| *** dims has quit IRC | 00:30 | |
| *** dims has joined #openstack-containers | 00:30 | |
| *** EricGonczer_ has joined #openstack-containers | 00:44 | |
| *** suro_ has quit IRC | 00:49 | |
| *** dims has quit IRC | 00:52 | |
| *** dims has joined #openstack-containers | 00:52 | |
| *** dims has quit IRC | 01:21 | |
| *** dims has joined #openstack-containers | 01:22 | |
| *** EricGonczer_ has quit IRC | 01:38 | |
| *** EricGonczer_ has joined #openstack-containers | 02:01 | |
| *** dims has quit IRC | 02:02 | |
| *** dims has joined #openstack-containers | 02:02 | |
| *** marcoemorais has quit IRC | 02:21 | |
| *** EricGonczer_ has quit IRC | 02:30 | |
| *** EricGonczer_ has joined #openstack-containers | 02:44 | |
| *** EricGonczer_ has quit IRC | 02:45 | |
| *** jeckersb is now known as jeckersb_gone | 02:47 | |
| *** dims has quit IRC | 02:50 | |
| *** dims has joined #openstack-containers | 02:50 | |
| *** yuanying has joined #openstack-containers | 03:18 | |
| *** adrian_otto has quit IRC | 03:19 | |
| *** EricGonczer_ has joined #openstack-containers | 03:23 | |
| *** EricGonczer_ has quit IRC | 03:30 | |
| *** adrian_otto has joined #openstack-containers | 03:37 | |
| *** yuanying has quit IRC | 03:46 | |
| *** yuanying has joined #openstack-containers | 03:46 | |
| *** subscope has joined #openstack-containers | 04:36 | |
| *** nshaikh has joined #openstack-containers | 05:32 | |
| *** suro_ has joined #openstack-containers | 06:50 | |
| *** subscope has quit IRC | 06:55 | |
| *** suro_ has quit IRC | 08:03 | |
| *** suro_ has joined #openstack-containers | 08:03 | |
| *** jgjl has joined #openstack-containers | 08:05 | |
| *** suro_ has quit IRC | 08:08 | |
| *** jgjl has quit IRC | 08:08 | |
| *** adrian_otto has quit IRC | 08:16 | |
| *** jgjl has joined #openstack-containers | 08:33 | |
| *** EricGonczer_ has joined #openstack-containers | 10:46 | |
| *** dims has quit IRC | 10:54 | |
| *** dims has joined #openstack-containers | 10:55 | |
| *** jgjl has quit IRC | 11:07 | |
| *** EricGonczer_ has quit IRC | 11:16 | |
| *** jgjl has joined #openstack-containers | 11:48 | |
| *** kitch_ has quit IRC | 12:23 | |
| *** nshaikh has left #openstack-containers | 12:47 | |
| *** kitch_ has joined #openstack-containers | 13:05 | |
| *** stannie has joined #openstack-containers | 13:07 | |
| *** jgjl has quit IRC | 13:54 | |
| *** jeckersb_gone is now known as jeckersb | 13:55 | |
| *** dims has quit IRC | 13:56 | |
| *** dims has joined #openstack-containers | 13:56 | |
| *** jgjl has joined #openstack-containers | 14:06 | |
| *** julim has joined #openstack-containers | 14:07 | |
| erw | dims: fyi, did you see my commit? | 14:16 |
|---|---|---|
| erw | dims: it's mostly a PoC for feeling out how the API should look, what we should expose, etc. | 14:16 |
| dims | erw: did not dig into it yet. wanted to get the CI straightened out with the github code. waiting on adrian_otto now, he is the only one with core karma. | 14:18 |
| dims | erw: up early? :) | 14:18 |
| erw | dims: back in the USA | 14:18 |
| dims | erw: ah ok | 14:18 |
| erw | arrived last night | 14:18 |
| erw | dims: my PoC doesn't have to be merged. I was playing with the API, figuring out the usage model, etc. | 14:19 |
| dims | erw: cool | 14:19 |
| *** jeckersb is now known as jeckersb_gone | 14:19 | |
| kitch_ | erw: Link to your commit? | 14:20 |
| erw | https://review.openstack.org/#/c/133211/ | 14:23 |
| *** thomasem has joined #openstack-containers | 14:33 | |
| *** EricGonczer_ has joined #openstack-containers | 14:39 | |
| *** jeckersb_gone is now known as jeckersb | 14:43 | |
| *** ycombinator_ has joined #openstack-containers | 14:51 | |
| *** ycombinator_ has quit IRC | 14:52 | |
| *** jlb13 has joined #openstack-containers | 15:04 | |
| kitch_ | erw: So containers don’t exist outside of a pod? | 15:17 |
| erw | kitch_: yes. | 15:19 |
| erw | kitch_: I had discussed with slower about just calling a pod a "container" and containers "processes"... but we agreed that while that might make more sense, it's not the terminology people are using today | 15:20 |
