IRC chat logs for #ltsp on irc.libera.chat (webchat)
Channel log from 27 June 2010 (all times are UTC)
| 01:44 | shogunx has joined #ltsp | |
| 01:57 | shogunx has quit IRC | |
| 02:35 | bobby_C has joined #ltsp | |
| 02:44 | <sean74> hi alkisg
| |
| 02:49 | <alkisg> Hello
| |
| 02:50 | ogra has quit IRC | |
| 02:52 | ogra has joined #ltsp | |
| 02:55 | <sean74> alkisg: again cups advices? hehe
| |
| 02:56 | alkisg: if i didnt login the client, the cups wont start
| |
| 02:56 | <alkisg> sean74: you mean the RC_SCRIPT I gave you the other day?
| |
| 02:56 | <sean74> alkisg: how can i start the cups once the device was start?
| |
| 02:56 | yup
| |
| 02:56 | <alkisg> Isn't it on the irc logs?
| |
| 02:59 | (06:50:26 μμ) alkisg: Create a file named /opt/ltsp/i386/etc/ltsp/start-cups
| |
| 02:59 | (06:50:32 μμ) alkisg: Put the following contents into it:
| |
| 02:59 | (06:50:38 μμ) alkisg: #!/bin/sh
| |
| 02:59 | (06:50:38 μμ) alkisg: /usr/sbin/invoke-rc.d cups start
| |
| 02:59 | (06:50:53 μμ) alkisg: Then, make it executable: sudo chmod +x /opt/ltsp/i386/etc/ltsp/start-cups
| |
| 02:59 | (06:51:14 μμ) alkisg: Finally, make it start automatically from lts.conf:
| |
| 02:59 | (06:51:19 μμ) alkisg: RCFILE_01=/etc/ltsp/start-cups
| |
| 02:59 | (06:51:34 μμ) alkisg: On the next reboot, the client will see the printer if you point cups to localhost.
| |
| 02:59 | (06:52:27 μμ) sean74: do i need to rebuild the image?
| |
| 02:59 | (06:52:40 μμ) alkisg: Yes
| |
| 03:04 | Eghie has joined #ltsp | |
| 03:13 | ogra_cmpc has quit IRC | |
| 03:15 | <sean74> ok, let me redo this again
| |
| 03:15 | thanks
| |
| 03:18 | alkisg: err, the lts.conf in /opt/ltsp/i386/etc/lts.conf or /var/lib/tfptboot/ltsp/i386/lts.conf?
| |
| 03:19 | <alkisg> (06:56:57 μμ) sean74: which lts.conf? tftpboot? /opt/ltsp/i386/etc?
| |
| 03:19 | (06:57:05 μμ) alkisg: /var/lib/tftpboot/ltsp/i386/lts.conf
| |
| 03:19 | (06:57:09 μμ) alkisg: [Default]
| |
| 03:19 | (06:57:22 μμ) alkisg: RCFILE_01=/etc/ltsp/start-cups
| |
| 03:19 | <sean74> many thanks
| |
| 03:19 | <alkisg> Do keep logs ;)
| |
| 03:20 | <sean74> my bad, i'm too excited to try the thing.
