've hacked together | 16:41 |
| mgiles | although I guess it's better than nothing :) | 16:42 |
| Serlex | Can I suggest removing some of the "old" documentation to easy the confusion? | 16:42 |
| sdake | Serlex will move it to an archive for th emoment - want to keep it around as it is our best reference | 16:43 |
| Serlex | Roger | 16:43 |
| sdake | Serlex and parsing rst is a pita :) | 16:43 |
| sdake | i prefer reading docs.oo | 16:43 |
| *** satyar has joined #openstack-kolla | 16:43 | |
| prameswar | Hi all , can you please review this patch for kolla-kubernetes : https://review.openstack.org/#/c/422803/ | 16:44 |
| inc0 | I read it from github | 16:44 |
| kfox1111 | sdake: gits a reference. can delete it and it will still be in history. | 16:44 |
| inc0 | since I usually know which document I'm looking for and navigation is quicker | 16:44 |
| *** satyar_ has quit IRC | 16:46 | |
| *** v1k0d3n has quit IRC | 16:46 | |
| Serlex | I'm trying to get setup with k8s on github but I see "development env" "minikube quickstart" "quickstart" "kubernetes-setup" | 16:48 |
| Serlex | maybe I'm been a douche and very picky | 16:48 |
| *** jascott1_ has joined #openstack-kolla | 16:51 | |
| openstackgerrit | Steven Dake proposed openstack/kolla-kubernetes: Clean up mysql installed during test of helm https://review.openstack.org/423120 | 16:55 |
| openstackgerrit | Steven Dake proposed openstack/kolla-kubernetes: Install kubectl and helm on the host OS in docs https://review.openstack.org/423046 | 16:55 |
| jascott1_ | sdake still going I see :) | 16:55 |
| sdake | jascott1_ actually I crashed for about 3 hours again | 16:56 |
| sdake | get the crash cart! | 16:56 |
| jascott1_ | haha, cheating | 16:56 |
| sdake | working too much is the cheating part ;) | 16:56 |
| sdake | Serlex ya - the core reviewers are struggling to setup their environments | 16:57 |
| sdake | serlex so don't feel alone :) | 16:57 |
| jascott1_ | just need adrenaline IV | 16:57 |
| *** jascott1_ has quit IRC | 16:57 | |
| *** jascott1_ has joined #openstack-kolla | 16:58 | |
| jascott1_ | bouncer is taking my id but wont let me connect :/ | 16:59 |
| sbezverk | sdake: it sounds a bit strong about struggling ;) bunch of guys developing and running as far as I can see.. | 17:00 |
| sdake | sbezverk ok some devs are struggling :) | 17:00 |
| sdake | sbezverk and you hae to admit the docs are disorganized :) | 17:01 |
| sbezverk | sdake: we are in the development, some things we do not even know how it will be, so at this point I think it is normal.. | 17:01 |
| kfox1111 | sbezverk: portdirect said he hasn't been able to deploy yet. | 17:02 |
| kfox1111 | the docs really need fixing. :/ | 17:02 |
| kfox1111 | with the huge refactor, its not surprising the docs bitrotted. | 17:02 |
| Serlex | Chill guys, I didn't mean to stir anything, docs are cool, its the user, | 17:03 |
| Serlex | aka me :) | 17:03 |
| *** DanyC has joined #openstack-kolla | 17:04 | |
| *** salv-orlando has quit IRC | 17:04 | |
| *** fguillot has quit IRC | 17:04 | |
| mgiles | sdake Let me know if those instructions I sent work for you. If they do, I'm willing to post to a larger audience, but we should make sure they're tested by somebody other than just me first. | 17:05 |
| kfox1111 | Serlex: no, the docs are really really bitrotten. | 17:06 |
| kfox1111 | Serlex: we've been completely reworking the deployment mechanism, so the docs reflect the old way of doing things. | 17:06 |
| openstackgerrit | Edmund Rhudy proposed openstack/kolla: Add rabbitmq_install block to RabbitMQ Dockerfile https://review.openstack.org/423378 | 17:07 |
| openstackgerrit | Michal Jastrzebski (inc0) proposed openstack/kolla-kubernetes: WIP - local registry in gates https://review.openstack.org/418550 | 17:07 |
| *** DanyC has left #openstack-kolla | 17:08 | |
| *** jrich523 has quit IRC | 17:08 | |
| *** jrich523 has joined #openstack-kolla | 17:08 | |
| inc0 | hey guys., check it out https://review.openstack.org/#/c/422931/ | 17:09 |
| inc0 | this thing will land in main openstack docs as our deployment guide | 17:09 |
| inc0 | so let's make it count | 17:09 |
| *** lamt has quit IRC | 17:10 | |
| *** DanyC_ has joined #openstack-kolla | 17:10 | |
| kfox1111 | inc0: going to ask again. is kolla repo the right place for that? shouldnit there be one in kolla-ansible and one in kolla-kubernetes? | 17:11 |
| kfox1111 | kolla repo isn't concerned with deployment. | 17:11 |
| *** DanyC_ has left #openstack-kolla | 17:11 | |
| inc0 | kfox1111, we need to refactor docs, true, however I'd wait till we have real alternative frankly | 17:13 |
| erhudy | opinion question | 17:13 |
| *** matrohon has quit IRC | 17:13 | |
| inc0 | or at least someone willing to write down docs to clearly point to ansible/k9s | 17:13 |
| inc0 | k8s | 17:13 |
| kfox1111 | inc0: strongly disagree. there was a huge push to split repo's. still is. we need to stop resisting that split. :/ | 17:13 |
| SamYaple | ++ | 17:14 |
| *** lamt has joined #openstack-kolla | 17:14 | |
| kfox1111 | taking time to split old stuff is understandable. | 17:14 |
| inc0 | ok, fair point | 17:14 |
| kfox1111 | new stuff on the other hand, should go in the right place, to begin with. | 17:14 |
| erhudy | since there are multiple containers that pull get-pip.py from the interwebs (e.g. kolla-toolbox, some helm stuff), would anyone be mad if i introduced a pip-base container to consolidate the pip installation step and rebased affected containers to depend on it | 17:14 |
| inc0 | then I'll say, when someone write docs that clearly will help people to quickly install kolla, that'd be ok | 17:14 |
| kfox1111 | there should just be a pointer from kolla to docs in either repo. | 17:15 |
| inc0 | erhudy, that's a very good question, you're right imho | 17:15 |
| erhudy | independent of that, there are containers that also configure their own package installation stuff | 17:16 |
| erhudy | e.g. fluentd | 17:16 |
| erhudy | i'm not sure if it was intentional that they weren't included in the base image? | 17:16 |
| openstackgerrit | Merged openstack/kolla: Remove empty spaces dind dockerfile https://review.openstack.org/423249 | 17:16 |
| inc0 | also good catch, one I'm particulary not very fond of, this is a bug | 17:17 |
