| *** dimsum__ has joined #kolla | 00:16 | |
| openstackgerrit | Fang Fenghua proposed stackforge/kolla: Add openvswitch contianer https://review.openstack.org/191532 | 00:16 |
|---|---|---|
| *** dimsum__ has quit IRC | 00:21 | |
| openstackgerrit | Fang Fenghua proposed stackforge/kolla: Add openvswitch contianer https://review.openstack.org/191532 | 00:35 |
| *** dimsum__ has joined #kolla | 00:35 | |
| *** fangfenghua has joined #kolla | 00:41 | |
| *** fangfenghua has quit IRC | 01:04 | |
| *** erkules has joined #kolla | 01:24 | |
| *** erkules_ has quit IRC | 01:27 | |
| *** fangfenghua has joined #kolla | 01:42 | |
| *** dimsum__ has quit IRC | 02:14 | |
| *** Haomeng has joined #kolla | 02:17 | |
| *** Haomeng|2 has quit IRC | 02:20 | |
| *** fangfenghua has quit IRC | 02:40 | |
| *** dimsum__ has joined #kolla | 02:40 | |
| *** dimsum__ has quit IRC | 02:46 | |
| *** [Leeloo] has quit IRC | 03:06 | |
| *** fangfenghua has joined #kolla | 03:17 | |
| SamYaple | yo | 03:18 |
| SamYaple | anyone awake and around | 03:18 |
| *** dimsum__ has joined #kolla | 03:46 | |
| *** dimsum__ has quit IRC | 03:52 | |
| *** daneyon has quit IRC | 04:51 | |
| *** daneyon has joined #kolla | 04:55 | |
| *** fangfenghua has quit IRC | 04:59 | |
| *** fangfenghua has joined #kolla | 05:01 | |
| *** fangfenghua has quit IRC | 05:08 | |
| *** fangfenghua has joined #kolla | 05:11 | |
| bmace | yo sam :) | 05:15 |
| *** fangfenghua has quit IRC | 05:16 | |
| SamYaple | hows it going bmace | 05:16 |
| bmace | good, thanks much SamYaple hope you are having a good weekend. | 05:21 |
| SamYaple | its actually my workday | 05:21 |
| SamYaple | i work Sat-Tue 2200-0800 | 05:22 |
| bmace | ah. hope work is going well then :) | 05:22 |
| bmace | if you want me to look at anything just let me know | 05:22 |
| SamYaple | just waiting on the multispec stuff to drpo | 05:24 |
| *** nihilifer_ has joined #kolla | 05:24 | |
| bmace | did sdale post a re-review taking into account the feedback? last time i looked i didn't think so. though it feels like we have agreement all around at this point. | 05:26 |
| SamYaple | he has seen it, i think he just doesnt want to type the words..... | 05:26 |
| SamYaple | or he is busy | 05:26 |
| SamYaple | ;) | 05:26 |
| bmace | i'm betting on the busy part, with his travel last week, etc. | 05:27 |
| SamYaple | a suggestion was made to remove the "env" option and do a one-time-only copy of the config into the container | 05:28 |
| *** nihilifer has quit IRC | 05:28 | |
| bmace | yeah. i saw there was some discussion too about whether we really need to continue the compose support or not. | 05:28 |
| bmace | so maybe a couple of fine points left to discuss. | 05:28 |
| SamYaple | no not compose, the "env" method was never about compose | 05:29 |
| SamYaple | its the dynamic config gen script | 05:29 |
| SamYaple | for the ansible stuff compose has to go | 05:29 |
| bmace | i know, i mean as a separate issue, people brought up just going pure ansible, even for dev delpoy, etc. | 05:29 |
| SamYaple | what will probably happen is there will be a way to so aio deploys with ansible so you can do both methods and then compose slowly goes away | 05:30 |
| bmace | i don't care that much about having another / separate env option. at that point it is about who uses it vs. the cost to maintain it. | 05:30 |
| bmace | super looking forward to ansible multi-node deploy though :) | 05:31 |
| SamYaple | ive been waiting.... | 05:32 |
| SamYaple | get this spec approved and itll be a very close 1to1 copy from yaodu | 05:32 |
| SamYaple | stuff that i have been using for 4 months, so it is at least a _little_ bit solid | 05:32 |
| SamYaple | ;) | 05:32 |
