| *** zhiwei has quit IRC | 00:00 | |
| *** dwalsh has quit IRC | 00:03 | |
| *** sbezverk has quit IRC | 00:04 | |
| *** ravig has joined #openstack-kolla | 00:06 | |
| *** salv-orlando has joined #openstack-kolla | 00:06 | |
| *** dgonzalez has quit IRC | 00:08 | |
| *** salv-orlando has quit IRC | 00:13 | |
| *** dgonzalez has joined #openstack-kolla | 00:15 | |
| *** ravig has quit IRC | 00:16 | |
| *** Rodrigo_BR has quit IRC | 00:19 | |
| *** fragatina has joined #openstack-kolla | 00:23 | |
| *** fragatina has quit IRC | 00:27 | |
| *** diogogmt has quit IRC | 00:46 | |
| *** zhiwei has joined #openstack-kolla | 01:00 | |
| *** phuongnh has joined #openstack-kolla | 01:01 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Fix the native ovsdb_interace failed https://review.openstack.org/336352 | 01:02 |
|---|---|---|
| *** salv-orlando has joined #openstack-kolla | 01:10 | |
| *** daneyon has joined #openstack-kolla | 01:11 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla: TEST OVSDB CONNECTION SOCK https://review.openstack.org/337391 | 01:14 |
| *** daneyon has quit IRC | 01:16 | |
| *** JianqingJiang has joined #openstack-kolla | 01:17 | |
| *** zhurong has joined #openstack-kolla | 01:22 | |
| *** fragatina has joined #openstack-kolla | 01:24 | |
| *** salv-orlando has quit IRC | 01:28 | |
| *** fragatina has quit IRC | 01:29 | |
| *** ssurana has joined #openstack-kolla | 01:38 | |
| *** aernhart has quit IRC | 01:48 | |
| *** salv-orlando has joined #openstack-kolla | 01:56 | |
| *** diogogmt has joined #openstack-kolla | 02:07 | |
| *** Jianqing_ has joined #openstack-kolla | 02:09 | |
| *** JianqingJiang has quit IRC | 02:09 | |
| *** salv-orlando has quit IRC | 02:13 | |
| *** fragatina has joined #openstack-kolla | 02:24 | |
| *** fragatina has quit IRC | 02:29 | |
| *** haplo37 has quit IRC | 02:33 | |
| *** g3ek has quit IRC | 02:34 | |
| *** klint has joined #openstack-kolla | 02:37 | |
| *** haplo37 has joined #openstack-kolla | 02:38 | |
| *** g3ek has joined #openstack-kolla | 02:39 | |
| *** salv-orlando has joined #openstack-kolla | 02:40 | |
| *** williamcaban has joined #openstack-kolla | 02:44 | |
| *** britthouser has joined #openstack-kolla | 02:47 | |
| *** dwalsh has joined #openstack-kolla | 02:56 | |
| *** phuongnh has quit IRC | 02:56 | |
| *** salv-orlando has quit IRC | 02:58 | |
| *** coolsvap has joined #openstack-kolla | 02:59 | |
| *** daneyon has joined #openstack-kolla | 03:00 | |
| *** daneyon has quit IRC | 03:04 | |
| *** williamcaban has quit IRC | 03:09 | |
| *** salv-orlando has joined #openstack-kolla | 03:24 | |
| *** fragatina has joined #openstack-kolla | 03:25 | |
| *** fragatina has quit IRC | 03:30 | |
| *** dwalsh has quit IRC | 03:32 | |
| *** Jianqing_ has quit IRC | 03:36 | |
| *** salv-orlando has quit IRC | 03:40 | |
| *** g3ek has quit IRC | 03:48 | |
| *** haplo37 has quit IRC | 03:48 | |
| *** haplo37 has joined #openstack-kolla | 03:53 | |
| *** g3ek has joined #openstack-kolla | 03:54 | |
| *** ssurana has quit IRC | 03:55 | |
| *** ssurana has joined #openstack-kolla | 03:57 | |
| *** dwalsh has joined #openstack-kolla | 04:00 | |
| *** salv-orlando has joined #openstack-kolla | 04:06 | |
| *** ravig has joined #openstack-kolla | 04:12 | |
| *** JianqingJiang has joined #openstack-kolla | 04:17 | |
| *** salv-orlando has quit IRC | 04:18 | |
| *** dwalsh has quit IRC | 04:24 | |
| *** fragatina has joined #openstack-kolla | 04:26 | |
| *** zhurong has quit IRC | 04:26 | |
| *** fragatina has quit IRC | 04:32 | |
| *** salv-orlando has joined #openstack-kolla | 04:45 | |
| *** daneyon has joined #openstack-kolla | 04:48 | |
| *** daneyon has quit IRC | 04:52 | |
| *** g3ek has quit IRC | 05:01 | |
| *** haplo37 has quit IRC | 05:01 | |
| *** g3ek has joined #openstack-kolla | 05:01 | |
| *** zhurong has joined #openstack-kolla | 05:03 | |
| *** salv-orlando has quit IRC | 05:05 | |
| *** haplo37 has joined #openstack-kolla | 05:07 | |
| *** salv-orlando has joined #openstack-kolla | 05:10 | |
| *** fragatina has joined #openstack-kolla | 05:28 | |
| *** fragatina has quit IRC | 05:32 | |
| *** kaushikc has joined #openstack-kolla | 05:33 | |
| *** ravig has quit IRC | 05:39 | |
| *** coolsvap has quit IRC | 05:42 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Fix the overwrite sudoers file issue https://review.openstack.org/336789 | 05:49 |
| *** coolsvap has joined #openstack-kolla | 05:52 | |
| *** ssurana has quit IRC | 06:00 | |
| *** Jianqing_ has joined #openstack-kolla | 06:04 | |
| *** Jianqin__ has joined #openstack-kolla | 06:05 | |
| *** Jianqing_ has quit IRC | 06:05 | |
| *** Jianqing_ has joined #openstack-kolla | 06:07 | |
| *** Jianqin__ has quit IRC | 06:07 | |
| *** JianqingJiang has quit IRC | 06:07 | |
| openstackgerrit | Merged openstack/kolla: Use the disk size as the osd weight when osd_initial_weight is auto https://review.openstack.org/331395 | 06:14 |
| openstackgerrit | Merged openstack/kolla: Added unit test for image operation https://review.openstack.org/334917 | 06:18 |
| Jeffrey4l_ | coolsvap, please review this PS. https://review.openstack.org/336352 it block the gate. thanks | 06:19 |
| *** coolsvap has quit IRC | 06:20 | |
| *** coolsvap has joined #openstack-kolla | 06:22 | |
| *** coolsvap has quit IRC | 06:25 | |
| *** coolsvap has joined #openstack-kolla | 06:25 | |
| *** fragatina has joined #openstack-kolla | 06:28 | |
| openstackgerrit | Merged openstack/kolla: Updated from global requirements https://review.openstack.org/337331 | 06:31 |
| *** JianqingJiang has joined #openstack-kolla | 06:32 | |
| *** Jianqing_ has quit IRC | 06:32 | |
| *** fragatina has quit IRC | 06:33 | |
| *** salv-orlando has quit IRC | 06:34 | |
| *** daneyon has joined #openstack-kolla | 06:36 | |
| *** salv-orlando has joined #openstack-kolla | 06:36 | |
| *** daneyon has quit IRC | 06:41 | |
| *** salv-orlando has quit IRC | 06:41 | |
| *** salv-orlando has joined #openstack-kolla | 07:07 | |
| *** fragatina has joined #openstack-kolla | 07:10 | |
| *** mewald1 has joined #openstack-kolla | 07:14 | |
| Jeffrey4l_ | coolsvap, could u check this PS https://review.openstack.org/336352 it block the gate. thanks. | 07:14 |
| *** cmceniry has quit IRC | 07:18 | |
| *** fragatina has quit IRC | 07:18 | |
| *** cmceniry has joined #openstack-kolla | 07:18 | |
| *** fragatina has joined #openstack-kolla | 07:19 | |
| *** mewald1 has left #openstack-kolla | 07:19 | |
| *** salv-orlando has quit IRC | 07:19 | |
| coolsvap | Jeffrey4l_, I was checking the same this looks good the failures in ubuntu and oraclelinux gates seem unrelated | 07:19 |
| Jeffrey4l_ | coolsvap, yes. the ubuntu failed due to the python-libvirt/libvirt. it is wired. It shouldn't happen. I have no idea about this. | 07:20 |
| *** Jianqing_ has joined #openstack-kolla | 07:22 | |
| *** JianqingJiang has quit IRC | 07:22 | |
| coolsvap | yes | 07:22 |
| *** JianqingJiang has joined #openstack-kolla | 07:24 | |
| *** Jianqing_ has quit IRC | 07:24 | |
| *** JianqingJiang has quit IRC | 07:28 | |
| *** haplo37 has quit IRC | 07:33 | |
| *** g3ek has quit IRC | 07:34 | |
| *** Satya_ has joined #openstack-kolla | 07:37 | |
| Satya_ | Hi All | 07:37 |
| Satya_ | Need help on kolla deployment | 07:38 |
| Satya_ | i am trying a multinode deployment | 07:38 |
| Satya_ | while deploy getting a error at creating log volume | 07:38 |