| erhudy | okay, i just wanted to discuss it before i start lobbing more PRs up, since there may have been reasons for doing so | 17:18 |
| erhudy | being that i am new to the community and not steeped in all the tribal knowledge | 17:18 |
| inc0 | understandable, good to ask, but also both points are totally valid | 17:18 |
| *** fguillot has joined #openstack-kolla | 17:19 | |
| erhudy | alright, i'll get to work then | 17:19 |
| inc0 | thank you | 17:19 |
| inc0 | https://review.openstack.org/#/c/418550/ <- hmm...what did I do to haproxy conf? oO | 17:26 |
| *** DanyC has joined #openstack-kolla | 17:28 | |
| openstackgerrit | Michal Jastrzebski (inc0) proposed openstack/kolla-kubernetes: WIP - local registry in gates https://review.openstack.org/418550 | 17:29 |
| inc0 | nvm, I messed up big time last one;) | 17:29 |
| *** DanyC has left #openstack-kolla | 17:31 | |
| *** jascott1_ has quit IRC | 17:33 | |
| openstackgerrit | Mauricio Lima proposed openstack/kolla: Use install-pip macro in most source images https://review.openstack.org/413861 | 17:37 |
| *** aj_power has joined #openstack-kolla | 17:38 | |
| *** kristian__ has quit IRC | 17:38 | |
| *** jascott1_ has joined #openstack-kolla | 17:40 | |
| *** nrado has joined #openstack-kolla | 17:42 | |
| *** jascott1_ has quit IRC | 17:44 | |
| *** athomas has quit IRC | 17:45 | |
| *** jascott1_ has joined #openstack-kolla | 17:47 | |
| *** jascott1_ has quit IRC | 17:48 | |
| *** jascott1_ has joined #openstack-kolla | 17:48 | |
| *** dturner has quit IRC | 17:48 | |
| *** mgoddard has quit IRC | 17:52 | |
| *** sacharya has quit IRC | 17:54 | |
| openstackgerrit | Michal Jastrzebski (inc0) proposed openstack/kolla-kubernetes: WIP - local registry in gates https://review.openstack.org/418550 | 17:56 |
| *** hieulq has quit IRC | 17:56 | |
| sdake | mgiles will do as soon as I sort out how to get tiller to work correctly in vagrant | 18:01 |
| sdake | mgiles with the update to docker 1.13 that docker released, tiller is nonfunctiona | 18:01 |
| mgiles | sdake that's not good. guess I will try to avoid updating my vagrant images | 18:02 |
| sdake | mgiles portdirect said he would pin teh dev env to 1.12 | 18:02 |
| sdake | i'd do it myself, but no tquite sure how to do so | 18:02 |
| sdake | mgiles i may take a crack at it shortly though as i'm completely blocked | 18:03 |
| sdake | could core revieweers ack all related patchee shere plz: https://review.openstack.org/#/c/423039/ | 18:04 |
| sdake | these are doc improvements that get the dev env doc up to date and working perfectly (once docker-engine is pinned0 | 18:04 |
| kfox1111 | sdake: shouldn't have to do anython other then helm init | 18:08 |
| kfox1111 | to get tiller working. | 18:08 |
| sdake | kfox1111 nah portdirect had me pull a tiller from soe special location to get tiller working | 18:08 |
| kfox1111 | well, assuming a working k8s though. | 18:08 |
| kfox1111 | wha? | 18:08 |
| kfox1111 | why? | 18:08 |
| sdake | the tiller image from google.com is busted | 18:09 |
| sdake | with docker-engine 1.13 | 18:09 |
| kfox1111 | oh. :/ | 18:09 |
| sdake | the gate tests docker-engine 1.10 | 18:09 |
| sdake | or rather the red hat version of docker i think | 18:09 |
| kfox1111 | docker inc really needs to stop breaking stuff with every release! :/ | 18:09 |
| kfox1111 | yeah. readhat version.l | 18:09 |
| kfox1111 | cause I was fed up chasing bugs docker inc introduces. | 18:09 |
| kfox1111 | redhat's much better about stability. | 18:10 |
| jrich523 | .... i disagree or maybe i just dislike RHEL :) | 18:10 |
| sdake | red hat's version of docker doesn't pull from dockerhub by default iirc | 18:10 |
| *** adrian_otto has quit IRC | 18:10 | |
| sdake | also its not msater of docker-engine | 18:10 |
| erhudy | i think RHEL just pushed docker 1.12 to their repos today and clobbered 1.10, there was a lot of discussion internally about that | 18:11 |
| sdake | typically kolla runs the most recent stable release of everything | 18:11 |
| erhudy | i think some people at BB have been shouting at red hat since that happened | 18:11 |
| kfox1111 | yeah it does. | 18:11 |
| kfox1111 | you do a docker pull kolla/.... it comes from the hub as expected. | 18:11 |
| sdake | kfox1111 portdirect mentioend it in irc last night - the one thing i can confirm is in the dev env docker-engine 1.13 is used | 18:11 |
| sdake | and kolla-ansible uses docker engine 1.13 and that broke kolla-ansible's gate too | 18:12 |
| *** hieulq has joined #openstack-kolla | 18:12 | |
| kfox1111 | yeah. that I'll believe. | 18:12 |
| sdake | so Jeffrey4l_ pinned docker engine to 1.12.6 iirc | 18:12 |
| sdake | the key is to run the gate and dev env off the same binaries | 18:12 |
| sdake | and theoreticallyt he same binaires as kolla-ansible | 18:13 |
| sdake | so we can cross-debug problems | 18:13 |
| kfox1111 | or all suffer together. :/ | 18:13 |
| sdake | a pin is easy enough to introduce | 18:13 |
| sdake | group suffering by not unifying building blocks where we can is not ideal :( | 18:14 |
| kfox1111 | sure. just saying, as an op, I tend to prefer redhat's version over an upstream that has a track record of backwards incompatabile changes. | 18:14 |
| kfox1111 | redhat's goal is stability, which is a normal op goal as well. | 18:15 |
| kfox1111 | docker inc's goal is different. | 18:15 |
| sdake | altough we are an upstream not a downstream :) | 18:15 |
| *** caowei has quit IRC | 18:15 | |
| kfox1111 | we're kind of both. | 18:15 |
| kfox1111 | we pull from upstreams, and we are an upstream of others. | 18:16 |
| sdake | in this case both docker-engine and rhel's version of docker are upstreams | 18:16 |
| kfox1111 | true. | 18:16 |
| sdake | we should sort out which one we are going to use cross-deliverable | 18:16 |
| kfox1111 | yeah. ptg? | 18:16 |
| *** prameswar has quit IRC | 18:17 | |
| sdake | wfm | 18:18 |
| sdake | in the meantime need a working environment - can't wiat 4 weeks for that ;) | 18:18 |
| kfox1111 | yup. | 18:18 |
| kfox1111 | redhat docker known to work. | 18:19 |
| kfox1111 | pinning back a ver of docker inc's also known to work. | 18:19 |
| sdake | yup - if I knew immediately how to change it it would be done already :) | 18:19 |