| |
| 03:20 | and always late nite working, once the thing work i go to sleep
| |
| 03:21 | thanks
| |
| 03:21 | now i logged it
| |
| 03:23 | ogra_cmpc has joined #ltsp | |
| 03:48 | ogra_cmpc has quit IRC | |
| 03:54 | komunista has joined #ltsp | |
| 04:00 | ogra_cmpc has joined #ltsp | |
| 04:13 | bobby_C has quit IRC | |
| 04:23 | alkisg has quit IRC | |
| 04:33 | komunista has quit IRC | |
| 04:34 | komunista has joined #ltsp | |
| 04:39 | alkisg has joined #ltsp | |
| 04:40 | dobber_ has joined #ltsp | |
| 04:52 | shogunx has joined #ltsp | |
| 04:53 | alkisg has quit IRC | |
| 05:35 | Eghie has left #ltsp | |
| 06:06 | sean74 has quit IRC | |
| 06:12 | Selveste1 has joined #ltsp | |
| 06:13 | Egyptian[Home] has quit IRC | |
| 06:18 | Selveste1 has quit IRC | |
| 06:27 | Egyptian[Home] has joined #ltsp | |
| 06:45 | komunista has quit IRC | |
| 07:13 | bobby_C has joined #ltsp | |
| 07:42 | bobby_C has quit IRC | |
| 07:49 | pmatulis has joined #ltsp | |
| 08:07 | Faithful has joined #ltsp | |
| 08:59 | peter_tm has joined #ltsp | |
| 09:10 | komunista has joined #ltsp | |
| 09:16 | Faithful has quit IRC | |
| 09:35 | chupacabra has quit IRC | |
| 10:04 | Selveste1 has joined #ltsp | |
| 10:07 | Selveste1 has quit IRC | |
| 10:08 | Selveste1 has joined #ltsp | |
| 10:35 | Selveste1 has quit IRC | |
| 10:35 | Selveste1 has joined #ltsp | |
| 10:47 | Selveste1 has quit IRC | |
| 10:51 | Selveste1 has joined #ltsp | |
| 10:55 | Selveste1 has quit IRC | |
| 10:55 | Selveste1 has joined #ltsp | |
| 11:10 | vagrantc has joined #ltsp | |
| 11:48 | jammcq has joined #ltsp | |
| 11:48 | <jammcq> hey friends
| |
| 11:54 | * vagrantc waves | |
| 11:54 | <jammcq> hey hey
| |
| 11:54 | heading to NYC soon ?
| |
| 11:55 | <vagrantc> bout three weeks till i start the journey
| |
| 11:57 | <jammcq> hey, it's been a while since I used ltsp. but I have LTSP-5.2 set up at a customer site, and I really have to tell you, it rocks !
| |
| 11:57 | it's freaking amazing how good it has become
| |
| 11:58 | <vagrantc> indeed.
| |
| 11:58 | <jammcq> you guys have done such a fantastic job.
| |
| 11:59 | <vagrantc> just the other day i updated the sdm package in debian ... and that pointed out to me the huge differences with ldm.
| |
| 11:59 | <jammcq> I'm using it at a large doctors office where we have 5 terminals at the front desk. each terminal has a printer, 2 scanners and a signature pad connected
| |
| 11:59 | sane for the scanners rocks
| |
| 12:00 | for the sigpad, i had to write a daemon to interact with it, and I launch that via udev. so if someone unplugs the signature pad, then plugs it back in, the process automatically starts and connects to my database
| |
| 12:01 | <vagrantc> that part generalizeable?
| |
| 12:01 | <jammcq> I'd be happy to share what I have, but it's rather custom for my application
| |
| 12:01 | it's written in perl and C
| |
| 12:01 | and specific to a Topaz signature pad
| |
| 12:02 | we go live on tuesday with all this new stuff. Once everything is smooth, I'd like to write an article about how it all works with ltsp
| |
| 12:02 | that should bring some attention to the project, and we may get other people wanting to do similar things
| |
| 12:02 | <vagrantc> cool :)
| |
| 12:04 | "founder of LTSP sings it's praises"
| |
| 12:04 | <jammcq> heh
| |
| 12:04 | dobber_ has quit IRC | |
| 12:07 | <jammcq> the one thing i'm still wrestling with is determining which workstation a user is logged into
| |
| 12:07 | firefox is running on the ltsp server
| |
| 12:07 | the web application is running on a different server
| |
| 12:07 | I need my cgi scripts on the web server to be able to know which thin client the user is logged in to, so that I can know which scanners,printers,sigpad to interact with.
| |
| 12:08 | for now, I'll just have to ask the user when they log into my app, but there should be a better way
| |
| 12:08 | <vagrantc> any way to get the webapp to see environment variables from firefox?
