IRC chat logs for #ltsp on irc.libera.chat (webchat)
Channel log from 8 May 2008 (all times are UTC)
| 00:01 | monteslu has joined #ltsp | |
| 00:01 | generic has joined #ltsp | |
| 00:02 | captain_magnus has quit IRC | |
| 00:03 | captain_magnus has joined #ltsp | |
| 00:10 | achandrashekar has joined #ltsp | |
| 00:21 | StefanB has joined #ltsp | |
| 00:25 | jblack has joined #ltsp | |
| 00:26 | <jblack> So, I'm having a rough night with ltsp.
| |
| 00:27 | I've tried upgrading from gutsy to hardy twice now, and both times, ltsp breaks. I'm about done for the night, but I'd love some suggestions for what to look at tomorrow.
| |
| 00:27 | asac_ has joined #ltsp | |
| 00:28 | <johnny> ?
| |
| 00:28 | did you rebuild the chroot after upgrading?
| |
| 00:28 | <jblack> My symptoms are rather vague... the splashscreen comes up, but the login manager (gdm, I presume) never does. Instead, the monitor acts as if it's beign suspended. I have tried editing the lts.conf in the ftpboot dir.
| |
| 00:28 | <johnny> it's ldm..
| |
| 00:28 | you must rebuild the chroot
| |
| 00:28 | i found that out myself
| |
| 00:28 | <jblack> Tonights latest attempt was a fresh install of a hardy system, and installing most everything under the sun.
| |
| 00:28 | <johnny> oh
| |
| 00:29 | <jblack> Yeah. I created a fresh image and such
| |
| 00:29 | <johnny> well.. if you're not going to have time to sit in front of it to debug it.. you should prolly just come back tomorrow
| |
| 00:29 | it'll be easier that way
| |
| 00:29 | cuz it'll involve some poking and prodding
| |
| 00:29 | <jblack> I have the time. The guy that's in washington flipping the switch on the thin client is out. =)
| |
| 00:29 | <johnny> lol
| |
| 00:30 | you should make a local setup
| |
| 00:30 | install virtualbox
| |
| 00:30 | and then use another virtualbox vm to install a client
| |
| 00:30 | <jblack> I just don't have the hardware here.
| |
| 00:30 | <johnny> that's how i test :)
| |
| 00:30 | i use virtualbox vm as a client for easy easy easy testing
| |
| 00:30 | <jblack> I even had him hook up another monitor, in case it was bad sync rates.
| |
| 00:31 | <johnny> you should try what i said
| |
| 00:31 | <jblack> the client is booting up well enough that he can apt-get install ssh-server, and i can poke at the client.
| |
| 00:31 | <johnny> it'll make your life much easier :)
| |
| 00:31 | did you check /var/log/ldm.log on the client?
| |
| 00:31 | * johnny guesses not.. :) | |
| 00:31 | <jblack> I tried the equivilant of running the thin client from here once, and it took 35 minutes to get the full desktop.
| |
| 00:32 | I didn't see a ldm.log, but you can bet I'll be looking at it tomorrow.
| |
| 00:34 | <MacIver> i'm guessing you didn't change the flux capacitor
| |
| 00:34 | <jblack> now that you mention it...
| |
| 00:34 | <johnny> jblack, that's why you should try what i said
| |
| 00:34 | install hardy in a vm
| |
| 00:34 | and then use another vm to boot off of it
| |
| 00:35 | only thing it can't do is hardware.. and that is usually NOT the problem..
| |
| 00:35 | altho it certain can be..
| |
| 00:35 | but at least you'll know
| |
| 00:35 | and you can twiddle it without moving :)
| |
| 00:36 | <jblack> I'm pretty certain it's a configuration issue on the ltsp server.
| |
| 00:36 | <johnny> thus why you should do what i suggest :)
| |
| 00:36 | tux_440volt has quit IRC | |
| 00:36 | <jblack> you did catch the "35 minutes to start" that I mentioned a bit ago?
| |
| 00:36 | I'm in Pennsylvania. The server is in Spokane, Washington
| |
| 00:37 | makghosh has quit IRC | |
| 00:37 | <johnny> it won't if you do what i said
| |
| 00:37 | install hardy in vm locally
| |
| 00:37 | on YOUR computer
| |
| 00:37 | then set up one more vm
| |
| 00:37 | <jblack> I wish I could, but I don't have the resources here to do it
| |
| 00:37 | <johnny> that just netboots
| |
| 00:37 | why not?
| |
| 00:37 | you have a computer
| |
| 00:37 | your'e using it right now :)
| |
| 00:37 | <MacIver> iphone
| |
| 00:37 | <johnny> even the iphone might be able to lol.. in the future :)
| |
| 00:38 | <jblack> I'm lacking the memory, processor and disk space to pull it off usefully.
| |
| 00:38 | <MacIver> lol
| |
| 00:38 | <johnny> how much memory?
| |
| 00:38 | <jblack> ANyways, that's one wonderful suggestion.
| |
| 00:38 | <johnny> disk space is just like 3 ..
| |
| 00:38 | 3 gb
| |
| 00:38 | <jblack> Great. Can you loan me about 2.6 gigs?
| |
| 00:38 | asac has quit IRC | |
| 00:38 | asac_ is now known as asac | |
| 00:38 | <johnny> delete some crap :)
| |
| 00:38 | <jblack> ANyways, thanks for the suggestion.
| |
| 00:39 | <MacIver> 500 GB drive for 30-60$
| |
| 00:39 | <jblack> Your two... the ldm.conf, and trying to perform a local install.
| |
| 00:39 | <johnny> it'll save your hours of time.. which i assume is worth the upgrade
| |
| 00:39 | well you could do a local install.. if you're running hardy
| |
| 00:39 | and then the vm only needs 0 disk space..
| |
| 00:39 | <jblack> pardon, looking for the ldm log on the client.
| |
| 00:39 | Any other suggestions?
| |
| 00:39 | <johnny> and 128mb memory
| |
| 00:40 | can't really make any more until you do that
| |
| 00:40 | it's a poking and prodding process
| |
| 00:40 | <MacIver> ltsp-update-reflux
| |
| 00:40 | <johnny> check this.. check that, etc
| |
| 00:40 | <jblack> Dude, I appreciate the help, but get out of this rut. I'm not changing my laptop from gutsy to hardy to install xen to make a fake pxeboot machine to test ltsp 3000 miles from where the problem is.
| |
| 00:41 | Don't get me wrong, thanks for the help, but saying the same thing a 12th time isn't going to help more than the 2nd time you said it. =)
| |
| 00:41 | <johnny> xen is too difficult ..
| |
| 00:41 | <MacIver> why not?
| |
| 00:41 | <johnny> virtualbox is all point and click
| |
| 00:41 | jblack, the only actual suggestion i have is for you to check the ldm.log and xorg log
| |
| 00:42 | that's all there is
| |
| 00:42 | utnil you see what they say
| |
| 00:42 | <jblack> Yeah. I looked at the xorg.log. I couldn't find anything useful. The impression it gave me was that the card was driving a modeline that the monitor couldn't handle.
| |
| 00:43 | <johnny> that's often not an error
| |
| 00:43 | i see that all the time
| |
| 00:46 | <jblack> what? An xorg upgrade results in new frequencies being supported, thus resulting in a non-syncable display?
| |
| 00:46 | <johnny> seeing it try modelines..
| |
| 00:46 | and fail
| |
| 00:46 | before it settles on a good one
| |
| 00:47 | <jblack> Oh, I didn't see it try a bunch of modelines in the xorg log. I'm not even convinced the lts.conf in tftpbook/ltsp is being used.
| |
| 00:47 | The only indication I have that something is going on is that he reported the lcd light changing color
| |
| 00:48 | splash screen, monitor light changes color, but ctrl-alt-f1 to get a console works
| |
| 00:48 | he's not even seeing the X checkerboard.
| |
| 00:49 | <johnny> mine doesn't work in tftpboot directory.. but it seems like a dnsmasq issue
| |
| 00:49 | i don't use the standard dhcpd setup
| |
| 00:49 | <jblack> uhhh. hmmmmm
| |
| 00:50 | root-path "/opt/ltsp/i386 filename "/ltsp/i386/pxelinux.0"
| |
| 00:50 | that looks like a problem to me
| |
| 00:50 | <johnny> root-path is only used by nfs
| |
| 00:50 | hardy doesn't use nfs by default
| |
| 00:50 | the filename is really a concern to your tftp server anyways
| |
| 00:51 | <jblack> I figured that may be the base path that lts.conf is loaded from.
| |
| 00:51 | <johnny> no
| |
| 00:51 | it is not
| |
| 00:51 | <jblack> The more intersting one is filename. this system doesn't have a /ltsp, but an /opt/ltsp
| |
| 00:51 | <johnny> that's based on the tftp root
| |
| 00:52 | it's a path from /var/lib/tftpboot
| |
| 00:52 | <jblack> Ok. Sensible.
| |
| 00:52 | Maybe it's something simple, like "ltsp works on xen in gutsy, but not hardy"
| |
| 00:53 | or "2.6.24-16 is broken"
| |
| 00:53 | <johnny> you're using ltsp in xen ?
| |
| 00:53 | i never tried that
| |
| 00:53 | don't see why it would be an issue tho
| |
| 00:53 | <jblack> the server is on xen, yes.
| |
| 00:53 | <johnny> if you have made it so far in the boot process
| |
| 00:54 | <jblack> Yeah. He can even apt-get install ssh-server on the client once he drops down to tty1
| |
| 00:54 | <johnny> you could install it in the chroot and rebuild the image
| |
| 00:54 | so you don't have to keep reinstalling it eveyrtime
| |
| 00:54 | <jblack> Yup, I'm aware.
| |
| 00:56 | I don't suppose there's some good logging on the ltsp server that I can review.. I couldn't find much of anything, other than somethign called ldminfod
| |
| 00:56 | <johnny> your problem is mostly likely with x.. and that doesn't log remotely
| |
| 00:57 | <jblack> But it could be a ltsp/xen relationship issue... xen is kind of quirky when it comes to tty setup
| |
| 01:02 | daya has joined #ltsp | |
| 01:10 | <jblack> https://bugs.launchpad.net/ubuntu/+source/ltsp/+bug/224901/ might be my bug
| |
| 01:11 | <johnny> you would have had an issue earlier
| |
| 01:11 | wouldn't you have noticed ltsp-build-client failed?
| |
| 01:12 | deavid has joined #ltsp | |
| 01:12 | <jblack> I noticed various warnings on build-client. I fixed several of them by hand
| |
| 01:12 | <johnny> hmm.. iregenerated my chroot for hardy recently.. all seems well
| |
| 01:13 | <jblack> This was reported 4-30 (but was a duplicate of one on 4-25)
| |
| 01:14 | I don't remember seeing an outright failure at the end.
| |
| 01:15 | <johnny> then it is probably not the same bug
| |
| 01:15 | the only thing i had to do was fix the expired root account
| |
| 01:15 | which is a registered bug
| |
| 01:16 | <jblack> I'm doing a rebuild ow.
| |
| 01:16 | yeah, I fixed that one by hand too. :)
| |
| 01:17 | StefanB has quit IRC | |
| 01:17 | <jblack> in fact, I still have the shadow manpage up
| |
| 01:21 | Nope. Just debconf warnings and some font complaints.
| |
| 01:21 | And an expired root account, but no Big Explosion
| |
| 01:25 | It would help if I could run an xserver on the ltsp server. =)
| |
| 01:26 | <johnny> my ltsp server is at a coffeehouse..
| |
| 01:26 | and is also the point of sale machine
| |
| 01:27 | <jblack> Nice.
| |
| 01:27 | <johnny> custom open source software
| |
| 01:29 | <jblack> Quick question... he should have been able to run "X" on the client, and get the checkboard, no?
