=== wedgwood_away is now known as wedgwood === wedgwood is now known as wedgwood_away === udsbotu changed the topic of #ubuntu-uds-servercloud-2 to: Track: Cloud & Server | WebScale Packaging Review | Url: http://summit.ubuntu.com/uds-1303/meeting/21658/webscale-packaging-review/ [13:59] jcastro, ping? [14:00] starting shortly [14:00] Daviey, great! [14:01] mattyw: pong! [14:01] oh, sorry, that machine is rebooting [14:01] on my way! [14:01] anyone else joining? [14:01] Video "the live event will begin in a few minutes" :) [14:01] (the fishbowl) [14:01] nealmcb: just trying to get everyone seated [14:02] :) [14:02] Hello [14:02] we are getting started with WebScale Packaging Review session [14:03] jcastro, you're lightning talk about go juju later - are you going to talk about why it's been re written in go? [14:03] any folks here for that session [14:03] arosales, so whos running the hangout? [14:03] arosales, yes [14:03] jamespage: me [14:03] hi [14:03] arosales, URL? [14:03] https://plus.google.com/hangouts/_/f940f0bbd48c29bcb1f6f6d4646ea5dbcef5ed6d?authuser=0&hl=en [14:03] QUESTION: all three seem to be in the archive today, is this session about wrapping them in juju or something else? [14:04] oops, wrong channel :) [14:05] jamespage: will you be able to join? [14:05] arosales, yeah - getting and error joining - just trying with chrome [14:05] jamespage: ok, thanks. [14:07] arosales, OK _ so I can't join the hangout for some reason [14:07] I'll tune in and participate via IRC [14:07] great set of etherpad notes to start the session! [14:08] Video still not showing - is anyone talking yet? [14:08] o/ [14:08] nealmcb, yes, the video is working for me [14:08] arosales: I'm on it [14:09] still says "this live event will begin in a few moments". I'll try reloading again [14:09] nealmcb: reload the page [14:10] But please do note in irc each time the video goes live for a session [14:10] ok, got it [14:11] arosales: I poked mongodb enough last night to get 2.2.3 into raring [14:11] sans SSL support [14:11] jamespage: ah great to hear [14:11] IMHO nodejs itself moves far to fast to be in main [14:12] utlemming: ^^ [14:12] yeah and thats OK until we get 10x [14:13] its the support that would be a killer - esp security I think [14:13] I would actually prefer to see a series align set of PPA's for nodejs [14:13] 6.x 8.x 10.x etc. etc. [14:13] We are using Node for development, and the PPA version is sufficient, but not the released ones in distro. [14:14] for openstack what we have in distro 6.x is sufficient [14:14] Other deps, we get from npm [14:14] BUT we don't use it for openstack-dashboard runtime - just to generate the less assets statically [14:15] RE PPA - there is a maintained PPA already [14:15] https://launchpad.net/~chris-lea/+archive/node.js [14:17] maybe I'll write a less compressor in python and remove the dependency for openstack [14:17] :-) [14:18] mattyw: it'll mostly be an overview on feature parity. We haven't done a thorough benchmark quite yet. [14:18] reliance on specific node.js versions sounds like the messy world of java dependencies - ouch... [14:18] jcastro, ok thanks [14:18] I have tested the SSL support [14:18] A version of nodejs in backports would be better I think than PPA, or supply different versions for the user to decide. [14:18] had some issues on armhf [14:19] utlemming: agreed [14:19] FWIW, it's not clear that we cannot include it [14:19] GPL code linking to OpenSSL without exception [14:19] THere is a specific exception about distribution software with a distribution, IIRC [14:19] niemeyer: you are correct - we are following the Debian stance right now [14:20] We should talk to a lawyer [14:20] niemeyer: Ubuntu can elect todo something different [14:20] niemeyer: upstream don't believe there is an issue [14:21] jamespage-not-in: If we can legally ship it, it sounds like a big service we'd be doing to people [14:21] niemeyer: agreed [14:21] Making the ppa endorsed as "official" would probably have the same effect [14:22] Since we'd be matching precisely the exception