| |
| 12:08 | through some javascript or some such?
| |
| 12:09 | though i suppose javascript is prevented from doing that...
| |
| 12:10 | <jammcq> dunno yet. I've just starting looking at this in a serious way
| |
| 12:10 | gonna play with user-agent string in firefox, see if I can set that before firefox runs
| |
| 12:11 | maybe set it based on the SSH_CONNECTION info or something like that
| |
| 13:41 | Faithful has joined #ltsp | |
| 14:23 | danalt has joined #ltsp | |
| 14:26 | chupacabra has joined #ltsp | |
| 14:27 | danalt has quit IRC | |
| 14:27 | vmlintu has joined #ltsp | |
| 14:29 | danalt has joined #ltsp | |
| 14:40 | artista_frustrad has quit IRC | |
| 14:48 | danalt has quit IRC | |
| 14:53 | artista_frustrad has joined #ltsp | |
| 15:09 | alvin has quit IRC | |
| 15:15 | komunista has quit IRC | |
| 15:19 | peter_tm has left #ltsp | |
| 15:46 | sweetpi has joined #ltsp | |
| 15:47 | Billynkid has joined #ltsp | |
| 15:47 | <Billynkid> Hi guys
| |
| 15:47 | Trying to configure a ltsp server to play flash videos via a prism browser
| |
| 15:48 | the problem is it generates so much network traffic I can't support more than 5-6 clients
| |
| 15:48 | <vagrantc> !flash
| |
| 15:48 | <ltspbot> vagrantc: "flash" :: Yes, flash sucks. Make sure you have LDM_DIRECTX=True in your lts.conf file, or if it's just youtube you're after, try some flash replacing plugin like https://addons.mozilla.org/el/firefox/addon/161869/ (per user installation, gecko-mediaplayer is also needed).
| |
| 15:49 | <Billynkid> no its locally hosted flash files
| |
| 15:50 | the LDM_DIRECTX improve performance but also increases network traffic
| |
| 15:50 | which obvious reduces the total number of thin clients we can have running
| |
| 15:50 | I am using a PIV4 with 1.5Gb ram
| |
| 15:50 | <elias_a> Billynkid: What do you mean with locally hosted flash videos?
| |
| 15:50 | <Billynkid> the thin clients are neoware c15's
| |
| 15:51 | It's a copy of the RACHEL eductional repositiory
| |
| 15:51 | flash math and science videos
| |
| 15:51 | <elias_a> AFAIK the problem is not where the videos are hosted.
| |
| 15:51 | <Billynkid> we have tried xrdp as an alternative but the sound does not redirect
| |
| 15:52 | no the problem is the fact one node generates 50mbits of network traffic rendering one video. If I just connect via my normal laptop it generates 0.40 kbytes or so
| |
| 15:52 | <elias_a> The problem is that basically the work that usually would take place inside a PC, including bus activity, is in the network.
| |
| 15:52 | <vagrantc> Billynkid: try running the flash players as a localapp
| |
| 15:53 | <elias_a> Yes. Go for localapps approach.
| |
| 15:53 | <vagrantc> if your thin clients have the horseplayer to do so...
| |
| 15:53 | <Billynkid> have tried installing prism or firefox and adobe flash locally seems the apps run slower
| |
| 15:53 | vagrantc not sure they do
| |
| 15:53 | <vagrantc> well, you're in a tricky spot, then.
| |
| 15:53 | <elias_a> Billynkid: To me it seems you are trying to solve a nonsolvable problem.
| |
| 15:57 | <Billynkid> Ok is there a good xdcmp client build
| |
| 16:04 | alvin has joined #ltsp | |
| 16:12 | alkisg has joined #ltsp | |
| 16:12 | spectre- has joined #ltsp | |
| 16:12 | <spectre-> hi!
| |
| 16:13 | same problem as billynkid
| |
| 16:13 | is there a way to localize x11?
| |
| 16:13 | so that all this rendering traffic can stay local to the node?