| |
| 01:29 | <johnny> yes
| |
| 01:30 | <jblack> Then that narrows it down considerably
| |
| 01:30 | He didn't get a checkerboard when I ran X on him.
| |
| 01:38 | viking-ice_ has joined #ltsp | |
| 01:39 | itais has joined #ltsp | |
| 01:41 | <generic> hey guys
| |
| 01:41 | i have installed SAP gui on ubuntu now i want to make a link on each user destop how i can do this?
| |
| 01:41 | <jblack> Ok. so, thinking about it, I think 2 things are happening. Firstly, X --configure is reporting modelines that the monitor can't handle, that are different than gutsy, and second, that /var/lib.../i386/lts.conf is being ignored.
| |
| 01:41 | <generic> a soft link
| |
| 01:41 | <jblack> That's presuming he didn't manage to lay his hands on two lcd monitors that can't drive at 800x600
| |
| 01:44 | the former is sensible; via chipsets are probably much less common.
| |
| 01:44 | The latter... <shrug?>
| |
| 01:45 | The documentation I read earlier wasn't clear. what if there's a lts.conf in /var/.../i386 and in the pxeboot image? Which one takes precidence?
| |
| 01:47 | If it's the latter, than a well meaning person disabled the /var one on hardy by adding a lts.conf file in the image's etc directory that says "This one works, but don't use it".
| |
| 01:50 | Q-FUNK has joined #ltsp | |
| 01:52 | <jblack> I'll be happy to check the sources if someone can tell me what parses lts.conf
| |
| 01:53 | <johnny> the ltsp-client init scripts in the chroot
| |
| 01:53 | the one in the tftproot should take precedence
| |
| 01:54 | <jblack> I'll double check
| |
| 01:58 | cpunches has joined #ltsp | |
| 02:00 | <jblack> interesting. a "grep -r lts.conf /opt/ltsp/i386/etc" lists only one file that contains lts.conf, and that's lts.conf itself.
| |
| 02:00 | * jblack looks in boot | |
| 02:01 | Pascal_1 has joined #ltsp | |
| 02:01 | <johnny> yeah.. it's empty
| |
| 02:01 | with a comment
| |
| 02:01 | to tell you that you should use tfpboot directory :)
| |
| 02:01 | <jblack> COrrect.
| |
| 02:02 | However, I think the one int tftpboot is being ignored, thus wanting to double check that the precence as listed in the chrooted one is correct.
| |
| 02:02 | I can't find what in /etc actually parses either one. ;)
| |
| 02:03 | itais has quit IRC | |
| 02:04 | <jblack> Ok. This is _very_ interesting.
| |
| 02:05 | gdm in the ltsp _server_ matches a grep for lts.conf. The gdm in the ltsp _client_ does not
| |
| 02:06 | then again, there is no gdm in the chroot, at least not in usr/sbin
| |
| 02:09 | Yeah. I believe that lts.conf, at least as of hardy, ignores lts.conf
| |
| 02:10 | <johnny> ldm
| |
| 02:10 | it's not gdm
| |
| 02:11 | gdm is not used in ltsp
| |
| 02:11 | <jblack> I did a grep on /usr/sbin, within the chroot
| |
| 02:11 | What I found was that ldm does not reference lts.conf, but gdm does.
| |
| 02:12 | <johnny> the file that reads in the lts.conf is in the initramfs
| |
| 02:12 | <jblack> try it. on your ltsp server, cd into /usr/sbin, and grep lts.conf *. gdm will pop right up. No such luck on ldm.conf
| |
| 02:12 | pardon, for ldm.
| |
| 02:12 | <johnny> ldm does not read lts.conf
| |
| 02:12 | <jblack> Ok. I'll check the initrd then
| |
| 02:12 | <johnny> it takes environmentable variables
| |
| 02:13 | that are passed to it
| |
| 02:13 | via the screen-x-common script that ltsp uses
| |
| 02:16 | <jblack> My ltsp server doesn't have initrds at all. do you mean the i386.img squashfs image?
| |
| 02:17 | <johnny> no
| |
| 02:17 | i'm talking about in /var/lig/tftpboot/ltsp/chrootarch
| |
| 02:17 | <jblack> I forgot about those
| |
| 02:18 | Pascal_1 has quit IRC | |
| 02:21 | <jblack> Ok. No.
| |
| 02:21 | THey're ignored
| |
| 02:23 | I have no screen-x-common anywhere. I even unpacked the initrds. and the only thing I can find within the chroot, or files used to boot the chroot, that reference lts.conf is lts.conf itself.
| |
| 02:23 | <johnny> screen-x-common is in the chroot
| |
| 02:24 | only the tftping of lts.conf is done in the initrd
| |
| 02:24 | and the nbd mounting
| |
| 02:24 | <jblack> Not here it's not.
| |
| 02:24 | <johnny> hmm.. should be
| |
| 02:24 | lemme look in my server
| |
| 02:24 | <jblack> I'm telling you. I went through the chroot, it's not there. I loopback mounted the img, it's not there. I even ran the initrd through cpio to get at the contents, and it's not there either
| |
| 02:26 | of course, by there, I mean here.
| |
| 02:26 | <johnny> Binary file i matches
| |
| 02:26 | scripts/ltsp_nbd: # get the lts.conf via tftp
| |
| 02:26 | scripts/ltsp_nbd: (echo "get ${tftpath}/lts.conf"|tftp ${ROOTSERVER} >/dev/null 2>&1)
| |
| 02:26 | scripts/ltsp_nbd: if [ -s ./lts.conf ];then
| |
| 02:26 | scripts/ltsp_nbd: cp ./lts.conf ${rootmnt}/etc/
| |
| 02:26 | oh.. here's the command
| |
| 02:26 | root@emma:/var/lib/tftpboot/ltsp/i386/test# grep -r 'lts\.conf' *
| |
| 02:29 | <jblack> I see it in ltsp_nbd.com too.
| |
| 02:30 | itais has joined #ltsp | |
| 02:32 | <generic> helo
| |
| 02:32 | <jblack> Sorry I missed it. I really did look.
| |
| 02:32 | <generic> guys
| |
| 02:32 | how i make a link
| |
| 02:32 | <jblack> what kind of link? A symbolic link? (soft link) ?
| |
| 02:33 | <generic> and how i change nbd to use nfs and not use nbd
| |
| 02:33 | <johnny> why would you want to do that
| |
| 02:33 | <generic> ya i install sapgui
| |
| 02:33 | <johnny> there are instructions on the ubuntu wiki
| |
| 02:33 | for both questions
| |
| 02:33 | <generic> i like to make a link on my users dektop to use that to lauch sapgui
| |
| 02:33 | <johnny> but if i was you.. i woulldn't change to nfs
| |
| 02:33 | it's noticably slower
| |
| 02:33 | here
| |
| 02:34 | <generic> i have some works to do on
| |
| 02:34 | where on ubuntu wiki
| |
| 02:34 | send me link
| |
| 02:36 | <johnny> sorry..i can't
| |
| 02:36 | but i'm sure you can use a search engine
| |
| 02:37 | mikkel has joined #ltsp | |
| 02:39 | <jblack> there goes that theory
| |
| 02:40 | <johnny> ?
| |
| 02:40 | <jblack> The lts.conf being ignored w/ X --configure acting differently between distro releases
| |
| 02:41 | <johnny> lts.conf isn't even necessary these days.. unless you really need it
| |
| 02:41 | you could try passing your own known working X_CONF
| |
| 02:41 | but i think you should try the ldm log
| |
| 02:42 | and/or starting ldm manually after killing it from the console
| |
| 02:42 | <jblack> The system worked in gutsy, broke after upgrade to hardy. So I move the image aside, and tried a fresh hardy install
| |
| 02:42 | <johnny> your issue might not be related to ltsp.. but to other components
| |
| 02:42 | <jblack> Yeah. I'l try putting a braindead xorg.conf on it tomorrow.
| |
| 02:43 | Yeah. it may
| |
| 02:43 | <johnny> you could also set CONFIGURE_X=F
| |
| 02:43 | before making an xorg.conf
| |
| 02:43 | in lts.conf
| |
| 02:43 | <jblack> Yeah, I suppose I should. Otherwise, it'll autoconfigure it away
| |
| 02:43 | <johnny> well..
| |
| 02:43 | CONFIGURE_X is needed for you to manually set certain X parameters
| |
| 02:43 | <jblack> What keeps ringing in my ears is him telling me the monitor light changed.. "as it it were going to sleep".
| |
| 02:44 | <johnny> if you use CONFIGURE_X=F
| |
| 02:44 | it uses x's own built in detection
| |
| 02:44 | which might be more successful for you
| |
| 02:44 | me personally..
| |
| 02:44 | i've been working on the port of ltsp to gentoo
| |
| 02:44 | ltsp5*
| |
| 02:45 | and i forced CONFIGURE_X=F completely atm
| |
| 02:45 | so unlike ubuntu CONFIGURE_X defaults to F
| |
| 02:45 | <dberkholz> (yay!)
| |
| 02:45 | <johnny> cuz i could never get X --configure --novtswitch to succeed
| |
| 02:45 | <jblack> I wonder if CONFIGURE defaulted to F back in gutsy, and switched in hardy
| |
| 02:45 | <johnny> alwasy get a paln screen
| |
| 02:45 | no.. it didn't
| |
| 02:46 | it's a standard feature of ltsp to be able to set certain x parameters via lts.conf
| |
| 02:46 | with CONFIGURE_X=F you can't set those parameters
| |
| 02:47 | dberkholz, i finally figured out my issue .. or rather agaffney did
| |
| 02:47 | it was really dumb
| |
| 02:47 | i'll spare you the details
| |
| 02:47 | <dberkholz> issues usually are dumb
| |
| 02:47 | <johnny> i'm building a fresh one now
| |
| 02:47 | <jblack> GOtcha.
| |
| 02:49 | <johnny> xorg's built in detection has grown greatly
| |
| 02:49 | <jblack> Definitely. I remember the old 3.x days.
| |
| 02:49 | <johnny> when evdev advances, and better tools are available to manage screen placement via randr... i will never use an xorg.conf again prolly
| |
| 02:49 | <jblack> Back when x.org didn't exist. =)
| |
| 02:50 | <johnny> atm sadly i have to.. :(
| |
| 02:50 | but the day when it is unnecessary is on the horizon
| |
| 02:50 | <jblack> What's throwing me off the tail so easily was that as a gutsy system, things just worked. Moving to hardy, no X on the client.
| |
| 02:50 | <johnny> well.. i've ran into a few other nasties in hardy myself
| |
| 02:51 | but not that one
| |
| 02:51 | tux_440volt has joined #ltsp | |
| 02:51 | <jblack> My first thought was broken migration, so naturally, I set the logical volume aside, and did a fresh install.
| |
| 02:51 | <johnny> my network driver was totally screwed
| |
| 02:51 | <jblack> And walked into the same thing.
| |
| 02:51 | <johnny> it would die eveyr day
| |
| 02:51 | <jblack> ouch
| |
| 02:51 | <johnny> at least once..
| |
| 02:51 | killing all ltsp clients
| |
| 02:51 | i reverted to the gutsy kernel, and all is well
| |
| 02:52 | <jblack> That reminds me of the xen hardy nightmare that I just went through for a month.
