|01:32||Guerdal82_ has joined #ltsp|
|01:55||rptan has joined #ltsp|
hi, anyone familiar with the "failed to connect to NBD server" error for Ubuntu 10.04?
|02:12||toscalix has joined #ltsp|
does it not always say that?
rptan: that's usually because dhcp points the client to the wrong server
(or, more rare, because the client is too fast, e.g. core 2 duo)
alkisg: hm, what i did is install ltsp on a server using alternate cd on amd64
Hmmm maybe inetd.conf isn't pointing to the right file then
grep nbd /etc/inetd.conf
alkisg: it was okay when i tested with VMs, but on the physical machines they need i386
And see if the image file is actually present on the file system (/opt/ltsp/images/xxx)
alkisg: so i removed the directory and did a rebuild of the ltsp-client with the arch i386 option
If inetd is pointing to and amd64.img and you have an i386.img, then that's your problem...
|02:34||elias_a has joined #ltsp|
alkisg: ok let me check
alkisg: wow, ur right on target. do i need to restart anything after editing the conf file?
sudo invoke-rc.d openbsd-inetd restart
Mikael - are you arond?
alkisg: thanks alot, its working fine now.!
|02:54||toscalix has quit IRC|
|03:32||satellit_ has joined #ltsp|
|03:39||bobby_C has joined #ltsp|
|03:46||ogra_ has joined #ltsp|
|03:47||ogra has quit IRC|
|03:47||ogra_ is now known as ogra|
|03:47||ogra has joined #ltsp|
|03:48||ogra_ has joined #ltsp|
|03:48||tlze has joined #ltsp|
|03:49||ogra_ has quit IRC|
|03:51||otavio has quit IRC|
|04:12||bobby_C has quit IRC|
|04:32||rmcgougan has joined #ltsp|
|04:38||otavio has joined #ltsp|
Have been trying to set up a 64 bit alternate cd ltsp using Ubuntu lucid lynx as a ltsp server using one ethernet card to communicate with machines on my schools lan, our lan server is SME Server it allocates the dhcp for the lan. Have got client machines initiating the boot process from the i386 boot image on the ltsp server but I get an error saying nbd failure. Have tried dpkg-reconfigure nbd-server to create a config file in
/etc/nbd-server followed by sudo ltsp-update-image --arch i386 but still the thin clients cannot reach the log in page.
09:32 < alkisg> grep nbd /etc/inetd.conf
09:33 < alkisg> If inetd is pointing to and amd64.img and you have an i386.img, then that's your problem...
Heh, it seems that problem is quite common :)
I thought it was necessary to create an i386 image even though the actual ltsp server is 64 bit because the thin clients are i386 machines is this a foolish mistake on my part?
sudo ltsp-build-client --arch=i386
rmcgougan: so now you have an i386 image?
If you do, then you just need to change inetd.conf...
Thanks Appiah I did use sudo ltsp-build-client --arch=i386 and have created the i386 image, yes my problem is now how to change the inetd.conf I am a teacher just having a go at setting up an ltsp server in my school and do not know how to change inetd.conf to point at the i386.img
just open the file with a text editor
rmcgougan: sudo gedit /etc/inetd.conf
I can use the terminal and can edit conf files but need some step by step help to alter /etc/inetd.conf file - do i just change amd64.img to i386.img within the file?
Then, run this to restart inetd: sudo invoke-rc.d openbsd-inetd restart
Thanks very much for the help.
I have already tried using K12Linux Live Server media (and now I'm trying Ubuntu Lucid Lynx ltsp) and ran into problems with screen lock ups on the clients and server using K12Linux Live Server when two or three thin clients were running flash player based websites. I read some tips that indicated you can edit the servers lts.conf file to allocate more ram for some applications such as firefox e.g. instead of 32mb ram per client
for firefox change the allocation to 128mb ram for that application to help avoid screen lock ups with multiple thin clents on flash player websites.
Do you think this editing of the lts.conf is a good move for schools where children often are using flash palyer based websites?
What are your client specs?
*and network speed, 100 mbps or gigabit?