| bmace | an awesome start anyway | 05:32 |
| SamYaple | itll just be a matter of convincing people or my choices | 05:33 |
| SamYaple | or them hopefully having better ideas | 05:33 |
| bmace | i'm hoping we can get things in and fix them in place vs. blocking it getting in due to nits. | 05:33 |
| SamYaple | i agree. nothing i do changes the architecture so fix-in-place | 05:34 |
| SamYaple | would be my vote | 05:34 |
| SamYaple | before liberty-2 we could have fully functioning multi-node that way | 05:34 |
| SamYaple | albeit probably limited in its ability | 05:34 |
| bmace | yup. get the block of stone in place and then start chiseling to make it a beautiful statue | 05:35 |
| SamYaple | it looks like the community wants ceph containers in kolla too | 05:36 |
| bmace | really interested to see how some of the service dependency stuff is worked out, like service x depends on service y but you need to restart service y therefore perhaps you need to bring x down first, then y, then y back up, then x.. or do most of the ostk services handle the drop of their dependent services fairly well? | 05:36 |
| SamYaple | so ill pull those over | 05:36 |
| SamYaple | ansible runs everything serially, there is no dependacy list to run through | 05:36 |
| SamYaple | if mysql is down, the playbooks stop | 05:36 |
| SamYaple | if keystone is down the playbooks stop | 05:37 |
| SamYaple | etc | 05:37 |
| SamYaple | run the playbooks again and since they are idempotent they wont _redo_ things they have already done, just make new things work | 05:37 |
| bmace | serially per node, or per service? so say you have a bunch of rabbit nodes, it goes through the start on each? or can it fan out / async start at all? | 05:38 |
| SamYaple | basically its "run the playbooks until they work" | 05:38 |
| SamYaple | serially per task | 05:38 |
| SamYaple | so a "start" task would run on all nodes | 05:38 |
| SamYaple | at the same time* | 05:38 |
| bmace | ah, kk, that is good.. and all nodes need to pass for the task to pass? | 05:38 |
| SamYaple | (the number of nodes a task runs on at a time is also configurable) | 05:38 |
| SamYaple | no, that is also configurable | 05:39 |
| SamYaple | you can quit on _any_ failures, or quit if >50% fail, or if 3 nodes fail, etc | 05:39 |
| bmace | i assume you can specify, specifically, what services run on what nodes. that is definitely something someone may want to do, beyond just # of a particular service. | 05:39 |
| bmace | cool | 05:39 |
| SamYaple | yes, thats the inventory files job, determin what runs where | 05:40 |
| SamYaple | bmace: https://github.com/SamYaple/yaodu/blob/master/ansible/inventories/production | 05:40 |
| SamYaple | ignore from 12 down for a moment | 05:40 |
| *** nihilifer_ has quit IRC | 06:20 | |
| *** nihilifer has joined #kolla | 06:21 | |
| bmace | SamYaple that is an interesting format. especially that support[01:03] is inside of [support] that is either some poor naming or some sort of crazy set inception type thing? :) | 06:30 |
| SamYaple | it expands to support01, support02, support03 | 06:31 |
| SamYaple | they can also be spcified as such on multiple lines | 06:31 |
| bmace | kk, so there are 3 nodes with those names that are uposed for everything | 06:32 |
| SamYaple | in the case of _that_ inventory file, yes | 06:32 |
| SamYaple | but thosoe 4 sections about line 12 are more reference/metadata sections | 06:32 |
| SamYaple | the lines below line 12 are the actual groups ansible uses | 06:33 |
| SamYaple | so if oyu wanted to expand the keystone servers _only_ you could add entries under the keystone section | 06:33 |