| |
| 02:52 | <johnny> i'm using the b44 driver.. which is really hard to find info for.. run into too much nonsense about wireless driver issues for a similiarly named driver :(
| |
| 02:52 | <jblack> They broke the netfront driver in 2.2.24, months ago, ignored bugs reports, ignored fixes, etc etc
| |
| 02:52 | <johnny> i think ignored is wrong
| |
| 02:52 | i just don't think they can handle the load
| |
| 02:53 | <jblack> Fair enough
| |
| 02:53 | <johnny> it's really hard to do stable releases at this point for any distro
| |
| 02:53 | too much is changing in the infrastructure of the distros
| |
| 02:53 | in the gnome,kernel, and x realms
| |
| 02:53 | gnome 2.22 had gvfs merged ... but most likely prematurely
| |
| 02:53 | even though it is great
| |
| 02:53 | tux_440volt has quit IRC | |
| 02:54 | <jblack> Yeah. I used to yell at distro every time we met for throwing the latest crap in. (I used to work there)
| |
| 02:54 | <johnny> sure.. but if you don't
| |
| 02:54 | nobody will support you
| |
| 02:54 | because they don't run that old stuff
| |
| 02:54 | that's the kinda support you have to pay for
| |
| 02:54 | that's why i upgrade to new ubuntu releases right away
| |
| 02:54 | <jblack> Your right, of course.
| |
| 02:54 | <johnny> because in all my other work.. i'm using the new stuff
| |
| 02:55 | as i'm on gentoo
| |
| 02:55 | even if i have to deal with some bugs
| |
| 02:55 | i give it a few days.. if it looks manageable.. i jump in
| |
| 02:55 | <jblack> For distro, it's a damned if you do, damned if you don't type situation. You're going to get burned either way, all you get to pick is the type of pain
| |
| 02:55 | <johnny> yep
| |
| 02:55 | yeah.. it's cuz we're still matching parity to other systems at this point
| |
| 02:56 | let a few things catch up..and everything will be fine for awhile :)
| |
| 02:56 | <jblack> Then again, I think part of the responsibility for that mess falls into sabdfl's lap because of his big push for "daily crack"
| |
| 02:57 | <johnny> i imagine alot of folks disagree
| |
| 02:57 | <jblack> Sure, even back before ubuntu was called no-name-yet, the clock was already speeding up.
| |
| 02:57 | <johnny> sure.. but the ball isnt' in ubuntu's court
| |
| 02:57 | it's in the x and gnome peeps hands
| |
| 02:58 | <jblack> I don't think the clockhands would be spinning quite so quickly.
| |
| 02:58 | <johnny> distros are too far out of date for me within 3 months of the release cycle
| |
| 02:58 | err within 3 months post release i mean
| |
| 02:58 | <jblack> Aspects of it, sure... But consider... If other distros weren't chasing ubuntu's tail... Debian isn't exactly known for high speed, nor were the redhat types... Perhaps gentoo
| |
| 02:59 | <johnny> debian is too slow :)
| |
| 02:59 | 6 month release cycles just arne't long enough.. :)
| |
| 02:59 | <jblack> That's rather my point. Prior to someone deciding he wanted his own distro, most distros _were_ slow.
| |
| 02:59 | <johnny> they need to go with 7 and a half..
| |
| 02:59 | that's not been my experience
| |
| 03:00 | but i've only been involved since 98
| |
| 03:00 | <jblack> And I just a couple years before that.
| |
| 03:00 | <johnny> i was using redhat then
| |
| 03:00 | <jblack> But seriously... Fedora? Redhat, Debian? Slackware? Who's the speed demon of that group
| |
| 03:00 | <johnny> the distros are try to catch up to the actual projects
| |
| 03:00 | the projects themselves are moving faster
| |
| 03:01 | the users drive demand for new stuff
| |
| 03:01 | too many people were compiling their own software to get what they needed
| |
| 03:01 | so distros came about that got you there faster
| |
| 03:01 | etc
| |
| 03:01 | it just needs to settle out first
| |
| 03:01 | <jblack> I don't think that the majority of users have built their own software.. at least since 2003 or so. And I think that truly slowed development.
| |
| 03:02 | <johnny> but then they were using somebody else's built software :)
| |
| 03:02 | <jblack> If 3/4 of your users are complaining about bugs in your last stable, you're going to end up paying more attention to that.
| |
| 03:02 | <johnny> some random rpm or deb they found
| |
| 03:02 | <jblack> sure, you're always going to want to work on head, but you'd get anoyed to death if you didn't release a patch for the stable release.
| |
| 03:03 | <johnny> not all developers have time for that
| |
| 03:03 | <jblack> True.
| |
| 03:03 | <johnny> so then the distro maintainers have to make their own
| |
| 03:03 | amongst their normal bug load
| |
| 03:04 | <cyberorg> that is where https://build.opensuse.org/ come into picture, it builds packages for all top distro and archs from same source tarball
| |
| 03:04 | <johnny> plus there is a lack of general knowledege sharing .. too much useful information scattered across too many bug systems
| |
| 03:04 | and mailing lists
| |
| 03:04 | it is not scaling
| |
| 03:04 | <jblack> I don't think having the world live on HEAD is a palatable solution.
| |
| 03:05 | <johnny> but the world wants the features of HEAD!!!
| |
| 03:05 | becaus they still cant' do certain things they can do on osx or win
| |
| 03:05 | <jblack> A substatial portion of it. Certainly more than half, but maybe not as much more as you'd expect.
| |
| 03:05 | <johnny> like group policies in active directory
| |
| 03:06 | in the server realm
| |
| 03:06 | thus.. systems were created to deal with it
| |
| 03:06 | but they are still in development
| |
| 03:06 | <jblack> I do't know whether you're cheering or lamenting the state of things...
| |
| 03:06 | <johnny> both
| |
| 03:06 | but more cheering
| |
| 03:07 | 70/30
| |
| 03:07 | <jblack> ANd though I'd like to see a change, I know no such thing will happen...
| |
| 03:07 | <johnny> change? change is coming
| |
| 03:07 | :)
| |
| 03:07 | <jblack> but inevitably, down this parth, lies in Windows ironically becoming the more stable operating system.
| |
| 03:07 | <johnny> i think we will see more useful aggregate info
| |
| 03:07 | which will create more useful communication
| |
| 03:07 | more distros are focusing on committing changes in upstream
| |
| 03:07 | thus sharing out the patch burden
| |
| 03:08 | <jblack> I don't know about you, but an MRI machine running windows because it's less likely to crash scares the shit out of me.
| |
| 03:08 | <johnny> we're almost to the point of seeing the fruits of that labor
| |
| 03:08 | <jblack> Not because they evetnually raise the bar, but because everyone else lowered it too much
| |
| 03:08 | <johnny> it's still in the adolescent phase
| |
| 03:08 | <jblack> Yeah. That's somewhere a certain mark should get credit for. He's reaching tightening that coordination up
| |
| 03:09 | <johnny> don't forget the fedora folks
| |
| 03:09 | stuff is finally happening on that front
| |
| 03:09 | previously redhat projects carried too many patches locally
| |
| 03:09 | * ogra humps jblack's leg | |
| 03:09 | <ogra> jblack, !!!
| |
| 03:09 | <jblack> ogra!!!!
| |
| 03:10 | * jblack smacks ogra down with a newspaper | |
| 03:10 | <johnny> ogra, i'm not tryin to make you look bad :)
| |
| 03:10 | <ogra> heh
| |
| 03:10 | <jblack> How is it, my friend?
| |
| 03:10 | <ogra> all fine :)
| |
| 03:10 | <johnny> i hope i'm representing things accurately
| |
| 03:10 | <jblack> johnny: You did fine.
| |
| 03:10 | <ogra> still doing the same
| |
| 03:10 | <jblack> still with the company?
| |
| 03:10 | <ogra> sure
| |
| 03:10 | <jblack> That's great news.
| |
| 03:10 | <johnny> you know this crazy guy jblack ?
| |
| 03:11 | lol
| |
| 03:11 | <jblack> Ogra? Yeah. We've drinks together a few times.
| |
| 03:11 | <johnny> oopss.. missed a comma there
| |
| 03:11 | <ogra> johnny, jblack spent me my first drink with ubuntu people ever :)
| |
| 03:11 | <jblack> I was his biggest cheerleader. :)
| |
| 03:11 | "bought" buddy
| |
| 03:11 | <johnny> jblack, ogra and i have worked together on a few things
| |
| 03:11 | trying to help him test ubuntu ltsp
| |
| 03:12 | and he also helps me with the gentoo integration
| |
| 03:12 | GiantPickle has quit IRC | |
| 03:12 | <jblack> heh. every time I see your nick, ogra, I get reminded of Bernd das brot. Don't know why
| |
| 03:13 | <johnny> ogra.. i'm getting a gigabit nic tomorrow/today
| |
| 03:13 | going to install it in my ltsp server ...
| |
| 03:13 | found out my switch has gigabit uplink
| |
| 03:13 | i didn't buy it.. it was here previously
| |
| 03:13 | <jblack> ogra: Oh, btw, I think ltsp gutsy->hardy is broken, on at least some system. I'll file a bug about it once I figure it out
| |
| 03:14 | <johnny> ogra, that gvfs update made hardy proposed right?
| |
| 03:14 | hardy-proposed*
| |
| 03:14 | <ogra> jblack, haha, every time i see the night trailer on kika i tell my GF hey, i always have to think about jblack when i see that ... she's getting tired of the mataro bar story :)
| |
| 03:14 | <jblack> I'll check if you like, I owe you one, johnny
| |
| 03:15 | lol
| |
| 03:15 | <johnny> i'm about to hit the bed
| |
| 03:15 | now that ogra is here :)
| |
| 03:15 | if you hve any questions.. he will be most helpful
| |
| 03:15 | :)
| |
| 03:15 | <ogra> jblack, you need to rebuild the client chroot, we do way to mny changes at built time that i cnt avoid to get wiped on upgrades
| |
| 03:15 | s/built/build
| |
| 03:16 | <johnny> ogra, it's possible that all the load on the network card is what caused my issues
| |
| 03:16 | <jblack> gfgvs 0.2.3-0ubuntu5 is latest available on pro..per.. whatachamacallit
| |
| 03:16 | ogra: Of course.
| |
| 03:17 | <ogra> well, the gvfs fix is actually hidden in glib :)
| |
| 03:17 | <jblack> I haven't figured it out quite yet (thus why I haven't filed anything yet), but it's fundamental to break running "X" on the console and getting a checkerboard
| |
| 03:17 | whatever it is, it both breaks after upgrade from gutsy to hardy, and exists on a fresh hardy install too.
| |
| 03:18 | I suspect it's an edge case with via chipsets, but I haven't nailed it down yet
| |
| 03:18 | <johnny> evil via stuff
| |
| 03:18 | uggh.. sleepytime
| |
| 03:18 | <jblack> I didn't even know they made video cards.
| |
| 03:18 | <ogra> hum, i have a via client upstairs that appears to work fine
| |
| 03:19 | <jblack> oh. hrmm
| |
| 03:19 | <ogra> what exactly do you mean with checkerboard ?
| |
| 03:19 | <jblack> Have you ever run X manually? not startx, or gdm, or anything?
| |
| 03:20 | Just "X". Normally, you'll get back and white squares. The higher the resolution, the smaller the squares. Just looks grey on high resolutions.
| |
| 03:21 | <ogra> sounds like yuo use xdmcp
| |
| 03:21 | <jblack> ANyways, the xserver is breaking badly enough on the client that X isn't starting at all. The monitor status light is changing colors ,as if it's getting a freq that it can't handle.
| |
| 03:21 | Nope.
| |
| 03:21 | <ogra> hmm
| |
| 03:22 | <jblack> Which tells me that X --configure might not be handling the card well. Perhaps it treated it as vesa in gusty, and now it handles it as a via, but it's using a resolution/refresh that's too high for the monitor.
| |
| 03:23 | It could be anything at this point. I haven't even managed to rule out martians.
