IRC chat logs for #ltsp on irc.freenode.net (webchat)


Channel log from 28 July 2009   (all times are UTC)

00:16Jenna has joined #ltsp
00:50
<Jenna>
hey all.
00:50
is this SAMSUNG SyncMaster 730XT compatible withe ltsp
00:52
or can anyone suggest a suggest a sub $200 ltsp compliant terminal ?
00:52
<johnny>
anything that can pxe boot usually
00:52
depends on if you want local apps or not.. then you need 512mb ram
00:52
but for normal thin client usage you need at least 128mb ram
00:53
<Jenna>
yeah with ltsp5 I do need more ram. that I gather.
00:53
<johnny>
it's not ltsp5 that needs it per se
00:53
it's the tools of linux itself
00:53
ltsp5 needs more cpu by default..
00:53
<Jenna>
oh. but I have found ltsp 5 to be a little beefer client that ltsp 4.X ?
00:53
<johnny>
since it tunnels X over ssh
00:54
sure.. but it doesn't have much specific to ltsp.. it uses more standard distro stuff
00:54
so..don't blame ltsp :) it's the tools
00:54
<Jenna>
IC. what is the least Mhz that ltsp 5 requires ?
00:55
<johnny>
i'm not sure
00:55
you can disable the ssh tunneling tho.. and should make it run a bit better
00:55
<Jenna>
I have been running ltsp 4.2 on 233 Mhz (optiplex) pcs for 3 years now. but now I want to move onto ltsp 5 for sounds sake.
00:55
<johnny>
that's probably unlikely
00:56
even if you disable the ssh tunneling
00:56
ltsp 4 was a custom distro
00:56
ltsp5 uses the host distro for more
00:58
<Jenna>
hmm. I need to setup a ltsp lab for underpriviliged . I am havving hard time in choosing the terminal clients
00:59
the funding it tight. cant find $100 - $150 ltsp compliant clients
01:00
<johnny>
i guess it depends on if you can rely on craigslist..
01:00
or try to get some machines from companies who are on their upgrade cycle
01:01
others here might have suggestions, but this is the wrong time
01:01
you might want to come back 12 hours from now
01:02
or post to ltsp-discuss if that isn't a possibility
01:04
<Jenna>
naah. I can be here
01:04
what sort of client ur using ?
01:04
<johnny>
donated machines
01:05
but they are all pentium 3 or 4
01:05
nothing under 600mhz
01:05
not by choice.. that's just how it came down
01:06
<Jenna>
hmm. I wanted to cut down on power costs as well. so I was hoping for those slicker/thinner kinds
01:08vagrantc has joined #ltsp
01:31japerry has joined #ltsp
01:34cyberorg has joined #ltsp
01:35nubae has joined #ltsp
01:46
<Ahmuck>
jenna, i located some 150 clients on newegg the other day
01:47
<Jenna>
ah. what brand ?
01:48
<Ahmuck>
http://www.newegg.com/Product/ProductList.aspx?Submit=ENE&N=50002136&Description=atom%20barebone&name=FOXCONN
01:48
foxconn
01:50Ahmuck has quit IRC
01:51Ahmuck has joined #ltsp
01:51
<Jenna>
and I suppose they work out-of-the-box with ltsp5
01:51
?
01:52
<Ahmuck>
ack, firefox is closing randomly recently
01:52
anywho, haven't tested it myself with ltsp5 thin clients
01:53
been thinking bout it
01:53
no cdrom, and have to buy mem. but i speced mem out at 20.00 for a G iirc
01:54
though i can say i don't prefer that case. i prefer the older cases
01:54
<Jenna>
hmm.
