| klindgren_ | no it doesn't | 00:00 |
|---|---|---|
| SamYaple | are you saying they are not both active at the same time so it doesnt? | 00:00 |
| klindgren_ | you top the existing rpc service, you start the new container. As long as hostname is set in the config file its fine | 00:00 |
| klindgren_ | stop* | 00:00 |
| SamYaple | ok concern relieved there | 00:00 |
| SamYaple | i think, with the rename stuff, this could work for kolla-ansible without as much work as i initially stated, yes | 00:01 |
| klindgren_ | the rpc flow is test rpc connectivity, stop _active container - start _new container if start fails.... remove _new container, start _active | 00:01 |
| SamYaple | i thought, at first, you wanted multiple active old and new on the same host, but that workflow seems good | 00:01 |
| klindgren_ | no - only api containers are active/active for some period. | 00:02 |
| SamYaple | well that cant work because kolla-ansible uses host networking | 00:02 |
| SamYaple | youd have two apis on the same ip:port | 00:02 |
| klindgren_ | some period == amount of time to test new api node, and time to drain connections to existing running api container. | 00:02 |
| klindgren_ | it works because we use a static set of ip's that we map between api containers | 00:03 |
| klindgren_ | in a different network | 00:03 |
| SamYaple | i mean for kolla-ansible | 00:03 |
| klindgren_ | which the local haproxy is configured to use. | 00:03 |
| SamYaple | oh. hmm. youre saying there will be another layer of haproxy in the mix here? i thought you meant doing the draining at the main haproxy container | 00:04 |
| klindgren_ | so you just configure the container services to listing on 0.0.0.0 so when you map in a new ip, the service just works on that ip automatically | 00:04 |
| klindgren_ | So - the way we currently do it is every api node is fronted by a local haproxy | 00:05 |
| klindgren_ | so we run all the api servers on the same set of servers. | 00:05 |
| SamYaple | one of kollas design principles is "dont touch the host", so swapping in new ips wont work. the other is a limiting access so it doesnt use 0.0.0.0 (anywhere that can be avoided) | 00:05 |
| SamYaple | the haproxy per api host is a tougher sell in my opinion | 00:06 |
| SamYaple | but its doable | 00:06 |
| klindgren_ | why wont new ip's work? | 00:06 |
| klindgren_ | its jsut another docker network | 00:06 |
| klindgren_ | that you create? | 00:06 |
| SamYaple | we dont use docker networks | 00:06 |
| SamYaple | host-networking only | 00:06 |
| klindgren_ | so you are exposing ports in docker? | 00:07 |
| SamYaple | no | 00:08 |
| SamYaple | the container sees allthe host interfaces | 00:08 |
| SamYaple | docker is unaware of anything related to kolla-networking | 00:08 |
| klindgren_ | you mean docker-networking? | 00:11 |
| klindgren_ | sorry - I guess I jsut dont see how using docker networks breaks the tenant of "dont touch the host"? | 00:12 |
| SamYaple | klindgren_: no thats a different thing | 00:13 |
| SamYaple | i said that because i thought you meant add ips to the host interfaces | 00:13 |
| SamYaple | kolla-ansible doesnt use docker networking for two reasons | 00:13 |
| SamYaple | first, back inthe day when kolla came out, docker networking was dirt slow | 00:13 |
| SamYaple | second, there is a bug with iscsi and network namespaces and qemu that breaks iscsi containers | 00:14 |
| SamYaple | errr nova-compute containers | 00:14 |
| SamYaple | anyway, thepoint is that kolla-ansible doesnt use container networking, it uses host networking | 00:14 |
| klindgren_ | so the way that I use docker networks, its just as fast as host networking and the docker binary is not inline for any network traffic. 2.) so your not willing to do a much better container rollout strategy because of a single edge case in a single service? which isn't even an API service, so it could use host-networking anyway? | 00:17 |
| SamYaple | klindgren_: i left kolla-ansible almost 2 years ago. im not blocking anything | 00:18 |
| SamYaple | im just letting you know what you canexpect | 00:18 |
| klindgren_ | kk | 00:19 |
| *** jtriley has quit IRC | 00:19 | |
| kolla-slack | <inc0> I think it's good idea but as Sam said it would have to be done on haproxy | 00:25 |
| SamYaple | dont listen to him, he wont be PTL much longer | 00:26 |
| kolla-slack | <inc0> Ha ha ha... | 00:26 |
| kolla-slack | <inc0> /slowlaugh | 00:26 |
| SamYaple | inc0: using container networking in kolla-ansible instead of host-networking... possible? | 00:26 |
| kolla-slack | <inc0> Uh not now | 00:26 |
| inc0 | I can't think of compelling reason to do it | 00:27 |
| SamYaple | inc0: klindgren_ was saying thats how this method works | 00:28 |
| SamYaple | i feel most of what klindgren_ wants to do can be done.... with the exception of running multiple active api hosts | 00:28 |
| *** jascott1 has quit IRC | 00:28 | |
| inc0 | well...multiple active api hosts on same host | 00:28 |
| *** jascott1 has joined #openstack-kolla | 00:29 | |
| inc0 | and same port | 00:29 |
| SamYaple | yep | 00:29 |
| inc0 | you could modify default port and deal with it on haproxy | 00:29 |
| inc0 | I think it would be easier to handle than rewiring our networking | 00:29 |
| inc0 | not to mention that docker networking itself is well...uh. | 00:29 |
| inc0 | maybe just for API nodes? | 00:29 |
| SamYaple | but at that point, without docker networking, local haproxy isnt needed since we can do that on the other haproxy | 00:29 |
| *** yangyapeng has quit IRC | 00:30 | |
| inc0 | well you can do this on regular haproxy too assuming different port | 00:30 |
| SamYaple | weve been talking about it for year, maybe klindgren_ can finally implemetn something | 00:30 |
| inc0 | you'll manage backends dynamically | 00:30 |
| *** yangyapeng has joined #openstack-kolla | 00:30 | |
| inc0 | well on that note, using some networking underlay like calico or weave (but ansible driven, not k8s) might also be something worth looking at | 00:31 |
| klindgren_ | its possible to not do active/active api's - thats jsut how I was able to do it | 00:31 |
| klindgren_ | I jsut want some agreement on the general idea. | 00:31 |
| inc0 | well active/active apis on same node is not possible today | 00:31 |
| klindgren_ | since I already have code to bring - I was hoping I could get started on a spec and start trying to get some stuff landed. | 00:32 |
| SamYaple | i dont think youll get opposition on the end goal, just the imeplmentation like using local haproxy and docker networking | 00:32 |