| |
| 03:23 | <ogra> create /var/lib/tftpboot/ltsp/i386/lts.conf
| |
| 03:23 | add
| |
| 03:23 | [default]
| |
| 03:23 | <jblack> I just spent an hour and a half (mistakenly) trying to convince poor johnny that lts.conf was being ignored.
| |
| 03:23 | <ogra> SCREEN_02=shell
| |
| 03:24 | then you get a login shell on tty2
| |
| 03:24 | <jblack> Oh, there's a login shell on tty1
| |
| 03:24 | <ogra> for deugging
| |
| 03:24 | there are no users you could log in with
| |
| 03:24 | <jblack> sure there are.
| |
| 03:24 | <ogra> the lts.conf option will log you in
| |
| 03:25 | did you create users in the chroot ?
| |
| 03:25 | <jblack> I edited the image and rewrote it so that root could log in.
| |
| 03:25 | <ogra> ah
| |
| 03:25 | k
| |
| 03:25 | <jblack> I even fixed the expired flag someone put in
| |
| 03:25 | toscalix has joined #ltsp | |
| 03:25 | <jblack> I'm not going to be able to work on it. The clients are something like 3,000 miles away. Litrally
| |
| 03:25 | <ogra> that not put in deliberately, its how the default is apparetnyl
| |
| 03:26 | <jblack> The person flipping the switches for me and playing monkey went to sleep
| |
| 03:26 | yeah. I saw the bug on the bts
| |
| 03:26 | and I saw you all over ltsp on the bts too. You're still a frigging machine, man. :)\
| |
| 03:27 | <ogra> heh, well, i only do ltsp and all edu related stuff in ubuntu :)
| |
| 03:27 | <jblack> Know what would be nice?
| |
| 03:28 | <ogra> and with the last release the edu duties dropped a lot, we dont have an edubuntu install CD anymore so i dont have much to care for there
| |
| 03:28 | <jblack> A ltsp-server package just installs a minimally working X for a headless server providing ltsp. =)
| |
| 03:28 | Yeah, so it'll just be a task from now on?
| |
| 03:28 | * jblack goes outside for a smoke | |
| 03:28 | <ogra> a metapackage and addon-cd
| |
| 03:29 | * ogra joins the smoking | |
| 03:32 | <jblack> uch better.
| |
| 03:32 | yeah, a meta package
| |
| 03:34 | <Q-FUNK> ogra: got any info about Prague, so that I can register and book my hotel?
| |
| 03:34 | <ogra> wrt via ... there are three drivers you can user with via chipsets (totaly messy situation)
| |
| 03:34 | <jblack> via, vesa, and?
| |
| 03:35 | <ogra> via, openchrome and unichrome
| |
| 03:35 | Q-FUNK, http://towers.corinthia.cz/hotel/en/hotel-location/
| |
| 03:35 | <jblack> four then
| |
| 03:36 | <Q-FUNK> ogra: that's where everyone is booked? and for registering for the event?
| |
| 03:36 | <ogra> unichrome is in universe though
| |
| 03:36 | <stgraber> Q-FUNK: UDS is on LP, FOSSCamp on www.fosscamp.org (if you attend it)
| |
| 03:36 | <jblack> That's ok. I trust universe fine, because I know you rule there.
| |
| 03:36 | <ogra> Q-FUNK, https://launchpad.net/sprints/uds-intrepid
| |
| 03:37 | jblack, thats long ago, dholbach is the he-man nowadays :)
| |
| 03:37 | <jblack> Daniel is? He rocks too
| |
| 03:37 | <ogra> yeah
| |
| 03:37 | and he switched from monarchy to democracy, we have a motu council now :)
| |
| 03:38 | so a bunch of he-mans :)
| |
| 03:38 | jblack, what are *you* doing now ? i didnt ask yet
| |
| 03:38 | <jblack> Is it organized in about the same way the ubuntu council was when I worked there?
| |
| 03:38 | <ogra> yeah
| |
| 03:38 | <jblack> these days, I'm doing a telecommuting sysadmin ting.
| |
| 03:38 | <ogra> its all split up into smaller councils to handle the immense growth
| |
| 03:39 | <jblack> xen, drbd, perhaps 20 virtual servers... and ltsp.
| |
| 03:39 | asterisk
| |
| 03:39 | <ogra> nice !
| |
| 03:39 | i was planning to get audio capturig working in intrepid
| |
| 03:39 | then you can have asterisk ltsp clients in callcenters :)
| |
| 03:40 | <jblack> Yeah, it's been great so far. They told me what they wanted to do, I told them what I thought they wanted, and they said "sounds good. whatever you think".
| |
| 03:40 | <ogra> hehe
| |
| 03:40 | <jblack> So it's a nice little setup, complete with ldap
| |
| 03:41 | <ogra> so do you get any hints about the X breakage in Xorg.6.log on the client ?
| |
| 03:42 | <Q-FUNK> ogra: is there any shcedule for UDS? I'm only likely to cone for fosscamp and a couple of days of UDS and I'd preferably attend only embedded and LTSP sessions
| |
| 03:42 | <jblack> I didn't look at Xorg.6.log. Thought it was cruft, so I looked at Xorg.1.log.
| |
| 03:42 | <ogra> we stopped relying on dpkg-reconfigure xserver-xorg with gutsy, it makes everything slow to pull the whole debconf database over the net every thime ...
| |
| 03:43 | well, by default X starts on :6.0 ... if you manually start it on :1.0 then tht log applies indeed
| |
| 03:43 | <jblack> The I also somehow missed that there was a lts log too. I'm looking forward to getting another look at the client. I'm hoping something obvious shows.
| |
| 03:44 | Ok. The 1 log didn't have anything useful, and looked as if X started just fine.
| |
| 03:44 | <ogra> yeah, there are only two logs on the client, ldm.log and the Xorg one
| |
| 03:44 | <jblack> which had me convinced, along with the monitor indicator changing color, that it's putting out a mode that the monitor can't handle.
| |
| 03:45 | <ogra> you can try ddcprobe or xresprobe to see what the montor tells X about its capabilities
| |
| 03:45 | <jblack> however, setting XRESOLUTION or whatever down to 800x600 didn't seem to have the expected result. lts.conf seemed ignored.
| |
| 03:45 | <ogra> X will only pick up what it gets from there
| |
| 03:45 | X_MODE_0=800x600 is the right option
| |
| 03:46 | and the lts.conf sits in the tftp dir nowadays
| |
| 03:46 | <jblack> Yeah.
| |
| 03:46 | I set it according to documentation Ifound online. It was a few hours ago, so I'm not longer clear about what I set.
| |
| 03:48 | I'll hit it again tomorrow night.
| |
| 03:49 | ogra has quit IRC | |
| 03:50 | indradg_ has quit IRC | |
| 03:51 | ogra has joined #ltsp | |
| 03:51 | * ogra curses daily reconnects | |
| 03:52 | <Q-FUNK> daily reconnects?
| |
| 03:52 | <ogra> yeah, once a day my line drops
| |
| 03:52 | common german practice
| |
| 03:52 | <jblack> heh
| |
| 04:41 | <laga> very silly, though
| |
| 04:41 | <ogra> yeah, and i even pay a business line, not the common dsl crap you get here
| |
| 04:43 | <laga> get a static IP, the disconnects don't matter that much then
| |
| 04:43 | <ogra> i have a static IP
| |
| 04:44 | <laga> oh :/
| |
| 04:44 | <ogra> its still annoying since my builder scripts that upload stuff run ver night i have to make sure the reconnect happens thorugh my business hours
| |
| 04:45 | <laga> i've got a semi-static IP and no disconnects. although customer support is a bit lacking sometimes, i'm quite satisfied with my cable line ;)
| |
| 04:46 | <ogra> yeah, cable in .de rocks .... they usually overzize their backbones by design so you rarely get slowdowns or anything
| |
| 04:50 | tsurc has joined #ltsp | |
| 04:51 | indradg has joined #ltsp | |
| 04:58 | ace_suares has joined #ltsp | |
| 04:59 | <ace_suares> gooood morninggg !
| |
| 04:59 | !seen ogra
| |
| 04:59 | <ltspbot`> ace_suares: ogra was last seen in #ltsp 12 minutes and 55 seconds ago: <ogra> yeah, cable in .de rocks .... they usually overzize their backbones by design so you rarely get slowdowns or anything
| |
| 04:59 | <ace_suares> ogra: is that security issue as nasty as I think it is !?
| |
| 05:00 | <ogra> ace_suares, yes, but its fixed :)
| |
| 05:00 | instructions are in the USN
| |
| 05:01 | http://www.ubuntu.com/usn/usn-610-1
| |
| 05:01 | <ace_suares> yeah great I just saw that!
| |
| 05:01 | Wow !
| |
| 05:01 | okay, well that's a first, isn't it...
| |
| 05:01 | <ogra> send your beer or flowers to daduke :)
| |
| 05:02 | <ace_suares> daduke = chris herzog ?
| |
| 05:02 | yeah. Kudo's daduke !
| |
| 05:02 | <ogra> noe of the ltsp sevs that regulary looked at the code ever saw it .... and it was relly obvious
| |
| 05:02 | *none
| |
| 05:02 | *devs
| |
| 05:02 | <ace_suares> that's the many eyes theory
| |
| 05:02 | <ogra> heh, yeah
| |
| 05:02 | failed here apparently
| |
| 05:03 | <ace_suares> if enough eyes look at the same code the obvious errors will still get overlooked ;-)
| |
| 05:03 | in the end it didn't fail tough
| |
| 05:03 | <ogra> well, we have more eyes now with gentoo, opensuse and fedora entering the game
| |
| 05:03 | <ace_suares> oh ?! Great!
| |
| 05:07 | <tsurc> hows the best way to go about diagnosing dhcpd weirdness. I've a device on our network that is giving out discovery packets but the dhcpd server isn't offering a lease even though i've configured it to assign a specific ip.
| |
| 05:08 | <ace_suares> ogra, as far as I understand, an automatic upgrade doesn't do the trick...
| |
| 05:08 | so, should there be another way to alert possible system managers ?
| |
| 05:08 | like on the homepage of edubuntu ?
| |
| 05:09 | just asking.
| |
| 05:12 | <The_Code> cyberorg: thanks for making this clear
| |
| 05:13 | <cyberorg> The_Code, hiya
| |
| 05:13 | ogra, meet The_Code our SOC guy working on LTSP GUI
| |
| 05:13 | ace_suares has quit IRC | |
| 05:14 | <The_Code> cyberorg: what if make the setup of the server a module that can be there, but doesn't need to
| |
| 05:15 | <cyberorg> The_Code, yeah that would be good, ogra doesn't want to use it, he is more interested in lts.conf management
| |
| 05:16 | <The_Code> reading documentation i get the impression that everybody makes it in a different way
| |
| 05:16 | so if there is a script, you can install the module and setup your server with the GUI
| |
| 05:16 | if not you can edit your lts.conf
| |
| 05:16 | <cyberorg> yeah, so the GUI need to call just the scripts
| |
| 05:17 | <The_Code> what scripts you mean
| |
| 05:17 | <cyberorg> The_Code, no, lts.conf doesn't have a gui editor yet, and lts.conf is same everywhere
| |
| 05:18 | <The_Code> if there is no install script you don't install the module for setting up the server
| |
| 05:18 | you can then "only" use LTSP GUI to edit and manage your lts.conf
| |
| 05:19 | <cyberorg> The_Code, i mean, ltsp-build-client would be there on all distros, ltsp-server-initialize on whoever wants it
| |
| 05:20 | The_Code, we can have SHOW_MODULES = ltsp-build-client ltsp-server-initialize lts-conf-editor in GUI's configuration
| |
| 05:20 | <The_Code> and ltsp-build-client only builds the chroot stuff?