| Satya_ | FAILED! => {"changed": false, "failed": true, "module_stderr": "", "module_stdout": "Traceback (most recent call last):\r\n File \"/root/.ansible/tmp/ansible-tmp-1467704065.6-20533520538734/kolla_docker\", line 2686, in <module>\r\n main()\r\n File \"/root/.ansible/tmp/ansible-tmp-1467704065.6-20533520538734/kolla_docker\", line 674, in main\r\n module = generate_nested_module()\r\n File \"/root/.ansible/tmp/ansible- | 07:38 |
| Satya_ | any help? | 07:38 |
| *** mewald1 has joined #openstack-kolla | 07:39 | |
| *** g3ek has joined #openstack-kolla | 07:39 | |
| *** haplo37 has joined #openstack-kolla | 07:40 | |
| *** salv-orlando has joined #openstack-kolla | 07:46 | |
| *** kaushikc has quit IRC | 07:55 | |
| *** salv-orlando has quit IRC | 08:04 | |
| *** tyrola has joined #openstack-kolla | 08:06 | |
| *** Serlex has quit IRC | 08:09 | |
| mewald1 | I could find some Zookeeper related code in Kolla. What's the context here? What does Zookeeper have to do with everything? | 08:10 |
| *** Serlex has joined #openstack-kolla | 08:10 | |
| *** zhurong has quit IRC | 08:11 | |
| *** shardy has joined #openstack-kolla | 08:12 | |
| openstackgerrit | Dave Walker proposed openstack/kolla: Build docker images for OpenStack Watcher https://review.openstack.org/337362 | 08:24 |
| *** daneyon has joined #openstack-kolla | 08:25 | |
| *** daneyon has quit IRC | 08:29 | |
| *** salv-orlando has joined #openstack-kolla | 08:31 | |
| *** zhiwei has quit IRC | 08:32 | |
| *** zhiwei has joined #openstack-kolla | 08:33 | |
| openstackgerrit | Dave Walker proposed openstack/kolla: Build docker images for OpenStack Watcher https://review.openstack.org/337362 | 08:38 |
| *** zhiwei has quit IRC | 08:38 | |
| *** zhiwei has joined #openstack-kolla | 08:39 | |
| *** pbourke_ has joined #openstack-kolla | 08:45 | |
| *** pbourke has quit IRC | 08:48 | |
| *** pbourke_ is now known as pbourke | 08:48 | |
| *** salv-orlando has quit IRC | 08:48 | |
| *** g3ek has quit IRC | 08:49 | |
| *** haplo37 has quit IRC | 08:49 | |
| openstackgerrit | Dave Walker proposed openstack/kolla: Add Watcher ansible roles and templates https://review.openstack.org/337556 | 08:52 |
| *** g3ek has joined #openstack-kolla | 08:55 | |
| *** haplo37 has joined #openstack-kolla | 08:55 | |
| openstackgerrit | Dave Walker proposed openstack/kolla: Build docker images for OpenStack Watcher https://review.openstack.org/337362 | 08:56 |
| *** athomas has joined #openstack-kolla | 09:00 | |
| openstackgerrit | Dave Walker proposed openstack/kolla: Add Watcher default settings https://review.openstack.org/337559 | 09:02 |
| Daviey | Is gate currently broken? | 09:14 |
| *** salv-orlando has joined #openstack-kolla | 09:15 | |
| *** narasimha_SV has joined #openstack-kolla | 09:19 | |
| narasimha_SV | what is the difference between specifying [heat] and [heat-base] in kolla-build.conf ??? | 09:20 |
| openstackgerrit | Takeaki Matsumoto proposed openstack/kolla: Add enable_iscsi option to etc/kolla/globals.yml https://review.openstack.org/337571 | 09:26 |
| openstackgerrit | Takeaki Matsumoto proposed openstack/kolla: Add enable_iscsi option to etc/kolla/globals.yml https://review.openstack.org/337571 | 09:28 |
| openstackgerrit | Dave Walker proposed openstack/kolla: Build docker images for OpenStack Watcher https://review.openstack.org/337362 | 09:28 |
| *** Satya_ has quit IRC | 09:29 | |
| *** salv-orlando has quit IRC | 09:30 | |
| narasimha_SV | what is the difference between specifying [heat] and [heat-base] in kolla-build.conf ??? | 09:31 |
| pbourke | narasimha_SV: there is no [heat] section | 09:36 |
| pbourke | narasimha_SV: there's also no image named just "heat" so adding [heat] will do nothin | 09:36 |
| pbourke | *nothing | 09:36 |
| narasimha_SV | no after adding [heat] i am providing my local git repo to build | 09:39 |
| narasimha_SV | I am trying to build heat for my environment with my local git server code I have | 09:39 |
| narasimha_SV | but I am unable to see the git code which I used to create Image in the deployed container | 09:40 |
| pbourke | narasimha_SV: update the [heat-base] section | 09:40 |
| pbourke | with your own repo | 09:40 |
| narasimha_SV | ok | 09:41 |
| narasimha_SV | then what about the heat components ? | 09:41 |
| narasimha_SV | could you please tell me exact things needed to be placed in kolla-build.conf | 09:42 |
| pbourke | it's documented in kolla-build.conf | 09:52 |
| pbourke | which part is not clear? | 09:52 |
| *** salv-orlando has joined #openstack-kolla | 09:56 | |
| *** tyrola has quit IRC | 10:05 | |
| *** bootsha has joined #openstack-kolla | 10:11 | |
| *** daneyon has joined #openstack-kolla | 10:13 | |
| *** Mech422 has quit IRC | 10:14 | |
| *** Mech422 has joined #openstack-kolla | 10:15 | |
| *** salv-orlando has quit IRC | 10:16 | |
| *** daneyon has quit IRC | 10:18 | |
| mewald1 | I want to deploy with Kolla stable/mitaka on Ubuntu Xenial. How can I install an older version of Ansible than 2.0.0 which is shipped with xenial? I tried "pip install ansible==1.9.6" but that doesnt give me ansible and ansible-playbook commands | 10:23 |
| *** mewald1 has quit IRC | 10:24 | |
| openstackgerrit | Paul Bourke proposed openstack/kolla: Add a check before wiping Ceph disks in bootstrap https://review.openstack.org/337594 | 10:26 |
| *** mewald1 has joined #openstack-kolla | 10:27 | |
| openstackgerrit | Paul Bourke proposed openstack/kolla: Add a check before wiping Ceph disks in bootstrap https://review.openstack.org/337594 | 10:28 |
| openstackgerrit | Paul Bourke proposed openstack/kolla: Add a check before wiping Ceph disks in bootstrap https://review.openstack.org/337594 | 10:29 |
| sean-k-mooney | mewald1: you can use pip | 10:30 |
| sean-k-mooney | mewald1: pip install -U ansible\<2.0.0.0 | 10:30 |
| sean-k-mooney | having said that i found it better to build the images using stable mitaka then swap to master then i deployed | 10:31 |
| *** dims has joined #openstack-kolla | 10:38 | |
| *** berendt has joined #openstack-kolla | 10:40 | |
| narasimha_SV | pbourke: after keeping heat-base in kolla-build.conf | 10:41 |
| narasimha_SV | i can see that heat-base is getting built from the code | 10:41 |
| narasimha_SV | even I see heat component images getting created | 10:42 |
| narasimha_SV | but when I an deploying these images I see a different code in side the heat folder inside venv/lib/python2.7/site-packages/heat | 10:42 |
| narasimha_SV | I dont see the code which I have used to build the images | 10:42 |
| *** salv-orlando has joined #openstack-kolla | 10:43 | |
| sean-k-mooney | narasimha_SV: this may be a stupid question but you are doing a source build right? | 10:44 |
| *** bootsha has quit IRC | 10:44 | |
| sean-k-mooney | narasimha_SV: you should have something like this in your kolla-build in /etc/kolla | 10:48 |
| sean-k-mooney | [heat-base] | 10:48 |
| sean-k-mooney | # | 10:48 |
| sean-k-mooney | # From kolla | 10:48 |
| sean-k-mooney | # | 10:48 |
| sean-k-mooney | # Source location type (string value) | 10:48 |
| sean-k-mooney | # Allowed values: local, git, url | 10:48 |
| sean-k-mooney | type = git | 10:48 |
| sean-k-mooney | # The location for source install (string value) | 10:48 |
| sean-k-mooney | location = file:///path/to/local/heat/repo | 10:48 |
| sean-k-mooney | # Git reference to pull, commit sha, tag or branch name (string value) | 10:48 |
| sean-k-mooney | reference = dev/branch | 10:48 |
