[05:23] mooning [05:45] Good morning [14:39] Morning all [14:53] hi, sbalneav [14:57] Morning HedgeMage [16:17] !info ltsp-docs lucid [16:17] ltsp-docs (source: ltsp-docs): LTSP Documentation. In component universe, is extra. Version 0.99+bzr91-1 (lucid), package size 365 kB, installed size 800 kB [17:16] anybody using ubuntu cloud computing? [17:29] Ahmuck: Nope. Can't. [17:30] sbalneav: can't in edubuntu? [17:31] No, I can't no matter what I run. [17:31] I work in the Legal industry. [17:32] sbalneav: what does that have to do with cloud computing? [17:32] I get called to court by the judge to give a chain-of-custody testimony on some electronic evidence. Judge says: "So, who had access to this data?" [17:32] *snark* [17:33] "Potentially anyone in the cloud" isn't the correct answer. [17:33] that is the problem with the cloud [17:33] no privacy [17:33] i know big corps are pushing "internet only" apps, and cloud computing, because they want control of "all your data", but i can't see this working in the long run [17:34] my situatuion is that i was looking at teaching a class using hugin [17:34] and wondered if there was a way to offload the processes and stitching of hugin into the cloud [17:34] private clouds [17:34] are no different than private servers [17:38] Cloud computing, IMHO, is the biggest freaking scam ever. It's nothing being marketed as something. [17:38] it's something [17:38] it's just not what it's being hyped as being [17:38] Yeah? What is it. [17:38] Tell me. [17:38] it's flexible deployment [17:38] and allocation [17:38] That's another buzzword [17:39] it's being able to double your available resources in short order [17:39] Servers can be flexibly deployed. And allocated. [17:39] How? [17:39] and back it down again when it's not needed [17:39] How? [17:39] sbalneav: yes, they can be [17:39] and when there is a software product that does this for you, they call it cloud [17:39] What software product? [17:39] apt-get install ? [17:40] ecalyptus [17:40] eucalyptus [17:42] Which requires that I re-write entire apps to handle this Cloud thing. [17:42] which is distributed computing [17:42] which I studied in university 20 PLUS YEARS AGO. [17:43] like I say, it's nothing that hasn't been around since forever. [17:43] and it's useless to ANYONE who actually cares about their data. [17:43] sbalneav: not quite distributed computing [17:43] a cloud doesn't appear as one big instance [17:43] it's more like dynamically-managed clustering [17:44] Which has been around forever as well. [17:44] it's just the latest buzzword for technologies that have been around for ages. [17:44] i.e., it's nothing. [17:45] Anyway, that's how I see it. YMMV :) [17:45] mhall119|work: so, would the software cloud do something like hugin? [17:45] ah, apps have to be re-written? [17:45] so in reality, there are prolly few apps that could even leverage the cloud environment? [17:45] Well, when it reaches the masses as "boot your device and you have everything without installing anything", it'll be something :) [17:46] Ahmuck: I'm not familliar with hugin [17:46] You could even have "pay per use" programs there. No more extreme licensing fees for something that you use once in a while [17:46] mhall119|work: it's a program to paste multiple pictures together into a panorama. [17:46] sbalneav: it's not a new idea, no, it's just something that is now going mainstream [17:47] Ahmuck: in a cloud, you get an OS on each "node", your apps just have to be able to work in a clustered setup [17:47] sbalneav: ah, ok [17:47] Ahmuck: think of Apache clustering [17:47] which has been around for a while [17:48] all that cloud computing lets you do is add and remove nodes from the cluster without having to reconfigure or restart anything [17:49] And management would be done centrally, so users won't have to install stuff, protect themselves from antivirus, backup etc etc [17:49] It's a nice idea, I hope it's implemented and marketed properly... [17:49] it's new in it's implementation, not it's design [17:50] Well selling apps in a different way is a different design [17:50] for example, you can buy a cluster of 5 virtual machines from Amazon to run hugin, and then when it get's featured on Slashdot or Fark, you can tell Amazon to give you 20 or 50 or 100 instances instead [17:50] until the demand dies back down, then you tell Amazon to go back to 5 [17:52] Actually cloud has a lot in common