rule [14:23] Recompiling wont work when you run "juju add-unit -n 10" [14:23] You end up with manual work or different version in different machines [14:23] just around mongodb [14:23] mariusko_: that's was my point about repeatability. [14:23] 2.4 is around the block [14:23] Do we have plans for including it? [14:23] arm support is currently a distro patch (again) [14:24] niemeyer: yes [14:24] jamespage-not-in: Super, thanks [14:24] japp, about nodejs [14:25] utlemming: something with arm support would be good [14:25] chris's PPA does not have that [14:25] Regarding the mongo&SSL, isn't upstream breaking license too? [14:25] mariusko_: I don't think so [14:25] A PPA is a PPA.. blessing it or not doesn't make a lot of difference in practice it seems [14:26] There will still be a package in main, that will be out of date [14:26] Upstream cannot ship it [14:26] We can [14:27] (apparently) [14:27] We should ship it if that's true [14:27] niemeyer: mainly I think folks are looking for a way to have confidence in a package and get the latest in terms of node.js. [14:28] utlemming: kind of [14:28] utlemming: yes [14:28] +1 [14:28] elmo: do you have a different take on this? [14:29] the distinction is the wording of 'major essential component' [14:29] and whether or not openssl counts as such for Ubuntu [14:29] I'm not qualified to say whether or not it does [14:29] but it's not about whether or not we actually ship libssl [14:29] Debian does too [14:29] it's about whether we/lawyers are happy for that to be considered a major essential component [14:29] +1 [14:30] More information from the Gnome Folks: http://people.gnome.org/~markmc/openssl-and-the-gpl.html [14:30] anyway, this is details, and I doubt we can resolve it here [14:30] Can I help with the task of "Identify specific advantages of nginx over apache"? [14:30] OK - lets move on from mongodb [14:31] I've requested a OpenSSL license exception from upstream to resolve this cleanly [14:31] None here :) [14:33] no disagreements, that was [14:34] What about JBoss? [14:34] utlemming: would you like to pickup the work items for nginx? [14:34] mariusko_: hmm - massive amout of work to package [14:35] some parts of jboss is in universe from Debian. aiui, there is a lot that is not there [14:35] utlemming: I don't think so [14:35] oh, I am wrong then. :) [14:35] they are extremely out of date [14:35] does nginx have a publicly visibile VCS yet? (supportability issue) [14:35] utlemming: its not [14:35] utlemming: ^ [14:35] arosales: only some parts [14:35] sbeattie: yes, it does [14:35] you can't run jboss from the archive [14:35] howdy [14:35] mdeslaur: ah okay [14:35] JBoss version in Ubuntu is outdated (4.2.x) [14:35] FPI ended so others are joining [14:36] some of its components are present - but just to support other java packages [14:36] the effort to package it fully is not worth the effort [14:36] its better charmed [14:36] it has regular security updates from RedHat [14:36] Current release is 7.1 [14:36] it would be non-trivial to support [14:37] yes, it does, I checked [14:37] I haven't looked [14:37] utlemming: I was talking about jboss there, not nginx [14:37] I would completely support jdstrand on this [14:37] I have opinions on nginx, but they are probably out of date [14:37] (the lag between IRC and the hangout is awesome) [14:38] http://trac.nginx.org/nginx/browser [14:38] code: svn://svn.nginx.org/nginx [14:38] sbeattie: ^ [14:38] http://wiki.debian.org/JBossPackaging [14:38] arosales: thanks [14:39] I looked at jboss 2 years ago and ruled it out as a possibility to have effective packaging [14:39] mariusko, look at MIR for jboss? [14:39] MIR? [14:39] Main INclusion Request in ubuntu [14:39] arosales: thats not the right question [14:39] he may not be familiar [14:40] the question is - who wants to package Jboss [14:40] I'm not sure right now. Most users install it manually I think, but for Juju it would be important to have it included. [14:40] Not I. [14:40] * jamespage-not-in stands back [14:40] jboss is java right? AFAIK jamespage