The specs vary but the machines generally are old with cpu 433-999, 128-512mb ram, vga -not sure but basic and 6-10 years old, network 100mb lan, ltsp server gigabit network cards x2, tried to use ltsp server with a switch but was only 100mb speed switch no gigabit connections. My current network set up is just using one gigabit ethernet card from ltsp server, with dhcp from our separate lan server pointing to the static ip of ou
r ltsp server to let the client find the boot image.
|05:19||rptan has quit IRC|
" My current network set up is just using one gigabit ethernet card from ltsp server" ==> but that is connected to a 100mbps switch, right?
That's a little low for LTSP, a gigabit connection from the server to the switch with flow control=off is required for acceptable flash playback
Current set up just goes from ltsp server using one Gigabit ethernet connection on ltsp server straight to lan connection which is all 100mb cabling.
So it's a 100mbps connection, too low for flash.
|05:30||tlze has quit IRC|
The best thing to do is to upgrade your switch, or at least use multiple cards on the server (https://help.ubuntu.com/community/UbuntuLTSP/Trunking)
shouldn't flash / browser be runned as a local-app ?
With 10-year old PCs with 128 RAM? Not a chance... :)
heh, in that case =)
new switch would be perfect to be used
Yeah, a 16 port switch with 1 gigabit port here costs about 60 €, which isn't much..
The LTSP server has a gigabit ethernet connection going from it, but our lan is 100mb cabling in our school. So all the client machines are sending data through 100 mb cable.
rmcgougan: but is the switch gigabit?
rmcgougan: the bottleneck link is the one which is between switch and server
If it doesn't have a gigabit port, then the connection from the server to the switch is 100 mbps...
No the switch is not gigabit, sounds like I definately need to use a switch with at least one gigabit port connected directly to the server.
switch must have at least one gigabit port and the server must be connected to it. in optimal case 10 clients can communicate with 100Mb/s speed with the server
yah, that is what you want =)
rmcgougan: as a tip, one good program to monitor current bandwidth usage in linux is bwm-ng. very lightweight and clear software. just shows you how much stuff is going in the network interfaces of the server =)
Thanks very much your help, that has cleared this up for me. I will invest in a new switch.
rmcgougan: make sure to read that page about flow control
Otherwise you may buy a gigabit switch and STILL have 100mbps bandwidth
(i.e. you may just waste your money if flow control is on)
alkisg I appreciate the fine tuning about the flow control I have bookmarked the https://help.ubuntu.com/community/UbuntuLTSP/FlowControl page thanks again you've helped me e a lot:-)
also if you buy some switch which can be configured, it could be possible to disable flow control in the switch. never done that though, but it could also work
Yup it works, it's something one needs to check before buying a switch.
mm.. after a quick check i'm not sure after all is it as effective as disabling flow control as explained in that wiki page
it would be worth to make an experiment =)
but 2pm. coffee coffee coffee ->
Mava: I've checked both ways (I wrote that page after some discussion in the mailing list), both are effective, but disabling flow control on the server NIC isn't always possible (depends on the NIC) and of course has some additional administrative overhead.
So the "get a good switch" way is always better, as long as one has the extra money
(the cheapest configurable switch that I found was 4-5 times more expensive than the unmanaged one)
|06:19||rmcgougan has left #ltsp|
alkisg: ahaa... feeling kind a confused, so... the client cant process data enough fast and it sends PAUSE to server right ?
Right, so the server pauses instead of sending data to another client
I've benchmarked with flow control off = 990 mbps and with flow control on = 90 mbps with the same switch (by changing flow control on the nic)
and the pause really is tcp connection ?`
No, it's an ethernet pause
sorry, not the pause, but the connection is using tcp protocol ?
I've tried that with netperf
layer 2 level pause ?