| inc0 | https://review.openstack.org/423403 <- heads up, you might have to deal with myself for next few months;) | 18:20 |
| sbezverk | sdake: I do not get the rational to jump on new docker version. it is not that we are missing some features | 18:20 |
| sdake | sbezverk ya - it made sense in the past | 18:20 |
| sbezverk | or there are some security vulnerabilities | 18:20 |
| sdake | sbezverk people complain about red hat's versoin of docker for various reasons - hence we use docker-engine | 18:20 |
| sdake | the reason it made sense in the past is because we needed all of the featuers of 1.10 and we don't like to introduce top pins in dependencies | 18:21 |
| sdake | != pins are ok | 18:21 |
| openstack | sdake: Error: "=" is not a valid command. | 18:21 |
| sdake | top ins not ok | 18:21 |
| sdake | top pins not ok | 18:21 |
| sbezverk | sdake: hmmm, why then you complain about docker now ;)?? | 18:21 |
| sdake | sbezverk pretty sure I bitched loudly about 1.8.3 | 18:21 |
| sdake | that is why SamYaple wrote kolla_docker.py | 18:21 |
| sdake | since 1.8.3 docker hasn't managed to break anything | 18:22 |
| sbezverk | sdake: so far the one from red hat worked like a charm, seee no reason to change it.. | 18:22 |
| *** jascott1_ has quit IRC | 18:22 | |
| sdake | although from erhudy 's comments it appears centos version is busted too | 18:22 |
| sdake | sbezverk i'd challenge you to reubild your environment and make the same startemnt :) | 18:22 |
| kfox1111 | sdake: if thats the case, we should see it in the gate any time now. | 18:23 |
| sbezverk | sdake: no prob, especially since my env runs all latest updates for cenots | 18:23 |
| sbezverk | kfox1111: +1 | 18:23 |
| sdake | inc0 ballstic is not th ebest choice of verb ;) | 18:24 |
| inc0 | haha | 18:24 |
| erhudy | sdake: not sure if centos is affected, but rhel-7-server-extras-rpms got an unexpected upgrade | 18:24 |
| inc0 | well it does have a nice ring to it;) | 18:25 |
| sdake | erhudy ahh - thanks for that | 18:25 |
| inc0 | I might change it to something better, thanks | 18:25 |
| inc0 | went skyrocket would be better? | 18:25 |
| erhudy | people saw it go from 1.10.3 to 1.12.5 when RH promised they wouldn't do that | 18:25 |
| *** sayantani01 has quit IRC | 18:25 | |
| erhudy | docker-latest is supposed to track the latest upstream version, is what i was told | 18:25 |
| kfox1111 | erhudy: unexpected upgrades are one thing. did it actually break anything? is another. | 18:25 |
| erhudy | idk, historically i've treated docker point releases like major version changes after getting burned a couple times | 18:26 |
| erhudy | but that was pretty early on in 1.x | 18:27 |
| kfox1111 | yeah. docker wasn't very careful about api levle breakages back then. | 18:27 |
| kfox1111 | I know several times docker renamed/removed cli switches. | 18:27 |
| kfox1111 | redhat pinned their version then to avoid breaking existing stuff. | 18:27 |
| *** msimonin1 has quit IRC | 18:28 | |
| kfox1111 | if redhat released a 1.10 -> 1.12 though, Id be surprised if they hadn't tested that though. | 18:28 |
| *** msimonin has joined #openstack-kolla | 18:28 | |
| *** mgoddard has joined #openstack-kolla | 18:28 | |
| erhudy | people who spend a lot more time with docker than i do say 1.12 had some significant API changes | 18:28 |
| *** jascott1_ has joined #openstack-kolla | 18:29 | |
| kfox1111 | yeah, but never trust redhat's version numbering meaning what you think it does. ;) | 18:29 |
| kfox1111 | they could have added patches to make it fully backwards compatible again. | 18:29 |
| kfox1111 | they do that kind of thing. | 18:29 |
| erhudy | i think we're just giving red hat grief because they told us their policy was X and then went and did Y | 18:29 |
| inc0 | sdake, this I changed ballistic | 18:29 |
| inc0 | I looked it up and yeah, not what I wanted to say;) | 18:30 |
| *** adrian_otto has joined #openstack-kolla | 18:30 | |
| sdake | i guess its not a verb | 18:31 |
| sdake | but you get the idea ;) | 18:31 |
| kfox1111 | thats the great thing about english. anything can be a verb! :) | 18:31 |
| erhudy | anything can be verbed | 18:31 |
| kfox1111 | exactly! :) | 18:31 |
| sdake | erhudy i disagree, upstream is not a verb ;) | 18:32 |
| sdake | you can't upstream a boat. | 18:32 |
| kfox1111 | I upstreamed that.... yup. its a verb. :) | 18:32 |
| kfox1111 | cant a boat. but can a patch. :) | 18:33 |
| * sdake tears out what little hair he has left | 18:33 | |
| * kfox1111 chuckes. | 18:33 | |
| *** jascott1_ has quit IRC | 18:33 | |
| kfox1111 | dev's have their own special version of english. :) | 18:33 |
| inc0 | well, I still can't get idea of "ask" as a noun | 18:34 |
| inc0 | I mean, this sounds silly | 18:34 |
| kfox1111 | so do physisists, but I don't speak that version. :) | 18:34 |
| *** sacharya has joined #openstack-kolla | 18:34 | |
| *** jascott1_ has joined #openstack-kolla | 18:35 | |
| *** schwicht has quit IRC | 18:36 | |
| *** schwicht has joined #openstack-kolla | 18:37 | |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: WIP: Support rolling upgrades of horizon https://review.openstack.org/333996 | 18:37 |
| britthouser | “ask” is business speak for “request" | 18:37 |
| kfox1111 | such as, "an ask" ? | 18:37 |
| kfox1111 | yeah. | 18:37 |
| kfox1111 | I guess I've heard it used that way. | 18:38 |
| britthouser | b/c request is too long! =P | 18:38 |
| kfox1111 | hehe. | 18:38 |
| britthouser | and hard to spell | 18:38 |
| kfox1111 | cs folks just make acronyms insteadl. :) | 18:38 |
| inc0 | rqst | 18:39 |
| kfox1111 | YAY! | 18:39 |
| kfox1111 | https://review.openstack.org/#/c/408450/ | 18:39 |
| inc0 | and letters are closser to each other | 18:39 |
| kfox1111 | I can finally start to consider k2k federation. | 18:39 |
| kfox1111 | k2k merged a long time ago, but really not very useful without horizon support. | 18:39 |
| *** jascott1_ has quit IRC | 18:40 | |
| erhudy | what's the right way to override a macro? template override? | 18:40 |
| inc0 | erhudy, that's...something we never did | 18:40 |
| erhudy | womp womp | 18:40 |
| *** esmiurium has quit IRC | 18:40 | |
| inc0 | what's the issue? | 18:40 |
| erhudy | i'm looking at https://review.openstack.org/#/c/413861/ now and part of what it promises is that uniform use of the macro will allow doing things like setting a particular index, trusted-url, whatever | 18:41 |