| kitch_ | I would agree with that. | 15:21 |
| kitch_ | Just need to think through the consumption model for users who don’t care about pods | 15:21 |
| kitch_ | Does pod neccessarily indicated all it’s containers are on a single host? | 15:21 |
| kitch_ | s/it’s/its | 15:22 |
| erw | kitch_: all pods are in a single "bay" (term is... flexible), where a bay == nova instance == host | 15:22 |
| erw | a pod may contain multiple containers | 15:23 |
| kitch_ | erw: understood, but we are stating that all containers in a pod are located in a single bay | 15:23 |
| erw | kitch_: really, we could just collapse that and say that containers might contain containers and eliminate pods. The problem is that we can't necessarily manage systems that are N-level nested | 15:24 |
| kitch_ | erw: Do we think that pods are a logical construct, or a real construct, in that a pod, is represented by a container in the real world | 15:25 |
| kitch_ | Or, third option, depends on the “driver” | 15:26 |
| erw | kitch_: a pod is a real thing and a container by another name. | 15:26 |
| kitch_ | ok | 15:26 |
| erw | specifically, it's a container with a specific image (for the selected driver) that allows nesting | 15:26 |
| kitch_ | So when I call the api to create a pod. a container is being created in a bay (on a host) | 15:26 |
| kitch_ | got it | 15:26 |
| erw | the idea here is that a kubernetes pod definition, fig definition, or docker multi-host metdata could be fed directly into magnum | 15:27 |
| kitch_ | right | 15:27 |
| erw | or someone could use the magnum api to programmatically manage nesting, rather than have a leaky abstraction where the user would need to speak directly to Docker and supply their own specialized images for pod management | 15:28 |
| kitch_ | Sure. It’s all just a way of prescribing good usage | 15:29 |
| kitch_ | You could also argue that a pod is just a speciallized bay | 15:29 |
| kitch_ | etc, etc | 15:29 |
| erw | even in the code, iirc, the Pod is a subclass of Container | 15:29 |
| erw | well, they're all containers of some sort, yes? | 15:29 |
| kitch_ | Agreed. | 15:30 |
| erw | the problem with exposing that to a user is that the UX gets confusing | 15:30 |
| kitch_ | agreed | 15:30 |
| dims | Inception :) | 15:30 |
| kitch_ | So then my question is this. Many users won’t care about pods. They just want to get a pile of containers, distributed across available hosts. | 15:30 |
| kitch_ | They may not care or know about pods, do we need to provide an entry point for them? | 15:31 |
| kitch_ | Or do we just force them to swallow pods | 15:31 |
| erw | my perspective was that the 'PUT pod' could easily create a pod with a single container | 15:32 |
| erw | and they would list their pods. | 15:32 |
| erw | i.e. that's the top-level entry | 15:32 |
| kitch_ | ok, so basically there is a pod type: container | 15:33 |
| kitch_ | for those that don’t want pod type: k8s, or pod type: fig | 15:33 |
| erw | something like that | 15:33 |
| erw | by doing this, the bay can be split up into multiple units of compute... and users aren't restricted from adding new siblings to their containers later, if they decide they need it (but it wasn't needed at spawn) | 15:34 |
| erw | otherwise, if containers were top-level and a user had an unforeseen need to run another process on the same IP address, they could simply do it without having to respawn | 15:35 |
| kitch_ | Have you guys done any performance analysis to show that there is no overhead of nesting containers? | 15:35 |
| kitch_ | or other limitations | 15:36 |
| erw | *er... I wrote that wrong, but I hope you get it? | 15:36 |
| kitch_ | I get it | 15:36 |
| erw | kitch_: some have and it seemed fine. Kubernetes folks are doing it... | 15:36 |
| erw | kitch_: but it's certainly a path being paved, rather than a well-traveled road | 15:36 |
| kitch_ | Day 0: I just want my container to live all on it’s own. Day 5: Oh man, I wish I could have a monitoring container in a pod with my original app | 15:36 |
