Monday, 2016-03-07

*** toddjohn_ has joined #openstack-app-catalog00:22
*** spzala has joined #openstack-app-catalog00:48
*** toddjohn_ has quit IRC00:50
*** spzala_ has joined #openstack-app-catalog02:15
*** spzala has quit IRC02:17
*** spzala_ has quit IRC02:25
*** spzala has joined #openstack-app-catalog02:25
*** spzala has quit IRC02:30
*** spzala has joined #openstack-app-catalog02:38
*** spzala has quit IRC02:43
*** spzala has joined #openstack-app-catalog03:15
*** spzala has quit IRC04:02
*** spzala has joined #openstack-app-catalog04:02
*** spzala has quit IRC04:03
openstackgerritSahdev Zala proposed openstack/app-catalog: Add TOSCA assets
openstackgerritSahdev Zala proposed openstack/app-catalog: Add TOSCA assets
*** rmoe has quit IRC06:50
*** pcaruana has joined #openstack-app-catalog08:25
*** hogepodge has quit IRC10:56
openstackgerritOmar Shykhkerimov proposed openstack/app-catalog: Create form for making assets
*** hogepodge has joined #openstack-app-catalog11:34
*** toddjohn_ has joined #openstack-app-catalog13:04
*** toddjohn_ has quit IRC13:09
*** toddjohn_ has joined #openstack-app-catalog13:10
*** toddjohn_ has quit IRC13:14
*** toddjohn_ has joined #openstack-app-catalog13:18
*** rmoe has joined #openstack-app-catalog13:24
*** toddjohn_ has quit IRC13:52
*** toddjohn_ has joined #openstack-app-catalog13:58
openstackgerritSahdev Zala proposed openstack/app-catalog: Add TOSCA assets
*** spzala has joined #openstack-app-catalog14:37
kfox1111so, whats the plan with tosca resources an the horizon plugin?17:40
kfox1111how are we going to get that integrated?17:40
*** toddjoh__ has joined #openstack-app-catalog18:29
*** toddjohn_ has quit IRC18:31
openstackgerritOmar Shykhkerimov proposed openstack/app-catalog: Create form for making assets.
*** kzaitsev_mb has joined #openstack-app-catalog18:34
*** toddjohn_ has joined #openstack-app-catalog18:36
*** toddjoh__ has quit IRC18:39
kzaitsev_mbwanted to give an update on =) Omar is an intern in Mirantis, and I have partial access to his time =) I've tasked him with making a simple uploading form for assets, while I'm tinkering with auth.18:45
kzaitsev_mbhaven't had time to review his work yet, though — it's a holiday today in Russia and my wife and family demanded most of my attention today =). The joys of working in a multi-location company.18:46
kzaitsev_mbI've commented in the review, but had though it would be right to keep you, docaedo, kfox1111 updated here too. ^^18:47
*** spzala has quit IRC18:52
*** spzala has joined #openstack-app-catalog18:52
*** spzala has quit IRC18:57
*** spzala has joined #openstack-app-catalog18:58
*** kzaitsev_mb has quit IRC19:07
docaedokzaitsev_ws: hi, I have looked at it but have not provided feedback. I basically agree, there's no need to merge PoC code that won't be used in production, and I don't see how that form would be used in production without a *lot* of additional work around it19:10
docaedoso makes sense to have it for the PoC and also useful to share that work, but not sure about this approach19:11
openstackgerritSahdev Zala proposed openstack/app-catalog: Add TOSCA assets
*** kzaitsev_mb has joined #openstack-app-catalog19:54
kfox1111kzaitsev_mb: yeah. I'm ok with it being poc and maybe perpetially WIP'ed forever. but just want to make sure it doesn't go farther then that. Its kind of a dead end. Don't want to spend much review time on it knowing its doa.19:57
kfox1111tosca question from above still needs discussion.19:58
*** kzaitsev_mb has quit IRC19:59
kfox1111not sure how to handle it. I believe the patch is valid, but its not very useful to accept dead code with no current path.19:59
kfox1111I'm kind of on the fence, but just over on the side where I think every new asset type needs to be supportable by the horizon plugin through some means or another. And I'm not sure what that would look like with tosca.20:01
kfox1111has that been discussed while I've been away?20:01
kfox1111Is it reasonable to block the new asset type until that discussion has been had?20:04
kfox1111do we have a blueprint or something for tosca support?20:08
docaedoThere is a blueprint for tosca stuff20:19
docaedoand I don't think it's reasonable to block it waiting for the horizon plugin to handle it, as it's not too difficult to add it to the web site20:19
docaedoand as great as the horizon work is, it's not the primary way people are consuming things from :)20:20
openstackgerritSahdev Zala proposed openstack/app-catalog: Add TOSCA assets