| *** sudipto has quit IRC | 18:41 | |
| *** sudipto_ has quit IRC | 18:41 | |
| erhudy | but if it can't actually be overridden except by editing macros.j2 then :sad: | 18:41 |
| *** jascott1_ has joined #openstack-kolla | 18:41 | |
| inc0 | well we can write it to be possible to override...I think | 18:42 |
| erhudy | it's relevant to consolidating the get-pip.py installations because those images all want to install stuff from pip, which will still try to escape to the internets | 18:42 |
| inc0 | this is some jinja wizardry I haven't use before | 18:42 |
| inc0 | you can try to include your own macro file with new macro under same name | 18:42 |
| inc0 | but I have no idea how that works | 18:42 |
| inc0 | if that works that is | 18:42 |
| erhudy | really getting into the weeds with jinja here | 18:45 |
| kfox1111 | hmm... k8s is acting weird on reboot... | 18:45 |
| inc0 | yeah, by extending block you could potentially override macro | 18:45 |
| kfox1111 | oh.... I wonder if this is an iptables thing... | 18:47 |
| kfox1111 | I don't think i persisted iptables... | 18:47 |
| *** esmiurium has joined #openstack-kolla | 18:47 | |
| kfox1111 | yeah... that was it. | 18:48 |
| erhudy | alternative solution would be to add an option to copy a pip.conf file into the container and have the macro use that, i guess | 18:48 |
| inc0 | that woudl be better | 18:49 |
| inc0 | I think you could just override default pip.conf in header | 18:49 |
| erhudy | yeah | 18:49 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: WIP: test rebooting in the gate https://review.openstack.org/413193 | 18:49 |
| inc0 | if that doesn't work, I'd change macro to use this conf file if it exists | 18:50 |
| *** TxGirlGeek has quit IRC | 18:50 | |
| erhudy | not sure about where it would need to go in order for pip in the venv to pick it up and use it, though, whether it will still check /etc/pip.conf even in a venv | 18:50 |
| erhudy | something to test, i guess | 18:50 |
| *** rhallisey has quit IRC | 18:50 | |
| kfox1111 | sdake: gate-kolla-kubernetes-deploy-centos-binary-t-ceph-multi-nv shouldn't have been promoted yet. | 18:51 |
| kfox1111 | its not even implemented yet. | 18:51 |
| openstackgerrit | Merged openstack/kolla: Fix manila-base dockerfile format https://review.openstack.org/423089 | 18:53 |
| kfox1111 | sbezverk: https://review.openstack.org/#/c/422932/ is ready. | 18:54 |
| kfox1111 | hmm... I take that back. | 18:54 |
| *** JoseMello has quit IRC | 18:55 | |
| kfox1111 | got a bad paste.... | 18:55 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: Add a simple test to see if horizon is running https://review.openstack.org/422932 | 18:57 |
| openstackgerrit | Edmund Rhudy proposed openstack/kolla: Add pip-base image https://review.openstack.org/423411 | 18:58 |
| *** Serlex has quit IRC | 18:58 | |
| sbezverk | kfox1111: nice curl ;) | 18:59 |
| kfox1111 | sbezverk: yeah. :/ | 19:00 |
| * kfox1111 hangs his head in shame | 19:00 | |
| erhudy | i need gerrit to tell me every time there's a PR that has "curl" or "wget" in it | 19:01 |
| erhudy | so i can run in and shout THIS IS ACCESSING THE INTERNET | 19:01 |
| qwang | sdake: I am reviewing your ps: https://review.openstack.org/#/c/423046/ but I wonder why kubectl and helm are required by the host OS. | 19:01 |
| qwang | sdake: I was able to install dev env without kubectl and helm installed on my host | 19:01 |
| *** sacharya has quit IRC | 19:02 | |
| kfox1111 | erhudy: not nessisarily. ;) | 19:02 |
| erhudy | not in the line you added, but right below it! | 19:02 |
| erhudy | curling in cirros | 19:02 |
| erhudy | smh | 19:02 |
| kfox1111 | curl 127.0.0.1 isn't. | 19:02 |
| kfox1111 | so does yum install. :) | 19:02 |
| kfox1111 | usually pulls stuff from the internet. | 19:02 |
| kfox1111 | pip install... | 19:03 |
| erhudy | which is why i have embarked on this sisyphean task of finding and crushing all those areas | 19:03 |
| erhudy | or at least putting them in blocks so that they can be swapped out, so that by the end of it all i'm going to have half a meg of template overrides going into my container build | 19:03 |
| kfox1111 | because the internet is bad, or because you want a way to cache all the things for sites that don't have internet access? | 19:03 |
| *** gfidente has quit IRC | 19:03 | |
| erhudy | the bloomberg walled fortress does not believe in the internet | 19:03 |
| kfox1111 | Im interested in the latter use case. :) | 19:04 |
| kfox1111 | yeah. I get that. | 19:04 |
| kfox1111 | I'm in the same boat. | 19:04 |
| kfox1111 | thats one of the great things about the container stuff. | 19:04 |
| kfox1111 | build is seperate from deploy. | 19:04 |
| sdake | qwang the model is that you launch the devenv, then use get-k8s-creds and leave the virtual env alone from there on out | 19:04 |
| sdake | qwang and then use the host os for all interaction | 19:04 |
| *** jtriley has quit IRC | 19:04 | |
| erhudy | caching is good, and i want to be sure i can use local caches for all my builds | 19:05 |
| erhudy | e.g. pull from bloomberg's internal ubuntu package mirror, or artifactory for pip, etc | 19:05 |
| kfox1111 | erhudy: +1 | 19:05 |
| kfox1111 | some places require all data to come in on cd. | 19:06 |
| kfox1111 | and then it doesn't leave. ever. :) | 19:06 |
| erhudy | yeah, we're not nearly that bad | 19:06 |
| erhudy | and philosophically i would support parameterizing everything to allow using local repos even if direct internet access were available | 19:07 |
| *** rhallisey has joined #openstack-kolla | 19:07 | |
| qwang | sdake: or we can connect to a node in the virtual env and manage the cluster there. what's the benefit of using host os for all interactions? | 19:07 |
| sdake | qwang you can do that as well - nobody forces it upon you :) | 19:10 |
| kfox1111 | bbiab. | 19:10 |
| kfox1111 | erhudy: I wonder though, if its enough just to build the containers with internet connectivity, then mirror them to a private place. | 19:11 |
| kfox1111 | rather then try and build them in total isolution. | 19:11 |
| kfox1111 | bbiab. | 19:11 |
| qwang | sdake: I see. Thank you. I just want to know if a way is better than the other for operators. | 19:12 |