| klindgren_ | its possbile if not using host networking. | 00:32 |
| inc0 | totally, but let's disect idea and split it into smaller chunks of functionality | 00:32 |
| inc0 | and fwiw I think APIs could be exempt from our host-network-only policy | 00:33 |
| klindgren_ | so I would push for running local haproxy, because if people use real lb's with dsr for loadblaancing - you have no way without local haproxy of maitnaenancing a node | 00:33 |
| klindgren_ | been there - done that. | 00:33 |
| inc0 | it *has* to be there for stuff like neutron agents | 00:33 |
| inc0 | but other than that...it was just easy to handle and that was written in time when docker network was less than ideal | 00:33 |
| *** jascott1 has quit IRC | 00:33 | |
| inc0 | so ad "check if API is alive" we had this sanity test tasks | 00:34 |
| inc0 | that did pretty much what you explained | 00:34 |
| *** yangyapeng has quit IRC | 00:34 | |
| inc0 | https://github.com/openstack/kolla-ansible/blob/d474987ad9722c63bf6bf964ca80d810b89a313e/ansible/roles/glance/tasks/check.yml | 00:35 |
| inc0 | but it wasn't ever finished | 00:35 |
| SamYaple | sanity checks told me everything i needed to know about you inc0 | 00:36 |
| *** jtriley has joined #openstack-kolla | 00:36 | |
| inc0 | don't blame me to use words that I'm familiar with...intimately | 00:37 |
| *** k_mouza has quit IRC | 00:43 | |
| *** tovin07_ has joined #openstack-kolla | 00:45 | |
| inc0 | anyway, I'm off for 2 weeks | 00:49 |
| inc0 | see you all aroind! | 00:50 |
| inc0 | I should be reachable by mail | 00:50 |
| *** inc0 has quit IRC | 00:53 | |
| *** pengdake has joined #openstack-kolla | 00:56 | |
| *** zhaochao has joined #openstack-kolla | 00:56 | |
| *** threestrands has quit IRC | 00:58 | |
| *** k_mouza has joined #openstack-kolla | 00:58 | |
| *** itlinux has joined #openstack-kolla | 01:01 | |
| *** ljjjustin has joined #openstack-kolla | 01:02 | |
| *** threestrands has joined #openstack-kolla | 01:02 | |
| *** k_mouza has quit IRC | 01:02 | |
| *** threestrands has quit IRC | 01:03 | |
| *** threestrands has joined #openstack-kolla | 01:04 | |
| *** mdnadeem has joined #openstack-kolla | 01:10 | |
| *** threestrands has quit IRC | 01:11 | |
| sbezverk | kfox1111: ping | 01:18 |
| *** yangyapeng has joined #openstack-kolla | 01:21 | |
| *** numans has quit IRC | 01:24 | |
| *** threestrands has joined #openstack-kolla | 01:24 | |
| *** threestrands has quit IRC | 01:24 | |
| *** threestrands has joined #openstack-kolla | 01:24 | |
| *** numans has joined #openstack-kolla | 01:25 | |
| *** threestrands has quit IRC | 01:25 | |
| *** threestrands has joined #openstack-kolla | 01:26 | |
| *** numans has quit IRC | 01:29 | |
| *** jascott1 has joined #openstack-kolla | 01:30 | |
| *** caowei has joined #openstack-kolla | 01:30 | |
| *** numans has joined #openstack-kolla | 01:35 | |
| *** jtriley has quit IRC | 01:38 | |
| *** duonghq has joined #openstack-kolla | 01:43 | |
| *** zhurong has joined #openstack-kolla | 01:45 | |
| *** dave-mccowan has joined #openstack-kolla | 01:46 | |
| duonghq | morning guys | 01:46 |
| kiennt26 | duonghq: morning :) | 01:48 |
| *** tone_z has joined #openstack-kolla | 01:49 | |
| duonghq | morning kiennt26 | 01:50 |
| *** threestrands has quit IRC | 01:54 | |
| *** mm6021 has quit IRC | 01:56 | |
| *** threestrands has joined #openstack-kolla | 01:56 | |
| *** hnl has joined #openstack-kolla | 02:16 | |
| *** gkadam has joined #openstack-kolla | 02:18 | |
| *** jianghuaw_ has quit IRC | 02:29 | |
| *** jianghuaw has joined #openstack-kolla | 02:29 | |
| *** unicell has quit IRC | 02:35 | |
| openstackgerrit | Merged openstack/kolla-ansible stable/ocata: Config my_ip to api_interface_address in cinder.conf https://review.openstack.org/513606 | 02:41 |
| *** gkadam has quit IRC | 02:42 | |
| *** chmarkus has quit IRC | 02:58 | |
| *** chmarkus has joined #openstack-kolla | 02:59 | |
| *** numans has quit IRC | 03:05 | |
| *** numans has joined #openstack-kolla | 03:07 | |
| *** itlinux has quit IRC | 03:19 | |
| *** ljjjustin has quit IRC | 03:24 | |
| *** dave-mccowan has quit IRC | 03:33 | |
| *** larsks has quit IRC | 03:56 | |
| *** larsks has joined #openstack-kolla | 03:56 | |
| *** zhurong has quit IRC | 04:03 | |
| *** unicell has joined #openstack-kolla | 04:07 | |
| *** goldyfruit has quit IRC | 04:13 | |
| *** itlinux has joined #openstack-kolla | 04:32 | |
| *** caowei has quit IRC | 04:43 | |
| *** absubram has joined #openstack-kolla | 05:23 | |
| *** absubram has quit IRC | 05:23 | |
| *** Herbert_Guan has joined #openstack-kolla | 05:32 | |
| *** caowei has joined #openstack-kolla | 05:33 | |
| *** gkadam has joined #openstack-kolla | 05:34 | |
| *** zhurong has joined #openstack-kolla | 05:42 | |
| *** aqsa has joined #openstack-kolla | 05:45 | |
| aqsa | Hi all while im trying to deploy multinode i find that the images which got built are 6.0.0 and when i try to pull the images they fail with the error 404 saying cannot find image <image_name>5.0.2 | 05:47 |
| aqsa | any help or any idea why im using a stable pike branch | 05:47 |
| Jeffrey4l | aqsa, configure the "openstack_release" variable to the proper tag name in your env. | 05:48 |
| aqsa | why? Note: Im using a stable/pike branch ** | 05:48 |
| Jeffrey4l | aqsa, where images are you using? | 05:48 |
| aqsa | ook in the globals.yml right.. | 05:48 |
| aqsa | Jeffrey4l: i just used this `kolla-build --base centos --type source --registry 192.168.1.61:4000 --push ` | 05:49 |
| aqsa | im using `--base ubuntu ` sorry | 05:50 |
| Jeffrey4l | aqsa, where the images come from? | 05:50 |
| aqsa | as per my understanding we first build images right using `kolla-build --base centos --type source --registry 192.168.1.61:4000 --push` | 05:51 |
| aqsa | i have not used kolla tar ball | 05:52 |
| Jeffrey4l | aqsa, you need check what's the image tag really is. | 05:52 |
| aqsa | im presuming they come from docker hub right | 05:52 |
| aqsa | Jeffrey4l: so how do i check it .. | 05:52 |
| Jeffrey4l | no, you built it. | 05:52 |
| Jeffrey4l | show me the "docker images" on you node ran "kolla-build" | 05:53 |
| aqsa | https://hastebin.com/junayizazu.vbs I deleted all the images and i just built nova images and attached is the output | 05:53 |
| Jeffrey4l | aqsa, you can the TAG is 6.0.0, so check the "openstack_release" in /etc/kolla/globals.yml file to "6.0.0", then try to deploy again. | 05:54 |
| aqsa | yes so now i have to set the #openstack_release: "" option to 6.0.0 in the globals.yml | 05:56 |
| aqsa | if im not wrong to avoid that error | 05:56 |
