[06:54] kim0: second link on you blog goes to jono's blog rather than the list of formulae [06:58] good morning [08:02] Morning [08:02] AlanBell: thanks .. attacking === dholbach_ is now known as dholbach [12:38] mornin [12:42] morning o/ [13:41] czajkowski: around? [13:51] dholbach: did you take pics when we were at the meet and greet? [14:05] jcastro, no [14:06] kim0: found one! [14:06] http://instagr.am/p/GqT9i/ [14:06] yeah! [14:06] perfect [14:06] nice :) [14:13] http://ubuntu-news.org/2011/07/06/unity-progress-report-irish-edition/ [14:13] how's that? [14:14] I traveled over 1200 km in the past 3 days and its been raining everywhere I've been :/ [14:45] http://ec2-50-19-197-35.compute-1.amazonaws.com/ [14:45] ENSEMBLE IS AWESOME [14:46] :) [14:46] jcastro, you need to use the new theme / new WP or you look boring :-P [14:46] <3 the title of the blog [14:52] hey jono [14:52] hey dholbach [14:53] o/ [14:53] BACON [14:53] http://ec2-50-19-197-35.compute-1.amazonaws.com/ [14:53] YEAH! [14:54] jcastro: rock n roll :) [14:54] so why is that better than "sudo apt-get install wordpress" [14:54] kim0, how's the bbb formula coming on? :) [14:54] AlanBell: I don't need to buy a server! [14:54] well you do, from the video I saw you need three [14:55] AlanBell: the good part is that you need three :) [14:55] AlanBell: It's like sudo apt-ec2 install wordpress [14:55] can scale and uses best practices [14:55] one to control it one to run mysql, one to run php [14:55] apt-get install wordpress doesn't install a db [14:55] kim0: by default are these micro instances? [14:55] james_w: ensemble one does [14:55] jcastro: no small [14:55] kim0, I know, that's why it's good :-) [14:56] james_w: that was the question :) [14:56] right, but at the most basic level it has nothing to do with ec2, scaling, etc. [14:56] I think I understand what it does, I just don't get why I would want to pay by the second for three servers rather than one [14:56] It's like a unix pipe, across machines [14:57] apt-get install wordpress installs the wordpress code, you still have to install a db, point wordpress at the db and then configure wordpress [14:57] that's how awesome it is :) [14:57] plus it has great things like ec2 integration, scaling etc. [14:57] AlanBell: pretend you'll be able to do it bare metal [14:57] AlanBell, my understanding is that it will be able to run on one server soon [14:57] which it will [14:57] but for now, this gets people fired up on EC2 right away [14:57] "my school needs to try moodle", blam, 5 minutes. [14:58] no messing with actually installing it, etc. [14:58] AlanBell: well, when I talked to jamespaige, I got the feeling that wordpress is not per se the example of things that ensemble is written for. Its used for the create, use, and throw type servers [14:58] so if there was a need to scale your db to 10 slaves because of load and then scale back down, that's when you'd use ensemble [14:58] AlanBell: for example [14:58] if we had an etherpad formula ... [14:59] we could have just run it [14:59] and be done [14:59] nigelb: it could also be used for a specific wordpress site [14:59] mhall119: Of course it could, but I'm trying to say where ensemble's power is [14:59] I see the AWS console still makes me want to kill myself [14:59] Ensembles solves a different problem from say, what puppet tries to solve [15:00] jcastro: Happens to me every day. [15:00] kim0: ok now ensemble shutdown makes it all go away? [15:00] jcastro: yeah it should [15:00] don't forget the y [15:00] kim0: ok, next up I'll want to do persistance [15:00] I think I have S3 and all that stuff [15:00] jcastro, kim0, dholbach about ready? [15:00] but by default in amazon shutting things off loses all your data right? [15:00] jono, yep [15:00] jono: all set. [15:00] yeah [15:00] jcastro: yeah .. you can't save the data today [15:00] jcastro: shutdown != terminate until you set it that way [15:00] nigelb: for example, I'd like to make a loco-directory and summit ensemble formula, so anyone who wants to throw up a testing instance can do it easily [15:01] mhall119: I like that [15:01] mhall119, do we need to package it first? :) [15:01] mhall119: Though, personally, I'd rather use vagrant, have you used it? [15:01] dholbach: no, we can script in bzr commands [15:01] nigelb: nope [15:01] ah ok [15:01] It spins up custom virtualbox images which can be controlled wiht chef [15:01] don't package it [15:01] just run it out of bzr! [15:02] dholbach: mhall119 no need to package yeah [15:02] so we can spin up, test, shutdown, rinse, repeate [15:02] Its pretty awesome when you have complex things like Hbase, Mahout, etc [15:02] dholbach: I can make a loco-directory .deb, but I'm told that IS prefers deploying from branches [15:02] * dholbach nods [15:02] I have no idea when this channel went from #ubuntu-community-team to #ubuntu-server-hippies :-P [15:02] that way we can do bzr status and find files that were never checked in [15:02] nigelb: the nice thing about ensemble for summit/loco-directory is that, with your help, we could also spin up an instance of launchpad [15:03] YES. [15:03] mhall119: Ah, I talked to Lp team about that [15:03] dholbach: shouldn't you be writing a report? :p [15:03] but to each their quarter of an hour :-P [15:03] jcastro, tomorrow [15:03] dholbach: we've been trying to keep it in #ubuntu-website [15:03] they siad they could be problems, like the fact that publically accessible isntances cannot use the images [15:03] jcastro, I'm working on your graphs [15:03] jcastro, sorry, just kidding :-P [15:03] nigelb: technical or legal reasons? [15:03] mhall119, don't worry - I was kidding [15:03] mhall119: legal mostly [15:03] I like learning about this stuff [15:04] nigelb: then it should be okay if we're just using it for testing [15:04] mhall119: Talk to lifeless, he'll give emore details. [15:04] kim0: terminated is what I want right? [15:04] that means not being charged money? [15:05] mhall119: He's probably sleeping now btw [15:05] jcastro: ensemble shutdown ? then yeah [15:05] jcastro: and check the ec2 web UI has all boxes terminating [15:05] yep [15:05] then u're good [15:05] jcastro: I think you can just stop the instance, and not be charged [15:05] since you're only charged for CPU time [15:05] you'd still be charged for storage [15:06] oh, ok [15:22] jono: I got dropped [15:22] jcastro, you back online? [15:23] yeah [15:23] you need to dial me back in [15:23] jcastro, I just tried, it dropped [15:23] jcastro, you are not on skype [15:23] ok restarted [15:23] try now [15:36] jcastro: so is oneiric working well for ya ? if so I'll probably be upgrading [15:36] yeah [15:36] it's running well on 2 of my machines now [15:36] during the sprint it sucked for me [15:36] but now it's all gelling [15:36] nice .. ok then [15:42] nice answer! [15:42] I've added to it [15:42] any idea how to know what arguments it accepts? [15:43] like, how would I do "High-Memory Double Extra Large Instance" [15:43] default-instance-type: "High-Memory Double Extra Large Instance" can't be right, heh [15:45] or does it accept API name? [15:45] like m2.4xlarge [15:45] that would make more sense [15:48] jcastro: guess m2.4xlarge .. but I never really tested that :) [16:23] live booting oneiric, and my usb keyboard/mouse are dead [16:26] hah [16:26] alright my friends - I call it a day - see you all tomorrow! [16:28] AlanBell: ok so, I've thought of something to make the example better for you [16:28] for ensemble [16:28] something I would be interested in would be a set of conference formulas [16:29] so say, etherpad, mediawiki, summit, openfire (for chat), and say .... drupal. === salgado is now known as salgado-doctor [16:29] right now you'd have to set all those up, etc. [16:30] but if we had a set of formulas that set them all up, then you could just fire up your conference's "IT support in the cloud" [16:30] instead of "ok, who has a spare linode or knows someone with a colo?" [16:31] lunch bbl [16:31] I like that example better [16:44] that is getting there a bit [16:45] I just don't get the "fire up an empty mediawiki in a three tier configuration on 9 servers" use case [16:54] I think the examples need to be more real or you get this problem: http://xkcd.com/619/ [16:55] ensemble doesn't support smooth full-screen flash video decoding though [16:56] I'm not sure there's a joe-sixpack example for ensemble [16:57] or cloud computing itself, for that matter [16:57] so if I do an ensemble deployment of mediawiki, summit, openfire etc. for a conference, would I get single signon between the services? [16:57] if you fired up an instance of SSO [16:57] or used an external one [16:58] so does it configure that to all work together? [16:58] if your formulas are written to do that, yes [16:58] hmm [16:58] what happens is ensemble runs the SSO startup, which tells SSO that it's offering OpenID and a specific URL [16:59] then Summit asks ensemble where it can find an OpenID provider, and ensemble gives it the SSO url [16:59] likewise ensemble can fire up etherpad, and then when summit asks for an etherpad host, it'll get your new instance [17:00] the typical use-case is having a DB running on a separate server, and needing that info passed to your webapp [17:01] but you can do it for just about any service dependency [17:01] can you run multiple services on one server? [17:02] yes [17:02] so you say, I want a database back end, plus a blog and a content management system and a wiki, all on one box [17:02] in that case, you'd have one formula I think [17:03] but then you're not getting any advantage from using ensemble, really [17:03] and then you