| bmace | sure, in the case of this particular file there really didn't seem to be a need to even have network / storage/ compute sections since they all reference the same 3 nodes, but i appreciate that in most environments the actual systems for support / network / storage / compute will be different. | 06:34 |
| SamYaple | there is because i reference those 4 groups i nthe below files | 06:34 |
| bmace | the syntax is pretty simple | 06:35 |
| SamYaple | so i could expand from those nodes to proper seperate nodes | 06:35 |
| bmace | yeah, i see that :) | 06:35 |
| SamYaple | basically those are the 4 groups I could see yo uwould ever need and _most_ installations, even large ones, adopt the "all services on controller node" type model | 06:36 |
| bmace | yup, when i had broken them out before it had come down to control / storage / network / compute. | 06:36 |
| bmace | support is a good alternative name | 06:37 |
| SamYaple | i only choise it because it is 7 characters, which matches network storage compute :P | 06:37 |
| SamYaple | control didnt quite work for me | 06:37 |
| SamYaple | wow dvr is wierd | 06:40 |
| SamYaple | i can ping from any router namespace | 06:40 |
| SamYaple | but i cant ping from two router namespsace at the same time | 06:40 |
| SamYaple | it only returns to one | 06:40 |
| *** Haomeng|2 has joined #kolla | 07:15 | |
| *** Haomeng has quit IRC | 07:17 | |
| *** fangfenghua has joined #kolla | 07:17 | |
| *** fangfenghua has quit IRC | 07:25 | |
| *** fangfenghua has joined #kolla | 07:40 | |
| SamYaple | ah nvm | 07:44 |
| SamYaple | it was just arp_responder and l2pop nonsense | 07:45 |
| SamYaple | dvr is fully functioning | 07:45 |
| *** nihilifer has quit IRC | 08:21 | |
| *** fangfenghua has quit IRC | 08:29 | |
| openstackgerrit | Martin André proposed stackforge/kolla: Separate base images into different folders https://review.openstack.org/184222 | 08:48 |
| openstackgerrit | Martin André proposed stackforge/kolla: Separate base images into different folders https://review.openstack.org/184222 | 08:52 |
| SamYaple | hey mandre, you here? | 08:52 |
| mandre | hi SamYaple, not for long… what's up? | 08:53 |
| SamYaple | hey! just checking in about the patch agove | 08:53 |
| SamYaple | pdb has another blueprint simliar to that to seperate all common scripts | 08:54 |
| mandre | right, I just added my comments to it | 08:54 |
| SamYaple | https://blueprints.launchpad.net/kolla/+spec/refactor-base-image-layout | 08:54 |
| mandre | I suggested rebasing on top of Id63eac1eb9893ceb775c28bf22842dede85c214a, that's why I fixed the merge conflict :) | 08:55 |
| SamYaple | ah ok. i see that now. so the idea was the common scripts would have if/then clauses for the cases where things are different | 08:55 |
| SamYaple | i did some checking, there will be very few differences i believe | 08:55 |
| mandre | yeah, I'm not too concerned about it, we'll find a solution if there are really big differences between distros | 08:57 |
| mandre | the package install may be annoying, that's why I mentioned it in the review | 08:57 |
| SamYaple | the script could always just _not_ be a symlink for those cases | 08:57 |
| SamYaple | mandre: the dockerfiles are per distro | 08:57 |
| SamYaple | so no package installs | 08:57 |
| mandre | oh… you're correct, what was I thinking | 08:58 |
| SamYaple | cool so were all o nthe same page. i like it | 08:59 |
| SamYaple | can you update https://review.openstack.org/#/c/184222/ with the blueprint above? | 08:59 |
| mandre | yep, i think this is the right approach | 08:59 |
| mandre | sorry i really have to run now, i need to prepare dinner | 09:00 |
| SamYaple | mandre: its cool. tkae off! | 09:00 |
| mandre | feel free to update it, it's your commit after all ;) | 09:00 |