| |
| 05:20 | <cyberorg> yes
| |
| 05:22 | <ogra> hey
| |
| 05:22 | <The_Code> that's the way we should have it
| |
| 05:22 | ogra: hi
| |
| 05:22 | * ogra still doesnt get what ltsp-server-initialize | |
| 05:22 | <ogra> shal be
| |
| 05:23 | did you have a look at python-ltsp from ubuntu ?
| |
| 05:23 | <The_Code> yes i did
| |
| 05:23 | <cyberorg> ogra, it is fedora equal of kiwi-ltsp-setup -c http://forgeftp.novell.com/kiwi-ltsp/kiwi-ltsp-diagram.png
| |
| 05:23 | <ogra> its a module that collects all possible options in a chroo so you get value lists to feed to optins
| |
| 05:23 | its slightly unmaintained though
| |
| 05:24 | <The_Code> a little bit more documentation would have made it easier to understand, but i got it
| |
| 05:24 | <cyberorg> i am sure warren would not be changing the name, so i'll have a wrapper that calls kiwi-ltsp-setup -c :)
| |
| 05:24 | <The_Code> and i think it will help me a lot
| |
| 05:25 | <ogra> The_Code, its fully documented in the python help
| |
| 05:25 | but it needs love
| |
| 05:25 | for sure
| |
| 05:25 | i didnt touch it since over a year and surly a lot has changed so it will need adjustment
| |
| 05:25 | <The_Code> sorry, but i only opened the .py file
| |
| 05:25 | <ogra> cyberorg, in deban and ubuntu we wont use something else than the distro provided tools we have already
| |
| 05:26 | i.e. i will never accept any extra feature in ltsp-manager that manages interfaces
| |
| 05:26 | <cyberorg> ogra, yeah, so we make it modular and let the distro choose to not use that module
| |
| 05:26 | <ogra> rather have a button that starts the respectve distro tool for maintining such stuff
| |
| 05:27 | all *i* need in ubuntu is an lts.conf editor .... the other distros might have different needs
| |
| 05:28 | <The_Code> this could also be on option to have an entry in gui conf telling it to just run a script without any input
| |
| 05:28 | <ogra> what i will want is a dhcpd.conf editor since ltsp in ubuntu uses a dedicated dhcpd.conf, i wuldnt accept such a tool if it touched /etc/dhcp/dhcpd.conf though but rather rely on a tool our server team provides for that
| |
| 05:28 | thats what i meant wen we initially talked about it with "lets do babysteps" :)
| |
| 05:29 | one thing we all have in common is lts.conf
| |
| 05:29 | <cyberorg> ogra, paths where you want config files placed would be customizable, so you can have GUI generate templates and use those if you wish or edit distro files manually or with distro tools
| |
| 05:29 | <ogra> and we have a peoper parser for that we use everywhere (getltscfg)
| |
| 05:29 | which is essentially an .ini file parser
| |
| 05:29 | paths should be set at build time ...
| |
| 05:30 | <The_Code> and so the lts.conf editor module will be there everytime, everything else distro dependend
| |
| 05:30 | <cyberorg> ogra, yeah, packagers can include their config files
| |
| 05:30 | <ogra> if someone wants to use nfs instead of images he will default to some path inside the chroot ... others will want it from tftp by default
| |
| 05:30 | The_Code, right
| |
| 05:31 | make it modular enough that distros can easily plug in their extras
| |
| 05:31 | <The_Code> this is the goal from now on
| |
| 05:31 | <ogra> but for a start (and i know how short SoC is) lets concentrate on lts.onf and have a proper plugin setup
| |
| 05:32 | * ogra has unfinished SoC projects all over ubuntu from the last two years that didnt finish because they wanted to implement the world... rather starts small :) | |
| 05:32 | <ogra> s/starts/start/
| |
| 05:33 | <cyberorg> ogra, other things like ltsp-build-client and ltsp-server-initialize are nothing but buttonts that call shell scripts
| |
| 05:33 | *buttons
| |
| 05:33 | <The_Code> configured in the gui conf
| |
| 05:34 | <ogra> cyberorg, right, something like that
| |
| 05:35 | the intresting bit will be the setup for multiple servers, you need a very good front/backend design for that
| |
| 05:35 | <cyberorg> ogra, we are not even thinking about that now :)
| |
| 05:35 | <ogra> so try to design the app from the begining into a service that listens for tls encrypted connects and a frntend
| |
| 05:36 | for the beginning just as a loopback thing
| |
| 05:36 | cyberorg, it will get in your way if you dont do it from the beginning
| |
| 05:36 | thats essentially why i dropped work on ltsp-manager
| |
| 05:37 | <cyberorg> lts.conf manager that has agents running on other servers?
| |
| 05:37 | <ogra> its design would never allow more than loca management and it would be way to much effort to redesign (a rewrite is faster here)
| |
| 05:37 | cyberorg, no, it hasnt, thats wat i say
| |
| 05:37 | its a design failure
| |
| 05:37 | take that into account from the beginnign
| |
| 05:38 | <cyberorg> i mean, we design this GUI with agents running on other servers in mind ?
| |
| 05:38 | <ogra> split ui and the actual code that touches the config in two apps that talk to each other
| |
| 05:38 | for the start just talking over loopback
| |
| 05:39 | <cyberorg> hmm, good idea
| |
| 05:39 | <ogra> but with the opportunity to put tls encrypted remote cnnections in place later
| |
| 05:39 | <The_Code> so that i can have the conf app on another server and talk to it?
| |
| 05:39 | <ogra> else you end up with something like ltsp-manager
| |
| 05:40 | The_Code, right, have a UI part and a client that listens on a port
| |
| 05:40 | <The_Code> the world is growing
| |
| 05:40 | <ogra> as long as its loopback you can likely omit any encryption and stuff
| |
| 05:41 | <cyberorg> The_Code, your world will grow very rapidly over next few months :)
| |
| 05:41 | <ogra> but it should be expandable by that
| |
| 05:41 | The_Code, thats why i say keep the list of features as small as possible
| |
| 05:41 | the planning and speccing is the essdential bit
| |
| 05:41 | *essential
| |
| 05:42 | <The_Code> encryption as an option first only loopback, on schedule
| |
| 05:43 | <ogra> and for the UI, do tests with your mother while developing it ;)
| |
| 05:43 | <The_Code> will do
| |
| 05:43 | <ogra> ask her to enable some certain thing in teh config and see if she manages , the your UI is good
| |
| 05:44 | s/the/then/
| |
| 05:44 | <cyberorg> :)
| |
| 05:44 | <ogra> (doesnt essentially need to be your mother indeed ;) but you know what i mean, a computer illiterate person for testing)
| |
| 05:45 | <The_Code> ogra: i got you
| |
| 05:45 | <ogra> ltsp is used by many teachers with no or not much knowledge about computers
| |
| 05:46 | <cyberorg> ogra, i have conveyed that thought already :)
| |
| 05:46 | <The_Code> i know a school teacher, i will ask him to test when the time has come
| |
| 05:46 | <ogra> great
| |
| 05:53 | * The_Code has to go to university now | |
| 05:59 | Pascal_1 has joined #ltsp | |
| 06:02 | daya has quit IRC | |
| 06:08 | <ogra> cyberorg, oh, and pretty please dont use XML (i just saw the ML discussion) lts.conf is an ini file, no need to explicitly use a format conversion back and forth here
| |
| 06:09 | <cyberorg> ogra, how do you suggest we handle backups, rollbacks, porting it to another server etc?
| |
| 06:10 | <ogra> bzr for rollbacks, tgz for backups and porting over
| |
| 06:10 | no need to add compilcated extras *just because we can*
| |
| 06:11 | keep it easy ad at a level that you can still hadle it from commandline if something fails
| |
| 06:11 | lts.conf has the right format already, no need for additional complication layers on top of that
| |
| 06:12 | <cyberorg> that is the idea, commandline and directly editing any file would be possible, and XML backend will sync with whatever manual modifications
| |
| 06:12 | <ogra> thats why scott selected an ini file parser to handle it
| |
| 06:12 | no, keep both at the same level
| |
| 06:12 | <cyberorg> we are not use XML just for lts.conf changes, we will be storing stuff we do to other config files too
| |
| 06:13 | * ogra hopes that isnt true | |
| 06:13 | <ogra> what *other* config files ?
| |
| 06:13 | <cyberorg> dhcpd.conf?
| |
| 06:13 | <ogra> i surely dont want the tool to touch *any* of the system files
| |
| 06:13 | dhcpd.conf only if its dedicated
| |
| 06:14 | <cyberorg> ogra, of course, you can choose just to create templates and modify distro files manually
| |
| 06:14 | <ogra> like we do in ubuntu with a sepearate dhcpd.conf that overrides the one in /etc/dhcp
| |
| 06:14 | itais has left #ltsp | |
| 06:16 | <cyberorg> bzr will not give flexibility of reverting one field or per client basis, we cannot group clients etc
| |
| 06:17 | nor can we copy over one client's config to selected clients
| |
| 06:17 | <ogra> sigh
| |
| 06:17 | why would you copy over anything to a single client
| |
| 06:17 | its read from lts.conf
| |
| 06:18 | it might make sense to split it up on a per client base vs general server settings, but i'm totally resstant to the XML idea
| |
| 06:19 | <cyberorg> ogra, for example if user wants to multiple clients grouped, so any change made to that group applies to all?
| |
| 06:19 | <ogra> thats fine
| |
| 06:19 | as i said, split the file
| |
| 06:20 | put it these sit out pieces in group dedicated subdirs ... at the end assemble a stadard lts.conf from it
| |
| 06:20 | have bzr maintain the whole dir setup so you are able to roll back single configs
| |
| 06:21 | trivial and straightforward
| |
| 06:21 | <cyberorg> ogra, wouldn't XML simplify that rather than split into folders and merge?
| |
| 06:21 | <ogra> not for proper rollback
| |
| 06:21 | and you give the admi an easy to understand setup
| |
| 06:22 | subir has quit IRC | |
| 06:23 | hersonls has joined #ltsp | |
| 06:23 | <cyberorg> ogra, that would require the GUI to become version control client too, it would be easier for the gui to keep its own storage and do rollbacks/reverts/exports
| |
| 06:25 | we can list out pros/cons and take it from there :)
| |
| 06:28 | <ogra> running bzr commit in the backend doesnt require the gui to be a version control client
| |
| 06:28 | <cyberorg> ogra, for reverting to last change or selected change does?
| |
| 06:29 | <ogra> the GUI just needs a "rollback" button going back a step at a time with the option to pick single client configs
| |
| 06:30 | keep it simple
| |
| 06:30 | dont overcomplicate something thats already easy
| |
| 06:30 | lts.conf format was chosen for a reason
| |
| 06:31 | <cyberorg> i'd be saving this log for The_Code to read later :)
| |
| 06:31 | * ogra doesnt want a yast in ltsp | |
| 06:31 | <cyberorg> yast doesn't use xml
| |
| 06:32 | <ogra> but it duplicates a lot of stuff that shuld be better handled directly
| |
| 06:32 | <cyberorg> it would be easier to do this in yast, we get ncurses mode for free :)
| |
| 06:32 | <ogra> whiptail isnt tat hard to use :)
| |
| 06:33 | <cyberorg> ogra, i agree about the duplication, but we want the school teacher who doesn't know anything about computers to be able to set up a server
| |
| 06:33 | <ogra> right
| |
| 06:33 | which he should be able to do *without* and lts.conf anyway
| |
| 06:33 | <cyberorg> i don't want to point them to different tools to set up different services required to get ltsp running on their distro
| |
| 06:34 | <ogra> ltsp is moving towards no lts.conf since ltsp5 exists
| |
| 06:34 | thats one of the main targets
| |
| 06:34 | <cyberorg> yeah, that is one good thing, i said that in last meeting, lts.conf would be minimal skeleton
| |
| 06:34 | <ogra> you dont need to set up anything in a ubuntu install nor in a debian install
| |
| 06:34 | by design
| |
| 06:35 | <cyberorg> ogra, network, dhcp, nfs (if required), firewall ports, activating xinetd services - tftp etc is required
| |
| 06:35 | <ogra> if you guys in suse or fedora decide to add extra stuff on top, like server side initscripts, go for it... you wont see me using such stuff
| |
| 06:35 | basanta has joined #ltsp | |
| 06:36 | <ogra> cyberorg, nothing of the above needs to be touched for an initial ubuntu setup
| |
| 06:36 | its all dont by the packages
| |
| 06:37 | <cyberorg> ogra, you can't have services running all by themselves and firewall ports open without user telling the system?