| Jeffrey4l | aqsa, technically, it is not. before we can not guess what's the image tag name on the kolla-ansible side. | 05:58 |
| aqsa | Jeffrey4l: now how do i do it ?? | 06:13 |
| aqsa | I mean how can i fix this ?? | 06:13 |
| *** hnl has quit IRC | 06:20 | |
| *** hnl has joined #openstack-kolla | 06:21 | |
| Jeffrey4l | aqsa, you can the TAG is 6.0.0, so check the "openstack_release" in /etc/kolla/globals.yml file to "6.0.0", then try to deploy again. | 06:26 |
| *** absubram has joined #openstack-kolla | 06:26 | |
| *** harlowja has quit IRC | 06:35 | |
| *** hnl has quit IRC | 06:42 | |
| *** gkadam has quit IRC | 06:45 | |
| *** hnl has joined #openstack-kolla | 06:46 | |
| *** jbadiapa has joined #openstack-kolla | 06:56 | |
| *** hnl has quit IRC | 06:57 | |
| *** daidv has quit IRC | 06:57 | |
| *** duonghq has quit IRC | 06:57 | |
| *** daidv has joined #openstack-kolla | 06:58 | |
| *** duonghq has joined #openstack-kolla | 06:58 | |
| *** hnl has joined #openstack-kolla | 06:58 | |
| openstackgerrit | Duong Ha-Quang proposed openstack/kolla-ansible master: Check enable_cinder before deploy iscsi and tgtd https://review.openstack.org/533099 | 06:58 |
| *** jascott1 has quit IRC | 06:58 | |
| *** jascott1 has joined #openstack-kolla | 06:58 | |
| *** hieulq has quit IRC | 06:58 | |
| *** kiennt26 has quit IRC | 06:58 | |
| *** tovin07_ has quit IRC | 06:58 | |
| *** hieulq has joined #openstack-kolla | 06:59 | |
| *** tovin07_ has joined #openstack-kolla | 06:59 | |
| *** kiennt26 has joined #openstack-kolla | 06:59 | |
| *** harlowja has joined #openstack-kolla | 07:10 | |
| *** harlowja has quit IRC | 07:10 | |
| *** feldmann has joined #openstack-kolla | 07:14 | |
| *** itlinux has quit IRC | 07:20 | |
| *** Angeh has joined #openstack-kolla | 07:24 | |
| *** pcaruana has joined #openstack-kolla | 07:29 | |
| openstackgerrit | Merged openstack/kolla-ansible stable/pike: Fix dhcp_agent configuration for vmware DVS https://review.openstack.org/531691 | 07:31 |
| *** cah_link has joined #openstack-kolla | 07:32 | |
| *** egonzalez has joined #openstack-kolla | 07:44 | |
| *** rajivk has joined #openstack-kolla | 07:45 | |
| *** rajivk has left #openstack-kolla | 07:45 | |
| openstackgerrit | Xinliang Liu proposed openstack/kolla master: Consume RDO packaged mariadb version https://review.openstack.org/468632 | 07:47 |
| *** Mr_Broken has joined #openstack-kolla | 07:48 | |
| *** unicell has quit IRC | 07:51 | |
| openstackgerrit | Zhaokun Fu proposed openstack/kolla-ansible master: FIX inject password error https://review.openstack.org/531949 | 07:55 |
| egonzalez | aqsa, if have image tags 6.0.0, means you builded queens (master branch) images instead of stable/pike | 07:56 |
| *** threestrands has quit IRC | 07:56 | |
| egonzalez | from stable/pike would be 5.0.* | 07:56 |
| egonzalez | aqsa, if have internet access from all hosts and not have any custom change in images, not build them. Use dockerhub images | 07:56 |
| *** cah_link has quit IRC | 08:05 | |
| *** gkadam has joined #openstack-kolla | 08:10 | |
| *** zhangfei has joined #openstack-kolla | 08:13 | |
| duonghq | Jeffrey4l, ping | 08:13 |
| duonghq | Jeffrey4l, about this patch: https://review.openstack.org/#/c/533099/1 I left ironic untouch, just fix cinder-related logic | 08:14 |
| openstackgerrit | Duong Ha-Quang proposed openstack/kolla-ansible master: Check enable_cinder before deploy iscsi and tgtd https://review.openstack.org/533099 | 08:15 |
| *** wojdec has joined #openstack-kolla | 08:15 | |
| *** shardy has joined #openstack-kolla | 08:24 | |
| *** unicell has joined #openstack-kolla | 08:24 | |
| Jeffrey4l | egonzalez, duonghq could you review this patch https://review.openstack.org/508407 thx. | 08:28 |
| *** wojdec1 has joined #openstack-kolla | 08:30 | |
| *** wojdec has quit IRC | 08:31 | |
| openstackgerrit | Merged openstack/kolla-ansible master: Ensure docker running before execute the destroy action https://review.openstack.org/531688 | 08:32 |
| *** knsahm has joined #openstack-kolla | 08:33 | |
| *** wojdec has joined #openstack-kolla | 08:33 | |
| *** wojdec1 has quit IRC | 08:36 | |
| *** lvdombrkr has joined #openstack-kolla | 08:44 | |
| *** k_mouza has joined #openstack-kolla | 09:10 | |
| egonzalez | are there free tickets for PTG to contributors? | 09:11 |
| *** lvdombrkr89 has joined #openstack-kolla | 09:13 | |
| *** lvdombrkr has quit IRC | 09:15 | |
| *** wojdec has quit IRC | 09:18 | |
| *** wojdec has joined #openstack-kolla | 09:19 | |
| *** zhurong has quit IRC | 09:20 | |
| jianghuaw | egonzalez: I have the same question:-) At least it should be free for cores (like you) who contributed a lot. | 09:21 |
| hrw | tone_z: elo | 09:21 |
| hrw | morning | 09:21 |
| egonzalez | jianghuaw, if request travel support program, maybe can only request just the ticket | 09:21 |
| jianghuaw | egonzalez, don't know if the support program also include the ticket. | 09:23 |
| hrw | egonzalez: good point. no mails about it compared to previous events | 09:23 |
| egonzalez | jianghuaw, includes the ticket, iirc last time i got rejected they sent me a free code ticket | 09:23 |
| *** robbbe has joined #openstack-kolla | 09:24 | |
| *** feldmann has quit IRC | 09:24 | |
| jianghuaw | ok. The ticket costs much less than a full travel support. If allow to request just the ticket, it can cover more people. | 09:25 |
| hrw | I had discount for Barcelona, Boston and Sydney summits. Did not used any as I did not went there | 09:25 |
| jianghuaw | :-) | 09:26 |
| hrw | jianghuaw: I do not have a luck with CfP | 09:26 |
| jianghuaw | hrw: CfP? | 09:26 |
| hrw | Call for Papers | 09:27 |
| jianghuaw | ah. got it:-) | 09:27 |
| hrw | jianghuaw: was co-speaker for Barcelona and Sydney talks. Both were not accepted. | 09:27 |
| jianghuaw | hrw, wish you'll have a good luck for the next summit. | 09:28 |
| hrw | ;d | 09:28 |
| * hrw off | 09:28 | |
| egonzalez | hrw, talk title should be something like "Stop using OpenStack and move to k8s" in order to get accepted | 09:28 |
| *** dougsz has joined #openstack-kolla | 09:28 | |
| kolla-slack | <aslancimbom100> That's actually funny :slightly_smiling_face: | 09:29 |
| jianghuaw | egonzalez, lol | 09:29 |
| *** aqsa has quit IRC | 09:29 | |
| *** feldmann has joined #openstack-kolla | 09:30 | |
| *** Angeh has quit IRC | 09:33 | |
| jianghuaw | egonzalez, the variable of xenapi_facts_file defined at: https://review.openstack.org/#/c/531640/7/etc/kolla/globals.yml@415 | 09:33 |
| jianghuaw | will be used by both nova and neutron. | 09:33 |