| erw | kitch_: and there is the whole 'nsexec' thing, too. I put "execute" into the api for a container, but I'm honestly not sure it belongs | 15:38 |
| erw | (and I'd argue if it belongs in *docker*, either) | 15:38 |
| erw | imho, 'docker exec' should just be a form of 'docker run' | 15:39 |
| kitch_ | erw: Agreed, although I would note that there are plenty of users that love docker/containers for the “image” portability rather than the container “way of life”. So it comes down to how much you want to entertain those users | 15:40 |
| erw | kitch_: a certain part of me thought, well, that's what Nova (and nova-docker) is for... docker image portability without the microservices/12factor design | 15:42 |
| erw | kitch_: and that magnum could be the api for microservices (and if OpenVZ or LXC doesn't want to do microservices, then the Nova api is enough?) | 15:43 |
| kitch_ | erw: Absolutely, these are users that would be very happy with nova-docker, magnum would have little appeal I think. | 15:43 |
| kitch_ | I wasn’t sure if you were speaking of docker execute vs docker run in the context of core docker or in the context of magnum | 15:44 |
| erw | kitch_: both | 15:44 |
| kitch_ | erw: I suppose that’s not the end of the world, for those that want to use docker as packaging they are most likely running ssh or some other way of hooking back into their container | 15:45 |
| *** EricGonczer_ has quit IRC | 15:48 | |
| *** EricGonczer_ has joined #openstack-containers | 15:50 | |
| erw | kitch_: well, one idea is that you don't need ssh except for compatibility reasons if you can get a 'docker exec' or 'docker run' into the picture | 15:57 |
| erw | but certainly, you're not prevented from running it, either. | 15:58 |
| kitch_ | erw: Sure, it’s just a comfort thing for some traditional IaaS consumers | 15:58 |
| erw | well, plus, it appears relatively secure. TLS, not so much. | 15:59 |
| *** diga has joined #openstack-containers | 16:03 | |
| *** kebray has joined #openstack-containers | 16:09 | |
| diga | Hello dims | 16:11 |
| dims | hi diga | 16:12 |
| diga | PYTHONPATH=. pecan serve magnum/config.py | 16:12 |
| diga | AttributeError: 'pecan.conf' object has no attribute 'app' | 16:13 |
| diga | this error I am still getting | 16:13 |
| diga | not able to trace the fix | 16:13 |
| diga | any pointers ?? | 16:13 |
| diga | r u running this in virtualenv | 16:14 |
| dims | diga: get a fresh git clone of magnum repo from git.openstack.org and then apply the 2 reviews here - https://review.openstack.org/#/q/status:open+project:stackforge/magnum+branch:master+topic:initial-squashed-commit,n,z | 16:14 |
| diga | ok | 16:14 |
| *** kebray has quit IRC | 16:16 | |
| dims | diga: please see erw's review - https://review.openstack.org/#/c/133211/ - this is unrelated to the initial import but was something erw and Slower came up with on friday i believe | 16:21 |
| dims | diga: i can confirm that a fresh checkout of magnum and the 2 reviews i pointed earlier works for me again | 16:21 |
| diga | ok | 16:21 |
| diga | ok | 16:22 |
| diga | Sorry dims, the actual issue is at my side | 16:22 |
| diga | just fixed | 16:22 |
| diga | Thanks | 16:22 |
| dims | ah ok, what did you have to do? | 16:22 |
| diga | I upgraded my machine to 14.04, so some packages are missing | 16:23 |
| dims | k | 16:23 |
| diga | yes, I am looking at erw's review | 16:24 |
| diga | but as I am seeing, they have used flask | 16:25 |
| diga | Hi erw | 16:28 |
| diga | you there ? | 16:28 |
| *** EricGonczer_ has quit IRC | 16:30 | |
| *** diga has quit IRC | 16:34 | |
| *** adrian_otto has joined #openstack-containers | 16:37 | |
| erw | pong | 16:38 |
| *** diga has joined #openstack-containers | 16:38 | |
| diga | dims: got disconncted | 16:39 |
| erw | diga: it's a PoC we threw together. I used flask because it was quick and easy. | 16:39 |
| diga | ok | 16:39 |
| diga | NP | 16:40 |
| diga | I think myself & dims will continue our work on API using pecan+wsme | 16:44 |
| diga | because its a time taking work | 16:44 |
| diga | we will integrate your stuff once you are done with the POC to this API module | 16:45 |