| erhudy | kfox1111: part of the problem is there's no place i can build things internally where i can get to the internet via the gateway IP | 19:12 |
| erhudy | any internet access is via proxy servers with SSL intercept | 19:13 |
| erhudy | proxy servers alone are a pain to deal with, but having to add extra CA certs to multiple chains just means it's easier to build using internal resources | 19:13 |
| *** unicell has joined #openstack-kolla | 19:15 | |
| *** unicell is now known as Guest59079 | 19:16 | |
| *** sayantani01 has joined #openstack-kolla | 19:17 | |
| *** david-lyle has joined #openstack-kolla | 19:17 | |
| *** jtriley has joined #openstack-kolla | 19:22 | |
| *** david-lyle has quit IRC | 19:24 | |
| sdake | qwang this is a dev enviroment, not targeted at operators | 19:26 |
| qwang | sdake: right. ;) | 19:26 |
| openstackgerrit | Michal Jastrzebski (inc0) proposed openstack/kolla-kubernetes: WIP - local registry in gates https://review.openstack.org/418550 | 19:27 |
| *** Guest59079 has quit IRC | 19:32 | |
| *** kristian__ has joined #openstack-kolla | 19:34 | |
| *** sambetts is now known as sambetts|afk | 19:36 | |
| *** mgoddard has quit IRC | 19:36 | |
| *** jascott1_ has quit IRC | 19:39 | |
| *** jascott1_ has joined #openstack-kolla | 19:39 | |
| *** kristian__ has quit IRC | 19:39 | |
| *** sdake has quit IRC | 19:47 | |
| *** schwicht has quit IRC | 19:49 | |
| *** schwicht has joined #openstack-kolla | 19:50 | |
| *** kristian__ has joined #openstack-kolla | 19:52 | |
| *** unicell has joined #openstack-kolla | 19:58 | |
| *** jascott1_ has quit IRC | 19:59 | |
| *** jascott1_ has joined #openstack-kolla | 19:59 | |
| *** TxGirlGeek has joined #openstack-kolla | 19:59 | |
| *** TxGirlGeek has quit IRC | 19:59 | |
| *** mliima has quit IRC | 19:59 | |
| *** TxGirlGeek has joined #openstack-kolla | 20:00 | |
| *** jascott1_ has quit IRC | 20:04 | |
| *** jascott1_ has joined #openstack-kolla | 20:06 | |
| *** TxGirlGeek has quit IRC | 20:06 | |
| *** jtriley has quit IRC | 20:07 | |
| *** jascott1_ has quit IRC | 20:08 | |
| *** jascott1_ has joined #openstack-kolla | 20:12 | |
| *** nrado has quit IRC | 20:14 | |
| *** esmiurium has quit IRC | 20:15 | |
| *** jascott1_ has quit IRC | 20:15 | |
| *** jtriley has joined #openstack-kolla | 20:16 | |
| *** jascott1_ has joined #openstack-kolla | 20:19 | |
| *** TxGirlGeek has joined #openstack-kolla | 20:20 | |
| *** jascott1_ has quit IRC | 20:25 | |
| *** adrian_otto has quit IRC | 20:27 | |
| *** jascott1_ has joined #openstack-kolla | 20:27 | |
| *** adrian_otto has joined #openstack-kolla | 20:28 | |
| *** TxGirlGeek has quit IRC | 20:29 | |
| *** jascott1_ has quit IRC | 20:30 | |
| *** MasterOfBugs has joined #openstack-kolla | 20:30 | |
| *** esmiurium has joined #openstack-kolla | 20:31 | |
| *** pramodrj07 has joined #openstack-kolla | 20:32 | |
| *** adrian_otto has quit IRC | 20:33 | |
| *** dave-mccowan has joined #openstack-kolla | 20:34 | |
| *** salv-orlando has joined #openstack-kolla | 20:38 | |
| *** jascott1_ has joined #openstack-kolla | 20:40 | |
| *** salv-orlando has quit IRC | 20:43 | |
| *** jascott1_ has quit IRC | 20:44 | |
| *** adrian_otto has joined #openstack-kolla | 20:46 | |
| *** david-lyle has joined #openstack-kolla | 20:47 | |
| *** cliles has quit IRC | 20:47 | |
| *** jascott1_ has joined #openstack-kolla | 20:47 | |
| *** adrian_otto has quit IRC | 20:48 | |
| *** adrian_otto has joined #openstack-kolla | 20:50 | |
| *** jascott1_ has quit IRC | 20:52 | |
| * portdirect reads scrollback | 20:55 | |
| inc0 | kfox1111, up and about? | 20:56 |
| portdirect | kfox1111: its not tiller thats busted, but the docker in centos at the moment cannot *pull* come images. | 20:56 |
| portdirect | so to unblock sdkae i simply ran: docker tag gcr.io/kubernetes-helm/tiller:v2.1.3 docker.io/port/tiller:v2.1.3, and then got him to use the image from dockerhub. | 20:57 |
| portdirect | (i think its the dash in the repo/image name that messes with docker) | 20:58 |
| inc0 | really? | 20:59 |
| *** adrian_otto1 has joined #openstack-kolla | 20:59 | |
| kfox1111 | back | 20:59 |
| kfox1111 | inc0: whats up? | 20:59 |
| inc0 | I mean...is this 1.13 niceness? | 20:59 |
| inc0 | https://review.openstack.org/#/c/418550/ <- gate fails are strangew | 21:00 |
| inc0 | but I think it's no longer issue of registry | 21:00 |
| kfox1111 | did docker 1.13 block gcr.io then? | 21:00 |
| portdirect | inc0: well a bug in centos's docker | 21:00 |
| portdirect | kfox1111: no | 21:00 |
| portdirect | but its having troble | 21:00 |
| kfox1111 | cause centos's docker is pulling fine as far as I can tell. you mean docker inc's centos build? | 21:00 |
| *** adrian_otto has quit IRC | 21:00 | |
| portdirect | i think (and conjecture here) it may be the '-' in the namespace | 21:01 |
| portdirect | as dockerhub does not have them | 21:01 |
| kfox1111 | ah. interesting. | 21:01 |
| portdirect | has affected machines I have set up this week | 21:01 |
| inc0 | did you guys talk to helm people about that? they'd be interested | 21:01 |
| portdirect | yeah - I tryy and find the bug report again | 21:02 |
| *** jascott1_ has joined #openstack-kolla | 21:02 | |
| openstackgerrit | Edmund Rhudy proposed openstack/kolla: Add pip-base image https://review.openstack.org/423411 | 21:03 |
| inc0 | erhudy, so I'm not huge fan of adding this as separate image | 21:03 |
| inc0 | that will potentially break backwards compatibility | 21:03 |
| inc0 | if people would for example have well defined build list, this change will mess them up | 21:04 |
| inc0 | I'd rather put installation of pip to base | 21:05 |
| inc0 | since we do it in openstack-base and few others | 21:05 |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: WIP: test rebooting in the gate https://review.openstack.org/413193 | 21:06 |
| *** sayantani01 has quit IRC | 21:06 | |
| sbezverk | kfox1111: reconfiugring is working!! :-) | 21:08 |
| erhudy | inc0: ok | 21:08 |
| kfox1111 | sbezverk: nice. :) | 21:08 |
| erhudy | inc0: if i'm putting it in the base image will you be opposed if i add an extra config option to copy a pip.conf into the build context | 21:09 |
| erhudy | instead of requiring multiline echo hackery in the header/footer to insert it | 21:10 |