| |
| 06:37 | <ogra> and focus of ltsp5 is to integrate it *with the distro* if that means changesw to the nfs server package of your distro these have to be dont first instead of adding additional scripts or stuff
| |
| 06:37 | s/dont/done/
| |
| 06:38 | thats the design of ltsp5 ... if you want to change it for your specific distro, feel free, but the definition says *integrate with the existing distro packages*
| |
| 06:39 | that also applies for the server side, not only for the client chroot
| |
| 06:39 | the_code_ has joined #ltsp | |
| 06:39 | basanta has quit IRC | |
| 06:41 | <ogra> cyberorg, things like nbdrootd and nbdswapd are already violating that for example ... both should be functions nbd-server should be able to provide on its own ...
| |
| 06:42 | <cyberorg> ogra, yeah, we install both of those enabled by default, not sure it would get past suse's security team
| |
| 06:43 | <ogra> see, there we have a fundamental distro difference :)
| |
| 06:43 | <cyberorg> ogra, fedora also doesn't allow any xinetd service to be on upon installation
| |
| 06:43 | <ogra> in deb based systems policy says a service needs to run immediately after install in a default config :)
| |
| 06:44 | so i understand that you need something to enable it in your distro (and fedora too)
| |
| 06:44 | still, such things arent necessary in ubuntu or debian, so we'll never have it
| |
| 06:45 | <cyberorg> the_code_, http://pastebin.com/d1869811b ogra's views on XML
| |
| 06:45 | the_code_ has quit IRC | |
| 06:46 | <cyberorg> ogra, we'll make it modular, so the GUI shows only things that packager want to be seen
| |
| 06:46 | <ogra> yes, yu will need a lot of build time setup to please everyone
| |
| 06:48 | <cyberorg> a config file ;)
| |
| 06:48 | <ogra> thats why i say start off only with lts.conf :) getting that properly specced will eat half of your SoC time already if you want to do it right :)
| |
| 06:48 | Pascal_1 has joined #ltsp | |
| 06:48 | <ogra> right, something like that
| |
| 06:49 | but a config file implies you wil need a lot of special casing in your code as well
| |
| 06:49 | which is hard to get right in a fashion that it stays maintainable later if more features come
| |
| 06:49 | <cyberorg> ogra, no, we leave all that to distro's script to handle
| |
| 06:50 | except of lts.conf configuration that would be handled by the GUI everything else would be by distro's scripts if required by them
| |
| 06:50 | <ogra> that wont work ... one distro will handle things differently to the other .... so you always need something that hooks in to the different handlers in your code
| |
| 06:50 | and that needs to be maintained
| |
| 06:51 | the more special cases you have the more extra work you have to do if you change a single option
| |
| 06:51 | <cyberorg> ogra, names of the scripts would be same, there are just two scripts ltsp-build-client and ltsp-server-initialize, rest is all lts.conf management
| |
| 06:51 | <ogra> well
| |
| 06:52 | <cyberorg> ogra, we wont get into single option configuration of various other config files
| |
| 06:52 | for that distro tools should be used or vi ;)
| |
| 06:52 | <ogra> you will want to have a statusbar in your gui showing a lottle green or red dot for example to indicate all server processes are fine
| |
| 06:52 | upstart handles status differently to sysvinit
| |
| 06:52 | sysvinit handles status differently in fedora than in debian
| |
| 06:53 | etc
| |
| 06:53 | <cyberorg> ogra, those things we would do it in the end after we have lts.conf worked out nicely
| |
| 06:54 | netstat would work same on all distros?
| |
| 06:54 | <ogra> in ubuntu we ship a service manager by default, i would only want a button that starts this app to manage services and a change to the service app that allows to only show related sevices
| |
| 06:54 | suse might want to start yast for that
| |
| 06:54 | etc
| |
| 06:54 | all this needs to be reflected in your code
| |
| 06:55 | if you dont want to drive it against the wall in two years because it became unmaintainable, goos planning i the most essential thing here
| |
| 06:55 | s/goos/good/
| |
| 06:55 | <cyberorg> ogra, you misunderstood this part, we are just going to put a button that runs shell scripts to set up services, not create something that lets user manage those
| |
| 06:56 | <ogra> right, wont work for me
| |
| 06:56 | my services are set up already
| |
| 06:56 | <cyberorg> so you can leave those buttons out in your package :)
| |
| 06:56 | <ogra> all i need is something to montor or start/stop them
| |
| 06:56 | the_code_ has joined #ltsp | |
| 06:57 | <ogra> just try to not underestimate maintenance of what you design is what i say
| |
| 06:57 | <cyberorg> we are not planning anything complicated, just call some shell scripts, for monitoring if you want we can put a button that calls ltsp-server-status which each distros would have to write
| |
| 06:58 | <ogra> that should be built into the backend
| |
| 06:58 | i.e. ltsp-manger-backend-query --status :)
| |
| 06:59 | <cyberorg> ogra, we would just be making use of two scripts that we have now, nothing fancy, those scripts are not installed/maintained by the GUI
| |
| 06:59 | deavid has quit IRC | |
| 06:59 | <ogra> no, the gui (in my example) should just call the bove command and get true or false back :)
| |
| 06:59 | and show something according to that
| |
| 07:00 | but your backend will need to handle the different status systems
| |
| 07:00 | like upstart, sysv or whatever else
| |
| 07:00 | (will become fun once someone starts to implement ltsp5 on slackware :) )
| |
| 07:00 | <cyberorg> ogra, yeah, it could if such script already exist, it would be beyond the scope of this SOC to write and maintain it for all distros
| |
| 07:01 | hersonls has quit IRC | |
| 07:01 | <ogra> no, but the code produced by this SoC should result in something thats usable for enhancement so such things need to be taken into account at plannin phase
| |
| 07:02 | thats why i say planning will be 80% of the SoC work to get it right
| |
| 07:02 | <cyberorg> ogra, yup, thats true, but the prime goal is lts.conf management, everything else is bonus
| |
| 07:02 | <ogra> acking that together in a usable manner takes an experienced guy a week
| |
| 07:02 | (i wrote ltsp-manager as it in in two days)
| |
| 07:02 | s/as it/as is/
| |
| 07:03 | <cyberorg> we've got till september :)
| |
| 07:03 | <ogra> right
| |
| 07:03 | so put some thought into the design so we can still use it in three years without getting a headdache if someone wants to change a single feautre or add another distro
| |
| 07:04 | thats all i'm saying :)
| |
| 07:05 | <cyberorg> yeah, that is why we intend to use shell scrips for distro specific stuff, all distros can add their own if required
| |
| 07:05 | <the_code_> ogra: will do it modular so that you can change the part you want to change
| |
| 07:05 | <ogra> ubuntu has only a handfull of flavours (k/ed/x/ubuntu) you wouldnt belive how many manhours it burns to change something that applies to all of them
| |
| 07:06 | anyway, i need to get some merge wok done :)
| |
| 07:07 | *work
| |
| 07:07 | <cyberorg> ogra, thanks for all your great inputs :)
| |
| 07:08 | <ogra> lets see what comes out :)
| |
| 07:16 | jammcq has quit IRC | |
| 07:17 | the_code_ has quit IRC | |
| 07:18 | the_code_ has joined #ltsp | |
| 07:22 | indradg has quit IRC | |
| 07:24 | TelnetManta has joined #ltsp | |
| 07:25 | hersonls has joined #ltsp | |
| 07:27 | exodos has joined #ltsp | |
| 07:31 | sepski has joined #ltsp | |
| 07:32 | <sepski> i am testing sound on a thin client, but i get the error sound requested but /dev/dsp missing,
| |
| 07:32 | <tarzeau> sepski: you need to use ALSA not OSS
| |
| 07:33 | <sepski> i only have SOUND=Y in lts.conf
| |
| 07:34 | so the error is irrelevant then
| |
| 07:35 | <tarzeau> which program?
| |
| 07:35 | you need to tell your program to not use OSS
| |
| 07:35 | <sepski> i just tried kde initially
| |
| 07:36 | and the error i mentiond was from thin client boot up
| |
| 07:38 | <ogra> is that ltsp5 ?
| |
| 07:43 | <sepski> yes. but it seams the user i tested with had edited kde sound settings, worked on another user
| |
| 07:44 | <ogra> kde needs to have "use arts through esd" or something like that in kcontrol iirc
| |
| 07:45 | Pascal_1 has quit IRC | |
| 07:45 | <ogra> the dsp message is irrelevant ... its from ancient times where oss was used
| |
| 07:45 | <sepski> good to know
| |
| 07:45 | <ogra> you will likey see it on every client nowadays since we use alsa by default
| |
| 07:46 | <sepski> atlest its not wrong just confusing :)
| |
| 07:47 | <ogra> yeah
| |
| 07:47 | * ogra doesnt have these probs in ubuntu, users by default only see usplash :) | |
| 07:48 | <ogra> so i dont really care about obsolete bootmsgs :)
| |
| 07:51 | <sepski> ogra, we have usplash as well, but i read syslog :)
| |
| 07:51 | since i had no sound on a test users
| |
| 07:51 | <ogra> ah :)
| |
| 07:52 | <sepski> thanks for the clearifications all, guess i somehow have to check the sound settings of 700 users...
| |
| 07:53 | sepski has quit IRC | |
| 07:54 | mhterres has joined #ltsp | |
| 08:02 | bobby_C has joined #ltsp | |
| 08:03 | GiantPickle has joined #ltsp | |
| 08:08 | slidesinger has joined #ltsp | |
| 08:12 | soneyka has quit IRC | |
| 08:14 | Nuba1 has joined #ltsp | |
| 08:14 | soneyka has joined #ltsp | |
| 08:16 | soneyka has quit IRC | |
| 08:17 | soneyka has joined #ltsp | |
| 08:19 | <Nuba1> is there a reason why the user-manager would not seem to work? I'm trying to change user settings on the server directly, and any changes are not reflected in passwd and shadow... where does user-manager store its data?
| |
| 08:23 | soneyka has quit IRC | |
| 08:24 | soneyka has joined #ltsp | |
| 08:45 | Nuba1 has quit IRC | |
| 08:50 | Pascal_1 has joined #ltsp | |
| 08:51 | Pascal_1 has quit IRC | |
| 08:55 | mccann has joined #ltsp | |
| 09:02 | Nuba1 has joined #ltsp | |
| 09:07 | Gadi has joined #ltsp | |
| 09:12 | the_code_ has quit IRC | |
| 09:16 | indradg has joined #ltsp | |
| 09:29 | Q-FUNK has quit IRC | |
| 09:45 | soneyka has quit IRC | |
| 09:45 | K_O-Gnom has joined #ltsp | |
| 09:46 | soneyka has joined #ltsp | |
| 09:46 | Q-FUNK has joined #ltsp | |
| 09:48 | deavid has joined #ltsp | |
| 09:48 | soneyka has quit IRC | |
| 09:49 | soneyka has joined #ltsp | |
| 09:54 | jammcq has joined #ltsp | |
| 10:03 | tsurc has quit IRC | |
| 10:04 | hersonls has quit IRC | |
| 10:05 | * gvy is back (gone 29:53:58) | |
| 10:06 | cliebow has joined #ltsp | |
| 10:08 | achandrashekar has quit IRC | |
| 10:12 | mccann_ has joined #ltsp | |
| 10:14 | The_Code has quit IRC | |
| 10:15 | mccann has quit IRC | |
| 10:19 | primeministerp has joined #ltsp | |
| 10:23 | achandrashekar has joined #ltsp | |
| 10:25 | gonzaloaf_work has quit IRC | |
| 10:25 | <achandrashekar> hello. looking for some help with an ltsp roll out for 50 clients p3s with 383 megs of ram. Anyhow. we have 2 athlon x2 systems available and a 10/100 setup. However, in the initial setup of on emachine with 2gb of ram we get very slow response when using edubuntu 8.04. Ive hear dof mille-xterm, but is there a way to utilize what we have to roll out more nodes, and not diminish performance?