I'm not sure about the level. But there's also flow control in the tcp protocol, and that is NOT a problem
I.e. if the client has to drop data because it receives it very fast, the sender lowers the rate because of the tcp protocol itself
ah, now it starts to feel much more understandable.. first i thought it was tcp pause, and tcp congestion control is different =)
and of course, once there happens this flow controlling in the lower levels it stops or lowers the whole ethernet so it affects also to another client connections
and now once flow control is disabled the congestion control in tcp protocol looks after the client machine =)
|06:53||pths has quit IRC|
|06:54||pths has joined #ltsp|
|07:09||pmatulis has joined #ltsp|
Mava, right, exactly
|07:16||jhutchins_lt has quit IRC|
|07:16||Appiah has quit IRC|
|07:16||morfic has quit IRC|
|07:16||korcan has quit IRC|
|07:16||dba has quit IRC|
|07:17||Faithful has quit IRC|
|07:18||jhutchins_lt has joined #ltsp|
|07:18||Appiah has joined #ltsp|
|07:18||morfic has joined #ltsp|
|07:18||korcan has joined #ltsp|
|07:18||dba has joined #ltsp|
|07:24||ruxpin has joined #ltsp|
|07:29||cliebow has joined #ltsp|
|07:36||otavio has quit IRC|
|07:47||Guerdal82_ has quit IRC|
|08:08||komunista has joined #ltsp|
|08:23||Gadi has joined #ltsp|
|08:51||otavio has joined #ltsp|
|09:23||ogra has quit IRC|
|09:24||ogra has joined #ltsp|
|09:42||Guerdal82_ has joined #ltsp|
|09:44||highvoltage has quit IRC|
|09:46||thunsucker has joined #ltsp|
|09:49||highvoltage has joined #ltsp|
|09:56||evilx has joined #ltsp|
|10:14||pmatulis has quit IRC|
|10:14||staffencasa has joined #ltsp|
|10:16||Barbosa_ has joined #ltsp|
|10:19||Barbosa has quit IRC|
|10:47||pmatulis has joined #ltsp|
|10:52||atkuepker has joined #ltsp|
|11:05||vizz has joined #ltsp|
which components i need, to get sound working at debian ltsp?
|11:07||japerry has quit IRC|
vizz: do your thin clients have sound cards?
vizz: then sound should work for them
vizz: does it not?
it didn't, i have installed pulseaudio on server and client and SOUND=yes in lts.conf
vizz: are you using any screen scripts? like rdp script
vizz: if you are not using screen script, try leaving sound=yes out of lts.conf
no screen scripts, i tried without lts.con to
here are some hacks, i will try that http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=537526
vizz: so you get no sound at all?
vizz: on the client, did you go into sound settings and look around?
in the chroot?
on the tc desktop
vizz: you running squeeze?
vizz: sorry i'm not super familiar with that setup but if that stuff that you linked in that bug report might help
vizz: of course you could always hang out til vagrantic is here, he wrote that bug report
vizz: sorry couldn't help more
the only other thing i can suggest
|11:29||Selveste1 has quit IRC|
|11:38||japerry has joined #ltsp|
|11:50||japerry has quit IRC|
|12:12||komunista has quit IRC|
More Fun with Intel Video: Still having problems with Intel Video on our stock Ubuntu 9.10 installs. Switching to Metacity using gconftool didn't fix it. Any further suggestions?
I did notice that out version of xserver-xorg-video-intel is "2:2.9.0"
|13:16||alkisg1 has joined #ltsp|
|13:16||alkisg has quit IRC|
|13:27||* _UsUrPeR_ tips his hat|
can anybody tell me how ltsp-remoteapps works, or how it would be used from an ltsp environment?
in client terminal (via ltsp-localapps xterm) ltsp-remoteapp oocalc
|13:32||* _UsUrPeR_ attempts|
would open up oocalc on the server
if your distro has it packaged..
not sure which version of ubuntu includes it.. or if any do
|13:33||alkisg1 has quit IRC|
hmm. Ok, so base Desktop installation is what I am using
it does what it says .. run an app remotely
on the localapps-opened xterm session, remoteapps exists, however when I attempt to run, nothing happens. is there a log file in /var/log on the server/client that reflects any problems?
i can imagine some useful cases might be, if a local app triggers an app that isn't truly a dep (and thus not installed) you could tell it to handle it on the server
yeah, such as opening an ooO doc from localapps firefox session
sure, but it probably only works on things with .desktop files.. or those that handle them
|13:35||atkuepker has quit IRC|
at least directly, otherwise you'd have to create your own symlinks
or other aliases
i don't really know much about it other than that.. never needed it
_UsUrPeR_, look up who committed it, and ask them
yeah, definitely special cases for remoteapps
ok, thanks for your help johnny
|13:36||akuepker has joined #ltsp|
|13:39||komunista has joined #ltsp|
|13:43||johnny has left #ltsp|
|13:43||mischko has joined #ltsp|
|13:44||johnny has joined #ltsp|
|14:04||vagrantc has joined #ltsp|
|14:13||kusznir has joined #ltsp|
Hi all: I've got an issue that suddenly started this morning on my terminals. The terminals will netboot just fine, but stop at a console prompt. About 2-5 minutes later, it will load the graphical login, and things appear to progress normally.