| jianghuaw | https://review.openstack.org/#/c/525999/16/ansible/roles/neutron/tasks/config.yml@151 | 09:33 |
| jianghuaw | https://review.openstack.org/#/c/531640/7/ansible/roles/nova/tasks/bootstrap_xenapi.yml@8 | 09:34 |
| *** jascott1 has quit IRC | 09:35 | |
| *** jascott1 has joined #openstack-kolla | 09:35 | |
| egonzalez | jianghuaw, better put it in group_vars (shared for various roles) but not that generic as globals.yml | 09:36 |
| *** duonghq has quit IRC | 09:36 | |
| jianghuaw | egonzalez, Agreed. Surely will rework it accordingly. thanks. | 09:37 |
| *** jascott1 has quit IRC | 09:39 | |
| openstackgerrit | Jianghua Wang proposed openstack/kolla-ansible master: XenAPI: Support compute node bootstrap https://review.openstack.org/531640 | 09:44 |
| openstackgerrit | Jianghua Wang proposed openstack/kolla-ansible master: XenAPI: Support neutron configuration https://review.openstack.org/525999 | 09:44 |
| egonzalez | PTG confirmed o/ | 09:45 |
| jianghuaw | egonzalez, confirmed for a free ticket? | 09:46 |
| egonzalez | paying :( | 09:46 |
| jianghuaw | :( | 09:46 |
| jianghuaw | egonzalez, ^reworked the 1st patch and rebased the 2nd one. | 09:47 |
| *** k_mouza_ has joined #openstack-kolla | 09:48 | |
| *** k_mouza has quit IRC | 09:48 | |
| *** wojdec1 has joined #openstack-kolla | 09:50 | |
| jianghuaw | berendt, I see you've +2 on the first one. Thanks very much:-) | 09:51 |
| *** wojdec has quit IRC | 09:52 | |
| jianghuaw | berendt, What's the benefit for the additional ml2_conf_xenapi.ini? | 09:55 |
| jianghuaw | https://review.openstack.org/#/c/525999/17/ansible/roles/neutron/tasks/config.yml@158 | 09:55 |
| berendt | jianghuaw: yes, i think it makes no sense, you can use a separate folder for a xen host | 09:56 |
| hrw | egonzalez: k8s is now that mandatory word? | 09:56 |
| egonzalez | hrw, been in last 3 summits | 09:57 |
| berendt | this way you can inject ml2 parameters for xen hosts and non xen hosts without mixing them in a global ml2 config | 09:57 |
| *** gfidente has joined #openstack-kolla | 09:57 | |
| *** gfidente has quit IRC | 09:57 | |
| *** gfidente has joined #openstack-kolla | 09:57 | |
| hrw | egonzalez: shit. that's why my talk for devconf.cz got rejected... too old buzzwords used | 09:57 |
| jianghuaw | berendt, got your point. Thanks. | 09:57 |
| hrw | I had containers, docker, openstack in title but no k8s | 09:57 |
| berendt | hrw: lol | 09:58 |
| berendt | jianghuaw: but i thinks an additional custom ml2_conf_xen.ini is not required, changed to +2 | 09:59 |
| jianghuaw | berendt, see it. Thanks very much. | 10:00 |
| hrw | berendt: "Kolla - deploying OpenStack using containers" | 10:01 |
| berendt | sounds very unspectacular | 10:02 |
| berendt | but I'm really bad at writing good titles and abstracts | 10:03 |
| *** kevinz has quit IRC | 10:05 | |
| *** numans has quit IRC | 10:05 | |
| hrw | so do I | 10:06 |
| *** pengdake has quit IRC | 10:06 | |
| *** numans has joined #openstack-kolla | 10:07 | |
| jianghuaw | berendt, will you attend the Dublin PTG? | 10:07 |
| *** wojdec has joined #openstack-kolla | 10:07 | |
| *** kiennt26 has quit IRC | 10:07 | |
| *** wojdec1 has quit IRC | 10:09 | |
| *** daidv has quit IRC | 10:10 | |
| *** tovin07_ has quit IRC | 10:11 | |
| *** nathharp has joined #openstack-kolla | 10:12 | |
| *** pbourke has quit IRC | 10:16 | |
| *** pbourke has joined #openstack-kolla | 10:18 | |
| *** sambetts|afk is now known as sambetts | 10:23 | |
| *** ktibi has joined #openstack-kolla | 10:23 | |
| ktibi | Hi kolla :) | 10:24 |
| hrw | hi ktibi | 10:24 |
| ktibi | any news about kubernetes in kolla ? | 10:25 |
| ktibi | I guess not for Q release. | 10:25 |
| sean-k-mooney | ktibi: k8s in kolla or kolla-kubernetes? | 10:37 |
| *** jascott1 has joined #openstack-kolla | 10:38 | |
| sean-k-mooney | ktibi: i think queens should be the first production release fo kolla-kubernets at least that was a cycle goal | 10:38 |
| *** caowei has quit IRC | 10:38 | |
| ktibi | sean-k-mooney, kolla-kubernetes yes | 10:38 |
| ktibi | ho great :) | 10:39 |
| sean-k-mooney | ktibi: as i said i dont know if that goal was achived but that was the intent at least | 10:40 |
| sean-k-mooney | kolla-ansible had a couple of cycles head start so its more feature complete and stable but the k8s port has been comming along nicely too | 10:41 |
| *** serlex has joined #openstack-kolla | 10:49 | |
| *** shardy has quit IRC | 10:50 | |
| *** tone_z has quit IRC | 10:51 | |
| *** shardy has joined #openstack-kolla | 10:52 | |
| *** pabbi has joined #openstack-kolla | 10:54 | |
| sean-k-mooney | egonzalez: pbourke o/ | 11:04 |
| sean-k-mooney | egonzalez: pbourke QQ if you are using images form the docker hub and you want to do an minor upgrade to you just do a pull then upgrade since the tag (pike) is the same | 11:06 |
| egonzalez | sean-k-mooney, deploy or reconfigure | 11:06 |
| egonzalez | upgrade will do db syncs which are not needed | 11:06 |
| *** k_mouza_ has quit IRC | 11:10 | |
| *** k_mouza has joined #openstack-kolla | 11:12 | |
| sean-k-mooney | egonzalez: oh ok so is upgrade only needed for major version updates then? | 11:16 |
| *** zhangfei has quit IRC | 11:22 | |
| egonzalez | sean-k-mooney, yep | 11:23 |
| *** pabbi has quit IRC | 11:30 | |
| *** nathharp has quit IRC | 11:34 | |
| *** zhangfei has joined #openstack-kolla | 11:39 | |
| *** lpetrut has joined #openstack-kolla | 11:47 | |
| *** nathharp has joined #openstack-kolla | 11:59 | |
| chmarkus | is it possible to limit the 'kolla-ansible deploy' command for an existing deployment to a specific service? Let's say I want to redeploy a removed barbican container in an existing installation without having ansible going/skipping through all the existing services? | 12:08 |
| sean-k-mooney | chmarkus: you can use --tags for that | 12:09 |
| sean-k-mooney | it will still check if all the other roles should run but since the tag wont be in the list it will skip them exept for the tags you set so --tags barbican | 12:10 |
| *** semoac has quit IRC | 12:11 | |
| chmarkus | sean-k-mooney, beautiful! That did the trick - thanks a bunch! | 12:14 |
| chmarkus | I didn't find such tags in the playbook roles' *.yml files, is this some implicit ansible magic? | 12:15 |
| hrw | sean-k-mooney: thanks for hint | 12:15 |
| sean-k-mooney | chmarkus: no its done at the top level playbook here https://github.com/openstack/kolla-ansible/blob/master/ansible/site.yml#L75 when the role is imported | 12:17 |
| *** Zophar has quit IRC | 12:17 | |