kfox1111heh. I'm not sure how most users consume from the catalog honestly. do we have metrics?20:24
kfox1111thanks for the link.20:24
kfox1111yeah, the blueprint helps. thanks.20:24
kfox1111so, the first two workitems are done?20:25
docaedoyeah, first two work items on that BP are complete, need to update that section :)20:25
kfox1111doesn't look like there is a work item for this change. hello world template.20:25
kfox1111the reason I ask about the app catalog ui stuff, is how do we want to order things: * add new asset types to website, * add new asset types to horizon plugin, * merge unified python server code20:28
kfox1111or * merge unified python server code, * add new asset type to unified code20:28
docaedoah fair question - the way I see it, making changes to the web site to support TOSCA assets (the website as-is) will be relatively easy and even it their short term, it's a big benefit to get the catalog looking even slightly new before the next summit20:30
docaedoI know you were planning to work on using a common codebase between the site and horizon plugin, but no idea where that stands, or what the implementation is going to be like20:31
docaedonot to mention whether or not we're going to need to dig up a UI person or a designer to help make it pretty ;)20:31
kfox1111yeah. well, thats kind of related. if tosca really wants to get in quickly, they could push to get the merge to happen.20:32
kfox1111I'm guessing I'll have time mid august to work on the merge again. so maybe about a month out if no one can get to it before me.20:32
kfox1111the UI person is a seperate issue. iether way, we need that. ;)20:33
docaedomid-august is 6 months out?20:33
kfox1111sry. April.20:33
docaedobut also it doesn't seem reasonable to me to say to a community that wants to share assets with the app catalog "ok sure, but before we accept this new stuff you need to do some other relatively unrelated work first"20:34
docaedoto me, continuing to improve the web site as-is (cosmetic stuff, plus adjust the layout so it more easily handles an extra asset type or two) seems like it's not a bad way to make progress20:35
docaedomeanwhile there's going to be work with glare stuff happening concurrently, plus the "common codebase for horizon and website" work can progress also in parallel20:36
kfox1111heh. it hasn't stoped any other openstack project for blocking on cleanup.... they all do that. :/20:38
docaedotrue enough, luckily we're different :D20:38
kfox1111yeah. I get that. just trying to figure out how to make progress on the blocked common catalog task,20:38
docaedo(also nobody else is dependent on us, unlike other projects - which is a *great* thing IMO)20:38
kfox1111and the adding yet another thing that will be needed be ported makes the task take longer, not less time.20:38
kfox1111and the precident thing. if we get solum finally show up or some of the other projects, thats yet more that would need to go through.20:39
docaedoSure adds more time but should be incredibly minimal, because at the end of the day we are aiming to make the catalog a place for all/any assets, and the catalog shouldn't care too much about what the asset types are20:39
kfox1111at what point do we pay off the technical debt?20:39
kfox1111yeah, but the website today isn't built for infinite assets. really just 3.20:40
docaedoI agree that we need to work on the common codebase thing, and I think I could probably make time to start doing that. I still don't want to block adding TOSCA stuff though20:40
kfox1111we need to ensure our plumbing supports it. I've tried to make it more easy in the shared code to add more asset types. but even it has a few technical debt things that need paying down.20:40
kfox1111k. since they have shown up and willing to do some work, I won't block it. but how about this, no additional new types until the merge happens, and someone needs to commit to writing the website code first, before the asset is merged?20:41
kfox1111they need to at least think through how the asset type can be added to the plugin. as is now, I'm really not sure how that's going to work. its a cli at the moment, right?20:42
docaedoI can agree with no additional new types for the moment (especially since I think we are in zero danger of that happening) - is there a start to the common codebase work somewhere? I'll admit I haven't even peeked at that new repo20:43
kfox1111yeah. sec....20:43
docaedoTOSCA has CLI and horizon stuff AFAIK20:43