| |
| 16:14 | (x11,gdm,whatever)
| |
| 16:15 | <alkisg> fat clients?
| |
| 16:18 | <spectre-> sorta
| |
| 16:19 | except these are limited to 64mb
| |
| 16:19 | <alkisg> 64mb ram? You can't do much with them then
| |
| 16:19 | <spectre-> diskless nodes would be ideal
| |
| 16:19 | alexqwesa has quit IRC | |
| 16:19 | <spectre-> but the hp of these neowares is crap
| |
| 16:19 | <alkisg> There's no recent DE that runs on 64M
| |
| 16:19 | alexqwesa has joined #ltsp | |
| 16:19 | <spectre-> nah 128ram
| |
| 16:19 | 64mb flash space
| |
| 16:19 | for hd
| |
| 16:20 | the bottleneck seems to be the network
| |
| 16:20 | <alkisg> You don't need a HD, you can use them as diskless clients. But with 128 RAM you can't really do anything with them. If they had 1-2GB RAM... :)
| |
| 16:20 | <spectre-> we COULD upgrade to gigabit but it's not ideal
| |
| 16:20 | <alkisg> 128 RAM ==> only thin clients
| |
| 16:20 | <spectre-> :/
| |
| 16:20 | <alkisg> You don't need a full gigabit network
| |
| 16:21 | Just a switch with 1 gigabit port, and a gigabit nic on the server
| |
| 16:21 | <spectre-> just gigabit from switch to server
| |
| 16:21 | yeah
| |
| 16:22 | alright
| |
| 16:22 | well, thanks :p
| |
| 16:22 | <alkisg> Sure
| |
| 16:22 | You can also use bonding
| |
| 16:22 | <spectre-> what's bonding?
| |
| 16:22 | <alkisg> E.g. 3-4 NICs on the server can give you 300-400 mbps with a 100mbps switch
| |
| 16:22 | <spectre-> ahhh
| |
| 16:22 | <alkisg> https://help.ubuntu.com/community/UbuntuLTSP/Trunking
| |
| 16:23 | <spectre-> alright, good stuff. thanks!
| |
| 16:23 | <alkisg> That helps. A gigabit server <=> switch connection is better though
| |
| 16:23 | <spectre-> yeah gigabit is a pain because we're in Uganda
| |
| 16:23 | and it's for a rural education system
| |
| 16:23 | <alkisg> Ah. How many clients?
| |
| 16:24 | <spectre-> 5-10
| |
| 16:24 | <alkisg> Here (greece) it would cost about 50€...
| |
| 16:24 | <spectre-> can't even really get gbps switches here
| |
| 16:24 | <alkisg> Can't you shop online?
| |
| 16:24 | <spectre-> no :)
| |
| 16:25 | <alkisg> Sucks :)
| |
| 16:25 | <spectre-> it's uganda
| |
| 16:25 | anything gigabit essentially needs to come here via someones suitcase
| |
| 16:25 | some newer consumer systems come with gbps nics
| |
| 16:25 | but gbps switches don' t exist here yet
| |
| 16:26 | <alkisg> I see... yeah well trunking would be a good choice then.
| |
| 16:26 | <spectre-> yeah, indeed, so thanks for the link :)
| |
| 16:26 | <alkisg> I've worked at a school with 8 clients and 4 nics on the server for a year
| |
| 16:27 | It was ok, with 400mbps overall traffic.
| |
| 16:27 | <spectre-> we're also thinking of converting flv's to ogg's
| |
| 16:27 | to cut down on traffic
| |
| 16:27 | <alkisg> !flash
| |
| 16:27 | <ltspbot> alkisg: "flash" :: Yes, flash sucks. Make sure you have LDM_DIRECTX=True in your lts.conf file, or if it's just youtube you're after, try some flash replacing plugin like https://addons.mozilla.org/el/firefox/addon/161869/ (per user installation, gecko-mediaplayer is also needed).
| |
| 16:27 | <alkisg> With a plugin like that you won't see any difference
| |
| 16:27 | As long as the player uses XV acceleration, the video format doesn't matter.