| inc0 | so I'd be ok with putting pip into base image and add logic to pass custom configuration to it | 21:12 |
| erhudy | how about we deal with the pip configuration issue separately for now, i'll just move get-pip installation into base | 21:13 |
| erhudy | could i get your eyes on https://review.openstack.org/#/c/422249/ since the get-pip curl command would go into an area already being modified by that PR and it'll avoid a rebase | 21:14 |
| *** sayantani01 has joined #openstack-kolla | 21:15 | |
| kfox1111 | the new docker 1.13 --squash feature is interesting. | 21:16 |
| inc0 | it is kfox1111, however if docker says it's experimental | 21:19 |
| kfox1111 | sure. | 21:19 |
| inc0 | and docker is something that routinely breaks on upgrades of stuff that's supposed to be stable\ | 21:19 |
| inc0 | I'm afraid. | 21:19 |
| kfox1111 | yup. | 21:19 |
| kfox1111 | thats why I prefer the centos docker stuff. | 21:19 |
| kfox1111 | have had far fewer problems with that. | 21:20 |
| inc0 | erhudy, I'll merge this change in a second, but for future reference | 21:20 |
| inc0 | to avoid rebasing you can have a branch with more than one change | 21:20 |
| inc0 | and gerrit will handle it fine | 21:21 |
| inc0 | like, you can make a branch with block for dump-init and then cherry-pick pip change | 21:21 |
| inc0 | and when you do git review, it will post both patches | 21:21 |
| inc0 | and all gates and everything will work with correct patch series | 21:21 |
| erhudy | okay, i haven't gotten into Advanced Gerrit yet | 21:22 |
| inc0 | also following patch won't merge until previous patch merges | 21:22 |
| erhudy | i'll work on my gerrit-fu | 21:27 |
| *** jascott1_ has quit IRC | 21:27 | |
| *** jascott1_ has joined #openstack-kolla | 21:27 | |
| sbezverk | kfox1111: busy? | 21:28 |
| *** jascott1_ has quit IRC | 21:29 | |
| *** jascott1_ has joined #openstack-kolla | 21:29 | |
| erhudy | actually, why is it that the binary pip packages get installed but then get-pip.py is installed anyway | 21:30 |
| *** jascott1_ has quit IRC | 21:31 | |
| erhudy | if get-pip.py goes into the base image, it'll clobber binary pip everywhere, so it makes sense to then remove python-pip packages, but some people might not like mixing apt/yum-packaged python stuff with get-pip.py if it's a source install | 21:31 |
| erhudy | yakshaving begins | 21:31 |
| erhudy | if it's NOT a source install, i mean, if it's all binary | 21:32 |
| *** l4yerffeJ_ has joined #openstack-kolla | 21:34 | |
| *** satyar has quit IRC | 21:34 | |
| *** Jeffrey4l__ has joined #openstack-kolla | 21:35 | |
| *** Jeffrey4l_ has quit IRC | 21:35 | |
| *** l4yerffeJ has quit IRC | 21:35 | |
| *** cliles has joined #openstack-kolla | 21:36 | |
| *** TxGirlGeek has joined #openstack-kolla | 21:37 | |
| *** jascott1_ has joined #openstack-kolla | 21:38 | |
| *** salv-orlando has joined #openstack-kolla | 21:39 | |
| *** kristian__ has quit IRC | 21:39 | |
| *** kristian__ has joined #openstack-kolla | 21:40 | |
| *** jascott1_ has quit IRC | 21:41 | |
| *** jascott1_ has joined #openstack-kolla | 21:43 | |
| *** salv-orlando has quit IRC | 21:44 | |
| *** kristian__ has quit IRC | 21:44 | |
| kfox1111 | sbezverk: back. | 21:44 |
| kfox1111 | whats up? | 21:44 |
| *** david-lyle has quit IRC | 21:45 | |
| sbezverk | kfox1111: sent you email wit hthe log.. strange thing happens | 21:45 |
| kfox1111 | sbezverk: oh. I see you sent a mail. sec. | 21:45 |
| erhudy | inc0: i guess i still favor pip-base for that reason because people may not like a non-packaged version of pip being installed in the base image even for binary builds, though i find i'm not a great judge of what people might object to :v | 21:46 |
| *** klindgren_ has joined #openstack-kolla | 21:46 | |
| kfox1111 | sbezverk: still got it running? | 21:47 |
| *** jascott1_ has quit IRC | 21:48 | |
| kfox1111 | so we can poke at it a bit? | 21:48 |
| sbezverk | kfox1111: yep here is http://paste.openstack.org/show/595845/ | 21:48 |
| *** jascott1_ has joined #openstack-kolla | 21:48 | |
| *** l4yerffeJ_ has quit IRC | 21:48 | |
| sbezverk | it still points to old configmap | 21:48 |
| *** klindgren has quit IRC | 21:49 | |
| kfox1111 | I've never tried doing a rolling upgrade with just one replica before. | 21:49 |
| *** l4yerffeJ_ has joined #openstack-kolla | 21:49 | |
| kfox1111 | and without an explicit upgrade policy. | 21:49 |
| *** jascott1_ has quit IRC | 21:49 | |
| *** jascott1_ has joined #openstack-kolla | 21:49 | |
| kfox1111 | maybe do a 'kolla rollout status <deployment-name>" ? | 21:50 |
| kfox1111 | cause the deployment can be upgraded, but the switching from old/new rs could be stuck. | 21:50 |
| sbezverk | kfox1111: neither have I I was hoping to get technosophos involved | 21:51 |
| sbezverk | I do not use cli | 21:51 |
| sbezverk | helm commands are generated by the operator | 21:51 |
| *** jascott__ has joined #openstack-kolla | 21:52 | |
| *** jtriley has quit IRC | 21:52 | |
| kfox1111 | the deployment policy is a k8s object level thing though. | 21:52 |
| kfox1111 | helm's just passing the upgrade command and the new template. | 21:52 |
| kfox1111 | see the spec.strategyrollingUpdate in helm/kolla-common/templates/_common_api_python_deployment.yaml | 21:53 |
| kfox1111 | aso, see what 'kubectl get deployments' shows | 21:53 |
| kfox1111 | it should display: | 21:54 |
| kfox1111 | NAME DESIRED CURRENT UP-TO-DATE AVAILABLE AGE | 21:54 |
| *** jascott1_ has quit IRC | 21:54 | |
| sbezverk | kfox1111: one sec let me check | 21:54 |
| *** jascott__ has quit IRC | 21:54 | |
| kfox1111 | if UP-TO-DATE is != DESIRED, that may be an indicator. | 21:54 |
| sbezverk | kubectl get deployment memcached-deployment-154b6510b2d9 -n kolla | 21:55 |
| sbezverk | NAME DESIRED CURRENT UP-TO-DATE AVAILABLE AGE | 21:55 |
| sbezverk | memcached-deployment-154b6510b2d9 1 1 1 1 29m | 21:55 |
| kfox1111 | hmm... | 21:55 |
| kfox1111 | what about kubectl describe deployment memcached-deployment-154b6510b2d9 | 21:55 |
| sbezverk | kfox1111: it is strange, it says updated: http://paste.openstack.org/show/595847/ | 21:56 |
| kfox1111 | hmm... looks like the default is 1 max unavailable, 1 max surge. so it should make progress under that.. | 21:57 |