| pbourke | narasimha_SV: make sure you remove the images from the target nodes first | 10:48 |
| pbourke | narasimha_SV: kolla-ansible cleanup | 10:48 |
| sean-k-mooney | narasimha_SV: be carefull with kolla-ansible cleanup as it will delete customised files in /etc/kolla | 10:49 |
| sean-k-mooney | narasimha_SV: the global.yml, passwords.yml and kolla-build will not be touched but the rest will be delete if i remember correctly | 10:50 |
| sean-k-mooney | i like to keep my inventory file in /etc/kollla when im doing multinode deployments and it always gets deleted when i run kolla-ansible cleanup | 10:51 |
| narasimha_SV | thanks guys I will try | 10:52 |
| pbourke | sean-k-mooney: that's strange it shouldnt touch anything on the operator node | 10:57 |
| pbourke | sean-k-mooney: unless you have your operator node as part of the cluster? | 10:57 |
| sean-k-mooney | yes it is | 10:58 |
| sean-k-mooney | i tend to do all in one then convert to multinode to make sure the images are working first | 10:59 |
| sean-k-mooney | pbourke: i tend to assume that people using kolla would not keep a seperate node for building/deploying and colocate it with one of there controler nodes. | 11:01 |
| *** salv-orlando has quit IRC | 11:01 | |
| pbourke | sean-k-mooney: id assume the opposite :p | 11:01 |
| pbourke | in my experience the operator node is always separate | 11:02 |
| pbourke | but we should fix the cleanup | 11:02 |
| pbourke | bad to be removing peoples configs | 11:02 |
| *** mewald has left #openstack-kolla | 11:02 | |
| *** salv-orlando has joined #openstack-kolla | 11:04 | |
| sean-k-mooney | pbourke: where the operator node is seperate would it typically be a vm or phyical node? | 11:04 |
| *** beekhof is now known as beekhof-ng-v2 | 11:04 | |
| pbourke | either I suppose | 11:04 |
| pbourke | depends if you're doing a vm or bare metal based deploy | 11:04 |
| sean-k-mooney | well you could use a vm for either | 11:05 |
| *** salv-orl_ has joined #openstack-kolla | 11:05 | |
| sean-k-mooney | my assumtion that the operator node is not seperate was based on small deployment of <10 nodes where you would not justify the expecse of deicatin 10% just for an operator cloud | 11:06 |
| sean-k-mooney | i can see it running in a vm in such a case | 11:06 |
| sean-k-mooney | /deicatin/dedicateing/ maybe? spelling is not my strong suit | 11:07 |
| Jeffrey4l_ | coolsvap, see the comments. https://review.openstack.org/331420 | 11:09 |
| *** salv-orlando has quit IRC | 11:10 | |
| openstackgerrit | Paul Bourke proposed openstack/kolla: Add a check before wiping Ceph disks in bootstrap https://review.openstack.org/337594 | 11:10 |
| pbourke | hey Jeffrey4l_ | 11:10 |
| Jeffrey4l_ | pbourke, hi | 11:10 |
| pbourke | thanks for your input on the ceph disk stuff, my patch would have made the partition based stuff difficult | 11:11 |
| pbourke | Jeffrey4l_: I took another look at Mech422's patch and shelling out to sgdisk solves the problem where udev is old | 11:11 |
| pbourke | Jeffrey4l_: could you take another look? https://review.openstack.org/#/c/326609/ | 11:11 |
| patchbot | pbourke: patch 326609 - kolla - Read GPT label using sgdisk rather than udev | 11:11 |
| coolsvap | Jeffrey4l_, sorry for the recheck, thank you for the explanation, i think we need to effectively use Depends-On | 11:12 |
| Jeffrey4l_ | does Depends-On work in the one project? I thought it only work on cross project. | 11:12 |
| Jeffrey4l_ | in one project, if u you depends-on some, just rebase on it works. | 11:13 |
| Jeffrey4l_ | pbourke, yup | 11:13 |
| Jeffrey4l_ | coolsvap, ^ | 11:14 |
| sean-k-mooney | pbourke: regarding you cpeh change you should also need to do the same check for KOLLA_CEPH_OSD_CACHE_BOOTSTRAP | 11:14 |
| Jeffrey4l_ | pbourke, do u have the PS link from Mech422? | 11:15 |
| Jeffrey4l_ | sorry. found it. | 11:15 |
| pbourke | sean-k-mooney: i thought I am? | 11:15 |
| pbourke | sean-k-mooney: (missed it in the first patchset) | 11:15 |
| * Jeffrey4l_ is checking. | 11:15 | |
| coolsvap | Jeffrey4l_, it should be fine if we do rebase, the only thing is rebase on base change makes it difficult to manage the next change sets but either way works if we get the information | 11:16 |
| sean-k-mooney | pbourke: i acidentally put my cache partition on the same drive as my os the first time i deployed cepth with kolla | 11:16 |
| sean-k-mooney | pbourke: actully you are | 11:16 |
| sean-k-mooney | pbourke: its just not mentioned in the commit message or in the release note | 11:16 |
| pbourke | sean-k-mooney: got you, will update | 11:17 |
| Jeffrey4l_ | coolsvap, so what should i do now? add a depends-on? | 11:17 |
| coolsvap | Jeffrey4l_, nothing as of now | 11:17 |
| Jeffrey4l_ | ok. | 11:17 |
| openstackgerrit | Dave Walker proposed openstack/kolla: Build docker images for OpenStack Watcher https://review.openstack.org/337362 | 11:17 |
| coolsvap | i think we need better mechanism for communication if we have dependencies in project | 11:18 |
| Jeffrey4l_ | coolsvap, don't rebase or depends-on work? | 11:19 |
| coolsvap | Jeffrey4l_, as i said either way works :) | 11:19 |
| Jeffrey4l_ | so why u say, we need better ... in my understanding, this mean: current solution are not good( rebase/depends-on), so we need think up a better one. :p | 11:21 |
| coolsvap | Jeffrey4l_, current solutions are good but not used | 11:22 |
| coolsvap | :) | 11:22 |
| Jeffrey4l_ | ok | 11:22 |
| Jeffrey4l_ | thanks for explain | 11:22 |
| Jeffrey4l_ | pbourke, https://review.openstack.org/326609 comments are made. | 11:27 |
| coolsvap | Jeffrey4l_, I prefer Depends-On since we can manage updates to individual changes more efficiently | 11:28 |
| Jeffrey4l_ | coolsvap, but, this PS depend on the un-exist ones, which need a backport PS. So actually, nothing can be done now. :( | 11:29 |
| coolsvap | Jeffrey4l_, agreed thats why i said nothing to do now :) | 11:30 |
| Jeffrey4l_ | understand | 11:30 |
| *** williamcaban has joined #openstack-kolla | 11:33 | |
| pbourke | Jeffrey4l_: thanks! | 11:33 |
| openstackgerrit | Dave Walker proposed openstack/kolla: Build docker images for OpenStack Watcher https://review.openstack.org/337362 | 11:35 |
| openstackgerrit | Dave Walker proposed openstack/kolla: Build docker images for OpenStack Watcher https://review.openstack.org/337362 | 11:36 |
| Jeffrey4l_ | :D | 11:37 |
| *** zhiwei has quit IRC | 11:38 | |
| openstackgerrit | Paul Bourke proposed openstack/kolla: Add a check before wiping Ceph disks in bootstrap https://review.openstack.org/337594 | 11:38 |
| pbourke | sean-k-mooney: so when you say you accidentally put the cache partition on the OS drive, did it wipe your drive? | 11:43 |
| *** zhiwei has joined #openstack-kolla | 11:44 | |
| sean-k-mooney | pbourke: yep | 11:47 |
| pbourke | sean-k-mooney: woops :) | 11:47 |
| sean-k-mooney | linux is awsome that way | 11:47 |
| sean-k-mooney | most operating systems wont delete the partion table for the dirve its running from but linux has no issues with this | 11:48 |
| *** zhiwei has quit IRC | 11:49 | |
| sean-k-mooney | pbourke: my assumtion when i first tried to use the ceph support was since i was using partition lable to identify the partition to use it would not touch the rest | 11:50 |
| *** zhiwei has joined #openstack-kolla | 11:50 | |
| pbourke | sean-k-mooney: yeah that's exactly the thinking that prompted this change | 11:50 |