with LTSP ;) === Ack-Jr is now known as Ahmuck [17:52] (cluster) [17:52] alkisg: how so? [17:53] People having dumb terminals and connecting to applications servers [17:53] well, that's what i thought [17:54] any chance of offloading office, etc. to the cloud ? [17:54] or anything for that matter, using ltsp server as a passthrough? [17:54] ie, while keeping ltsp as a server for other apps? [17:54] Ahmuck: think of a cloud as a vague collection of individual computers [17:55] they don't act like a single computer, like distributed computing [17:55] agreed [17:55] but the actual number and location of the computers isn't important to you [17:55] Well I haven't read of any marketing plans, but I guess the future is in server based computing. So you'll have some dumb terminal at home and every app you want will be provided by some app server. Sure, LTSP can do that, and I hope it does... [17:56] that might be a problem for privacy [17:56] though quickbooks now uses google to do the searcing inside of it [17:56] Not if proper encryption is used [17:56] which i consder monsterarilsy dangerous [17:56] Why? [17:56] big brother [17:56] With encryption? [17:57] If your data is uploaded in an encrypted disk on the cloud, noone could read it without your key [17:57] But, if the company is untrusted, of course they could fake the encryption :D [17:57] well, my overactive imagination and paranoia kicks in here. in that i don't believe anything is really encrypted [17:57] only believed to be encrypted [17:58] Right. Someone should check in depth all of those web space providers... [17:58] the biggest point about cloud computing that is often missed is that your cloud doesn't have to be hosted by a 3rd party [17:58] anywho, i know the plan for computing for the last 10 years has been to move users to "cloud" and web based apps. i did run across that paper somewhere [18:01] mhall119|work: light bulb just went on [18:01] in other words, one could build a cluster in house to offload to the server [18:02] what do you mean "to offload to the server"? [18:02] ie, what's happening then is ltsp server offloads it's load to the cluster in house [18:02] offload from the server [18:02] "in house" [18:02] uh, not quite [18:02] sorry, i've been irc speaking and im speaking for too long, as a result it breaks my normal conversation up [18:03] no? [18:03] can you cluster LTSP servers? [18:03] or is there always just one? [18:03] ltsp client --> resource hungry app --> ltsp server --> server load --> internal cloud (cluter) [18:04] er, cluster [18:04] iirc, i think you can cluster now, no? [18:04] for me it's just one [18:04] if you can cluster LTSP, then here is how you would use the cloud: [18:04] You have a 20 seat lab [18:04] so you have enough LTSP server instances to manage 20 seats [18:05] suddenly they need 100 seats for a special week long event [18:05] so you expand your cloud with enough LTSP servers to handle 100 seats [18:05] at the end of the week, you shrink your cloud back down again [18:05] k, that makes sence [18:05] by what? Returning the servers to the vendor? :) [18:05] sbalneav: no [18:06] by making those servers available to something else [18:06] i was thinking more along the lines, i have a 20 seat lab which is used for OO.o, firefox, edu apps. [18:06] then i want to teach class2 of hugin, the part where we take each user's set of photos for a gigapan, ie, 180 photos, [18:06] here's where cloud computing is going: [18:06] upload them to the server, [18:07] say you have an idea for a cool new webapp, like hugin [18:07] and then start the control point finding and stitching [18:07] you get some startup funding, and buy yourself a couple of heft servers [18:07] this requires huge memory and processing resources [18:07] you can handle 1000 sessions a day [18:07] suddenly you're the hot new thing, and you're getting 1,000,000 sessions a day [18:08] now, you have a private cloud on servers you manage [18:08] so, offloading this to the another "cluster" to do the work, which can be done with text .pto files would work [18:08] but, in the near future, you can grown that cloud onto Amazon's servers too [18:08] and wouldn't affect the other 10 clients in the other classroom [18:08] Ahmuck: I don't think any cloud offerings currently do that for you [18:09] you'd have to manage the offloading yourself [18:13] mhall119|work: actually your explanation helped out a lot [18:14] Elasticity [18:14] Applications can dynamically use more resources within the cloud when required ensuring users needs are met immediately. [18:14] right [18:14] Bursting [18:14] Overloaded applications running on your private cloud can expand to use resources from the public cloud. [18:14] this was from ubuntu's cloud page [18:14] that's not available yet, last I heard [18:14] which made me wonder if it was possible [18:14] but is coming [18:15] that's what i was looking for [18:15] so long as, and this is key here, your application is written using a very specific set of libraries. [18:15] again, though, it doesn't offload them [18:15] it just adds someone else's servers to your cloud [18:15] try putting 10 students into a room doing a monthly journal, etc. and let them upload photos on each page [18:15] it comes to a grinding halt [18:15] sbalneav: you can run a generic Ubuntu Server instance in the cloud [18:16] the only specialization your software needs is the ability to manage a cluster of itself [18:16] using scribus [18:17] hrm, could that mean that a cluster managment plugin could be written that was universel enough that it could be dropped in applications without much trouble? [18:17] mhall119|work: So I can create this cloud, and users log into their ltsp server, click on openoffice, and of the 30 machines in the cloud, each of them is only running 1-2 openoffice sessions? [18:17] Ahmuck: if LTSP cna be clustered, and you have some available resources, then cloud computing would let you make a bigger cluster during that class [18:17] sbalneav: I don't know about LTSP specifically, how it does server clustering [18:18] Ahmuck: not really, no [18:18] clustering is pretty implementation-specific [18:18] By having multiple ltsp servers, and round-robining entire sessions amongst all the boxes. [18:18] but that's clustered LTSP servers [18:18] i recall attempting a bewolf cluster with some 500mhz machines way back in the day [18:18] sbalneav: round robin through DNS? [18:19] we're talking about a cloud that I *just* want for one application. Openoffice. [18:19] All my other apps run fine on one box. [18:19] cloud instances run an OS, not individual applications [18:19] clustering software iirc [18:19] So they're virtual servers, then. [18:20] No different from having a KVM or vmware pool, like I have today. [18:27] Doesn't cloud also contain software as a service, not only platform as a service? [18:27] sbalneav: correct [18:28] in fact, Eucalyptus uses KVM instances [18:28] all cloud does is give you management tools on top of them [18:30] mhall119|work: thank you. Cloud == all the stuff I already have + nice frontend. :) [18:30] sbalneav: yes [18:30] And that's my point. Cloud's nothing but all the stuff we already have, just a shiny new management interface. [18:31] Well, the web is also just apps & data, which we already had :) [18:31] Oh, and people putting up public servers for nothing, so long as you don't mind/don't have legislation against storing customer/student/your data on publicly available machines. [18:31] here's my problem with the whole thing: [18:32] it's being sold to people as a panacea that will solve all their problems. Guys like Ahmuck. [18:32] :p [18:32] No money? CLOUD No resources? CLOUD better expandability? CLOUD!!!!!! [18:32] heh, the less time i have to figure is the more time i have to take panographs :) [18:33] then, when you actually figure it out, it's the same thing we've had all along. [18:34] I'd bet $20 Ahmuck's got legislation covering student data that will prevent him from migrating ONE CPU CYCLE off of stuff he controls into the public cloud. [18:34] so, he's forced to build his own cloud. [18:34] but "his own cloud" doesn't (really) solve any problems for him. [18:34] he's still gotta buy machines [18:34] still gotta set 'em up [18:35] still gotta fix 'em when they break. [18:35] sbalneav: look after our retirement, some decades later. Dumb terminals. No administration at all. Power on. Select application or site. Power off. No viruses to worry about, no data loss to worry about, no formats etc. It'll happen. [18:36] THAT's what I object to. The marketing that "Cloud computing will revolutionize everything" when cloud computing ISN'T anything revolutionary. [18:36] alkisg: No it wont. [18:36] It isn't. It's just a product, not something for scientific research... [18:36] Who's going to pay for all those