01:55
what about thincan. have any experience on those
01:57
<Ahmuck>
interesting
01:57
haven't seen thincan before
01:58
<Jenna>
they seem slick
02:07Ahmuck has quit IRC
02:08vagrantc has quit IRC
02:10Ahmuck has joined #ltsp
02:25hanthana has joined #ltsp
02:27l0C0 has joined #ltsp
02:33Jenna has quit IRC
02:36vagrantc has joined #ltsp
02:42primeministerp has quit IRC
02:52bobby__C has joined #ltsp
02:54nahita has joined #ltsp
02:58bobby__C has quit IRC
02:59nahita_ has joined #ltsp
03:02cyberorg has quit IRC
03:02sutula has quit IRC
03:02stgraber has quit IRC
03:02flokuehn has quit IRC
03:02sep has quit IRC
03:02daduke has quit IRC
03:03makghosh has joined #ltsp
03:03cyberorg has joined #ltsp
03:03sutula has joined #ltsp
03:03stgraber has joined #ltsp
03:03flokuehn has joined #ltsp
03:03sep has joined #ltsp
03:03daduke has joined #ltsp
03:09bobby_C has joined #ltsp
03:12vagrantc_ has joined #ltsp
03:18artista_frustrad has quit IRC
03:19vagrantc has quit IRC
03:19artista_frustrad has joined #ltsp
03:20tux_440volt has joined #ltsp
03:21tux_440volt is now known as subhodip
03:22bobby_C has quit IRC
03:23nahita has quit IRC
03:38makghosh has quit IRC
03:38subhodip has quit IRC
03:42tux_440volt has joined #ltsp
03:45makghosh has joined #ltsp
03:47vagrantc_ has quit IRC
03:55makghosh has quit IRC
04:09ajohnson has quit IRC
04:29nahita_ has quit IRC
04:38
<nubae>
highvoltage, ping
04:38F-GT has quit IRC
04:39F-GT has joined #ltsp
04:49
<highvoltage>
nubae: pong
04:59tux_440volt has quit IRC
05:08tux_440volt has joined #ltsp
05:15subhodipbiswas__ has joined #ltsp
05:15tux_440volt has quit IRC
05:16subhodipbiswas__ has quit IRC
05:18ogra has quit IRC
05:19ogra_ has quit IRC
05:20ogra has joined #ltsp
05:20ogra_ has joined #ltsp
06:01otavio has quit IRC
06:03otavio has joined #ltsp
06:05CAN-o-SPAM has joined #ltsp
06:19pmatulis has joined #ltsp
06:58intelliant has joined #ltsp
07:03GodFather has joined #ltsp
07:16gate_keeper__ has quit IRC
07:16gate_keeper__ has joined #ltsp
07:18artista_frustrad has quit IRC
07:19artista_frustrad has joined #ltsp
07:40cliebow has joined #ltsp
07:43lucascoala has joined #ltsp
07:49nubae has quit IRC
07:56denisesballs has quit IRC
07:58primeministerp has joined #ltsp
08:03hanthana has quit IRC
08:13ajohnson has joined #ltsp
08:24max9ai has joined #ltsp
08:24rbla1 has joined #ltsp
08:27
<max9ai>
Anybody here?
08:31
<cliebow>
just wallflowers like me'
08:32
<max9ai>
:)
08:33
Did you tried to install ltsp server under gentoo? :)
08:33
<_UsUrPeR_>
this place was boring yesterday too.
08:34
<max9ai>
o_O
08:37
<_UsUrPeR_>
^^ no one was here and talking
08:37
max9ai: I have not tried ltsp out w/gentoo yet
08:38
<max9ai>
I can't to emerge it... There are ebuild's bugs or I'm a blit. :(
08:39
But I can't to fix it... or may be I'm doing something wrong...
08:40GodFather has quit IRC
08:43
<_UsUrPeR_>
max9ai: What?
08:43
I recommend you try something that is not gentoo.
08:44grantk has joined #ltsp
08:44
<_UsUrPeR_>
Ubuntu, Fedora, OpenSUSE
08:44
<cliebow>
there "were" gentooites around somewhere..
08:45
<ogra>
johnny does the gentoo port
08:45
and i think there are known issues
08:46
<_UsUrPeR_>
huh. That's genuine news to me
08:47
<grantk>
Hello
08:47
Anyone using kde 4.2 with ltsp 5? I am having quite a few stability issues with it.
08:48Blinny has joined #ltsp
08:49
<grantk>
I have been looking at my Xsession errors from various users and it appears that it could be related to firefox and it's love of memory
08:50
looked into NBD swap and enabled it. But so far I was expecting it to create swap directories under the /tmp dir but
08:51
I have not seen anything there, anyone know of a good way to actually tell if NBD swap is actually being used?
08:53vagrantc has joined #ltsp
08:56
<ogra>
dont use kde4 with ltsp
08:56
<vagrantc>
that bad, eh?
08:56
<ogra>
well, i have only seen people where it simply didnt work yet
08:57
<_UsUrPeR_>
I was trying to come up with a nice way to say that...
08:57
<grantk>
In a normal work day I have some people crash about 4-5 times
08:57
then some machines it runs great
08:57max9ai has left #ltsp
08:58
<_UsUrPeR_>
I have had a brain fart, does anyone know the name of the pam/ldap conversion tool for authentication?
08:58
google's not helping either :)
09:00
<vmlintu>
_UsUrPeR_: conversion tool? You mean how to get pam authenticate using ldap?
09:00
<_UsUrPeR_>
vmlintu: yeah, that's the one
09:00
<vmlintu>
pam_ldap.so
09:00
<ace_suares>
vmlintu: yep, found that on google too :-)
09:01
<vmlintu>
libpam-ldap on debian/ubuntu
09:03
<_UsUrPeR_>
grr :)
09:03
thanks
09:07
<rbla1>
Hello, I have about 12 fat clients running off an Ubuntu 9.04 LTSP server. Every morning about 10 or so have hung up. When I switch to TTY1 I get the message:
09:07
Kernel call returned: Broken pipe Reconnectiong
09:07
Negotiation: ..size =720320KB
09:07
Ioctl/1.1a failed: bad file descriptor
09:07
Some googling has pointed me to some other people who have had the same problem, but the only solution I saw was to switch from NBD to NFS. I did this, but the clients will not even boot with this configuration. Can anyone give me any insight into why this is happening/how to fix it?