| pbourke | sean-k-mooney: I didnt think it would ever happen but there we go | 11:50 |
| sean-k-mooney | well at least there is a warning in the docs now. | 11:51 |
| pbourke | only if you use the parted command as shown by the docs | 11:51 |
| sean-k-mooney | pbourke: your also working on support for non gpt partions too correct? | 11:53 |
| sean-k-mooney | pretty much all os installers still create msdos partion tables by default | 11:54 |
| sean-k-mooney | having to use gpt partions if you wanted to share a driver between the os and as a cache/journal drive was always a bit of a pain because of that | 11:55 |
| *** zhiwei has quit IRC | 11:55 | |
| pbourke | sean-k-mooney: i was more working on a "non udev" approach which would have the side affect of not needed GPT. But I sidelined that for now as its hard to support all the different disk layouts without having the user specify a lot of convoluted info | 11:57 |
| pbourke | i think Jeffrey4l_ mentioned msdos labels are possible though | 11:57 |
| sean-k-mooney | pbourke: yes i used an msdos lable when kolla nuked my os dirve... so they work | 11:58 |
| Jeffrey4l_ | pbourke, do not like msdos. it is too old. and with it, we can not use extend journal, too. | 11:58 |
| sean-k-mooney | the issue is that the labels we use are too long | 11:58 |
| Jeffrey4l_ | *external | 11:59 |
| Jeffrey4l_ | why we need support msdos? | 11:59 |
| sean-k-mooney | Jeffrey4l_: it is the default partion table created by all linux distros that kolla supports | 11:59 |
| sean-k-mooney | e.g. if you install centos/rhel or ubuntu on the host it will create an msdos partion table via its installer | 12:00 |
| Jeffrey4l_ | sean-k-mooney, but use gpt has no issue right? any side effect? | 12:00 |
| sean-k-mooney | Jeffrey4l_: the installer dont let you select it | 12:01 |
| sean-k-mooney | at least not ubuntu | 12:01 |
| sean-k-mooney | i cant remember if centos/fedora do but i dont think so | 12:01 |
| *** zhiwei has joined #openstack-kolla | 12:01 | |
| sean-k-mooney | as a result you have to install the os then convert to a gpt partion table and reinstall grub | 12:02 |
| Jeffrey4l_ | sean-k-mooney, no. we do not require all the partition table are gpt. | 12:02 |
| Jeffrey4l_ | all the partition/all the disk | 12:02 |
| Jeffrey4l_ | sean-k-mooney, what's your use case? | 12:03 |
| openstackgerrit | Merged openstack/kolla: Fix publicURL endpoint not found issue in CI https://review.openstack.org/327146 | 12:03 |
| sean-k-mooney | Jeffrey4l_: i want to share the os drive which is an ssd with ceph to use as an external journal for the spinning disk | 12:03 |
| Jeffrey4l_ | hmm.. ok | 12:04 |
| Jeffrey4l_ | it is a case. But, honestly, share the OS disk with the ceph disk is not a good idea. | 12:05 |
| *** zhiwei has quit IRC | 12:06 | |
| *** rhallisey has joined #openstack-kolla | 12:06 | |
| sean-k-mooney | well it depend on your setup. im hoping to build our a ceph setup for ci soone but the server i will be useing only have 3 drive bays so 1 ssd for os and journal and 2 4TB hardrives fro storage each | 12:07 |
| sean-k-mooney | having a deicated drive for external journals or a cache there would reduce my capacity alot | 12:08 |
| *** dwalsh has joined #openstack-kolla | 12:08 | |
| sean-k-mooney | for what its worth though im just going to use a gpt partion table but its a little extra work to create the iniall os image | 12:10 |
| *** zhiwei has joined #openstack-kolla | 12:12 | |
| Jeffrey4l_ | OK. let me think about this. | 12:15 |
| *** zhiwei has quit IRC | 12:17 | |
| sean-k-mooney | actully it look like ubuntus install can create a gpt partition table its just not clear at first | 12:17 |
| *** openstackgerrit has quit IRC | 12:19 | |
| *** openstackgerrit has joined #openstack-kolla | 12:19 | |
| *** rhallisey has quit IRC | 12:19 | |
| *** rhallisey has joined #openstack-kolla | 12:23 | |
| sean-k-mooney | it looks like centos/fedroa anaconda installer can use gpt partion tables but you need to append inst.gpt to the commandline someware for disks less then 2TB to use GPT | 12:25 |
| *** ravig has joined #openstack-kolla | 12:29 | |
| *** zhiwei has joined #openstack-kolla | 12:29 | |
| Daviey | Anyone else seeing gate deploy jobs, booted instances hittng error state? I'm seeing it in all flavors... http://logs.openstack.org/62/337362/8/check/gate-kolla-dsvm-deploy-oraclelinux-source/edc0231/console.html#_2016-07-05_12_12_19_759366 | 12:32 |
| *** zhiwei has quit IRC | 12:34 | |
| *** ssurana has joined #openstack-kolla | 12:37 | |
| *** ssurana has quit IRC | 12:42 | |
| *** zhiwei has joined #openstack-kolla | 12:51 | |
| *** klint has quit IRC | 12:54 | |
| *** zhiwei has quit IRC | 12:56 | |
| mewald1 | To deploy with stable/mitaka is it ok to run xenial on the hosts? | 13:00 |
| *** zhiwei has joined #openstack-kolla | 13:08 | |
| sean-k-mooney | yes you should be able to do that though now that 16.04 is now useing systemd the docker configuration is closer to that of centos | 13:11 |
| *** zhiwei has quit IRC | 13:13 | |
| pbourke | Jeffrey4l_: I dont fully understand your comment on https://review.openstack.org/#/c/337594/ - can you explain a little more? | 13:18 |
| patchbot | pbourke: patch 337594 - kolla - Add a check before wiping Ceph disks in bootstrap | 13:18 |
| Jeffrey4l_ | pbourke, why u restrict to work with external_journal=false? how about external_journal=true? | 13:19 |
| pbourke | Jeffrey4l_: when external_journal=true it doesn't zap the disk | 13:20 |
| *** zhurong has joined #openstack-kolla | 13:20 | |
| Jeffrey4l_ | hmm. maybe i mixed up. let me check again. | 13:21 |
| *** banix has joined #openstack-kolla | 13:25 | |
| *** zhiwei has joined #openstack-kolla | 13:25 | |
| *** williamcaban has quit IRC | 13:28 | |
| *** zhiwei has quit IRC | 13:29 | |
| *** ravig has quit IRC | 13:30 | |
| *** dasTor has joined #openstack-kolla | 13:30 | |
| mewald1 | I just deployed to my network nodes. The servers crashed when deploying metadata-agent. They are not reachable on the network nor can I access them via lights out management. The lom console shows something like "NMI watchdog: BUG: soft lockup - CPU#5 stuck for 23s! [neutron-lbaasv2:21550]" for all containers as far as I can see | 13:35 |
| mewald1 | any ideas? | 13:35 |
| *** dwalsh has quit IRC | 13:40 | |
| *** jtriley has joined #openstack-kolla | 13:43 | |
| *** ayoung has joined #openstack-kolla | 13:43 | |
| *** williamcaban has joined #openstack-kolla | 13:46 | |
| *** zhiwei has joined #openstack-kolla | 13:47 | |
| *** diogogmt has quit IRC | 13:47 | |
| *** diogogmt has joined #openstack-kolla | 13:49 | |
| openstackgerrit | Paul Bourke proposed openstack/kolla: Use IP address rather than hostname for Ceph bootstrap https://review.openstack.org/337673 | 13:49 |
| *** williamcaban has quit IRC | 13:51 | |
| *** zhiwei has quit IRC | 13:52 | |
| *** ravig has joined #openstack-kolla | 13:55 | |
| *** britthou_ has joined #openstack-kolla | 13:59 | |
| *** britthouser has quit IRC | 13:59 | |
| *** huikang has joined #openstack-kolla | 14:01 | |
| *** diogogmt has quit IRC | 14:02 | |
| *** zhiwei has joined #openstack-kolla | 14:04 | |
| *** zhiwei has quit IRC | 14:08 | |
| *** ravig has quit IRC | 14:12 | |
| *** zhurong has quit IRC | 14:18 | |
| *** mdnadeem has joined #openstack-kolla | 14:20 | |
| *** zhiwei has joined #openstack-kolla | 14:21 | |