boxes? [18:37] And who's going to guarentee your data's integrity? [18:37] Companies will emerge which will only have the hardware [18:37] backups? [18:37] Other companies will rent hardware and run software on them [18:37] Who's backing up your gmail? [18:37] Don't have gmail [18:38] OK, but you do understand what I mean... :) [18:38] Sure. [18:38] till you lose your gmail. [18:38] sbalneav: i was looking for this: overloaded applications running on your private cloud can expand to use resources from the public cloud, Applications can dynamically use more resources within the cloud when required ensuring users needs are met immediately. [18:38] just like those cell phone customers at dangerous. [18:38] those two things [18:38] If I'm paying, I won't lose it [18:38] alkisg: HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA [18:39] Customers always pay. They'll pay less for hardware, and more for software... [18:39] just like those cell phone customers were paying? [18:39] You're funny. [18:39] I don't know the story... [18:39] But why, do you lose your back account data? [18:39] I wouldn't deposit money on banks then... [18:40] Right, but the bank HAS to have my data, I'm not expected to. [18:40] Who's responsible for the "cloud"? [18:40] ur taxes [18:40] My taxes? [18:40] Who's responsible for the internet? [18:40] My taxes in canada are going to yahoo's cloud servers? [18:40] really? [18:41] actually there is a good case here [18:41] UPS has electronic billing [18:41] and if you don't pick up your invoice within a specified time period, they won't let you have it [18:41] which is bunk [18:41] and Ahmuck, like I say, I bet there's NO WAY from a legislation point of view you're ALLOWED to put student information on a cloud device, for Student Privacy concerns. [18:41] sbalneav: in some part, international trade agreements [18:42] personal private information [18:42] ie, personal details [18:43] but i doubt it would cover the use of the cloud to do proc and mem intensive things like control point finding and stitching [18:43] sorry, catching up [18:43] So if the kid wants to make a hugin wallpaper of himself and his family at the field, and some pedo manages to get ahold of that, and sees what the kid looks like, and snatches him.her... [18:44] there are so many other ways to get photos. kids are uploading photos of themselves without parent's consent all the time [18:44] and you're up in front of the judge/your district IT manager, and he/she says "So what did you do to ensure the privacy opf this kids data" [18:44] you say.... What? [18:44] that's why we have a sheet that says you can do this or that [18:44] if they go outside those rules, the responsiblity falls back on them [18:44] same as if they decided to throw a rock in a window [18:45] ok, but there's a difference between the kid choosing to upload to facebook, and you, as the responsible admin who's supposed to know better, providing school services which FORCE the kid to upload information to public servers. [18:45] whose going to gain access to that public server? [18:45] encryption iirc, was mentioned [18:46] ANYONE ON THE INTERNET [18:46] Orly? [18:46] what encryption? [18:46] Do you know it's safe? [18:46] how good is it? [18:46] How do we know that our bank won't steal our money from us? [18:47] Can you stand up, in a court of law, and say YOU KNOW BEYOND A SHADOW OF A DOUBT that NO ONE had access to the data? [18:47] no, but i doubt you could do that with private servers [18:47] alkisg: because we get printed statements, and we're supposed to keep track of our finances. [18:47] heh heh, sbalneav the banks did steal money from us. via a 700 billion dollar bailout [18:48] then they took the homes back [18:48] Ahmuck: You can if they're not connected to the internet. [18:48] then they bulldozed some of them [18:48] So if they print that you withdrew 1000$ with your card, how are you going to prove that you didn't? [18:48] sbalneav: no, some really bright kid gets past security, and you'll find it in the www. it happens all the time [18:49] there is no such thing a security, only the veil of it [18:49] * sbalneav shrugs [18:49] Hey, you guys do what you want, not my problem, or my job. [18:49] But take it from someone who's BEEN in front of judges for chain of custody matters. [18:50] i don't disagree, i prefer private information on private servers, however, i don't have an issue with