| |
| 10:26 | i suppsoe its an architectural question.
| |
| 10:27 | toscalix has quit IRC | |
| 10:31 | <cyberorg> achandrashekar, add more ram to the server
| |
| 10:32 | BGomes has joined #ltsp | |
| 10:33 | <cyberorg> achandrashekar, while you are testing, give prebuilt image from http://en.opensuse.org/LTSP a run too, I would like to know how it handles many clients
| |
| 10:33 | staffencasa has joined #ltsp | |
| 10:33 | <achandrashekar> cyberorg: im, really loooking for a cluster type setup...and mosix is dead
| |
| 10:33 | andax has joined #ltsp | |
| 10:34 | <achandrashekar> cyberorg: i think mille-xterm is there...but besides dhcp-failvoer which is documented...what else is out there to use the computing power of the nodes??
| |
| 10:34 | <cyberorg> ah, but a single server with about 3GB ram should be powerful enough to run 30 - 40 TC
| |
| 10:34 | <achandrashekar> this is serious roll out...not just a measly 10 systems and call it a day roll out.
| |
| 10:35 | cyberorg: image all systems being hit at the same time with kids hammering away at it in openoffice...recipe for disaster.
| |
| 10:35 | image=imagine...sorry
| |
| 10:35 | <cyberorg> achandrashekar, RAM is the only constraint, processors is ok
| |
| 10:36 | <achandrashekar> okay..the damn systems only have two banks..perhaps what i can do is to upgrade to 4gb..i think though the kernel may have a limit and ill have to do something special..in edubuntu
| |
| 10:36 | and try again..
| |
| 10:37 | <cyberorg> one of the ltsp 4.2 set up we have here is over 40 clients on p4 server with 2 Gig ram
| |
| 10:37 | <achandrashekar> is that running k12ltsp?
| |
| 10:37 | and running gnome or kde?? or jsut simple x terminal?
| |
| 10:37 | <cyberorg> achandrashekar, if you go for 4GB you need x86_64 OS to take full advantage of it
| |
| 10:38 | <achandrashekar> yeah...and then build-client for i386...of which flash and java issues
| |
| 10:38 | darn.
| |
| 10:38 | <cyberorg> achandrashekar, it is running fedora gnome, full office use
| |
| 10:38 | <achandrashekar> and usage of clients per machine...at the same time?
| |
| 10:38 | <cyberorg> you can first try with 3GB and see how it goes
| |
| 10:39 | <achandrashekar> okay...might just have to go that route..
| |
| 10:39 | something really wierd though...if and when a person hits a flash site...it kills the damn systems
| |
| 10:39 | flash is a resource hog
| |
| 10:40 | and you are absolutely correct about the memory allocation
| |
| 10:40 | cpu is not affected...but the memory takes a hit.
| |
| 10:40 | I show of the 2gb 383Mib being used at all times.
| |
| 10:40 | is that correct??
| |
| 10:40 | <cyberorg> firefox is a memory hog too, although 3 is ok
| |
| 10:41 | <achandrashekar> yes...im on 8.04, and even with 3, i only had 4 systems fired up...and the system was coughing...
| |
| 10:41 | <cyberorg> achandrashekar, the way linux works is it always uses up all available memory, most of it in cache and buffer, so don't worry if it is always 100% used
| |
| 10:42 | <achandrashekar> i see... okay
| |
| 10:42 | its the performance that is scaring me....let me give you an example...
| |
| 10:42 | i addded an account
| |
| 10:42 | <cyberorg> i just ran 11 clients on 2 GB machine and did a openoffice training for one client, all clients were hammering away on ooo
| |
| 10:42 | <achandrashekar> went to a node...logged in no problem
| |
| 10:42 | fire up 6 more nodes..
| |
| 10:43 | go to one of other nodes... 4 mins to simply login
| |
| 10:43 | just seemed like an inordinate amount of time..
| |
| 10:43 | and then from there performance browsing...etc...was excrutiatingly slow
| |
| 10:44 | so that is why i figured..what are my options here..or am i doing something wrong..
| |
| 10:44 | and in fact it may be memory.
| |
| 10:44 | <cyberorg> achandrashekar, make sure you don't have any network bottlenecks, like bad switch/cables
| |
| 10:44 | <achandrashekar> and in fact this could be..
| |
| 10:44 | everything is connected to a 10/100 3com switch.
| |
| 10:45 | all nodes at 100full
| |
| 10:45 | but the cabling etc...maybe i need to go through each ad every-single node.
| |
| 10:46 | <cyberorg> i'd give 150M ram on server per client, it performs extremely well, processor is not such a big constraint
| |
| 10:46 | <achandrashekar> anyhow...from what you are saying with about 3 gb of memory..i can still run the i386 server version...and go from there...
| |
| 10:47 | cyberorg...how do i set the 150m per ram per system bit...
| |
| 10:47 | <cyberorg> achandrashekar, you don't just rule of thumb advise :)
| |
| 10:48 | <achandrashekar> i see...so its a "rule"..not a setting ;)
| |
| 10:48 | lol
| |
| 10:48 | okay...i thought you could actually allocate that...
| |
| 10:49 | cyberorg: any issues with running the i386 version on a x64bit system?i was advised to go with the i386 build for flash and java purposes.
| |
| 10:50 | <cyberorg> achandrashekar, no issue at all
| |
| 10:51 | <achandrashekar> okay...I will try once more...I do think though that you may be absolutely right about the network faux-pas that we have....and i will test every line.
| |
| 10:54 | thanks for your help
| |
| 10:54 | as for large infrastructure build..what clustering options are there though?
| |
| 10:56 | <ogra> http://doc.ubuntu.com/edubuntu/edubuntu/handbook/C/multiple-server-setup.html
| |
| 10:56 | amd64 and flash/java isnt an issue anymore in 8.04, nspluginwrapper should handle that flawlessly
| |
| 10:57 | <achandrashekar> ogra: yep read that one...with dhcp-failover correct.. that is at the server level....what about the nodes?
| |
| 10:57 | <ogra> what about them ?
| |
| 10:57 | instead of failover just use a roud-robn mechanism that connects every other client to a different server
| |
| 10:57 | <achandrashekar> ogra: my experience with nspluginwrapper...ran very slow with the 64bit operating system...so i rolled back to i386....and no issues.
| |
| 10:58 | <ogra> hmm
| |
| 10:58 | <achandrashekar> at least with one node...that is..
| |
| 10:58 | and as for th multi-server setup...the guide is excellent..
| |
| 10:58 | cliebow has quit IRC | |
| 10:58 | <achandrashekar> i did that also in one lab...
| |
| 10:58 | <ogra> well, if you want to use i386 with so many clients, use in any case the -server kernel package
| |
| 10:58 | it adresses above 3G
| |
| 10:58 | <achandrashekar> but the smb-ldap portion is riddled with mistakes..
| |
| 10:59 | <ogra> its outdated (sorry for that)
| |
| 10:59 | <achandrashekar> and the auto-installer...that the guide speaks of breaks the authentication..because new 8.04,7.10 configs use ldap.conf now.
| |
| 10:59 | <ogra> i'll meet moquist in prague in two weeks, we'll discuss that
| |
| 11:00 | (at UDS)
| |
| 11:00 | <achandrashekar> ogra: no issues...the other guide...that is on source-forge for smb-ldap for edubuntu is solid
| |
| 11:00 | <ogra> ah, k
| |
| 11:00 | <achandrashekar> ogra: id let him know to simply re-write the script for ubuntu to use that.
| |
| 11:00 | <ogra> i think hs time is very limited
| |
| 11:01 | thats the reason for it lagging behind
| |
| 11:01 | <achandrashekar> ogra: i really wish though...that some focus would be given BACK to the nodes...rather the server..with open-mosix being dead..someone has to capitilize on the nodes again..which often just sit..
| |
| 11:01 | rather than the server that is..
| |
| 11:02 | <ogra> well, we will meet in portland for the ltsp hackfest in july, i hope we can focus at localapps for that
| |
| 11:02 | <achandrashekar> there seems to have been some work done early on...but the project is either dead...or the mille-xterm seems to be the one to use...but im no good with frecnh :)
| |
| 11:02 | <ogra> there is proof of concept work done already
| |
| 11:02 | it just needs proper implementation
| |
| 11:02 | BGome1 has joined #ltsp | |
| 11:02 | BGomes has quit IRC | |
| 11:03 | <ogra> achandrashekar, fgiraldeau speaks english very well :) and i think he'd be happy to help
| |
| 11:03 | <achandrashekar> ogra: yep..i agree...the concept is LARGE infrastructure roll .
| |
| 11:03 | does he frequent?
| |
| 11:03 | <ogra> 50 clients isnt large :)
| |
| 11:03 | <achandrashekar> this channel?
| |
| 11:03 | <ogra> he is here from time to time but not an actual IRC fan
| |
| 11:04 | <achandrashekar> yeah..i have to be honest with you its for a large csu university..
| |
| 11:04 | and we are looking to do a roll out of over 1000 clients. :)
| |
| 11:04 | <ogra> see pm
| |
| 11:04 | <achandrashekar> k
| |
| 11:04 | got it thanks.
| |
| 11:04 | <ogra> you cn mail him, i think he'll happily help you
| |
| 11:04 | <achandrashekar> awesome...will do.
| |
| 11:05 | <ogra> mille just operates on top of the current ltsp5 infrastructure, its quite good for big rllouts
| |
| 11:05 | (and i'm talking five digits and above here :) )
| |
| 11:05 | <achandrashekar> wow
| |
| 11:05 | gonzaloaf_work has joined #ltsp | |
| 11:06 | <achandrashekar> i will definitely give him a call then.
| |
| 11:06 | brb
| |
| 11:06 | achandrashekar has quit IRC | |
| 11:08 | Pascal_1 has joined #ltsp | |
| 11:13 | Pascal_1 has quit IRC | |
| 11:15 | <Nuba1> so I just set up seperate exam profiles using sabayon and pessulus and it all worked wonderfully, I'm impressed
| |
| 11:16 | <ogra> Nuba1, send flowers to johnny he put quite some work into sabayon during hardy development
| |
| 11:16 | <Nuba1> its a bit fragile, but in the end it works...
| |
| 11:16 | <johnny> Nuba1, i'd love for it to be less fragile.i just haven't been able to afford to put more time into it
| |
| 11:17 | let's hope i can put a bit more effort into it before 2.24 release
| |
| 11:17 | backports should be possible luckily
| |
| 11:17 | <Nuba1> yeah, think its also because there's a lot for it to do on multiple accounts, so it tends to hang a bit
| |
| 11:17 | <johnny> as long as gio/gvfs bindings come out before 2.24
| |
| 11:17 | for python
| |
| 11:18 | Nuba1, i'll see what i can do to get it to use xephyr instead of xnest
| |
| 11:18 | <Nuba1> but important thing is it works, I was able to create a seperate profiles for our exam candidates with very specific profiles and it all worked ok
| |
| 11:19 | yeah xnest is what ends up get zombied sessions
| |
| 11:20 | <johnny> this is funny ..