| kfox1111 | probably should set it to 1,0 though for memcached. | 21:57 |
| *** aj_power has quit IRC | 21:57 | |
| *** QuentinM has quit IRC | 21:58 | |
| *** QuentinM has joined #openstack-kolla | 22:00 | |
| kfox1111 | the logs look good. the state looks like it did something... | 22:00 |
| kfox1111 | kubectl get replicaset -n kolla | 22:00 |
| sbezverk | kfox1111: maybe for such changes rollingupgrade must be configured in deployment spec | 22:01 |
| *** adrian_otto1 has quit IRC | 22:02 | |
| sbezverk | kfox1111: http://paste.openstack.org/show/595848/ | 22:02 |
| kfox1111 | hmm.. that looks a bit strange. | 22:03 |
| kfox1111 | usualy I thought there would be two rs's. | 22:03 |
| kfox1111 | one for the old and one for the new. | 22:03 |
| *** sdake has joined #openstack-kolla | 22:03 | |
| kfox1111 | so rollback would work. | 22:03 |
| kfox1111 | maybe its a side effect of not having an explicit policy. | 22:04 |
| kfox1111 | but again, the policy on the describe deployment looks ok to me... | 22:04 |
| kfox1111 | hmm.. but http://paste.openstack.org/show/595847/ has no OldReplicaSets defined. | 22:05 |
| kfox1111 | how very weird. | 22:05 |
| sbezverk | kfox1111: glad that I am not the only one puzzled ;) | 22:05 |
| kfox1111 | how about: kubectl get rs memcached-deployment-154b6510b2d9-3719714807 -o yaml | 22:06 |
| kfox1111 | -n kolla | 22:07 |
| sbezverk | kfox1111: http://paste.openstack.org/show/595849/ | 22:08 |
| openstackgerrit | Michal Jastrzebski (inc0) proposed openstack/kolla-kubernetes: WIP - local registry in gates https://review.openstack.org/418550 | 22:09 |
| kfox1111 | configmap name is 001 there... | 22:09 |
| kfox1111 | how about kubectl get deployment memcached-deployment-154b6510b2d9 -o yaml -n kolla | 22:09 |
| kfox1111 | it feels like ti didnt actually update the rs. | 22:10 |
| sbezverk | kfox1111: right creation time and update time is the same | 22:12 |
| sbezverk | http://paste.openstack.org/show/595850/ | 22:12 |
| sbezverk | so it is helm did not push update even though it seems was sucessful | 22:13 |
| *** QuentinM has quit IRC | 22:13 | |
| kfox1111 | configmap name is wrong there too.... | 22:13 |
| kfox1111 | yeah. | 22:13 |
| *** QuentinM has joined #openstack-kolla | 22:14 | |
| kfox1111 | hmm... | 22:14 |
| kfox1111 | doh. | 22:14 |
| kfox1111 | I missed it the first time: | 22:14 |
| kfox1111 | helm upgrade memcached-deployment-154b6510b2d9 http://helm-repo:8879/memcached-deployment-0.4.0-1.tgz --debug --dry-run --namespace kolla --set "operator_configmap=memcached-002" | 22:14 |
| kfox1111 | --dry-run | 22:14 |
| *** shardy has quit IRC | 22:22 | |
| *** mgiles has quit IRC | 22:22 | |
| sbezverk | kfox1111: DAMN!! | 22:23 |
| openstackgerrit | Merged openstack/kolla: Add block for dumb-init installation https://review.openstack.org/422249 | 22:24 |
| *** QuentinM has quit IRC | 22:32 | |
| *** jheroux has quit IRC | 22:34 | |
| *** kklimonda has quit IRC | 22:34 | |
| sbezverk | kfox1111: it is working now, new pod points to the updated config map.. | 22:36 |
| sbezverk | that was a very good catch with --dry-run :) | 22:36 |
| *** salv-orlando has joined #openstack-kolla | 22:40 | |
| *** goldyfruit has quit IRC | 22:43 | |
| *** salv-orlando has quit IRC | 22:44 | |
| kfox1111 | thanks. :) | 22:44 |
| kfox1111 | it just felt like something wasn't actually running. :) | 22:45 |
| * kfox1111 has been bitten by --dry-run like options more then he'd like to admit. | 22:45 | |
| *** dave-mccowan has quit IRC | 22:46 | |
| *** kklimonda has joined #openstack-kolla | 22:49 | |
| *** QuentinM has joined #openstack-kolla | 22:50 | |
| openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes: WIP: test rebooting in the gate https://review.openstack.org/413193 | 22:51 |
| *** aj_power has joined #openstack-kolla | 22:52 | |
| kfox1111 | bbiab. | 22:52 |
| *** msimonin has quit IRC | 22:57 | |
| *** masber has quit IRC | 22:59 | |
| *** masber has joined #openstack-kolla | 23:00 | |
| cliles | will any look at kolla bug 1654133? I would really like a working dashboard again... | 23:02 |
| openstack | bug 1654133 in kolla "horizon dashboard broken in master with centos binaries" [Undecided,New] https://launchpad.net/bugs/1654133 | 23:02 |
| *** schwicht has quit IRC | 23:04 | |
| openstackgerrit | Serguei Bezverkhi proposed openstack/kolla-kubernetes: WIP memcached operator https://review.openstack.org/418073 | 23:05 |
| *** schwicht has joined #openstack-kolla | 23:09 | |
| *** schwicht has quit IRC | 23:10 | |
| *** adrian_otto has joined #openstack-kolla | 23:16 | |
| *** portdirect has quit IRC | 23:18 | |
| sdake | sbezverk kfox1111 ca nyou pleaee ack this and related changes: https://review.openstack.org/#/c/423046/ | 23:21 |
| sdake | srwilkers_ ^^ | 23:21 |
| sdake | inc0 ^^ | 23:21 |
| sdake | it has been verified by mgiels | 23:22 |
| sdake | mgiles | 23:22 |
| *** portdirect has joined #openstack-kolla | 23:25 | |
| *** sayantani01 has quit IRC | 23:26 | |
| *** TxGirlGeek has quit IRC | 23:27 | |
| sbezverk | sdake: why pin to 1.5.1? | 23:28 |
| sbezverk | 1.5.2 is out | 23:28 |
| sdake | ok - news to me | 23:28 |
| sbezverk | I think you need to make it more generic | 23:28 |
| sdake | suggestions - I want people to be able to copy and paste into their terminal | 23:29 |
| *** v1k0d3n has joined #openstack-kolla | 23:29 | |
| sdake | in the meantime, i'd prefer to just modify the docs on changes | 23:29 |
| sdake | i'll take on responsibility for updating the docs as needed | 23:29 |
| sdake | is there a "latest-release" uri I can use? | 23:29 |
| sdake | the gate is running 1.5.1 right? | 23:30 |
| sdake | not 1.5.2 | 23:30 |
| sdake | we want the docs to match the gate | 23:30 |
| sdake | sup v1k0d3n - been busy with work and kids - how are things going | 23:30 |
| sdake | sbezverk the goal here is to get the docs into shape as soon as possible | 23:31 |
| sdake | sbezverk not make them gramitcally beutiful :) | 23:31 |
| sbezverk | 1.5.2 | 23:31 |
| sdake | sbezverk and not to create perfection | 23:31 |
| sdake | sbezverk if the gate is using 15.2, i'll update the docs to reflect thati n a followon patch | 23:31 |
| sdake | (today) | 23:31 |
| sbezverk | but then in 2 years people reading this doc will still use 1.5.1 ;) | 23:31 |