offloading mathematical caculations to a server cluster [18:50] But I'd suggest checking with your relevant legislation and/or IT executive committee before "clouding" your it operation. [18:51] And possibly your personal lawyer :) [18:51] sbalneav: if your point is that putting information on servers you don't control raises privacy and secrecy concerns, then yes, you're right [18:51] Right [18:51] if your point is that clouds are more dangerous than existing server setups, I disagree [18:52] Never said that. [18:52] then an internal, private cloud is no more dangerous than an internal, private server [18:52] Correct. [18:52] But then, you don't need a cloud, if you're going to do it all internally [18:52] and your Flickr pictures are just as insecure if they use a standard cluster as they would be if they used a cloud [18:52] sbalneav: clouds can still be useful internally [18:53] I disagree. But I've had enough. I lose, you all win. :) [18:53] no! [18:54] I have analogies! [18:54] to cars! [18:54] you can't quit now [18:54] I've conceded [18:54] dammit [18:54] go play in your clouds [18:54] I got bugs to fix :) [18:54] * mhall119|work wishes he had a cloud [18:54] my server has no VMX [18:54] angels have a higher priorty than bugs :) [18:55] they play in the clouds :) [18:55] but angels don't need fixing [18:55] heh, wonder what angel is playing around in the ubuntu cloud today [18:55] :p [18:57] On matters of things you can implement/play with TODAY :p [18:57] I (with federico's help) have released the latest sabayon [18:57] 2.29.2 [18:57] I'll have official builds up in my ppa later today. [18:57] congrats! [18:58] i recently used italc from my office and montioring the labs. interesting feature [18:58] gotta get a lucid build there too. [18:58] This sabayon has manuals, and will be the first one that, other than a minor patch, basically runs upstream-clean on Ubuntu. [18:59] sbalneav: deserves an award :) [18:59] So, if anyone had their 'druthers, what burning bug/issue should I work on next. I've got some merges left to do, but out of the bugsquad bugs, what's a PITA for people? [19:00] make sabayon run on a cloud [19:00] sorry, couldn't resist [19:00] * sbalneav slaps mhall119|work with a trout. [19:00] though making it work with XFCE would be nice, but probably not feasible [19:01] or LXDE [19:01] XFCE/LXDE doesn't work with gconf, does it? [19:01] it should do SOME stuff for you [19:01] if they use xdg menus, they'll pick that up. [19:02] and any settings that get made in the home dir should get picked up. [19:03] one of those "overloaded" issues is windows app artrage [19:04] teaching art with it is nice, teaching art with it on underpowerd thin clients, not so nice [19:05] XFCE is changing their configuration settings daemon around [19:05] not sure what they use [19:05] "artrage" [19:05] ? [19:05] it was xfconf in 9.04 [19:06] but I think they were changing again [19:06] maybe to use dconf [19:07] sbalneav: keep your strength for dbus merging for LTSP ;) You're the only one that can do it! [19:08] alkisg: ugh. It sucks being the only real C programmer :( [19:08] sbalneav: yup, so stop trying to be a packager/maintainer [19:08] Do some *real* work :P :D [19:09] ...and leave the kid stuff for kids :D [19:09] And we have howmany packager/maintainers here on this merry little band of misfits? [19:09] :) [19:09] Erm... urm... :D [19:10] Exactly [19:10] alkisg: I've made you upstream, when are YOU gonna be a packager slackass :) [19:10] I mean, geez :) [19:10] Oh, yeah [19:10] I forgot [19:10] Hey, I've packaged more apps than anyone here [19:11] Many Gb of them [19:11] Yeah, but you don't have upload, right? [19:11] But they all used the same template :P :D [19:11] That's the problem, we need more people who can upload. [19:11] Sure, I'd like to go for MOTU like you one day [19:11] alkisg: lets both go through the process together. [19:11] then we;ve got 3 [19:11] hv, you and me. [19:12] Hmm... sounds good... [19:12] i'd do it *if* i had some direction and training [19:12] I'll teach what I know so far, later tonight, if anyone's interested. [19:13] sbalneav: did you apply somewhere? [19:13] I.e. put packages in revu? [19:13] Not yet, I have to get one more package done before I'm "a shoe-in" [19:14] Then I'm gonna submit them all. ogra and LaserJock have said they'll coach me though the approval process. [19:14] I want to get it by end of december if I