| |
| 11:20 | David sez, "Someone put a plastic bag over the camera cluster on one of Google's Street View cars, and the resulting imagery seems to have made it right through to Google Maps."
| |
| 11:20 | exodos has quit IRC | |
| 11:35 | BGome1 has quit IRC | |
| 11:35 | BGomes has joined #ltsp | |
| 11:36 | sepski has joined #ltsp | |
| 11:44 | BGomes has quit IRC | |
| 11:45 | gbolte has joined #ltsp | |
| 11:48 | K_O-Gnom has quit IRC | |
| 11:54 | mccann_ has quit IRC | |
| 12:00 | Q-FUNK has quit IRC | |
| 12:03 | BGomes has joined #ltsp | |
| 12:07 | deavid has quit IRC | |
| 12:08 | mhterres has quit IRC | |
| 12:08 | mhterres has joined #ltsp | |
| 12:12 | hersonls has joined #ltsp | |
| 12:15 | mccann has joined #ltsp | |
| 12:23 | <johnny> is there any way i can convice you guys to make getltscfg not give a syntax error on an empty config file?
| |
| 12:23 | <ogra> johnny, send a patch :)
| |
| 12:23 | i wrapped some script magic around it in hardy to not call it at all on empty files as a wrokaround
| |
| 12:24 | <johnny> aha.. no wonder it works :)
| |
| 12:24 | lol
| |
| 12:24 | <ogra> but indeed that doesnt fix the actual issue
| |
| 12:24 | <johnny> cuz i'm having trouble with it grabbing my tftpboot lts.conf
| |
| 12:24 | i had to put it in the chroot again
| |
| 12:24 | altho that is unchanged from gutsy
| |
| 12:25 | time for a quick rant
| |
| 12:25 | these guys are TOTAL BASTARDS
| |
| 12:25 | this is just pure evil
| |
| 12:25 | Computer attacks typically don't inflict physical pain on their victims. But hackers recently bombarded the Epilepsy Foundation's Web site with hundreds of pictures and links to pages with rapidly flashing images. The breach triggered severe migraines and near-seizure reactions in some site visitors who viewed the images.
| |
| 12:25 | <ogra> # Read in lts.conf entries.
| |
| 12:25 | if [ -f /etc/lts.conf ] && [ -n "$(grep -v ^# /etc/lts.conf)" ]; then
| |
| 12:25 | eval $(getltscfg -a) || true
| |
| 12:25 | fi
| |
| 12:26 | the if wrapper is what you want
| |
| 12:26 | especially the second check
| |
| 12:26 | <johnny> yes
| |
| 12:26 | <ogra> hmm, looking at the the first check is moot :)
| |
| 12:27 | not needed at all since grep will return false anyway if the file is missing
| |
| 12:31 | <johnny> actually i had that code
| |
| 12:31 | so guess that's not my problem
| |
| 12:31 | <ogra> do you call getltscfg anywhere else directly ?
| |
| 12:32 | <johnny> guess not
| |
| 12:32 | i was calling it manually to see why i might not have booted into the proper session
| |
| 12:32 | <ogra> ah
| |
| 12:33 | put an echo $(env) >/tmp/env.out into the ldm screen script
| |
| 12:33 | see what the vaues are :)
| |
| 12:33 | *values
| |
| 12:37 | BGomes has quit IRC | |
| 12:38 | wwx has quit IRC | |
| 12:38 | BGomes has joined #ltsp | |
| 12:45 | mccann has quit IRC | |
| 12:57 | mccann has joined #ltsp | |
| 12:58 | tux_440volt has joined #ltsp | |
| 13:06 | <gvy> ogra, ! :)
| |
| 13:06 | lns has quit IRC | |
| 13:17 | K_O-Gnom has joined #ltsp | |
| 13:20 | wwx has joined #ltsp | |
| 13:23 | bobby_C has quit IRC | |
| 13:28 | joel_ has joined #ltsp | |
| 13:29 | <joel_> Anyone had a problem with the boot process? Mine freezes during the PXE boot proces.
| |
| 13:31 | Any suggestions?
| |
| 13:33 | <gvy> last lines?
| |
| 13:40 | tux_440volt has quit IRC | |
| 13:42 | indradg is now known as angry_elephant | |
| 13:49 | <joel_> Were you talking to me gvy, a couple of minutes ago?
| |
| 13:49 | angry_elephant is now known as indradg | |
| 14:03 | <joel_> The system freezes about a quarter way through the load bar on the splash screen.
| |
| 14:04 | <gvy> joel_, err... yes
| |
| 14:04 | <joel_> I have also tried taking out the splash and quiet options in the PXELinux.cfg file.
| |
| 14:04 | <gvy> argh, splash.
| |
| 14:05 | <joel_> (pxelinux.cfg/default)
| |
| 14:05 | <gvy> vga=0 might suffice
| |
| 14:05 | or ...mmm... what was for 80x43...
| |
| 14:05 | <joel_> hold on!
| |
| 14:05 | * gvy listens | |
| 14:10 | BGomes has quit IRC | |
| 14:11 | BadMagic has quit IRC | |
| 14:12 | BadMagic has joined #ltsp | |
| 14:12 | <joel_> gvy...
| |
| 14:13 | <gvy> yep?
| |
| 14:13 | <joel_> The splash screen still shows and freezes up!
| |
| 14:14 | BadMagic has quit IRC | |
| 14:17 | BadMagic has joined #ltsp | |
| 14:18 | <joel_> Even after adding vga=0 to pxelinux.cfg...
| |
| 14:18 | K_O-Gnom has quit IRC | |
| 14:18 | <laga> joel_: try removing "quiet splash"
| |
| 14:18 | if that's ubuntu.
| |
| 14:20 | <joel_> Already did that...
| |
| 14:21 | <laga> and what do you see on your screen when you do that?
| |
| 14:24 | otavio_ has joined #ltsp | |
| 14:24 | otavio has quit IRC | |
| 14:26 | <gvy> joel_, well... i built altlinux-based image just yesterday
| |
| 14:27 | if all else fails, ftp://ftp.linux.kiev.ua/pub/Linux/ALT/beta/ltsp-server/
| |
| 14:29 | <joel_> Nothing really changed (visually) after removing the text from the file.
| |
| 14:31 | viking-ice_ has quit IRC | |
| 14:32 | otavio has joined #ltsp | |
| 14:32 | otavio_ has quit IRC | |
| 14:37 | bobby_C has joined #ltsp | |
| 14:51 | jammcq has quit IRC | |
| 14:55 | joel_ has quit IRC | |
| 14:55 | BGomes has joined #ltsp | |
| 14:57 | mccann has quit IRC | |
| 14:57 | Gadi has left #ltsp | |
| 15:01 | hersonl1 has joined #ltsp | |
| 15:01 | mccann has joined #ltsp | |
| 15:13 | hersonls has quit IRC | |
| 15:20 | TelnetManta has quit IRC | |
| 15:24 | lns has joined #ltsp | |
| 15:28 | Q-FUNK has joined #ltsp | |
| 15:37 | bobby_C has quit IRC | |
| 15:38 | bobby_C has joined #ltsp | |
| 15:40 | Egyptian[Home1 has joined #ltsp | |
| 15:41 | mikkel has quit IRC | |
| 15:43 | generic has quit IRC | |
| 15:44 | BGomes has quit IRC | |
| 15:45 | BGomes has joined #ltsp | |
| 15:48 | BGomes has quit IRC | |
| 15:54 | mccann has quit IRC | |
| 15:56 | Egyptian[Home] has quit IRC | |
| 16:23 | sepski has quit IRC | |
| 16:25 | hersonl1 has quit IRC | |
| 16:36 | slidesinger has quit IRC | |
| 17:03 | mhterres has quit IRC | |
| 17:04 | nantes_geek has joined #ltsp | |
| 17:09 | gregbrady has joined #ltsp | |
| 17:34 | bobby_C has quit IRC | |
| 17:56 | gregbrady has quit IRC | |
| 18:12 | alan has joined #ltsp | |
| 18:13 | <alan> Does anyone here have experience with setting up sound on Ubuntu Hardy
| |
| 18:14 | <gvy> -
| |
| 18:14 | nantes_geek has quit IRC | |
| 18:15 | alan has quit IRC | |
| 18:17 | * gvy is away: sleep() | |
| 18:23 | gbolte has quit IRC | |
| 18:29 | rjune_ has joined #ltsp | |
| 18:29 | gvy has quit IRC | |
| 18:47 | TelnetManta has joined #ltsp | |
| 18:50 | Q-FUNK has quit IRC | |
| 18:56 | staffencasa has quit IRC | |
| 19:10 | GiantPickle has quit IRC | |
| 19:32 | alan has joined #ltsp | |
| 19:34 | <alan> Can someone point me to a howto page to sound on ltsp? Using Ubuntu Hardy. Everything I read says 'works automatically". Unfortunately, that doesn't help troubleshooting.
| |
| 19:49 | :-!
| |
| 19:50 | <johnny> hmm.. yeah.. it just worked for me
| |
| 19:50 | at least i hear the desktop sound
| |
| 19:50 | when i plugged my laptop in
| |
| 19:50 | none of my real terminals have sound on them tho
| |
| 19:52 | J45p3r has joined #ltsp | |
| 19:56 | <alan> using bunch of Dell Optiplex gx110s with the HD pulled. work great, just no freaking sound
| |
| 19:57 | <johnny> you should test the sound on the server..
| |
| 19:58 | make sure that works
| |
| 20:00 | <alan> okay
| |
| 20:01 | give me a few minutes
| |
| 20:09 | ok I get a login sound when I logon to the server
| |
| 20:10 | when I logon to a client I do not get sound on the server...I had read that certain people would get all sounds through the server.
| |
| 20:18 | back in a minute
| |
| 20:18 | alan has left #ltsp | |
| 20:26 | alan has joined #ltsp | |
| 20:26 | <alan> that sucks...still nothing
| |
| 20:27 | gregbrady has joined #ltsp | |
| 20:32 | alan has left #ltsp | |
| 20:34 | captain_1agnus has joined #ltsp | |
| 20:35 | captain_magnus has quit IRC | |
| 20:35 | captain_1agnus is now known as captain_magnus | |
| 20:37 | alan has joined #ltsp | |
| 20:44 | <alan> johnny> thanks for your help....I'll keep looking
| |
| 20:48 | alan has left #ltsp | |
| 20:50 | alekibango has quit IRC | |
| 21:08 | J45p3r has quit IRC | |
| 21:46 | BGomes has joined #ltsp | |
| 21:53 | alekibango has joined #ltsp | |
| 22:19 | indradg has quit IRC | |
| 22:35 | captain_1agnus has joined #ltsp | |
| 22:39 | gregbrady has quit IRC | |
| 22:39 | gregbrady has joined #ltsp | |
| 22:40 | |Paradox| has quit IRC | |
| 22:42 | captain_magnus has quit IRC | |
| 22:42 | captain_1agnus is now known as captain_magnus | |
| 22:44 | jamesjames has joined #ltsp | |
| 22:45 | |Paradox| has joined #ltsp | |
| 22:49 | gregbrady has quit IRC | |
| 22:50 | gregbrady has joined #ltsp | |
| 22:55 | Nuba1 has quit IRC | |
| 22:56 | Nubae has joined #ltsp | |
| 23:00 | gregbrady has quit IRC | |
| 23:09 | BGomes has quit IRC | |
| 23:11 | jamesjames has quit IRC | |