| |
| 16:28 | <spectre-> we noticed ogg's ran around 2MB/sec
| |
| 16:28 | flv's consumed nearly 6
| |
| 16:28 | <alkisg> A normal video is about 50 mbps
| |
| 16:29 | <spectre-> flv was -- we got ogg to run at around 25
| |
| 16:29 | <alkisg> Nah
| |
| 16:29 | What dimentions / fps was the video?
| |
| 16:29 | <spectre-> i saw it on vnstat -l
| |
| 16:29 | 2000ish kB/se
| |
| 16:29 | c
| |
| 16:29 | alexqwesa has quit IRC | |
| 16:29 | <spectre-> don't recall but decent
| |
| 16:29 | <alkisg> The hard disk video compression doesn't matter
| |
| 16:30 | When it goes to the clients, it gets decompressed
| |
| 16:30 | <spectre-> well, i dunno, but the ogg video we tested ran 2000ish kB/sec and the flv was around 6000
| |
| 16:30 | <alkisg> So if a video is 640x480x24bppx24fps, you multiply all those to get the bandwidth
| |
| 16:31 | And if the user resizes it, you use the window size for the calculation
| |
| 16:31 | alexqwesa has joined #ltsp | |
| 16:31 | <alkisg> Unless XV acceleration is used
| |
| 16:31 | <spectre-> interesting
| |
| 16:31 | <alkisg> In this case, the resizing happens on the client
| |
| 16:31 | So even if they see the video full screen, it's still 640x480 for the bandwidth
| |
| 16:31 | <spectre-> yeah
| |
| 16:31 | <alkisg> And because YUV is used, it's 12 bits per pixel for each color, not 24
| |
| 16:32 | So it's much less ==> 50 mbps
| |
| 16:32 | The disk format (flv, avi, ogg, whatever) doesn't matter AT ALL
| |
| 16:32 | <spectre-> ok so it was a resolution difference?
| |
| 16:32 | <alkisg> How did you play the video?
| |
| 16:32 | <spectre-> app
| |
| 16:33 | flv was browser
| |
| 16:33 | <alkisg> Did you mount a network disk, or did you use an ltsp client (==remote X)?
| |
| 16:33 | <spectre-> it was hosted on the ltsp server machine
| |
| 16:33 | <alkisg> And you watched it from an ltsp client?
| |
| 16:33 | <spectre-> yes
| |
| 16:33 | <alkisg> Did you have LDM_DIRECTX=true in lts.conf?
| |
| 16:34 | <spectre-> for a later test yes but for the original no
| |
| 16:34 | <alkisg> Then a lot of frames were dropped because the client had 100% cpu usage
| |
| 16:34 | So you didn't see the actual bandwidth needed, but a cut-down version because of the lost frames
| |
| 16:34 | <spectre-> when that directx line was false we got decent fps
| |
| 16:34 | it was fairly smooth
| |
| 16:35 | <alkisg> Right. What bandwidth did you see then?
| |
| 16:35 | <spectre-> 2000ish
| |
| 16:35 | maybe 2500
| |
| 16:35 | kB
| |
| 16:35 | <alkisg> No, thats too low
| |
| 16:35 | <spectre-> seriously :P
| |
| 16:35 | <alkisg> Was that network bandwidth or disk bandwidth?
| |
| 16:35 | <spectre-> was vnstat -l running on the ltsp server
| |
| 16:35 | network
| |
| 16:36 | but the flv racked up like 6000/51mbps
| |
| 16:36 | <alkisg> Wait, KByte? I.e. 2 MByte per second?
| |
| 16:36 | <spectre-> and was choppy
| |
| 16:37 | <alkisg> With the same player?