| *** mdnadeem_ has quit IRC | 14:23 | |
| *** zhiwei has quit IRC | 14:25 | |
| *** ssurana has joined #openstack-kolla | 14:27 | |
| *** narasimha_SV has quit IRC | 14:28 | |
| mewald1 | We have neutron_external_interface set to a vlan interface on linux bond (bond0.203) Is there any reason why this should not work with OVS on top? | 14:30 |
| *** ssurana has quit IRC | 14:31 | |
| *** zhiwei has joined #openstack-kolla | 14:39 | |
| Daviey | mewald1: Good luck.. exact same config here... I couldn't get it to work and switched to Linuxbridge. Let me know if you get it working | 14:39 |
| mewald1 | we decided to switch to linux bridge now, too :D But I will keep digging in a different environment | 14:42 |
| *** daneyon has joined #openstack-kolla | 14:43 | |
| *** zhiwei has quit IRC | 14:44 | |
| *** daneyon has quit IRC | 14:48 | |
| *** papacz has joined #openstack-kolla | 14:50 | |
| *** zhiwei has joined #openstack-kolla | 14:56 | |
| *** diogogmt has joined #openstack-kolla | 14:58 | |
| openstackgerrit | Paul Bourke proposed openstack/kolla: Read GPT label using sgdisk rather than udev https://review.openstack.org/326609 | 14:58 |
| *** ppowell has joined #openstack-kolla | 15:00 | |
| *** zhiwei has quit IRC | 15:01 | |
| openstackgerrit | Paul Bourke proposed openstack/kolla: Use IP address rather than hostname for Ceph bootstrap https://review.openstack.org/337673 | 15:01 |
| *** ravig has joined #openstack-kolla | 15:06 | |
| *** stvnoyes2 has quit IRC | 15:07 | |
| *** stvnoyes1 has joined #openstack-kolla | 15:07 | |
| openstackgerrit | Paul Bourke proposed openstack/kolla: Improve task names for Ceph https://review.openstack.org/337718 | 15:08 |
| *** Jeffrey4l_ has quit IRC | 15:11 | |
| *** ppowell has quit IRC | 15:14 | |
| *** jogam has quit IRC | 15:14 | |
| *** zhiwei has joined #openstack-kolla | 15:18 | |
| *** zhiwei has quit IRC | 15:22 | |
| *** ppowell has joined #openstack-kolla | 15:31 | |
| *** openstackgerrit has quit IRC | 15:33 | |
| *** openstackgerrit has joined #openstack-kolla | 15:33 | |
| *** zhiwei has joined #openstack-kolla | 15:35 | |
| *** zhiwei has quit IRC | 15:40 | |
| *** harlowja has joined #openstack-kolla | 15:40 | |
| *** ppowell has quit IRC | 15:40 | |
| *** vhosakot has joined #openstack-kolla | 15:47 | |
| *** ravig has quit IRC | 15:47 | |
| *** ravig has joined #openstack-kolla | 15:48 | |
| *** jmccarthy has joined #openstack-kolla | 15:50 | |
| *** jmccarthy has left #openstack-kolla | 15:50 | |
| *** ppowell has joined #openstack-kolla | 15:56 | |
| *** zhiwei has joined #openstack-kolla | 15:57 | |
| kfox1111 | morning. | 15:58 |
| *** zhiwei has quit IRC | 16:01 | |
| *** sdake has joined #openstack-kolla | 16:06 | |
| *** sdake has quit IRC | 16:11 | |
| *** zhiwei has joined #openstack-kolla | 16:14 | |
| openstackgerrit | Paul Bourke proposed openstack/kolla: Use inventory_hostname for Ceph bootstrap https://review.openstack.org/337673 | 16:17 |
| *** zhiwei has quit IRC | 16:18 | |
| openstackgerrit | Paul Bourke proposed openstack/kolla: Use inventory_hostname for Ceph bootstrap https://review.openstack.org/337673 | 16:19 |
| *** athomas has quit IRC | 16:30 | |
| *** arizidon has left #openstack-kolla | 16:30 | |
| *** zhiwei has joined #openstack-kolla | 16:31 | |
| *** Serlex has quit IRC | 16:34 | |
| *** zhiwei has quit IRC | 16:35 | |
| Mech422 | Daviey: it works with OVS brigdes if you don't start the ovs database container | 16:38 |
| Mech422 | mewald1: is it a linux native vlan or ovs vlan ? | 16:39 |
| Mech422 | Morning | 16:40 |
| *** berendt has quit IRC | 16:45 | |
| Mech422 | sean-k-mooney: I think I just have a ansible task that converts my Ubuntu drive to gpt non-destructively with sgdisk | 16:47 |
| Mech422 | sean-k-mooney: my test machines only have 1 drive, so I'm doing root/ceph on the same device | 16:49 |
| *** zhiwei has joined #openstack-kolla | 16:53 | |
| *** zhurong has joined #openstack-kolla | 16:56 | |
| *** zhiwei has quit IRC | 16:57 | |
| *** pbourke_ has joined #openstack-kolla | 16:58 | |
| *** ayoung has quit IRC | 17:02 | |
| sean-k-mooney | oh interesting. at least form the ubuntu installer point of view it looks like i can create a gpt partition table vi the ui if you manually partition the disk | 17:03 |
| *** banix has quit IRC | 17:04 | |
| *** banix has joined #openstack-kolla | 17:05 | |
| *** ravig has quit IRC | 17:09 | |
| *** lyncos has joined #openstack-kolla | 17:10 | |
| *** zhiwei has joined #openstack-kolla | 17:10 | |
| *** lyncos has quit IRC | 17:11 | |
| Mech422 | sean-k-mooney: do you need a preseed file ? I have one for 14.04 - no 16.04 yet | 17:11 |
| sean-k-mooney | no i can create it myself. the lack of support for msdos partions was never a blocker for us we can use GPT it just was an extra step | 17:13 |
| *** zhiwei has quit IRC | 17:14 | |
| Mech422 | sean-k-mooney: I 'drive' kolla with another set of ansible playbooks | 17:18 |
| Mech422 | sean-k-mooney: I actually build the deployment host from a pre-deployment host, and patch various Kolla things during the build/deploy | 17:19 |
| Mech422 | sean-k-mooney: this lets me apply patches to use ceph on partitions, use OpenVSwitch for the native host networking, etc. | 17:20 |
| *** unicell has joined #openstack-kolla | 17:22 | |
| *** vinkman has joined #openstack-kolla | 17:22 | |
| *** daneyon has joined #openstack-kolla | 17:26 | |
| *** zhiwei has joined #openstack-kolla | 17:27 | |
| *** ravig has joined #openstack-kolla | 17:27 | |
| Mech422 | This makes me wonder what kind of president she'll be - knowing roughly 50% of the country thinks she's the devil? She seems like a vindictive witch... | 17:28 |
| Mech422 | I could see her being even worse then normal, just to 'teach us a lesson' | 17:29 |
| Mech422 | blah - sorry | 17:29 |
| *** daneyon has quit IRC | 17:30 | |
| *** zhiwei has quit IRC | 17:31 | |
| *** shardy has quit IRC | 17:31 | |
| *** fragatina has quit IRC | 17:34 | |
| *** zhurong has quit IRC | 17:47 | |
| *** zhiwei has joined #openstack-kolla | 17:48 | |
| kfox1111 | no worries. | 17:49 |
| openstackgerrit | Merged openstack/kolla-kubernetes: Allow kubectl to use its own configured endpoint for kube API https://review.openstack.org/335627 | 17:50 |
| *** zhiwei has quit IRC | 17:53 | |
| *** mewald1 has quit IRC | 17:55 | |
| *** zhiwei has joined #openstack-kolla | 18:05 | |
| *** ravig has quit IRC | 18:07 | |
| *** ravig has joined #openstack-kolla | 18:07 | |
| *** unicell has quit IRC | 18:09 | |
| *** zhiwei has quit IRC | 18:10 | |
| openstackgerrit | David Wang proposed openstack/kolla-kubernetes: Refactor, speed up, and fix bugs in jinja template dict generation https://review.openstack.org/336745 | 18:14 |
| *** unicell has joined #openstack-kolla | 18:14 | |
| -openstackstatus- NOTICE: Job instability resulting from a block storage connectivity error on mirror.iad.rax.openstack.org has been corrected; jobs running in rax-iad should be more reliable again. | 18:14 | |
| *** ayoung has joined #openstack-kolla | 18:16 | |
| *** mewald1 has joined #openstack-kolla | 18:17 | |
| *** zhiwei has joined #openstack-kolla | 18:22 | |
| *** zhiwei has quit IRC | 18:27 | |
| *** ravig has quit IRC | 18:31 | |
| *** ravig has joined #openstack-kolla | 18:31 | |
| *** ravig has quit IRC | 18:35 | |
| *** ravig has joined #openstack-kolla | 18:37 | |
| *** zhiwei has joined #openstack-kolla | 18:39 | |