| sean-k-mooney | a couple of things like precheck or gather fact are not tagged or set to tags:always as then are assumed to have run always before the other roles but all roles that actully change things on the system are tagged | 12:18 |
| chmarkus | sean-k-mooney, I see. Thanks for the link and the explanation - that is very helpful :) | 12:19 |
| sean-k-mooney | chmarkus: if we didnt compose the playbooks this way your other option would have been to create your own playbook and just import just the role you want to run | 12:20 |
| sean-k-mooney | kolla-ansible deploy just translates to ansible-playbook -e action=deploy -e@/etc/kolla/globals.yml ansible/site.yml | 12:21 |
| sean-k-mooney | when you add -i or --tags to kolla ansible deploy they just get added on to the ansible-playbook command | 12:22 |
| *** Herbert_Guan has quit IRC | 12:28 | |
| *** janki has joined #openstack-kolla | 12:41 | |
| *** nathharp has quit IRC | 12:50 | |
| *** nathharp has joined #openstack-kolla | 12:51 | |
| *** dave-mccowan has joined #openstack-kolla | 12:54 | |
| *** goldyfruit has joined #openstack-kolla | 13:03 | |
| *** goldyfruit has quit IRC | 13:05 | |
| ktibi | I can see prometheus image in kolla, anyone works on the integration of prometheus in kolla-ansible ? | 13:08 |
| *** absubram has quit IRC | 13:11 | |
| egonzalez | ktibi, WIP https://review.openstack.org/#/q/project:openstack/kolla-ansible+prometheus | 13:13 |
| *** goldyfruit has joined #openstack-kolla | 13:14 | |
| *** goldyfruit has quit IRC | 13:19 | |
| *** numans has quit IRC | 13:20 | |
| *** numans has joined #openstack-kolla | 13:20 | |
| *** yangyapeng has quit IRC | 13:32 | |
| *** feldmann has quit IRC | 13:49 | |
| *** david-lyle has quit IRC | 13:49 | |
| *** egonzalez has quit IRC | 13:49 | |
| *** spsurya has quit IRC | 13:53 | |
| *** yangyapeng has joined #openstack-kolla | 14:02 | |
| *** kbaegis has joined #openstack-kolla | 14:07 | |
| kbaegis | Morning gents | 14:08 |
| kbaegis | I had a question | 14:08 |
| kbaegis | I'm using ceph-ext in my environment and I was wondering if there was a convenient method to get logging and mairadb mapped to ceph rbds via kolla | 14:09 |
| kbaegis | something like appending -v /mnt/logs/:/var/log to each of the services | 14:09 |
| sean-k-mooney | kbaegis: i just deployed the same congiuration and have been thinking about the same problem | 14:10 |
| sean-k-mooney | the short answer is yes you can change the volumes | 14:10 |
| sean-k-mooney | the mariadb volumns are defined here https://github.com/openstack/kolla-ansible/blob/master/ansible/roles/mariadb/defaults/main.yml#L10-L14 | 14:11 |
| sean-k-mooney | in your globals.yml on in your inventory you can override mariadb_services:{mariadb:{volumes:[]}}} | 14:12 |
| sean-k-mooney | the syntax is wrong but you get the idea | 14:12 |
| sean-k-mooney | to make this work with external ceph you either need to use a cephfs fileshare per db or have systemd script automount the rbd volumn on system boot before docker starts | 14:14 |
| sean-k-mooney | kbaegis: this conficuration is not offically supported by kolla-ansible however and i have not tested it personally but that is how i was considering moving the the centralised logs to be stored in ceph | 14:15 |
| kbaegis | sean-k-mooney: Yeah, it would certainly be easier | 14:20 |
| *** goldyfruit has joined #openstack-kolla | 14:21 | |
| kbaegis | fstab runs before docker, correct? | 14:21 |
| sean-k-mooney | kbaegis: im sure people would be interested in a way to do this automatically in kolla-ansible in the future so if you do go this route and automate it it would be a nice addtion | 14:22 |
| sean-k-mooney | yes so if you go with cephfs then you can use that to auto mount it | 14:22 |
| sean-k-mooney | * fstab | 14:22 |
| sean-k-mooney | you cant use rbd volumes directly from fstab as far as i know as it only deals with filesystems not block devices hence systemd script with before docker.target set | 14:23 |
| kbaegis | Well- let's see. | 14:25 |
| kbaegis | You got your ceph rbd module, then it populates dev via the mapper file | 14:25 |
| kbaegis | Yeah, you might very well be right. You'd need an explicit init dependency for docker on ceph | 14:26 |
| sean-k-mooney | i have never had much luck using the rbdmapper service so i just write my own | 14:26 |
| kbaegis | But the rbd shouldn't care about the contents of fstab. It's just a module and a generic nod | 14:26 |
| kbaegis | Ahh | 14:26 |
| kbaegis | Yeah, I'm just doing a rbd map operation | 14:27 |
| kbaegis | Does cephfs even have ckfs yet? | 14:27 |
| sean-k-mooney | am not sure. i deployed it at home and had a 10% perfromance hit vs maping an rbd volume so i jsut stuck with rbd | 14:28 |
| sean-k-mooney | i have not tried it on real server hardware | 14:28 |
| kbaegis | Ouch. Yeah, and I thought cephfs still had jumbo 'experimental' label because there was a risk of losing your metadata contents and not being able to recover | 14:29 |
| sean-k-mooney | cephfs is easier to use though if you can make use of the added utility of having a shared network filesystem | 14:29 |
| kbaegis | my entire ceph setup is wonky. I'm using a partition rather than an entire device. I know that's not best practice, but it's working fine and I don't know the specific reason | 14:30 |
| sean-k-mooney | kbaegis: i think luminous allows active active metadata services now so not sure if that has been resolved | 14:30 |
| kbaegis | Unless I/O gets pegged by the system, in which case it's a cluster and can stand to move out/down a couple of nodes anyway | 14:30 |
| sean-k-mooney | kbaegis: ya same. although i deployed a new one yesterday which is nolonger doing that | 14:31 |
| kbaegis | What's the penalty to using a partition? I only have one disk currently per blade | 14:31 |
| *** knsahm has quit IRC | 14:31 | |
| kbaegis | One of my more junior guys wiped the OS installs on accident because he was unaware that ceph would gobble up the entire drive by default | 14:32 |
| kbaegis | Hence ceph-external | 14:32 |
| *** numans has quit IRC | 14:32 | |
| sean-k-mooney | my old developer cloud use 1 ssd for os + osd 2 journal + cache osd data + 1 hdd for osd data per node | 14:32 |
| sean-k-mooney | it worked well for the last 14 ish months | 14:32 |
| sean-k-mooney | howere i realsed because i was useing earsure encoded pools with a cache teir with replica count 3 i was writing all data 8.8 times total | 14:33 |
| sean-k-mooney | 3 cache journals + 3 caches + 1.4 x for ec pool journal +1.4x for ec pool backing hdds | 14:34 |
| kbaegis | Yeah, erasure kinda sucks imo. Gobbles up too much cpu, and I have shared storage/compute | 14:34 |