kfox1111where's the horzion stuff?20:43
kfox1111if that exists, then that would mostly answer the quiestion, and I'd feel alot better about it.20:44
docaedowait, I might be conflating tosca with mistral20:44
kfox1111yeah, mistral has a webui. not sure about tusca.20:45
kfox1111I think when we talked about it like a month or two ago, we were contimplating running the cli converter thingy on it to generate heat entries and maybe putting them in the assets instead of the raw tosca stuff.20:45
kfox1111then it would work everywhere.20:46
docaedoyep - tosca parser takes a tosca template and spits out a heat template, but the issue is that if you run it locally, the translator can make a heat template that's tailored to your environment, rather than being a generic template that you'll likely need to adjust/fix20:46
docaedoso that's where that landed last time I talked to anyone about, but I think spzala has more info20:47
kfox1111so the app catalog ui is totally client side, so you wouldn't be able to call the tosca cli from it. they probably need a horizon plugin.20:48
kfox1111or we'd have to make a service on to do the conversion with the cli tool. but that's really ugly. :/20:48
docaedoyeah I don't think it's something to worry about too much ATM, though I think there's potential value to having some services (for instance, have an additional link for TOSCA assets that says "run through translator to output heat template")20:50
docaedosimilarly, running disk image builder as a service, to create disk images on the fly20:50
docaedothough .. that might actually count as distributing the base OS, which I know Ubuntu frowns on20:50
spzalaHi kfox1111: docaedo: so we have two projects in tosca space, tosca-parser is a just parser for tosca template and heat-translator is another project that takes parsed template in the form of graph and converts it to Heat HOT20:51
docaedo(though I think that would be an incredibly valuable thing to have)20:51
spzalaheat-translator is a command line tool only. Last what we decided was for now, use app-catalog to host tosca templates (in yaml and CSAR format)20:52
spzalaand that's it for initial support and once that is done. If I remember correctly, heat-translator being only CLI tool that was a problem to run it straight from app-catalog20:53
spzala(i.e. '...and once that is done, we can discuss if heat-translator can be invoked from app-catalog.')20:54
kfox1111spzala: yeah. we have differering opinions in the group on how users are actually going to consume the app catalog long term. I think the majority of users are going to be consuming it via horizon.20:59
kfox1111copy and pasting things from a website into a cli is really not user friendly.20:59
kfox1111still better then nothing, but may cut out 90% of your potential users.21:00
docaedofor me part of the issue though is that the app catalog should always be an outward facing index of things for openstack21:00
spzalakfox1111: Thx. OK, agree. So for now, we can go with our initial plan of enabling hosting of tosca templates and binaries21:00
docaedoif people can only see that content after installing openstack and choosing to include the app-catalog-ui, then they're unlikely to see that21:01
kfox1111docaedo: at present, yeah. I agree that's a problem too. something we need to really keep pushing hard.21:02
kfox1111same goes for the website too. if they have to know to look for it, then we fail too.21:02
kfox1111if its right their in their face in horizon, then we get users.21:03
docaedosure but in that instance they are already openstack users. the app catalog website on it's own could (should!) bring in users of any cloud21:03
docaedoTBH the biggest problem is that we/openstack lacks a standard way of packaging an app21:04
docaedoso we're trying to solve the problem of getting attention as if we are a PaaS, but it's just on top of a IaaS and you still need to bring something else along to really use it like it's a PaaS21:05
kfox1111IMHO, the biggest problem openstack has is it doesn't want to admit users that aren't developers are users.... not that I'm bitter about that or anything. :/21:05
docaedohaha yeah I don't know if I'd agree it's the biggest problem, but lack of agreement on who the users are is definitely an issue21:06
kfox1111yeah. I think once magnum grows an api that allows you to launch templates on your launched docker orchestration engine, then it will be very interesting.21:06