| |
| 16:37 | <spectre-> no
| |
| 16:37 | browser
| |
| 16:37 | <alkisg> OK, in this case no XV was used
| |
| 16:37 | <spectre-> the ogg we played in movie player or whatever default app
| |
| 16:37 | flv was browser based
| |
| 16:37 | <alkisg> So your numbers maybe be ok if it was a small video
| |
| 16:37 | E.g. 320x200 or something like that
| |
| 16:37 | <spectre-> it was the example ogg ubuntu gives you
| |
| 16:38 | <alkisg> Anyway, one thing is certain: the disk format has absolutely no impact at all on the network bandwidth
| |
| 16:38 | <spectre-> yeah so prolly a resolution issue
| |
| 16:38 | <alkisg> So if you get any results that say otherwise, check them again, something else is happening
| |
| 16:38 | <spectre-> well, it's all rendered on the server, right?
| |
| 16:39 | <alkisg> Yes, and then it travel on the network after it's decoded
| |
| 16:39 | So it may impact CPU usage, but not network bandwidth
| |
| 16:39 | <spectre-> and the video output meant for the video card is redirected to the thinclient, right?
| |
| 16:39 | <alkisg> Yes
| |
| 16:39 | <spectre-> gotcha -- figured that after all this diagnosis we did :p
| |
| 16:40 | so essentially directx takes away ssh
| |
| 16:40 | right?
| |
| 16:40 | thus reducing encryption overhead?
| |
| 16:40 | or rather eliminating?
| |
| 16:40 | <alkisg> More or less. X traffic is unencrypted, but ssh authentication is still encrypted
| |
| 16:41 | <spectre-> ah
| |
| 16:41 | <alkisg> So the initial password doesn't travel over the network unencrypted...
| |
| 16:41 | <spectre-> and with that to true the data is also uncompressed and thus traffic is higher?
| |
| 16:41 | * alkisg uses iftop and iptraf for network monitoring | |
| 16:42 | <alkisg> Yes traffic is higher, but the clients have 100% cpu usage for video, thus dropping most of the frames
| |
| 16:42 | <spectre-> ah
| |
| 16:42 | <alkisg> (and for scrolling a firefox window etc etc)
| |
| 16:43 | So if security is not much of a concern, ldm_directx is much better if set to true.
| |
| 16:46 | spectre- has quit IRC | |
| 16:52 | peter_tm has joined #ltsp | |
| 17:05 | alkisg has quit IRC | |
| 17:29 | F-GT has quit IRC | |
| 17:42 | Selveste1 has quit IRC | |
| 17:46 | F-GT has joined #ltsp | |
| 18:06 | johnny has quit IRC | |
| 18:08 | johnny has joined #ltsp | |
| 18:58 | artista_frustrad has quit IRC | |
| 20:20 | peter_tm has quit IRC | |
| 20:31 | gentgeen__ has joined #ltsp | |
| 20:36 | gentgeen__ has quit IRC | |
| 20:58 | johnny has left #ltsp | |
| 21:13 | gentgeen__ has joined #ltsp | |
| 21:14 | gentgeen__ has quit IRC | |
| 21:15 | gentgeen__ has joined #ltsp | |
| 21:22 | pmatulis has quit IRC | |
| 21:24 | gentgeen__ has joined #ltsp | |
| 21:25 | gentgeen__ has quit IRC | |
| 21:37 | Egyptian[Home] has quit IRC | |
| 21:51 | Egyptian[Home] has joined #ltsp | |
| 22:03 | nuboon2age_ has joined #ltsp | |
| 22:27 | johnny has joined #ltsp | |
| 22:34 | gentgeen__ has joined #ltsp | |
| 22:35 | sweetpi has quit IRC | |
| 22:43 | gentgeen__ has quit IRC | |
| 22:52 | jammcq has quit IRC | |
| 22:56 | gentgeen__ has joined #ltsp | |
| 22:58 | gentgeen__ has quit IRC | |
| 23:13 | gentgeen__ has joined #ltsp | |
| 23:29 | gentgeen__ has quit IRC | |
| 23:33 | gentgeen__ has joined #ltsp | |
| 23:41 | moose-tek has joined #ltsp | |