can [19:14] Well I imagine it won't be much more trouble for them if they coach me too... [19:15] Sure. [19:15] * alkisg will be here later on to watch sbalneav's lessons :) [19:15] that's what I;m saying, more uploaders we have, the better. === vorian_ is now known as v === v is now known as vorian [20:52] OK, tarballs for sabayon-2.29.2 are out out out! [20:52] I'm gonna push a new version, gimme 10 [20:53] sbalneav: cool. planning on uploading anything to revu soon? [20:55] highvoltage: I've got one more merge I want to do. [20:56] then I'll blast 3 or 4 things up in one shot, and hit you and ogra up for help [20:56] Think I can make motu by end of december? [20:56] sbalneav: oh for sure! [20:57] sbalneav: oh sorry I thought you meant December 2010 :p [20:58] sbalneav: well there's lots of holidays, etc in December, I think there's a chance though. [21:16] sabayon_2.29.2-0ubuntu1~ppa3 [21:16] just uploaded to my ppa [21:16] With the exception of 1 *very tiny* patch to change the icon to a gksu sabayon, this one runs on ubuntu with no other modifications [21:16] \o/ [21:18] * alkisg applauses for the new master of the universe (to be) [21:18] cool. do you uuencode it or use the new fancy source package format that allows binary diffs? [21:18] erm [21:18] I use dput :) [21:19] is that choice a or b? :) [21:19] I meant for the icon patch :) [21:19] ah, no [21:19] it uses cdbs [21:19] for the patch. it's the .desktop file that gets patched. [21:19] ah ok [21:20] Sorry, I was less than clear. [21:24] np [21:33] Oooo [21:33] crumb [21:34] Rejected: [21:34] File sabayon_2.29.2.orig.tar.gz already exists in PPA for Scott Balneaves, but uploaded version has different contents. See more information about this error [21:34] +in https://help.launchpad.net/Packaging/UploadErrors. [21:34] Files specified in DSC are broken or missing, skipping package unpack verification. [21:34] grumble [21:34] gotta delete the one from my ppa first... [21:37] sbalneav: I don't think it'll accept it easily without bumping the version... [21:39] Well, we'll try. [21:40] Sigh, I bumped the local packaging tarball number. Shouldn't have done that. [21:40] * sbalneav facepalms [21:41] eurgh [21:41] yeah [21:41] just the tarball? [21:41] lemme see if I can ping ogra [21:41] yeah [21:41] I need to ditch the tarball in my ppa [21:41] you should probably dch -i and increment it in the actual package source and then rebuild it from there [21:41] anyway I can do that? [21:42] Well, here's what happened. [21:42] I don't know, increasing the version sounds like The Right Thing to do [21:42] sbalneav: If you don't want to increment the source version, I think it'll be easier if you just give me the package, so I upload it to my ppa, and you copy it from there :P :D [21:42] Heh [21:42] So, since I'm upstream for sabayon now, I've been working on the 2.29.2 version [21:43] so, I've been creating 2.29.2 source tarballs. [21:43] We've done a release, last step was to update news and read,e [21:43] so, we've now got a released tarball. [21:44] problem is, released tarball and my previous tarball are ever so slightly different. [21:44] and I want to release the package with the official upstream tarball [21:45] there's gotta be a way to ditch the existing tarball in my ppa. [21:45] sbalneav: so the upstream tarball changed without a change in version number? [21:46] well, *I* was producing the upstream tarball, yeah. [21:46] I suppose I should call them (when I work on them) 2.29.x-prerelease or something from now on [21:47] sounds like a good idea :) [21:47] Which is fine for future reference... [21:47] How do I fix this? :) [21:48] Shouldn't that be 2.29.x~prerelease ? [21:55] well, whatever I call my "working" tarballs for upstream releases. [21:55] alkisg: sbalneav is talking about the upstream package version number, not the package in the ppa's version number [21:56] I'll ask later in #launchpad tonight. [21:56] alkisg: and upstream can pretty much version just how they want to [21:56] Ah, so launchpad would accept a "lower" version of the upstream package? [21:56] *tarball [21:56] I'm sure someone there can help me ditch the tarball [21:56] OK, got it [21:57] sbalneav: I think also in #ubuntu-motu they may be able to help you... [21:57] could try there. [21:58] But when that happened to me, I just put the package to another ppa of mine, and copied the binaries :D === Ahmuck-Sr is now known as Ahmuck