| sdake | m0ar places need changes | 23:32 |
| sdake | sbezverk the docs are published live to docs.oo | 23:32 |
| sdake | sbezverk when people reference the docs, they will reference the latest version - not the version in the git repo :) | 23:32 |
| sdake | sbezverk trust me when I say docs.oo for kolla gets a ton of hits :) | 23:32 |
| sbezverk | well, I like copy and paste command lines | 23:32 |
| sbezverk | and this doc actually suggests by its formating | 23:33 |
| *** QuentinM has quit IRC | 23:33 | |
| sbezverk | user can easily copy and paste, if you define two vars at the beggining and assign them variable them further commands can be used with ANY future versions | 23:34 |
| sdake | sbezverk ya - don't sweat he version updates, we do that for kolla and kolla-ansible as wel | 23:34 |
| *** QuentinM has joined #openstack-kolla | 23:34 | |
| sdake | sbezverk cool i'll do that in a followon today | 23:34 |
| *** schwicht has joined #openstack-kolla | 23:34 | |
| sdake | sbezverk doing it now, just want the tested docs as they are to go into the repository | 23:35 |
| sdake | since a change will require a new sanity test from someone as docker -engine is busted | 23:35 |
| *** masuberu has quit IRC | 23:36 | |
| sbezverk | sdake: not sure how "busted" docker can prevent testing curl :) | 23:36 |
| sbezverk | but since I do not have busted docker, I can run curl for you to confirm | 23:37 |
| sdake | sbezverk ok - well just ack those - i am working on a new patch for your request | 23:37 |
| *** salv-orlando has joined #openstack-kolla | 23:41 | |
| openstackgerrit | Marcus Williams proposed openstack/kolla-ansible: [WIP] Split Openvswitch into own role https://review.openstack.org/416793 | 23:41 |
| *** v1k0d3n has quit IRC | 23:43 | |
| openstackgerrit | Steven Dake proposed openstack/kolla-kubernetes: Use environment variables to specify versions https://review.openstack.org/423555 | 23:44 |
| *** schwicht has quit IRC | 23:44 | |
| openstackgerrit | Steven Dake proposed openstack/kolla-kubernetes: Use environment variables to specify versions https://review.openstack.org/423555 | 23:45 |
| *** salv-orlando has quit IRC | 23:45 | |
| openstackgerrit | Steven Dake proposed openstack/kolla-kubernetes: Use environment variables to specify versions https://review.openstack.org/423555 | 23:45 |
| *** jascott1_ has joined #openstack-kolla | 23:47 | |
| sdake | sbezverk there are 5 patches waiting for acks, need reviews on all of them plz | 23:48 |
| *** QuentinM has quit IRC | 23:48 | |
| sdake | sbezverk i saw you -1'ed the version issue we just discussed on irc, that is covered in that last patch (which needs to be validated) | 23:48 |
| jascott1_ | kfox1111 you say my pm wrt to technosophos and relative paths in charts? | 23:49 |
| jascott1_ | say/saw | 23:49 |
| sbezverk | sdake: tried your new curl and it is failing | 23:49 |
| sbezverk | left a comment | 23:49 |
| sdake | you see, this is why we test eac hpatch indivudally :) | 23:49 |
| sdake | sbezverk can you ack the others pls :) | 23:49 |
| sdake | jascott1_ i didn't see it, what did he have to say :) | 23:50 |
| sdake | afternoon to ya btw jascott1_ :) | 23:50 |
| *** inc0 has quit IRC | 23:50 | |
| sbezverk | sdake: https://review.openstack.org/#/c/423120/2 has not addressed valid comment | 23:51 |
| jascott1_ | hes down to review the PR, they didnt add them originally because requirements.yaml becomes context based but if everyone is ok with that he is also | 23:51 |
| jascott1_ | and a fine afternoon to you to sdake! | 23:51 |
| jascott1_ | getting hyped up as we near the drinking/music playing time | 23:51 |
| sdake | sbezverk wierd I thought I fixed that | 23:51 |
| sdake | sbezverk thanks for pionting that out | 23:51 |
| sdake | sbezverk i'll have to do a git show -5 and make sure i didn't put it in the wrong patch ;) | 23:51 |
| sdake | sbezverk oddly its in my local copy - this should sync it up with gerrit properly | 23:54 |
| openstackgerrit | Steven Dake proposed openstack/kolla-kubernetes: Clean up mysql installed during test of helm https://review.openstack.org/423120 | 23:54 |
| openstackgerrit | Steven Dake proposed openstack/kolla-kubernetes: Use environment variables to specify versions https://review.openstack.org/423555 | 23:54 |
| *** QuentinM has joined #openstack-kolla | 23:54 | |
| openstackgerrit | Steven Dake proposed openstack/kolla-kubernetes: Clean up mysql installed during test of helm https://review.openstack.org/423120 | 23:54 |
| openstackgerrit | Steven Dake proposed openstack/kolla-kubernetes: Use environment variables to specify versions https://review.openstack.org/423555 | 23:54 |
| * sdake shaves the yak | 23:55 | |
| sbezverk | sdake: must be something with your editor, please compare line 430 and line 435 | 23:55 |
| sdake | try now | 23:55 |
| sbezverk | https://review.openstack.org/#/c/423120/3/doc/source/development-environment.rst | 23:55 |
| sdake | wrong link | 23:56 |
| sdake | /4 | 23:56 |
| sdake | patch set 3 was to reset gerrit's rebase | 23:56 |
| sdake | patch set 4 is where it is fixed | 23:56 |
| sbezverk | still the same | 23:56 |
| sdake | here is what I have: | 23:57 |
| sdake | helm install stable/mysql --name helm-test | 23:57 |
| sdake | helm ls | 23:57 |
| sdake | # and to check via kubectl | 23:57 |
| sdake | kubectl get all | 23:57 |
| sdake | # and finally remove the test mysql chart | 23:57 |
| sdake | helm delete helm-test -purge | 23:57 |
| sdake | should the syntax be different? | 23:57 |
| sbezverk | helm delete helm-test -purge | 23:57 |
| sbezverk | do you see here --purge?? | 23:57 |
| sdake | you mean use --purge rather then -purge? | 23:58 |
| *** sacharya has joined #openstack-kolla | 23:58 | |
| sbezverk | yes | 23:58 |
| *** kklimonda has quit IRC | 23:58 | |
| *** lamt has quit IRC | 23:58 | |
| openstackgerrit | Steven Dake proposed openstack/kolla-kubernetes: Clean up mysql installed during test of helm https://review.openstack.org/423120 | 23:58 |
| openstackgerrit | Steven Dake proposed openstack/kolla-kubernetes: Use environment variables to specify versions https://review.openstack.org/423555 | 23:58 |
| *** aj_power has quit IRC | 23:59 | |
| *** kklimonda has joined #openstack-kolla | 23:59 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!