| *** zhiwei has quit IRC | 18:43 | |
| *** vinkman has left #openstack-kolla | 18:47 | |
| *** ravig has quit IRC | 19:01 | |
| *** zhiwei has joined #openstack-kolla | 19:01 | |
| *** ravig has joined #openstack-kolla | 19:04 | |
| *** zhiwei has quit IRC | 19:05 | |
| *** ravig has quit IRC | 19:11 | |
| wirehead_ | Breaking news: NASA’s Juno probe encountered a mysterious black object in Jupiter orbit. The last transmission from the probe was “Oh my god, it’s full of Ansibles” | 19:11 |
| Mech422 | wirehead_: the probe found religion ? | 19:12 |
| wirehead_ | To deploy the universe, first you must make a playbook. | 19:13 |
| *** ravig has joined #openstack-kolla | 19:13 | |
| *** sogabe has quit IRC | 19:13 | |
| *** daneyon has joined #openstack-kolla | 19:14 | |
| *** daneyon has quit IRC | 19:18 | |
| *** berendt has joined #openstack-kolla | 19:19 | |
| *** lyncos has joined #openstack-kolla | 19:22 | |
| *** zhiwei has joined #openstack-kolla | 19:23 | |
| *** ravig has quit IRC | 19:25 | |
| *** zhiwei has quit IRC | 19:27 | |
| *** coolsvap has quit IRC | 19:32 | |
| mewald1 | I've heard that running MariaDB and RabbitMQ in a Docker container was not a very good idea with regards to stability. The person who said it couldn't really give any reasons. What is your experience with that? Do I need to be worried about that? | 19:36 |
| Mech422 | mewald1: you can use an existing MariaDB cluster if you like... | 19:37 |
| *** inc0 has joined #openstack-kolla | 19:37 | |
| Mech422 | mewald1: not sure about rabbitmq | 19:37 |
| Mech422 | Hey inc0 :-) | 19:37 |
| inc0 | hey, good afternoon | 19:37 |
| *** zhiwei has joined #openstack-kolla | 19:40 | |
| *** godleon has joined #openstack-kolla | 19:41 | |
| *** zhiwei has quit IRC | 19:45 | |
| mewald1 | Mech422: is there any description of how external mariadb works? | 19:51 |
| Mech422 | mewald1: I believe you just dont enable mariadb in the globals.yml, and provide the connection info. for your existing mariadb | 19:55 |
| *** zhiwei has joined #openstack-kolla | 19:57 | |
| *** ravig has joined #openstack-kolla | 20:00 | |
| *** ravig has quit IRC | 20:01 | |
| *** zhiwei has quit IRC | 20:02 | |
| *** ravig has joined #openstack-kolla | 20:08 | |
| mewald1 | Mech422: ahh makes sense. That should easily be possible for rabbitmq, too | 20:08 |
| *** ravig has quit IRC | 20:13 | |
| *** ravig has joined #openstack-kolla | 20:13 | |
| *** ravig has quit IRC | 20:22 | |
| *** fragatina has joined #openstack-kolla | 20:25 | |
| *** zhiwei has joined #openstack-kolla | 20:25 | |
| *** jogam has joined #openstack-kolla | 20:27 | |
| *** inc0 has quit IRC | 20:29 | |
| *** ravig has joined #openstack-kolla | 20:31 | |
| dcwangmit01_ | mewald1: Running MariaDB as a plain old docker container shouldn't be any less stable than running it as a normal process on a machine. It's basically the same, and running it as a docker container enables a whole lot of operational benefits. For example, sometimes system updating a library updates on a host system triggers a bunch of dependency updates which can mess up an environment for | 20:32 |
| dcwangmit01_ | a running process like mariadb. If mariadb is containerized, the chances of messing it up lessen because most of its dependencies are within the container. Upgrade becomes easier. I would argue that the Kolla project exists *because* of these operational benefits. | 20:32 |
| *** zhiwei has quit IRC | 20:33 | |
| dcwangmit01_ | ^ change "a running process" to "an installed service" | 20:33 |
| wirehead_ | yeah, I’m a little hazy on why it’s more ‘dangerous’ other than running MariaDB or RabbitMQ in general in a world where systems crash, core routers fail causing a netsplit inside of a datacenter, et al. Maybe advice left over from the early days of containers where it was hard to map persistant storage into a container properly? | 20:35 |
| *** ravig has quit IRC | 20:43 | |
| *** williamcaban has joined #openstack-kolla | 20:48 | |
| *** zhiwei has joined #openstack-kolla | 20:50 | |
| *** ravig has joined #openstack-kolla | 20:52 | |
| *** ravig has quit IRC | 20:53 | |
| openstackgerrit | Merged openstack/kolla-kubernetes: Update kolla-kubernetes installation instructions https://review.openstack.org/334616 | 20:54 |
| *** zhiwei has quit IRC | 20:55 | |
| *** williamcaban has quit IRC | 20:55 | |
| *** v1k0d3n has joined #openstack-kolla | 20:56 | |
| *** unicell has quit IRC | 20:56 | |
| *** lyncos has quit IRC | 20:58 | |
| *** rhallisey has quit IRC | 21:00 | |
| *** ppowell has quit IRC | 21:00 | |
| *** rhallisey has joined #openstack-kolla | 21:01 | |
| *** daneyon has joined #openstack-kolla | 21:02 | |
| openstackgerrit | Ken Wronkiewicz proposed openstack/kolla-kubernetes: Add documentation for api_interface_address https://review.openstack.org/327961 | 21:04 |
| *** unicell has joined #openstack-kolla | 21:04 | |
| *** rhallisey has quit IRC | 21:05 | |
| *** daneyon has quit IRC | 21:07 | |
| mewald1 | wirehead_: What were the problems of mapping persistent storage into containers? I am not using docker for long enough to know I guess :D | 21:11 |
| wirehead_ | mewald1: Well, if you are using containers for development, you don’t map any volumes. Thus, every time the container is restarted, the storage is cleared. | 21:12 |
| wirehead_ | s/restarted/reconstructed/g | 21:12 |
| openstackgerrit | Merged openstack/kolla-kubernetes: Document briefly how services in k8s work and why 0.0.0.0 is OK. https://review.openstack.org/333636 | 21:13 |
| wirehead_ | If I create a Docker Volume that puts persistant storage where RabbitMQ or MariaDB intends to store it’s on disk-files, then the storage will continue to function if I destroy a container and then re-create it. | 21:14 |
| wirehead_ | If you set up persistant storage for those docker volumes on your choice of distributed filesystem, then you can kill that container and start it up on a different host and everything is fine. | 21:14 |
| wirehead_ | When Docker first came out, it was really quite daring to use it for production, whereas these days it’s totally OK. | 21:15 |
| wirehead_ | If you peek under the covers and look at what Docker is doing, it’s just creating namespaces, control groups, and union filesystem mounts in the kernel to create the illusion of multiple system images on the same system. | 21:16 |
| *** fragatina has quit IRC | 21:19 | |
| wirehead_ | Given that, many years ago, I worked on a stock market platform and our queue would blow up when run atop a hypervisor, so to me, I’m actually more comfortable running RabbitMQ inside of a container, now that the technology has matured somewhat, than I am running it atop a hypervisor. Even though that’s entirely an irrational sort of guess at what would be more stable, just based on past experience. | 21:20 |
| *** jtriley_ has joined #openstack-kolla | 21:21 | |
| *** jtriley has quit IRC | 21:24 | |
| *** zhiwei has joined #openstack-kolla | 21:24 | |
| *** jtriley_ has quit IRC | 21:25 | |
| *** zhiwei has quit IRC | 21:28 | |
| *** ravig has joined #openstack-kolla | 21:34 | |
| dcwangmit01_ | wirehead_: Container storage doesn't auto-destruct unless you run with the docker "--rm" flag. So you don't need to tell mariadb to write to /var/lib/mysql in order to save it, but putting data files outside of docker is definitely safer | 21:40 |
| *** banix has quit IRC | 21:41 | |