| sean-k-mooney | so every 1BG writen was 7.4 to the ssds | 14:34 |
| kbaegis | It's super handy though because you can control replicas on a per-pool basis, so you just configure cinder with different drivers pointing to discrete pools | 14:35 |
| kbaegis | Not that xor is too expensive. | 14:35 |
| sean-k-mooney | the down side of haveing the ssd have the journsal + cache +os was when that wright heavy workload stated to casue them to fail it took the entire node with it | 14:36 |
| kbaegis | Did you have cascading failures, or was a node loss not such a big deal? | 14:36 |
| *** janki has quit IRC | 14:36 | |
| sean-k-mooney | well i was doing a convered openstack deployment so the node was both an openstack controler, ceph mon node, openstack networking node and a ceph data node so not fun | 14:37 |
| sean-k-mooney | i manganged to catch it before the drive completely failed and a simple dd of the drive to a spare ssd was all that was needed to fix it | 14:38 |
| sean-k-mooney | buy ya on erasuer pools i like to use them for nova vms and cinder volumes/backups and then put glance on an ssd-primary replicated pool | 14:39 |
| sean-k-mooney | ssd-primarly meaning 1 replica on ssd +2 on hdd | 14:39 |
| kbaegis | Yeah, I'd be doing that if I had a dedicated ceph cluster | 14:39 |
| kbaegis | As it stands, it's way cheaper to do a combined compute/storage pool | 14:40 |
| sean-k-mooney | well i combine them too | 14:40 |
| kbaegis | There's a cpu toll and latency added when using erasure isn't there? | 14:40 |
| sean-k-mooney | my newest deployment has 3 dedicated servers for openstack contol/network group and they are my ceph mons too. then all the computes are mixed ceph osd + compute node servers | 14:41 |
| kolla-slack | <aslancimbom100> what are you using to monitor them? | 14:42 |
| *** athomas has quit IRC | 14:42 | |
| sean-k-mooney | kbaegis: well one i work at intel so lets just say cpu capasity is not really a problem the servers im useing have 2 10 or 14 core cpus on the older systesm and i have a couple of 2X 22 core systems for compute nodes | 14:42 |
| *** athomas has joined #openstack-kolla | 14:43 | |
| sean-k-mooney | aslancimbom100 currently myself unfortunetly. i am hoping ot use collectd + aodh+ gnocchi in the futre but also looking at collectd+influx+grapana | 14:45 |
| kolla-slack | <aslancimbom100> i see, I had ceph-dash but no alerting configured | 14:47 |
| kolla-slack | <aslancimbom100> Haven't tried ceph mgr yet | 14:47 |
| kbaegis | Oh, hey that's cool. So there's a bridge from irc into slack? | 14:48 |
| sean-k-mooney | kbaegis: yep | 14:48 |
| glycerine | Under ceph luminous, I get a warning about "application not enabled for pools" Would I be correct in setting my pools (images, volumes, backups, vms) to 'rgw' | 14:49 |
| sean-k-mooney | i think it was set up to help people working on kolla-kubernetes easilly talk to the k8s dev team that work only on slack | 14:49 |
| sean-k-mooney | glycerine: no i dont think that would be correct | 14:49 |
| glycerine | I have these options 'cephfs', 'rbd', 'rgw', or freeform | 14:50 |
| kbaegis | Well, slack is just about as secure as IRC so :) | 14:50 |
| sean-k-mooney | kbaegis: it was not about security just ease of use. that is what k8s project chose to use | 14:51 |
| sean-k-mooney | glycerine: i would say rbd would be the better choice but maybe openstack would also make sense | 14:53 |
| sean-k-mooney | glycerine: http://ceph.com/community/new-luminous-pool-tags/ | 14:54 |
| glycerine | thanks I'll read up on it | 14:55 |
| kbaegis | <<Just learning about luminous | 14:55 |
| sean-k-mooney | volumes,backups and vms are all block devices so should be rbd and image is an object store style usecase so maybe rgw but rdb would also work. | 14:56 |
| kbaegis | BlueStore sounds neat. I'm wondering what it actually does :) | 14:57 |
| sean-k-mooney | today the application has no effect really so you could also use a custom openstack applicaiton name | 14:57 |
| sean-k-mooney | kbaegis: bluestore replaces the files system used by the osd with a custume minimal filesystem based on a leveldb data base and write ahead journal | 14:58 |
| sean-k-mooney | kbaegis: it get ride of some of the issue around makeing sure data is commited to disk transactionally by removing the intermediate filesystem layer and allowing ceph to manage the disk directly | 14:59 |
| kbaegis | rather than deploying under xfs? | 14:59 |
| kbaegis | gotcha | 14:59 |
| kbaegis | Neat :D | 14:59 |
| sean-k-mooney | that means more perfomance and better data integrty vs xfs/btrfs and it enable some new featres such as partail overries on ec pool that could not be done safely before | 15:00 |
| kbaegis | Sounds awesome. It also makes it much more difficult I would imagine to troubleshoot | 15:01 |
| kbaegis | I can't use xfs/btrfs utlities to mount and repair | 15:01 |
| kbaegis | But that's worth doing | 15:01 |
| *** spsurya has joined #openstack-kolla | 15:01 | |
| sean-k-mooney | well true but i dont think you could do that before safely | 15:01 |
| kbaegis | Yeah. And it's built for node failure anyways. | 15:02 |
| kbaegis | Thanks for the primer. | 15:02 |
| kbaegis | Anyone here have some handy scripts they're willing to share to roll off old docker image versions? | 15:03 |
| sean-k-mooney | kbaegis: well i am not a ceph expert but just learning the bits i need to deploy a cloud for my team at work based on kolla SamYaple and kfox1111 know a lot more about ceph then i do | 15:03 |
| kbaegis | I've read the rados whitepaper too | 15:05 |
| kbaegis | Just didn't know about their new custom db/fs | 15:05 |
| sean-k-mooney | kbaegis: SamYaple had some interesting articles on ceph on https://yaple.net but it seams to be down at the moment | 15:07 |
| *** jtriley has joined #openstack-kolla | 15:16 | |
| kbaegis | Looks like it's on archive.org still | 15:16 |
| sean-k-mooney | sam did the origial kolla ceph support so for a time his blog was the only documentation for external journals or cache teirs for kolla deployed ceph. | 15:17 |
| sean-k-mooney | he did some interesting experimets with bcache that i use at home but for work im using a different setup now | 15:18 |
| *** semoac has joined #openstack-kolla | 15:20 | |
| *** zhangfei has quit IRC | 15:26 | |
| openstackgerrit | Tudosoiu Marian proposed openstack/kolla-ansible master: Rework mariadb recovery tasks https://review.openstack.org/531122 | 15:31 |
| *** lvdombrkr89 has quit IRC | 15:35 | |
| *** david-lyle has joined #openstack-kolla | 15:38 | |
| *** robbbe has quit IRC | 15:39 | |