kfox1111completely sidesteps nova's unwillingess to recongize users.21:07
kfox1111until that happens, I'm more and more worried about the app catalog use case being ignored too much.21:08
kfox1111they have restricted the user definition to operators, and why do operators need an app catalog? they can deploy stuff themselves.21:08
kfox1111its non ops folks that need the catalog the most.21:08
docaedoI agree on that point, though I think there are plenty of people that want to satisfy the needs of users. The problem is most of those conversations are happening in the working groups, and there's not enough connection back to the core project devs21:11
docaedoSo you end up with the unfortunate appearance of a project manager from company X trying to tell a dev at company Y that they need to implement feature X ASAP21:12
docaedoer, make that "feature Z" I guess, in my clumsy analogy :)21:12
docaedopoint being, devs are not swayed too much by the project managers trying to convince them what they should work on21:12
docaedowhich is really a problem of open source software more than anything - lack of a dictator prioritizing stuff is what we see21:13
docaedo(with the natural dev response being "hey, it's open source - you want feature Z, jump in and write up some code!")21:13
kfox1111heh. the thing that kills me is,21:15
kfox1111I'm happy the rigt the code and write the feature. they just won't take it because they disagree with thedefinition of 'user'.21:15
kfox1111I've written the same feature 3 times now. and its still no where close to being accepted. :/21:15
docaedooh yeah, I'm not saying that trite answer actually works, because of exactly what you've experienced21:16
docaedo"heat should have conditionals!"21:17
docaedo"OK, write it up, we're busy."21:17
docaedo"Hey, look I wrote this thing, would be great!"21:17
docaedo"Hmm...  No thanks, we don't need conditionals."21:17
docaedoI think about this all the time :) it's unfortunately the worst aspect of OpenStack sometimes, the fact that we're so open21:17
kfox1111just frustrated and blowing off steam...21:18
kfox1111I think this is the nail that may close off openstack's coffin. :/21:18
kfox1111if the docker orchestration tools gain multitenancy, and is user friendly to non ops, openstack's niche shrinks drastickly.21:18
kfox1111I think both are likely to happen.21:19
docaedoyeah .. the thing that I think is the biggest problem is that all the projects get to do things however they want, and any enforced interoperability happens through API feedback and global requirements21:20
docaedoThe way we are doing it is the only way you can build a huge community of contributors, and continue to bring in more people (because they can see nobody is going to stop them from implementing stuff)21:21
docaedobut the side effect is what we have - an absurdly difficult to install and maintain IaaS21:21
docaedovs. cloudstack (don't use it or you'll be sorry!) but .. it's unified21:21
docaedofar few options, but it's easy to install, easy to maintain, and all the parts are working happily together21:21
kfox1111the problme is lack of common definitions amungst the whole though. since nova has defined users differently then other projects, it alows them to shrug off things that would make the "users" experience better by not acknologing the user exists rather then considering them.21:22
docaedothat's true and definitely has a big impact21:22
kfox1111then to work around it, lots of other projects get spawned to fill in the gaps, making the operators life hell to deploy them all,21:24
kfox1111or the cloud developers life hell waiding through all the clouds that have random collections of services installed on them.21:24
kfox1111tosca cli for example (not trying to pick on anyone)21:25
docaedotrue enough21:26
docaedoand yeah, that's the side effect, the only way to fill a gap is generally by making your own project21:26
kfox1111exactly. :/21:27
kfox1111and that really is hurting openstack.21:27
docaedohell, even look at the conversation around freezer/ekko21:27
* kfox1111 nods21:27
docaedoat least they're playing nice21:27
docaedobut still, easier to start ekko probably rather than try to convince whole freezer team to accept all the new stuff/ideas21:27
docaedohah now you're gettin' me all worked up! :D21:28
docaedowell, I'm sure we'll solve all the things in Austin, so we have that to look forward to ;)21:28
kfox1111heh yeah. ;)21:28
kfox1111I think if we can get much closer integration into magnum,21:28