| SamYaple | yea yea. i will just cant get these multicommits working right :/ | 09:01 |
| mandre | have a nice week end | 09:01 |
| SamYaple | same | 09:01 |
| *** bradjones has joined #kolla | 09:02 | |
| *** bradjones has joined #kolla | 09:02 | |
| *** bradjones_ has quit IRC | 09:02 | |
| openstackgerrit | Sam Yaple proposed stackforge/kolla: Separate base images into different folders https://review.openstack.org/184222 | 09:11 |
| *** dimsum__ has joined #kolla | 10:19 | |
| *** dimsum__ has quit IRC | 10:24 | |
| *** vbel has quit IRC | 10:38 | |
| *** vbel has joined #kolla | 10:39 | |
| openstackgerrit | Martin André proposed stackforge/kolla: Separate base images into different folders https://review.openstack.org/184222 | 11:43 |
| mandre | ^ SamYaple, restored links to new designate service that were removed by mistake in PS6 | 11:44 |
| SamYaple | mandre: i must have pulled down patchset4 before you uploaded 5, thanks! | 11:45 |
| mandre | np | 11:47 |
| SamYaple | man. basically the entire start.sh script in most services needs to be only used for the self-configuring container stuff | 11:48 |
| SamYaple | i dont know how I am going to get people to see this stuff | 11:49 |
| SamYaple | agree* to this stuff | 11:49 |
| *** zhiwei has quit IRC | 12:23 | |
| *** sdake has joined #kolla | 13:48 | |
| sdake | morning | 13:49 |
| *** fangfenghua has joined #kolla | 13:54 | |
| *** sdake has quit IRC | 13:55 | |
| *** dimsum__ has joined #kolla | 15:36 | |
| *** fangfenghua has quit IRC | 15:43 | |
| *** fangfenghua has joined #kolla | 15:44 | |
| *** fangfenghua has quit IRC | 15:53 | |
| *** diogogmt has joined #kolla | 16:38 | |
| *** sdake_ has joined #kolla | 17:01 | |
| harmw | sdake_: ping | 17:23 |
| sdake_ | harmw wound me | 17:26 |
| harmw | you should drop the _ | 17:26 |
| *** sdake_ is now known as sdake | 17:26 | |
| harmw | there | 17:26 |
| harmw | my underscore-alarm was going crazy for hours now | 17:27 |
| harmw | do we have a list of dependencies? | 17:34 |
| harmw | required binaries to run kolla, that is | 17:35 |
| sdake | not that I know of | 17:35 |
| harmw | hm hm, ok | 17:35 |
| *** diogogmt has quit IRC | 17:39 | |
| *** diogogmt has joined #kolla | 17:40 | |
| harmw | sdake: whats the proper notation for a bugfix commit? | 17:48 |
| sdake | Closes-Bug: #XYZ | 17:48 |
| harmw | ah, yes | 17:48 |
| openstackgerrit | Harm Weites proposed stackforge/kolla: Check if tools/kolla binary dependencies are met. https://review.openstack.org/191580 | 17:53 |
| *** fangfenghua has joined #kolla | 17:54 | |
| *** dims_ has joined #kolla | 18:00 | |
| *** dims_ has quit IRC | 18:01 | |
| *** dimsum__ has quit IRC | 18:03 | |
| *** fangfenghua has quit IRC | 18:04 | |
| *** dimsum__ has joined #kolla | 19:02 | |
| *** fangfenghua has joined #kolla | 19:03 | |
| *** bradjones has quit IRC | 19:05 | |
| *** dimsum__ has quit IRC | 19:06 | |
| *** bradjones has joined #kolla | 19:07 | |
| *** bradjones has joined #kolla | 19:07 | |
| *** fangfenghua has quit IRC | 19:08 | |
| *** dimsum__ has joined #kolla | 20:36 | |
| *** dimsum__ has quit IRC | 21:09 | |
| *** fangfenghua has joined #kolla | 21:10 | |
| *** fangfenghua has quit IRC | 21:19 | |
| *** fangfenghua has joined #kolla | 22:18 | |
| *** fangfenghua has quit IRC | 22:24 | |
| *** pdb has quit IRC | 22:43 | |
| *** pdb has joined #kolla | 22:43 | |
| *** dimsum__ has joined #kolla | 22:55 | |
| *** dimsum__ has quit IRC | 23:00 | |
| *** Haomeng|2 is now known as Haomeng | 23:24 | |
| *** fangfenghua has joined #kolla | 23:24 | |
| *** fangfenghua has quit IRC | 23:29 | |
| *** dimsum__ has joined #kolla | 23:59 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!