outgrow it and want to move the blog to a dedicated insance [17:03] and then your single formula breaks [17:03] or tell ensemble to fire up another instance of wordpress and do load balancing between them [17:04] like I said, ensemble is useful when running multiple instances that have to communicate with eachother [17:04] if you want everything on one server, you don't get much advantage [17:04] yes, I am not sure how you grow into that [17:04] ah, I see [17:05] I can't see anyone starting with a two server empty wordpress blog [17:05] yeah, there's a bit of a gap between the "small site, everything on one box" and the "medium site, separate services on seperate boxes" [17:05] AlanBell: I guess it depends on the blog [17:06] it's not for personal "all about my life" blogs, for sure [17:06] ok, a major celebrity or media service might want more from the start [17:06] but if you're starting up an OMG!Ubuntu!, you'd probably want to start out with separate webserver and db server boxes [17:07] popey: have they? [17:07] probably not, since they seem to have so much downtime [17:07] ensemble is made to let you wire up instances automatically [17:08] so you can fire up a new Apache instance and it automatically gets added to your wordpress cluster [17:11] yeah, I guess I am not quite understanding who needs to do that on a sufficiently regular basis to get the time saving [17:11] getting services to magically do SSO is a much more exciting proposition than getting them to use more instances for me [17:24] AlanBell: suppose you have a moderately sized blog, like OMG!Ubuntu! [17:24] then you post something awesome, and you get on Slashdot, Fark and CNN.com [17:24] yup [17:24] suddently the demand for your site increases 1000x [17:24] * AlanBell has been slashdotted several times, with a tiny VPS [17:25] you're single instance can't support the traffic, firing up and conofiguring a new one will take hours [17:25] turn on supercache, stare in astonishment at the figures [17:25] aye, I had mine documented, bodhi and I worked on my blog [17:25] if you had ensemble, with one command you can spin up 1, 2, 10 or 100 new instances and have them start load balancing [17:25] went from a few requests a second to over 2000 [17:25] then, when you're off slashdot's homepage, you start shutting them down [17:25] mhall119: some good ole' fashond optimizing goes a lot farther :) [17:26] no need to take a rocket launcher to it, when it could have been done with a hammer [17:26] yeah, a single machine can easily handle a slashdotting [17:26] mhall119: http://blog.bodhizazen.net/linux/optimize-wordpress-for-speed/ ← my results [17:27] Before hacking: Requests per second: 13.21 [#/sec] (mean) [17:27] After hacking: Requests per second: 1942.31 [#/sec] (mean) [17:27] interesting results paultag [17:28] I just did wp-supercache and it dramatically fixed it [17:28] AlanBell: oh, totally. I could see that rocking [17:28] didn't do any of the other stuff, I guess you didn't see so much benefit as you had already fixed the problems it fixes (db queries) [17:29] AlanBell: :) [17:30] it is odd when it happens though! [17:30] first time I was having my breakfast and wondering why thunderbird wouldn't connect to my mail server [17:30] AlanBell: when the box gets run down? Hell yeah! I got DDoS'd by some punk on freenode who was harrasing me. I did this and fixed up iptables and it *rocked* [17:31] had a look at slashdot whilst wondering what might be up with my email [17:31] AlanBell: hahahahaha, mine started lagging my irc shell out [17:31] hahahahahahaha [17:31] that's bloody brilliant [17:31] wonder if I can find the article :) [17:31] AlanBell: site:slashdot.org "url" :) [17:32] paultag: nice [17:32] however, the middle of a slashdotting isn't the righ time to start optimizing [17:33] at least, not if you have the ability to quickly fire up more load-balancing [17:33] one of the British talent shows does this [17:33] they have online voting, and dynamically scale up the number of servers during the voting period to satisfy demand [17:34] not sure if they use ensemble for that or not, but that's the demographic [17:35] I am gonna take some of your Wp optimizations though, just getting on OMG!Ubuntu! caused a noticable slowdown in my T1.micro [17:36] I also installed the S3 media plugin, so images get served from an s3 bucket instead of apache [17:38] http://politics.slashdot.org/story/07/12/04/0310208/OOXMLs-662-Resolutions that one we got a bit caught out [17:39] AlanBell: win :) [17:40] 128MB ram VPS is a little bit tight for a slashdotting, I think we went up to 512 at that point [17:41] I hate apache, it uses up so much RAM on a normal basis [17:41] I think it's time to write a webserver, and a language that can be compiled into bytecode [17:42] I bet I could have it operate well in 