All my DNS and the like seem to be working; no changes were made since last week (when everything worked well), and I've even restarted my dns server.
Any suggestions? What would cause that delay?
sounds like name resolution issues...
I'm not sure where else to look...The server's forward and reverse lookup succeed just fine; the client's doesn't generally look up at all.
possibly a stalling ldm/rc.d hook or ltsp/xinitrc.d hook
vagrantc: what would not be resolving? The DNS server is definitely working in a general sense; I'm not sure what to check for specifically.
the clients have access to the DNS server?
On the same network as the dns server itself, several other devices are working.
|14:17||akuepker has quit IRC|
|14:20||Lns has joined #ltsp|
Hmm...so I logged in as root on the thin client. DNS works fine (can ping google, it resolves it. The host command isn't installed, so its hard to do through testing, but it can get to and resolve the thin client servers)
dmesg just had some notes about block read errors on the squashfs (which have been there sinc ethe beginning, and others here told me to ignore it)
There is NOTHING in any of the log files in /var/log
I don't know where else to look.
btw: nothing has changed in the ltsp export in months...I haven't even rebuilt the nbd image in over a month.
kusznir: anything else change in the network
switches, routers, internet anything?
kusznir: once the graphical screen comes up, it works fine, though?
thunsucker: upstream, not downstream that I'm aware of.
kusznir: more specific please
I have my switch that all my stuff is connected to (clients, servers, and routers). I don't control stuff upstream (main university network that my router connects to), and they've been making changes (most of which they don't announce)
I know they replaced their switches, and I've also been seeing signs of sludgyness in places. However, my setup is supposed to be "self-contained".
|14:44||bobby_C has joined #ltsp|
kusznir: so your servers like dns are on your local switch
kusznir: what order do the dns servers go in? their dns, your dns?
I'm running pfsence for my router, which is plugged into my switch. pfsense handles dns, dhcp, and ip routing.
ok just out of curiousity
disconnect yourself from them
so you are 100% truly contained
my thin clients are only configured with one DNS server, and that's mine.
Unfortunately, i can't disconnect myself. This is production network with 20+ people on the clock, and doing that would down them all (no e-mail, web, ssh sessions to off-site, etc).
kusznir: if upstream there is sluggishness
it's possible that it's effecting your local users
of coruse do the normal stuff, reboot your server and router and etc.
What I don't get is the ONLY place I experience any sluggishness is the starting of the login. Once the login screen comes up, everything else (including full auth/session loading) works fine.
kusznir: yes, and unfortenly i'm not an ubuntu master to know exactly whats it doing
I've restarted services on my router. I want to reboot my server, but I've got a bunch of users on it in the middle of work. The sucky part of production systems....
maybe it's looking for something specific on bootup
kusznir: what distro you running?
ubuntu. The server booting my thin clients is 9.04. I also have a 9.10 thin client server, but its not seeing as much use due to stability problems with samba/winbind (that samba is claiming is ubuntu, and ubuntu is just ignoring)
kusznir: reboot everything tonight and see whbat happens
then try isolating them to see what happens
This does lead me to another question: For ease of troubleshooting/maintence, I'd like to create a dedicated vm server that JUST serves up the thin client images. That way I can have a variety of terminal servers for different needs/load balancing, and be able to reboot the terminal servers without "crashing" all the thin clients.
then report back :)
(that's the wort part about rebooting...if its what the clients are booting off of, then they all get in a wierd state that my users can't figure out, and I get tons of calls until I go through every single one and power cycle them.
kusznir: ltsp-load balancing :)
kusznir: in a couple of my setups, i do a weekly snapshop of the ltsp erver
It has to do with the nbd server restarting; aparently that causes the thin clients to crash.
if it freaks out
i power it off, power on the backup
and then test the "freaked out one"
kusznir: just pull the power breakers
kill everything :)
kusznir, talk to you later, hopefull you resolve it i'm going home for the night
|14:52||thunsucker has quit IRC|
|14:52||cliebow has quit IRC|
Thanks (unfortunately, I am not certified nor authorized to reset the 2,000A breaker to the building; and I have stuff spread around....
kusznir: are you sure that there is not another DHCP server?