| openstackgerrit | Mark Goddard proposed openstack/kolla-ansible master: Use bool filter with enable_ironic_pxe_uefi https://review.openstack.org/533214 | 15:52 |
| *** jschlueter has joined #openstack-kolla | 15:58 | |
| jschlueter | mandre: in reviewing recent changes I see sensu-base has a run command creating sensu.repo in it | 15:58 |
| *** nathharp has quit IRC | 16:00 | |
| jschlueter | mandre: I see it was injected by tripleo-common | 16:01 |
| *** nathharp has joined #openstack-kolla | 16:04 | |
| openstackgerrit | Pradeep Kilambi proposed openstack/kolla master: Drop ceilometer client package requirement https://review.openstack.org/533224 | 16:07 |
| *** mchlumsky has joined #openstack-kolla | 16:09 | |
| *** itlinux has joined #openstack-kolla | 16:12 | |
| *** mchlumsky has quit IRC | 16:14 | |
| jschlueter | mandre: ^^ | 16:14 |
| *** zhaochao has quit IRC | 16:14 | |
| *** mchlumsky has joined #openstack-kolla | 16:16 | |
| *** itlinux has quit IRC | 16:20 | |
| sdake | morning peeps | 16:25 |
| *** kbaegis1 has joined #openstack-kolla | 16:30 | |
| *** kbaegis has quit IRC | 16:30 | |
| *** jistr has quit IRC | 16:37 | |
| *** adrianreza has quit IRC | 16:37 | |
| *** berendt has quit IRC | 16:37 | |
| *** cloudnull has quit IRC | 16:37 | |
| *** ansiwen has quit IRC | 16:37 | |
| *** dosaboy has quit IRC | 16:37 | |
| *** dosaboy has joined #openstack-kolla | 16:37 | |
| *** cloudnull has joined #openstack-kolla | 16:37 | |
| *** adrianreza has joined #openstack-kolla | 16:38 | |
| *** berendt has joined #openstack-kolla | 16:38 | |
| *** chmarkus has quit IRC | 16:39 | |
| *** itlinux has joined #openstack-kolla | 16:39 | |
| *** jistr has joined #openstack-kolla | 16:40 | |
| *** serlex has quit IRC | 16:40 | |
| *** ansiwen has joined #openstack-kolla | 16:40 | |
| *** chmarkus has joined #openstack-kolla | 16:42 | |
| *** Mr_Broken has quit IRC | 16:43 | |
| -openstackstatus- NOTICE: Zuul has been restarted and lost queue information; changes in progress will need to be rechecked. | 16:45 | |
| *** wojdec has left #openstack-kolla | 16:47 | |
| *** mdnadeem has quit IRC | 16:48 | |
| sean-k-mooney | just as a sainity check since our merge_configs fuction calls template here https://github.com/openstack/kolla-ansible/blob/master/ansible/action_plugins/merge_configs.py#L94 i shoululd be able to add jinja vairbale suppsutiontion {{my_var}} to a config over file in /etc/kolla/config/... and set the var in my inventory correct? | 16:53 |
| *** nathharp has quit IRC | 16:55 | |
| *** absubram has joined #openstack-kolla | 17:00 | |
| *** pcaruana has quit IRC | 17:03 | |
| *** ntpttr_laptop has joined #openstack-kolla | 17:06 | |
| sean-k-mooney | sdake: pbourke kfox1111 ^ any idea if that will work? ill be testing it in a few minuts but trying to avoid having to create multiple nova.conf overrides | 17:09 |
| *** dciabrin_ has joined #openstack-kolla | 17:09 | |
| *** dciabrin has quit IRC | 17:11 | |
| SamYaple | sean-k-mooney: kbaegis1 yea sorry, blogs down at the moment. anythingi an help with? | 17:12 |
| sean-k-mooney | SamYaple: no i was chating to people about ceph and mention you had done the original ceph support in kolla and your blog. | 17:13 |
| sean-k-mooney | SamYaple: unless you know the answer to can i use jinja in template overrides | 17:13 |
| sean-k-mooney | SamYaple: you did the original merge_configs plugin too right | 17:14 |
| *** jascott1 has quit IRC | 17:15 | |
| *** jascott1 has joined #openstack-kolla | 17:15 | |
| *** jascott1 has quit IRC | 17:16 | |
| *** shardy has quit IRC | 17:28 | |
| *** gfidente is now known as gfidente|afk | 17:32 | |
| sdake | sean-k-mooney SamYaple wrote nearly half the original code base | 17:33 |
| sdake | so yes | 17:33 |
| *** k_mouza has quit IRC | 17:34 | |
| *** absubram has quit IRC | 17:34 | |
| sean-k-mooney | sdake: haha thats true. im almost ready to test it so ill just run the reconfigure and find out | 17:38 |
| sdake | sean-k-mooney sounds good | 17:38 |
| SamYaple | sean-k-mooney: yea sorry, template overrides were after my time(mostly) | 17:39 |
| sean-k-mooney | SamYaple: im refering to the merge config not template overrides but i just hit enter on the reconfigure so ill know soo enought | 17:40 |
| *** lpetrut has quit IRC | 17:40 | |
| SamYaple | oh no, you cant put jinja templating into the merge config | 17:40 |
| SamYaple | though that would be pretty easy to extend support for | 17:40 |
| klindgren_ | So as mentioned yesterday here is our current ansible role which does the flow chart stuff I talked about | 17:41 |
| klindgren_ | https://github.com/krislindgren/openstack-deploy | 17:41 |
| sean-k-mooney | SamYaple: damb i was hoping this would render the template https://github.com/openstack/kolla-ansible/blob/master/ansible/action_plugins/merge_configs.py#L94 | 17:41 |
| klindgren_ | I tireid to remove thee godaddy-ism's like we have an inventory system and dynamic inventory script that gives a bunch of the infromation over to ansible. The goal being that though shalt never have passwords anywhere but an encrypted vault. It also means that some information comes from that topology information. Which allows us to do things like, not having to login to memcache servers, to get facts, to use as a variable so | 17:44 |
| klindgren_ | mewhere else. We jsut pull the relevant information from out topology information. | 17:44 |
| klindgren_ | SamYaple, inc0, kfox1111 | 17:45 |
| kolla-slack | <inc0> All passwords are stored in passwords.yml | 17:47 |
| kolla-slack | <inc0> So if you put them to vault under these keys you should be fine | 17:47 |
| sean-k-mooney | SamYaple: actully i think it worked the way i wanted it to and templated it out correctly | 17:49 |
| klindgren_ | Thats fine - but the way we store password is not currently comptable with kolla. tl;dr is that we have our own set of topology information/refrences and we translate those into whatever tool we are using. IE we use the same base information for ansible/puppet/pythonscripts, (insert tool here) with render's for whatever we need to put out. We got tired of being slaves of our tools. So we went with a tool (ansible/puppet/salt | 17:53 |
| klindgren_ | ) agnostic format, and when we need something in a specific format for that tool we simply build a renderer for that tool. | 17:53 |
| sean-k-mooney | SamYaple: sdake alittle bit sanitzed but http://paste.openstack.org/show/643955/ works | 17:54 |
| *** dougsz has quit IRC | 17:55 | |
| sdake | @sean-k-mooney you shouldn't have to set virt type as the defaults are merged with your updates | 17:55 |