kfox1111and push hard to get the horizon plugin into all of the distro's,21:28
kfox1111we might be able to bring the others around.21:29
kfox1111so that might be my goal for this summit.21:29
kfox1111(if I'm going. still not sure yet)21:29
docaedoah that would be a bummer if you couldn't make Austin21:29
kfox1111I think I can make it, but not sure until tickets are in hand.21:30
kfox1111the aproval process around here can be quite long.21:30
docaedoI'll keep my fingers crossed just the same, would be a huge help to the catalog to have you there in Austin :)21:31
kfox1111yeah. I'll try my best.21:34
*** kzaitsev_mb has joined #openstack-app-catalog21:35
kfox1111there's also something else we can probably do...21:42
kfox1111we talked about pulling in docker like asset types.21:42
kfox1111if we continued to make the heat templates I have out on more generic,21:43
*** kzaitsev_mb has quit IRC21:43
kfox1111we could have a docker heat template that just takes in a container name and launches it.21:43
kfox1111then the asset could be a link to launch the docker container in the heat template for now,21:44
kfox1111and when magnum matures, in kubernetes, or swarm, or wherever the user wants.21:44
kfox1111tosca question... the converter converts them to heat templates.21:45
kfox1111and it may customize better if you have access to the service catalog...21:45
kfox1111is the default still usable though? how much better is it?21:45
kfox1111maybe we should just run the converter on all tosca assets, store them on and give them a link to the heat template instead of the tosca one if they say they don't have the cli or they are using the horizon plugin?21:46
*** kzaitsev_mb has joined #openstack-app-catalog21:50
docaedoI would argue there's plenty of value in sharing TOSCA stuff as-is because many people will want to use that as the basis for their own specialized template21:51
openstackgerritMark Vanderwiel proposed openstack/app-catalog-ui: Lint cleanups
kfox1111sure, but the same was said of heat, or murano, and how many entries have been added so far?21:55
kfox1111open source projects tend to get developers grow from a small percentage of users.21:56
kfox1111so say 100 users, one turns into a developer.21:56
kfox1111we need to gain enough users to turn more into developers to help feed the users. :/21:56
kfox1111and with the openstack projects making it very hard for those develoeprs to write generic things, I totally get why they haven't sprouted.21:57
kfox1111I still can't make a generic https server template thats secure! :/21:57
* kfox1111 grumbles21:57
docaedohave you tried making one that uses letsencrypt?21:58
kfox1111yeah. the plumbing isn't there....21:58
kfox1111how do I get a secret to the vm securely?21:59
kfox1111can't be done in openstack today. :/21:59
kfox1111well, not in an automated way.21:59
kfox1111sshing by hand and putting the secret in place does not count.21:59
kfox1111bbiab. meeting next.21:59
docaedoI guess in this case the secret is generated automatically the first time, then automatically signed by letsencrypt - but that probably solves a different issue than what you're talking about22:00
openstackgerritMark Vanderwiel proposed openstack/app-catalog-ui: Lint cleanups
*** spzala has quit IRC22:27
*** spzala has joined #openstack-app-catalog22:27
*** spzala has quit IRC22:32
*** spzala has joined #openstack-app-catalog22:39
kfox1111last I read, lets encrypt was kind of about getting your own CA for your subdomain.22:40
kfox1111not really at a per instance level.22:40
kfox1111but I could have misunderstood.22:40
docaedono it's just about giving out free signed SSL certs, that are accepted by any up to date browser22:49
kfox1111yeah, but the handshake I thought involved delegating the subdomain to you, so you can create the ssl certs you need for your own domain.22:49
kfox1111which woudl be trusted through their chain of trust.22:49
docaedoBUT to verify, the server requesting the cert has to be reachable (as the intention is to secure all the web servers)22:49
kfox1111not strictly true. alternately, you can add some entries to dns establishing you own the dns server.22:50
docaedoI don't know about multiple sub-domains, only case I've used them for is just getting a cert for a single domain at a time22:50
kfox1111the dns server at least has to be availabile in that case.22:50
*** toddjoh__ has joined #openstack-app-catalog23:58

Generated by 2.14.0 by Marius Gedminas - find it at!