packages all the Java stuff for Ubuntu [14:40] :) [14:40] mariusko, for Juju, you don't have to have it packaged. [14:40] haha [14:40] elmo: I've already expressed my position on this [14:40] that's why he's using a hidden nick today [14:41] from my limited experience with jboss, it's not a simple stack [14:41] Sure, you could have juju script that fetch it from upstream somehow [14:41] I'm not working on this full time for the next 12 months to get together packages which are not maintainable [14:41] Maybe RedHat would be interested in contributing to getting JBoss in. [14:41] I don't think "packaging jboss" is the issue. To do real jboss support we'd need like 2 more jamespages. [14:41] bwahahahaha. [14:42] i missed varnish discussion, but i'd be interested in helping with that [14:43] this is far more effectively dealt with using Juju and the release tarball from upstream [14:43] lets talk about varnish [14:43] is ivoks here? hmm - no [14:43] ah, i guess i didn't miss it :) [14:43] jamespage-not-in, the "effectively dealt" is w/r/t jboss, right? [14:43] med_: yes [14:43] jamespage-not-in: Are you suggesting a pool of upstream carried jar's is hard to work with, in packaging ? [14:43] I know the varnish packager guy [14:44] Daviey: don't make me cry [14:44] Daviey: OK so it could be packaged (binary style) for Canonical partner [14:44] varnish has a security history (3 CVEs) [14:44] Their webpage recommends to not use the distro packages btw: [14:45] arosales: no, but it has a security history (3 CVEs) [14:45] https://www.varnish-cache.org/installation/ubuntu [14:45] heh [14:45] so if upstream doesn't like the packages [14:45] makes it much harder [14:45] "Varnish is distributed in the Ubuntu package repositories, but the version there might be out of date, and we generally recommend using the packages provided by varnish-cache.org." [14:45] they say they only support LTEs [14:46] erm, LTSs [14:46] I think if we want to look at this we need to reach out to upstream to discuss [14:46] (Full disclosure): I have a work item to convince them to just do that in the charm. [14:46] mithrandir works for the company behind varnish, last I heard [14:46] arosales: this attitude from upstream gives me some pause. it suggests they don't support older versions, only the newest [14:46] looks like they just want to be able to keep it up to date out of band [14:46] elmo: still does last I checked [14:46] Juju charm does install it: http://jujucharms.com/charms/precise/varnish/hooks/install [14:46] arosales: which becomes difficult at the tail end of LTS [14:46] The distro version [14:47] also on that page: Please be advised that we only provide packages for Ubuntu's LTS releases, not all the intermediate releases. Note that the packages might still work fine on newer releases. [14:47] utlemming: right, but the idea was to get them to add a feature to the charm to use their version [14:47] i think the problem is that the version we have in quantal was released in 2011 [14:48] they don't have that much new versions [14:48] I'm less concerned about their packaging. a consideration is how the handle their own releases. that could be fleshed out in the MIR [14:48] s/much/many [14:48] either way, whoever we get to do the packages should probably just start by talking to upstream first to see what their concerns are about distro packages [14:48] aka "How can we ship Varnish that would meet your requirements?" [14:48] if there isn't a lot of code churn, upstream's attitude on their support becomes less of a concern (which I think ivoks was hinting at) [14:49] which is much better than their download page saying "don't use ubuntu's packages" [14:49] again, that is something that could be discussed in the MIR process [14:49] * jcastro nods [14:50] just as a side note we need to be better at evaluating whether a Minor Release Exception is appropriate for packages we MIR [14:50] note Debian also has varnish [14:50] http://packages.debian.org/squeeze/varnish [14:50] this might be appropriate for varnish [14:50] which might help oil