| *** zhiwei has joined #openstack-kolla | 21:41 | |
| mewald1 | thanks for all the input :) learned something | 21:43 |
| *** zhiwei has quit IRC | 21:46 | |
| dcwangmit01_ | mewald1: Well, it's an opportunity for us to regurgitate our understanding, and have people correct us where wrong. kfox1111 the other day corrected my understanding of petsets | 21:48 |
| *** ravig has quit IRC | 21:51 | |
| *** aernhart has joined #openstack-kolla | 21:52 | |
| *** britthou_ has quit IRC | 21:56 | |
| *** mewald1 has left #openstack-kolla | 21:57 | |
| *** zhiwei has joined #openstack-kolla | 21:58 | |
| *** ravig has joined #openstack-kolla | 21:58 | |
| *** ravig has quit IRC | 21:59 | |
| *** ravig has joined #openstack-kolla | 22:02 | |
| *** zhiwei has quit IRC | 22:03 | |
| openstackgerrit | Greg Herlein proposed openstack/kolla-kubernetes: updated hyperkube version to 1.3 https://review.openstack.org/337903 | 22:06 |
| *** ravig has quit IRC | 22:09 | |
| dcwangmit01_ | Is anyone having trouble kolla-building keystone and horizon? | 22:09 |
| dcwangmit01_ | They keep failing on: | 22:09 |
| dcwangmit01_ | INFO:kolla.image.build.horizon: xstatic-angular-fileupload-common.noarch 0:12.0.4.0-1.el7 | 22:09 |
| dcwangmit01_ | INFO:kolla.image.build.horizon:Failed: | 22:09 |
| dcwangmit01_ | INFO:kolla.image.build.horizon: httpd.x86_64 0:2.4.6-40.el7.centos.1 | 22:09 |
| dcwangmit01_ | INFO:kolla.image.build.horizon:Complete! | 22:09 |
| *** ravig has joined #openstack-kolla | 22:10 | |
| *** ravig has quit IRC | 22:20 | |
| *** zhiwei has joined #openstack-kolla | 22:20 | |
| dcwangmit01_ | INFO:kolla.image.build.keystone:Error unpacking rpm package httpd-2.4.6-40.el7.centos.1.x86_64 | 22:22 |
| dcwangmit01_ | INFO:kolla.image.build.keystone: | 22:22 |
| dcwangmit01_ | INFO:kolla.image.build.keystone: | 22:22 |
| dcwangmit01_ | INFO:kolla.image.build.keystone:error: unpacking of archive failed on file /usr/sbin/suexec: cpio: cap_set_file | 22:22 |
| dcwangmit01_ | INFO:kolla.image.build.horizon: | 22:22 |
| dcwangmit01_ | Ah crap. It's the Docker Centos AUFS bug. Nevermind | 22:23 |
| openstackgerrit | Greg Herlein proposed openstack/kolla-kubernetes: Clarify installtion instructions to use the quickstart guide. https://review.openstack.org/337909 | 22:24 |
| *** zhiwei has quit IRC | 22:24 | |
| *** huikang has quit IRC | 22:25 | |
| *** salv-orl_ has quit IRC | 22:36 | |
| *** vhosakot has quit IRC | 22:36 | |
| *** zhiwei has joined #openstack-kolla | 22:37 | |
| *** salv-orlando has joined #openstack-kolla | 22:37 | |
| *** zhiwei has quit IRC | 22:41 | |
| *** salv-orlando has quit IRC | 22:42 | |
| *** ravig has joined #openstack-kolla | 22:47 | |
| *** rhallisey has joined #openstack-kolla | 22:50 | |
| *** daneyon has joined #openstack-kolla | 22:51 | |
| *** daneyon has quit IRC | 22:55 | |
| *** zhiwei has joined #openstack-kolla | 22:59 | |
| *** zhiwei has quit IRC | 23:03 | |
| *** salv-orlando has joined #openstack-kolla | 23:06 | |
| *** salv-orlando has quit IRC | 23:12 | |
| kfox1111 | dcwangmit01_: you there? | 23:15 |
| dcwangmit01_ | hi | 23:16 |
| kfox1111 | hi. :) | 23:16 |
| dcwangmit01_ | wirehead_: just turned around and said I am being summoned | 23:16 |
| *** zhiwei has joined #openstack-kolla | 23:16 | |
| dcwangmit01_ | What' up? | 23:16 |
| kfox1111 | I'm just unclear if the ability to be able to put stuff in kolla-kubernetes/kolla/ansible/$service/role/main.yml was intended and should make it into your patch, or if it was an unintended thing and should be left out. | 23:16 |
| dcwangmit01_ | The file kolla-kubernetes/kolla/ansible/$service/role/main.yml is being read by my patch | 23:17 |
| dcwangmit01_ | Are you thinking it's not | 23:17 |
| dcwangmit01_ | ? | 23:17 |
| kfox1111 | I'm thinking its not. | 23:17 |
| dcwangmit01_ | Okay. Let me look it up. | 23:18 |
| kfox1111 | say you have kolla installed in /usr/share/kolla. | 23:18 |
| kfox1111 | then you want to put some kolla-kubernetes specific overrides in kolla-kubernetes/kolla/ansible/$service/role/main.yml | 23:18 |
| kfox1111 | I think the current code works with that, but your patch does not? | 23:18 |
| dcwangmit01_ | Look at line 61 on the right side: https://review.openstack.org/#/c/336745/7/kolla_kubernetes/service.py | 23:18 |
| patchbot | dcwangmit01_: patch 336745 - kolla-kubernetes - Refactor, speed up, and fix bugs in jinja template... | 23:18 |
| dcwangmit01_ | if service_name is not None: | 23:19 |
| dcwangmit01_ | files.append(os.path.join(kolla_dir, 'ansible/roles', | 23:19 |
| dcwangmit01_ | service_name, 'defaults/main.yml')) | 23:19 |
| dcwangmit01_ | Oh role main | 23:19 |
| dcwangmit01_ | Hold | 23:19 |
| kfox1111 | yeah. kolla_dir is defined to be '/usr/share/kolla' in that case. | 23:19 |
| kfox1111 | which covers line 85 in the before patch file. | 23:20 |
| kfox1111 | but lne 87 does an aditional thing, looking in kolla-kubernetes/kolla/ansible/$service/role/main.yml | 23:20 |
| *** zhiwei has quit IRC | 23:20 | |
| kfox1111 | I'm not really sure if it was ever intended to do that though. just that its doing it. | 23:20 |
| dcwangmit01_ | I don't believe the code was ever hitting line 92 on the left side. | 23:21 |
| dcwangmit01_ | I've run kolla-kub every which way | 23:21 |
| dcwangmit01_ | But I can add it in. Let me think a big | 23:21 |
| dcwangmit01_ | bit | 23:21 |
| kfox1111 | ah... | 23:21 |
| kfox1111 | only if someone wanted to make a kolla-kubernetes specific role not in kolla. | 23:22 |
| kfox1111 | So, probably never used. yeah. | 23:22 |
| kfox1111 | still, some coded it that way. I wonder why. | 23:23 |
| dcwangmit01_ | When running kolla-kube as full-install, local_dir used to resolve to /usr/local which is ../../ of /usr/local/bin/kolla-kubernets. Then, the old code searched /usr/local/ + kolla/ansible/roles/$service/defaults/main.yml. This location is covered in the new code by PathFinder.find_kolla_dir() + ansible/roles/$service/defaults/main.yml | 23:27 |
| dcwangmit01_ | ^ So I think it works because I think they are equivalent | 23:27 |
| kfox1111 | yeah. I think the regular case is handled properly with the patch. | 23:28 |
| dcwangmit01_ | I have some blind confidence because I spent so much time testing every permutation of full install, virtualenv, and editable installs | 23:28 |
| dcwangmit01_ | The code was a little bit hard for me to follow, and I really really wanted to simplify it. Hopefully it is simpler. | 23:29 |
| kfox1111 | yeah. I'll believe that. Out of the box, kolla-kubernetes has no kolla subdir, and wouldn't ever not work. | 23:29 |
| kfox1111 | But I think if someone has a setup where they have added aditional roles in their kolla-kubernetes/kolla directory, it kind of unions in the old code but doesn't in the new. | 23:29 |
| kfox1111 | I'm pretty sure that doesn't matter. but someone coded it that way, which is weird. | 23:30 |
| kfox1111 | do we know who wrote it in the first place? | 23:30 |
| dcwangmit01_ | I kept support for that by Line 120 in: https://review.openstack.org/gitweb?p=openstack/kolla-kubernetes.git;a=blob;f=kolla_kubernetes/common/pathfinder.py | 23:30 |
| dcwangmit01_ | I really feel like we should support less paths. We should remove all that we can, so that we are working in the same environmental setupps | 23:30 |
| kfox1111 | I agree with that. | 23:31 |
| kfox1111 | The reason I ask is this sort of thing: | 23:32 |