32 MB RAM, and under 64 max [17:42] given no stupid memory leaks by the author of the webapp [17:43] ugh, then again the HTTP spec is stupid huge, it would be a lot of work [17:58] "thanks to Bytemark for sorting it out, we now have 450MB of RAM, up from 128 this morning. It is serving up over three thousand hits per hour, about one hit per second on average, and they are complicated pages. I think I will probably install wp-cache or something, but right now it is working and I don't want to touch it!" [17:59] so that was just with enough ram, it went really fast when I installed the cache === salgado-doctor is now known as salgado === daker is now known as daker_ [18:26] jcastro: heh, just heard someone ask how to delete a wiki page, you should write a post on how to delete too! [18:27] you need to know how to do a tag on help.u.c [18:27] they just don't let people delete the page [18:27] ah [18:27] which is kind of a bummer but might as well have a team checking out deletions [18:27] yeah, that makes sense [18:27] http://askubuntu.com/questions/42503/what-team-do-i-need-to-talk-to-about-deleting-a-page-on-help-ubuntu-com [18:27] this was wiki.u.o though, and I said the delete actually is on the options menu [18:30] oh [18:30] yeah [18:30] unless the page is ACLed you should be able to delete on wiki.u.c [18:30] if you can't then someone has made the page like that on purpose [18:31] AlanBell, mhall119: you won't need a single formula for blog + db on one machine, you'll be able to have them running on one machine using the existing formulas soon. [18:31] it's just the software isn't perfect yet...what a surprise :-) [18:41] that would make it a much more compelling story [18:54] kim0: ugh crap I close that irc window with yout docs on how to contribute to ensemble docs [18:54] can you pastebin that conversation we had? [18:54] (the part with the bzr and whatnot) [18:57] jcastro: ah yeah [19:06] https://code.launchpad.net/~jorge/ensemble/docfix-instance-type [19:06] woo! [19:06] first one! [19:11] * kim0 hugs jcastro [19:11] jcastro: in the docs folder, you can 'make html' to check the result [19:13] yeah, I am unsure about changing the examples [19:13] since all the docs are written from the point of view that you're in the source tree [19:13] maybe for 11.10 when it's more ready to be used out of the distro [19:13] instead of like, a PPA, heh [19:14] kim0: also, sphinx is badass [19:15] Yeah rox hard [19:22] 18:06:50 < mhall119> but if you're starting up an OMG!Ubuntu!, you'd probably want to start out with separate webserver and db server boxes [19:22] 18:07:02 < AlanBell> popey: have they? [19:22] no [19:40] 13:08 < mhall119> probably not, since they seem to have so much downtime [19:46] I'm looking to get a printer for college. I'm thinking of going for a cheap laserjet (tired of wasting a fortune on ink). I was wondering if people had any recommendations [19:52] do you need to be able to carry it? [19:52] and do you want colour? [19:53] nhandler: yes [19:53] nhandler: you want any used laserjet 40x0 [19:54] 4000 series, they're cube shaped, you can get them used for cheap, and they're workhorses [19:54] generally getting a new laser is about the same price as refilling a used one [19:55] but you only have to do it once every few years unlike the inkjets [19:55] indeed [19:56] I got a colour laser because I didn't use the inkjet enough so it kept drying out and wasting ink cartridges [19:56] that can happen? [19:57] my inkjet finally kicked the bucket after 5 years of service. i broke it trying to clear a paper jam [20:01] AlanBell: I'd like it to be able to fit on a desk, color is not necessary [21:14] nhandler: I'm quite happy with my cheap Samsung b/w laser printer (using the open source 'splix' driver) [21:15] I have a cheap brother b/w laser (when I need color I spend $.35/sheet at the print shop down the street) [21:18] it certainly beats having to throw away expensive, 50% (or less) used dried out inkjet cartridges... [21:20] JanC, pleia2: Both of those worked well with linux? I had an old brother inkjet that flat out refused to work. I'd ideally like to have near plug-and-play support [21:21] nhandler: brother laserjets have official support for linux these days [21:21] I've had good luck with HP too [21:22] nhandler: like I said, there is a good open source driver for Samsung printers (certainly for b/w, I never tested with their colour lasers) [21:23] pleia2: official support is closed source though? (for both Samsung & Brother AFAIK, but for Samsung there is an open source driver too, which works well) [21:24] JanC: I don't know, I just plugged it in and it worked :) [21:24] well, the drivers are in multiverse... [21:25] unless some brother printers have open source drivers included with CUPS too === salgado is now known as salgado-afk