| erw | diga: you're welcome to incorporate the ideas or code now (or not at all) if you want | 16:48 |
| diga | ok | 16:48 |
| erw | diga: the point was to have something we could point at and say, "here is some progress", also, it was a good exercise for figuring out what made sense to be in magnum and what the design should look like -- even if the code proves to be ephemeral | 16:48 |
| diga | okay | 16:49 |
| diga | fine | 16:49 |
| dims | erw: so you ok to just borrow the REST API and re-code over what diga has? | 16:51 |
| dims | ah you already answered that :) | 16:51 |
| diga | :) | 16:53 |
| *** jgjl has quit IRC | 16:57 | |
| diga | dims: give me today, I'll send my patches from tomorrow due to formating of my laptop | 17:04 |
| *** marcoemorais has joined #openstack-containers | 17:14 | |
| *** nshaikh has joined #openstack-containers | 17:38 | |
| *** diga has quit IRC | 17:44 | |
| *** suro_ has joined #openstack-containers | 17:55 | |
| *** kebray has joined #openstack-containers | 18:02 | |
| *** diga has joined #openstack-containers | 18:15 | |
| dims | adrian_otto: ping | 18:18 |
| adrian_otto | hi | 18:19 |
| adrian_otto | dims: pong | 18:20 |
| dims | adrian_otto: you are the sole core on magnum, you can add more folks using gerrit. | 18:20 |
| dims | adrian_otto: the squashed set of commits from github is here - https://review.openstack.org/#/c/133811/ if you wish to +2/+A | 18:20 |
| *** diga has quit IRC | 18:21 | |
| adrian_otto | dims: thanks! | 18:21 |
| adrian_otto | I'll add core selection to our upcoming team agenda | 18:22 |
| adrian_otto | I'd like to get consensus from the team on the right starting group | 18:22 |
| dims | sounds good | 18:23 |
| adrian_otto | until then I will need to diligently review the code and respond to the +1 -1 votes from our reviewers | 18:23 |
| adrian_otto | I'll edit the agenda wiki right now so I don't forget | 18:23 |
| *** diga has joined #openstack-containers | 18:23 | |
| dims | sounds good, the url in gerrit for adding folks is here - https://review.openstack.org/#/admin/groups/473,members | 18:24 |
| adrian_otto | https://review.openstack.org/#/q/status:open+magnum,n,z | 18:27 |
| adrian_otto | ^^ please review and comment/vote on these | 18:27 |
| *** subscope has joined #openstack-containers | 18:36 | |
| *** diga has quit IRC | 18:44 | |
| *** EricGonczer_ has joined #openstack-containers | 18:48 | |
| *** EricGonczer_ has quit IRC | 18:52 | |
| *** EricGonczer_ has joined #openstack-containers | 18:59 | |
| *** kebray has quit IRC | 19:08 | |
| *** kebray has joined #openstack-containers | 19:12 | |
| *** marcoemorais has quit IRC | 19:44 | |
| *** marcoemorais has joined #openstack-containers | 19:44 | |
| *** sdake has joined #openstack-containers | 19:59 | |
| *** subscope has quit IRC | 20:01 | |
| sdake | adrian_otto are the container commits going to irc, and if not, would you like me to fix that? | 20:02 |
| *** thomasem has quit IRC | 20:57 | |
| adrian_otto | I think I configured that | 21:27 |
| adrian_otto | are new reviews not showing up here? | 21:27 |
| adrian_otto | sdake: if it needs fixing, please proceed | 21:27 |
| *** kitch_ has quit IRC | 21:34 | |
| *** thomasem has joined #openstack-containers | 21:37 | |
| *** kebray has quit IRC | 21:49 | |
| *** nshaikh has quit IRC | 21:53 | |
| *** adrian_otto has quit IRC | 22:04 | |
| *** adrian_otto has joined #openstack-containers | 22:13 | |
| *** jeckersb is now known as jeckersb_gone | 22:33 | |
| *** EricGonczer_ has quit IRC | 22:49 | |
| *** thomasem has quit IRC | 22:54 | |
| *** marcoemorais1 has joined #openstack-containers | 23:05 | |
| *** marcoemorais has quit IRC | 23:09 | |
| *** marcoemorais1 has quit IRC | 23:30 | |
| *** kebray has joined #openstack-containers | 23:30 | |
| *** marcoemorais has joined #openstack-containers | 23:31 | |
| *** EricGonczer_ has joined #openstack-containers | 23:51 | |
| *** dims_ has joined #openstack-containers | 23:54 | |
| *** dims has quit IRC | 23:57 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!