I'd be very surprised. The initial dhcp goes very quickly, and it reports all the right info. It boots correctly, and comes up with a correct IP in a correct range, and the correct dhcp server.
and the correct dns server
And no other devices (personal laptops on same network) are reporting problems.
the another DHCP don't must give the boot image
then first boot is OK
but after boot the thin clients gets IP again, without PXE image, and another DHCP can take place...
I log into the thin client's console after it boots to that level and log in; its got all the right IP settings at that point (which is when it should be launching the graphical login, but isn't)
have you remote syslog enabled? https://help.ubuntu.com/community/UbuntuLTSP/ClientTroubleshooting
Nope, I haven't taken any explicit steps to enable remote syslog.
stgraber: When attempting to run ltsp-remoteapps from an xterm session on an ltsp client, it appears nothing happens. Is there an lts.conf setting required for remoteapps to be enabled?
kusznir: try folow https://help.ubuntu.com/community/UbuntuLTSP/TCSyslogToLTSPServer
just in case
kusznir: reboot thin client and while reboot you can see in serrvers syslog...
*server's syslog :-)
|15:16||alkisg has joined #ltsp|
I'm logging into the server looking at the logs (or what I can find of them -- dmesg works fine), but not seeing much of anything helpful.
How do I manually kick off the graphical login?
(in my terminal so I can catch any stdout/stderr output)
|15:19||japerry has joined #ltsp|
then just "ldm"?
but it autostart again :-)
The problem I'm having is it doesn't seem to be doing that, or its taking a very long time to do that.
and here is problem with long time to start login dialog or to login?
long time to start login dialog.
ldm takes a long time to show up; it sits at the text login prompt for several minutes.
(text login prompt -> the thin client's console login)
try kill it and see if the long time to start ldm go again
How much ram does the client have?
I think a gig of ram.
You could install bootchart to watch where the delay is...
|15:46||johnny has left #ltsp|
|16:07||komunista has quit IRC|
|16:10||johnny has joined #ltsp|
|16:15||alkisg has quit IRC|
|16:20||bobby_C has quit IRC|
|16:20||vagrantc has quit IRC|
|16:24||vagrantc has joined #ltsp|
so wha'ts needed to enable the ntp service these days on ubuntu/ltsp?
just set TIMESERVER in lts.conf and install ntp on the server?
unless ntpdate isn't installed in the chroot
ok.. so now i'll just wait til i reboot em
|16:36||atkuepker has joined #ltsp|
|17:00||japerry has quit IRC|
|17:13||pmatulis has quit IRC|
|17:31||satellit_ has quit IRC|
|17:31||satellit_ has joined #ltsp|
|17:46||japerry has joined #ltsp|
|17:58||Gadi has left #ltsp|
|18:00||mischko has quit IRC|
|18:21||Selveste1 has joined #ltsp|
|18:36||staffencasa has quit IRC|
|18:46||Barbosa has joined #ltsp|
|18:47||Barbosa_ has quit IRC|
|18:49||Lns has quit IRC|
|18:59||vagrantc has quit IRC|
|19:53||Selveste1 has quit IRC|
|20:01||pmatulis has joined #ltsp|
|20:10||F-GT has quit IRC|
|20:20||F-GT has joined #ltsp|
|20:46||pmatulis has quit IRC|
|21:04||_UsUrPeR_ has quit IRC|
|21:19||lucascoala has joined #ltsp|
|21:55||johnny has left #ltsp|
|22:18||ogra_cmpc has quit IRC|
|22:31||ogra_cmpc has joined #ltsp|
|22:36||johnny has joined #ltsp|
|23:20||rjune has quit IRC|
|23:26||johnny has left #ltsp|
|23:27||johnny has joined #ltsp|
|23:33||abeehc has quit IRC|
|23:40||abeehc has joined #ltsp|
|23:47||abeehc has quit IRC|
|23:54||abeehc has joined #ltsp|