| sean-k-mooney | default should be DEFAULT though to match up with the template | 17:56 |
| *** ktibi has quit IRC | 17:56 | |
| sean-k-mooney | sdake: on ya i know it just force of habit | 17:56 |
| sean-k-mooney | the fact that you can in fact use jinja in your template overrides make doing tweeking like this much simpler as i can have templated template overrides | 17:58 |
| *** gkadam has quit IRC | 17:58 | |
| sean-k-mooney | sorry templated config overrides | 17:58 |
| SamYaple | klindgren_: there is all kinds of dynamic ansible inventory and vars stuff, surely you could write something to hook into kolla without much issue | 17:59 |
| *** robbbe has joined #openstack-kolla | 17:59 | |
| klindgren_ | We can and we will. I was just saying that if something is missing in the above thing from me, its probably because it comes from the dynamic inventory stuff/vars and I missed it in my attempt to try to make the module with less Godaddy-ism's | 18:01 |
| sean-k-mooney | klindgren_ i cant rembere who did it but i know someone wrote a custom dynamic inventory plugin that pull all the config varible form txt records in there dns system so others have hooked into kolla sucessfully in the past without needing to change the code | 18:01 |
| klindgren_ | to be clear, my comments were solely related to: https://github.com/krislindgren/openstack-deploy | 18:02 |
| *** david-lyle has quit IRC | 18:12 | |
| *** k_mouza has joined #openstack-kolla | 18:14 | |
| *** kbaegis has joined #openstack-kolla | 18:15 | |
| *** kbaegis1 has quit IRC | 18:17 | |
| *** k_mouza has quit IRC | 18:19 | |
| *** athomas has quit IRC | 18:20 | |
| *** spsurya has quit IRC | 18:23 | |
| *** ntpttr_laptop has quit IRC | 18:23 | |
| *** ntpttr_laptop has joined #openstack-kolla | 18:29 | |
| *** jtriley has quit IRC | 18:33 | |
| *** yee379 has joined #openstack-kolla | 18:36 | |
| *** yee37932 has quit IRC | 18:38 | |
| *** semoac has quit IRC | 18:51 | |
| *** ntpttr_laptop has quit IRC | 19:03 | |
| *** iclon has joined #openstack-kolla | 19:04 | |
| *** sambetts is now known as sambetts|afk | 19:09 | |
| *** ntpttr_laptop has joined #openstack-kolla | 19:19 | |
| *** bmace has quit IRC | 19:19 | |
| *** bmace has joined #openstack-kolla | 19:20 | |
| *** kbaegis has left #openstack-kolla | 19:25 | |
| *** ntpttr_laptop has quit IRC | 19:27 | |
| *** harlowja has joined #openstack-kolla | 19:27 | |
| *** david-lyle has joined #openstack-kolla | 19:33 | |
| *** sbezverk has quit IRC | 20:03 | |
| *** harlowja has quit IRC | 20:04 | |
| *** sbezverk has joined #openstack-kolla | 20:04 | |
| *** iclon has quit IRC | 20:11 | |
| *** goldyfruit has quit IRC | 20:14 | |
| *** robbbe has quit IRC | 20:15 | |
| *** robbbe has joined #openstack-kolla | 20:16 | |
| *** robbbe has quit IRC | 20:16 | |
| *** robbbe has joined #openstack-kolla | 20:18 | |
| *** robbbe has quit IRC | 20:18 | |
| *** robbbe has joined #openstack-kolla | 20:21 | |
| *** strgout has joined #openstack-kolla | 20:21 | |
| *** robbbe has quit IRC | 20:21 | |
| strgout | howdy all, i'm using the images from docker. I've noticed that ovsdb won't start because the unix socket.. is a directory. Removing the directory and restarting fixed the problem but how do i commit this change. More of a docker question i guess. | 20:23 |
| strgout | i don't have a local registry | 20:23 |
| *** robbbe has joined #openstack-kolla | 20:24 | |
| *** robbbe has quit IRC | 20:24 | |
| *** robbbe has joined #openstack-kolla | 20:27 | |
| *** robbbe has quit IRC | 20:28 | |
| *** robbbe has joined #openstack-kolla | 20:30 | |
| *** robbbe has quit IRC | 20:31 | |
| *** robbbe has joined #openstack-kolla | 20:31 | |
| *** robbbe has quit IRC | 20:32 | |
| *** robbbe has joined #openstack-kolla | 20:33 | |
| *** openstackstatus has quit IRC | 20:46 | |
| *** openstack has joined #openstack-kolla | 20:50 | |
| *** ChanServ sets mode: +o openstack | 20:50 | |
| sean-k-mooney | how do you start kolla ansible container again after running kolla-ansible stop. | 20:52 |
| sean-k-mooney | deploy and upgrade dont seam to work | 20:52 |
| *** jascott1 has joined #openstack-kolla | 20:52 | |
| *** lpetrut has joined #openstack-kolla | 20:53 | |
| sean-k-mooney | mariadb_recovery also fails with "kolla_toolbox container is not running." after starting the dbs | 20:56 |
| sean-k-mooney | well this explains it... https://bugs.launchpad.net/kolla-ansible/+bug/1714015 | 20:59 |
| openstack | Launchpad bug 1714015 in kolla-ansible "containers not started during deploy if are stopped" [High,In progress] - Assigned to Eduardo Gonzalez (egonzalez90) | 20:59 |
| *** itlinux has quit IRC | 21:02 | |
| *** Jeffrey4l has quit IRC | 21:03 | |
| *** Jeffrey4l has joined #openstack-kolla | 21:11 | |
| *** gfidente|afk has quit IRC | 21:17 | |
| *** jascott1 has quit IRC | 21:22 | |
| *** Jeffrey4l has quit IRC | 21:24 | |
| *** harlowja has joined #openstack-kolla | 21:26 | |
| *** goldyfruit has joined #openstack-kolla | 21:27 | |
| *** itlinux has joined #openstack-kolla | 21:32 | |
| *** jascott1 has joined #openstack-kolla | 21:34 | |
| *** Jeffrey4l has joined #openstack-kolla | 21:37 | |
| *** lpetrut has quit IRC | 21:39 | |
| openstackgerrit | Pierre Blanc proposed openstack/kolla-ansible master: Homogenize the topics conf variable on templates https://review.openstack.org/533310 | 21:43 |
| *** unicell1 has joined #openstack-kolla | 21:50 | |
| *** strgout has quit IRC | 21:59 | |
| *** itlinux has quit IRC | 22:10 | |
| *** Jeffrey4l has quit IRC | 22:11 | |
| *** Jeffrey4l has joined #openstack-kolla | 22:14 | |
| *** mchlumsky has quit IRC | 22:38 | |
| *** dave-mccowan has quit IRC | 22:39 | |
| *** robbbe has quit IRC | 22:59 | |
| *** robbbe has joined #openstack-kolla | 22:59 | |
| *** robbbe has quit IRC | 22:59 | |
| *** robbbe has joined #openstack-kolla | 23:00 | |
| *** robbbe has quit IRC | 23:00 | |
| *** robbbe has joined #openstack-kolla | 23:01 | |
| *** harlowja has quit IRC | 23:01 | |
| *** robbbe has quit IRC | 23:01 | |
| *** vskilet has joined #openstack-kolla | 23:04 | |
| *** vskilet has quit IRC | 23:06 | |
| *** owalsh has quit IRC | 23:12 | |
| *** owalsh has joined #openstack-kolla | 23:12 | |
| *** itlinux has joined #openstack-kolla | 23:22 | |
| *** itlinux has quit IRC | 23:39 | |
| *** itlinux has joined #openstack-kolla | 23:57 | |
| jascott1 | portdirect around? | 23:58 |
| kolla-slack | Action: portdirect hides | 23:58 |
| portdirect | sup dude? | 23:59 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!