things with upstream [14:50] "Varnish is distributed in the Debian package repositories, but the version there might be out of date, and we generally recommend using the packages provided by varnish-cache.org or packages from backports.debian.org." [14:50] heh, same issue [14:52] we should also leverage backports more in Ubuntu as well for this stuff [14:52] jamespage: +1 [14:52] arosales: yep, assign it away! [14:52] I intend to request backports of mongodb for precise [14:52] and quantal [14:52] and if they don't "in" the hangout they can't defend themselves. [14:53] go for it. === wedgwood_away is now known as wedgwood [14:54] Btw, the mongodb package is outdated yes, for quantal and precise [14:54] I question how exciting backports are [14:54] Was just updated in Raring. [14:54] They are as hard to use as a PPA, and have roughly equal distro support [14:55] * jdstrand has to go === udsbotu changed the topic of #ubuntu-uds-servercloud-2 to: Track: Cloud & Server | Juju Application Stacks | Url: http://summit.ubuntu.com/uds-1303/meeting/21643/servercloud-1303-juju-app-stacks/ [14:56] ta [14:56] session ended [14:56] ttfn [14:56] thanks for everyone's time. [14:57] we will never have 100% happy upstream since, by nature, we are moving slower than they are; if we would like to make everybody happy, we would have to drop LTS and move to always rolling release [14:57] and then no one would use the distro :) [15:00] Juju Application Stacks getting kicked off. [15:04] * marcoceppi raise? [15:10] * niemeyer wonders if jcastro moved back to CUba [15:10] Yay [15:10] \o/ [15:11] niemeyer: wrong flag! I am offended! [15:15] deploy all the node apps [15:17] hi [15:19] It's too early for that [15:19] arosales: worth trying I think. Good idea! But note that people following the stream have no hope of seeing anything in there unless you select it. [15:21] elmo: are you in this session? [15:21] Probably more useful when other charms want a memcached relation? [15:21] It's a key player, but it's not a large key player [15:21] jcastro: I just joined [15:21] jcastro: which charm sorry? [15:21] rbasak: right, if we focus on the IRC I will give it focus [15:21] arosales: makes sense! [15:21] jcastro: checking prodstack now, but I don't believe we're using it yet [15:22] jcastro: I think it's part of a lego stack [15:22] it doesn't feel like a standalone full 'stack' like e.g. rails would [15:23] jcastro: ah, so summit was using it [15:23] jcastro: so it's at least functional [15:23] and uccs.ubuntu.com (which is not yet live in prodstack, but is deployed) is using it [15:23] It's definitely functional, been using it myself for quite a bit [15:23] arosales: The IRC window works.. [15:24] I don't think any one service has a hard requirement for memcached, a lot use it in a supplementary nature [15:24] arosales: so, in theory, if your service can't function without a cache, your service is broken [15:24] arosales: or so lifeless tells me [15:24] elmo: +1 [15:25] arosales: I think it can be made on demand [15:25] elmo, marco-summit: Sure, but needing data in a cache != needing a cache service [15:26] niemeyer: sure, I was partially being flip [15:26] I think it's not a high-priority either way [15:26] niemeyer: I agree, some services are going to expect memcache to be available, but we've met that bar [15:26] jcastro: it's the remote login service that modern unity can use [15:27] ack [15:27] arosales: +1 for IRC idea. I think it might be particularly useful for those watching the recording later, too. [15:27] rbasak: yeah, I can see that being really helpful. Only downside is there's a bit of lag between what we hear and what they say [15:28] Yeah the lag is annoying but more fundamental to any live interaction with the hangout I think. IRC included. [15:29] jcastro: nope [15:29] some people use redis as a persistent store [15:29] which makes it slightly different case to memcache [15:29] would anyone here be interested in a neo4j charm? [15:29] jcastro: yeah, that's my reaction ;-) [15:29] I've used redis in production, it