| kfox1111 | https://review.openstack.org/#/c/334115/ | 23:32 |
| patchbot | kfox1111: patch 334115 - kolla - Horizon Frontend - (for kolla-kubernetes) | 23:32 |
| dcwangmit01_ | Le tme read | 23:32 |
| kfox1111 | Its a role that kind of is kubernetes specific. | 23:33 |
| kfox1111 | it could be housed in kolla-kubernetes instead of kolla, to make it easier to review. | 23:33 |
| kfox1111 | but would have to be unioned in somehow. which the old code seemed to kind of do. | 23:34 |
| dcwangmit01_ | Was the path here a typo? I'm not sure I follow the additional roles and union thing > But I think if someone has a setup where they have added aditional roles in their kolla-kubernetes/kolla directory, it kind of unions in the old code but doesn't in the new. | 23:35 |
| *** berendt has quit IRC | 23:36 | |
| dcwangmit01_ | I think a lot of the original kolla-kubernetes code was donated by someone from Mirantis | 23:36 |
| dcwangmit01_ | "git blame <file>" | 23:36 |
| dcwangmit01_ | Oh. You are making a "roles" directory within kolla-kubernetes | 23:37 |
| kfox1111 | for example, if I took review 334115, and put it in kolla-kubernetes/kolla I think it would be read in by the old code. | 23:37 |
| kfox1111 | right. | 23:37 |
| kfox1111 | while still reading in all the roles from /usr/share/kolla. | 23:37 |
| dcwangmit01_ | Oh. We can add it as a search path to the CONFIG_SEARCH_PATHS | 23:37 |
| dcwangmit01_ | So the searching for config and templates is really messed up in kolla-kube | 23:38 |
| kfox1111 | yeah. I think something like that might solve it too. | 23:38 |
| dcwangmit01_ | Let me elaborate. I didnt' cahnge any of this logic | 23:38 |
| kfox1111 | (I'm puzzled by that too. Mirantis decided to contribute the start of the kolla-kubernetes code, but then is making their own Fuel CCP too?) | 23:39 |
| dcwangmit01_ | The old and new code (logic unchanged) when locating service and bootstrap files locks onto the first path that exists. The old and new code when locating config files locks onto the first path where the file exists. | 23:41 |
| dcwangmit01_ | ^ Those are very different | 23:42 |
| dcwangmit01_ | So I don't see how the old code supported a union of kolla/ansible/roles and kolla-kubernetes/roles. Is that what you want, because I can figure out how to make that happen? | 23:45 |
| kfox1111 | yeah... I think I see what you mean. I guess its never worked so your patch works as is. maybe we discuss further if having kubernetes specific roles makes sense or not at some other point. | 23:47 |
| kfox1111 | I'd rather have the horizon-haproxy thing land in kolla anyway, but it might be difficult without an ansible implementation too. | 23:47 |
| kfox1111 | And I think an ansible implementation might be very hard. :/ | 23:47 |
| dcwangmit01_ | Okay. Just let me know if you need it. I have no probem adding it in. We just need to try not to make a 3rd way of pathfinding configs and templates | 23:48 |
| kfox1111 | What the solution there though is, I don't know. searching multiple places is just one way to solve it. | 23:48 |
| kfox1111 | k. sounds good. thanks. :) | 23:48 |
| dcwangmit01_ | I think we would special case it. https://review.openstack.org/#/c/336745/7/kolla_kubernetes/service.py Line 60, add the file if the file exists | 23:48 |
| patchbot | dcwangmit01_: patch 336745 - kolla-kubernetes - Refactor, speed up, and fix bugs in jinja template... | 23:48 |
| dcwangmit01_ | Isn't pretty | 23:48 |
| kfox1111 | yeah. I really don't like the magic of the service name. | 23:49 |
| kfox1111 | I wanted to pull in the haproxy image from a different service, but the magic of service name like that meant it was inaccessable. | 23:50 |
| dcwangmit01_ | I think we can override that now with this patch, in kolla-kubernetes | 23:50 |
| dcwangmit01_ | ./etc/kolla-kubernetes/kolla-kubernetes.yml | 23:50 |
| dcwangmit01_ | let me test it real quick | 23:51 |
| kfox1111 | I ended up with just making a new service in kolla and including the bits, ala https://review.openstack.org/#/c/334115/3/ansible/roles/horizon-frontend/tasks/main.yml so kolla-kubernetes could get at it. | 23:51 |
| patchbot | kfox1111: patch 334115 - kolla - Horizon Frontend - (for kolla-kubernetes) | 23:51 |
| kfox1111 | lools really ugly though. :/ | 23:51 |
| kfox1111 | I think we'll run into the issue as more things get kubernetesized. for example, I'd like to have a gallera cluster just for horizon. | 23:52 |
| kfox1111 | so I'd need access to the mysql role variables in a k8s rc file for that purpose. | 23:52 |
| dcwangmit01_ | Yeah, you can override only that one image if you use the patch we've been talking about. | 23:52 |
| dcwangmit01_ | Set mariadb_image_full: my_custom_image_location in ./etc/kolla-kubernetes/kolla-kubernetes.yml | 23:52 |
| dcwangmit01_ | vagrant@vagrant-ubuntu-trusty-64:~/Dev/kolla-kubernetes$ kolla-kubernetes jinja vars mariadb|grep mariadb_image_full | 23:53 |
| dcwangmit01_ | "mariadb_image_full": "gcr.io/annular-reef-121620/centos-binary-mariadb:3.0.0" | 23:53 |
| dcwangmit01_ | vagrant@vagrant-ubuntu-trusty-64:~/Dev/kolla-kubernetes$ kolla-kubernetes jinja vars mariadb|grep mariadb_image_full | 23:53 |
| dcwangmit01_ | "mariadb_image_full": "my_custom_image_location" | 23:53 |
| kfox1111 | yeah. that would work. just really ugly. :/ | 23:53 |
| dcwangmit01_ | I think the proper way to do it, which I found by reading in kolla. Is to set ./etc/kolla/global.yml the following varaibles: | 23:54 |
| kfox1111 | I think the only ones we may have multiple of are rabbit, mysql, and haproxy? Maybe we just always include those roles? | 23:54 |
| dcwangmit01_ | docker_registry: "gcr.io" | 23:54 |
| dcwangmit01_ | docker_namespace: "your_gcr_project" | 23:54 |
| dcwangmit01_ | Ohhhhh. I get what your problem is | 23:54 |
| dcwangmit01_ | Now I understand | 23:54 |
| kfox1111 | let me show you another review that uses the first... | 23:54 |
| dcwangmit01_ | To support 2 images, you don't want to edit the config each time. | 23:54 |
| kfox1111 | https://review.openstack.org/#/c/334122/ | 23:55 |
| patchbot | kfox1111: patch 334122 - kolla-kubernetes - Horizon Frontend | 23:55 |
| dcwangmit01_ | So that's why your review uses an array variable | 23:55 |
| kfox1111 | In that review, I make a haproxy just for horizon. | 23:55 |
| kfox1111 | it can/should just reuse the 'haproxy' container. | 23:55 |
| kfox1111 | yeah. | 23:56 |
| kfox1111 | I want the number of services run to be greater then the number of containers in kolla. | 23:56 |
| kfox1111 | so, maybe several haproxies, or several mysql's. | 23:56 |
| kfox1111 | rather then the one to one mapping that exists today. | 23:56 |
| dcwangmit01_ | Why did I think array variable. I can seem to find that in the Frontend nor Version patches | 23:57 |
| kfox1111 | so horizon-haproxy the rc would use the haproxy ansible role from kolla. | 23:57 |
| kfox1111 | it also plays with this patch set: https://review.openstack.org/#/c/333996/ | 23:58 |
| patchbot | kfox1111: patch 333996 - kolla-kubernetes - Horizon Versioning | 23:58 |
| kfox1111 | there are 3 total at the moment. THe last one lets you run 2 horizon clusters at the same time. The haproxy frontend lets you set a cookie on the browser side to determine which to talk to, and the third patch applies the config into kolla to allow kolla-kubernetes to get the haproxy image name. | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!