works along the lines of memcached. Simple to set up [15:29] mattyw: YES! give me a second to finish out this list [15:29] then we can move on to new work! [15:30] cham works as expected [15:30] jcastro: yep [15:34] jcastro, ummm [15:34] mattyw: :) [15:35] arosales, I've only ever played around with it [15:35] It's a graph database [15:35] arosales, but there's a company behind it, maybe they'd be interested in helping out? [15:35] Somewhat of a unique one [15:36] niemeyer, have you used it? I wrote a toy application to work my way around the london underground, but that was about it [15:36] mattyw: No.. I just follow along and watch from the distance [15:37] http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=685615 [15:37] Debian bug 685615 in wnpp "RFP: neo4j-community -- Neo4j is a high-performance, NOSQL graph database with all the features of a mature and robust databas" [Wishlist,Open] [15:37] It's a different class of database than any of the listed ones really [15:37] It's not about whether it's a webapp or not.. more about what's the use case [15:39] It's very useful for graph-related problems [15:39] Network of friends, etc [15:39] The real question in the context being debated, though, is how many people have we found that have an app using it [15:40] I could give it a go [15:40] I'm no expert though, but I'm happy to take a look [15:42] I have some real life experience with NodeJS charm, and I believe it has issues common with other user app charms, to get user code deployed [15:42] Huh? [15:42] See https://bugs.launchpad.net/charms/+source/node-app/+bugs?field.status:list=NEW [15:44] jcastro: The *data* is, sure.. [15:44] ack === udsbotu changed the topic of #ubuntu-uds-servercloud-2 to: Currently no events are active in this room - http://summit.ubuntu.com/uds-1303/servercloud-2/ - http://ubottu.com/uds-logs/%23ubuntu-uds-servercloud-2.log [15:45] cheers [15:46] Session ended [15:47] Thanks for everyone's time [15:49] Hello world! === udsbotu changed the topic of #ubuntu-uds-servercloud-2 to: Track: Cloud & Server | Improve Juju Documentation | Url: http://summit.ubuntu.com/uds-1303/meeting/21692/servercloud-1303-juju-docs/ [15:56] philipballew: hello [15:58] Getting started with http://summit.ubuntu.com/uds-1303/meeting/21692/servercloud-1303-juju-docs/ [15:59] arosales, I hope someone is going to invite me to the hangout :) [16:00] you should see the hangout URL when you load http://summit.ubuntu.com/uds-1303/meeting/21692/servercloud-1303-juju-docs/ [16:00] evilnickveitch: https://plus.google.com/hangouts/_/f87b2b24c412bee262b5e31d2c0633f24a1d9831?authuser=0&hl=en [16:03] roll call! \o [16:03] * marco-summit o/ [16:03] hi there [16:03] I'm here again [16:03] * philipballew waves [16:04] waves [16:06] https://stripe.com/docs btw is what I mean [16:07] as an example of awesome docs [16:08] melmoth, mattyw, mreed , hello :-) [16:08] marco-summit, hello :-) [16:16] markdown. [16:17] * marcoceppi is not biased [16:24] Most of all the javascript, if any, should probably be at the template level anyways [16:24] So contributors won't have to mess with it [16:25] jcastro.md [16:26] Everything should be in LaTeX :P [16:36] +1 definitely annotate the anatomy of a charm. There really isn't a good place for this [16:36] out there currently [16:40] evilnickveitch: I tried my hand at this a while ago: [16:40] https://docs.google.com/document/d/1p--PQprZI8-rGr68M_zW-4rdftaIu6-h6n_UGtfk-Ws/edit [16:40] As sort of "the docs really suck at being organized" [16:40] marco-summit, thanks ! [16:42] I'd be interested in helping contribute more up to date documentation when time permits [16:42] marco-summit, okay, thats cool, we should keep in touch [16:43] it would be good to know when the docs get updated [16:43] somehow???? [16:43] cool [16:43] cheers! [16:44] cheers! === udsbotu changed the topic of #ubuntu-uds-servercloud-2 to: Currently no events are active in this room - http://summit.ubuntu.com/uds-1303/servercloud-2/ - http://ubottu.com/uds-logs/%23ubuntu-uds-servercloud-2.log [17:05] waves