09:08bieb has joined #ltsp
09:09brendan` has quit IRC
09:09brendan` has joined #ltsp
09:11
<grantk>
is your /etc/exports file configured properly for nfs?
09:11
<rbla1>
the only entry is
09:11
<vagrantc>
rbla1: how far do they boot?
09:11
<rbla1>
/home_ltsp 192.168.0.0/24(rw,no_root_squash,async,no_subtree_check)
09:11intelliant has quit IRC
09:13
<rbla1>
They boot completely then sometime overnight they all hang up
09:14
<vagrantc>
that's a rather odd location for LTSP
09:14
normally it's installed in /opt/ltsp/i386
09:14
and it shouldn't be writeable
09:14Gadi has joined #ltsp
09:14
<vagrantc>
they all use the same filesystem, so it should be read-only
09:15
<grantk>
that is what I have on my old ltsp 4.2 boxes..../opt/ltsp/i386 192.168.0.0/255.255.0.0(ro,no_root_squash)
09:15
<_UsUrPeR_>
grant: have you tried localapps with firefox? If you're concerned about memory, you can always shift the memory demands to your clients...
09:16
<grantk>
thinclients unfortunatley
09:16
most are old pII machines
09:16
<_UsUrPeR_>
ahh.
09:16
<grantk>
just recently started to convince them to go newer, so the ones that run well are intel atom based clients with around 1gb mem
09:17
if I had an office full of those I would be happy
09:17
some of the old pII's can swing it though, it just the ultra low memory models
09:18
<bieb>
ok.. I have a wiring/server setup question.... I built and tested my ltsp server in my office with a small switch with 2 PCs (clients), the server has 2 NIC cards, eth0 was to the regular network here at the office eth1 was connected to the switch. Now that we are moving it into the actual server room (windows DHCP will be used, https://help.ubuntu.com/community/UbuntuLTSP/LTSPWindowsDHCP) do I assign a static IP to both eth0 and eth1? eth1 having a 192.
09:18
<_UsUrPeR_>
have you taken a look at http://www.disklessworkstations.com 's 1520 atom yet?
09:18* _UsUrPeR_ shamelessly plugs
09:18
<bieb>
_UsUrPeR_: you work there too?
09:19
<_UsUrPeR_>
yeah :D
09:19
<bieb>
cool.. I know CAN-o-SPAM does also
09:19
<_UsUrPeR_>
yeah, he's in my office.
09:20
<grantk>
seen those before, would love to get the bossman to throw down for some
09:20
<ltsplogbot>
I don't know who those is.
09:20
<bieb>
I have 5 of my new 1520's installed in the lab.. once I figure out the answer to the question above.. I will be able to complete my testing.. Once that is done, the other 15 clients will be installed
09:22
<_UsUrPeR_>
bieb: have you tried out the ubuntu alternate install cds?
09:22
bieb: wait, are you using ubuntu? :)
09:22
<bieb>
already running Ubuntu
09:22
<_UsUrPeR_>
the reason I ask about the alternate install CD: It's standard setup is as followd: eth0 = DHCP by default, eth1 = 192.168.0.1
09:23
<bieb>
I have tested and all is working.. but that was sitting in my office, with a dedicated switch
09:23
<_UsUrPeR_>
you don't *HAVE* to assign a static IP to eth0 if you just want to keep DHCP, but it works with a static IP as well
09:23
<bieb>
I just want to make sure its the same since DHCP will be handled by a Windows DHCP server
09:24
<rbla1>
vagrantc: what do you mean that is a rather odd location for LTSP? This is exporting a copy of the home directory with the clients. My LTSP image is located in /opt/ltsp/fat2
09:24
<_UsUrPeR_>
bieb: DHCP is only handed out on eth1 by default when installed with the alternative installation
09:25
<vagrantc>
rbla1: ah, i i thought you meant that it was the LTSP filesystem...
09:25
<bieb>
_UsUrPeR_: in that document I linked, where it says: 066 Boot Server Host Name: <ip address> that IP should be the 192.168 address correct?
09:25* _UsUrPeR_ looks at the link
09:26
<vagrantc>
rbla1: you'll probably want an entry in /etc/exports like so: /opt/ltsp *(ro,no_root_squash,async,no_subtree_check)
09:26
and restart nfs-kernel-server
09:26
<_UsUrPeR_>
vagrantc: oh. I see. So you will be using windows' dhcp server to give IP addresses to your clients
09:27
<bieb>
_UsUrPeR_: you mean bieb
09:27
:D
09:27
<_UsUrPeR_>
err... yeah
09:27* _UsUrPeR_ saw that too late
09:27
<bieb>
correct
09:27
<rbla1>
vagrantc: is the only fix for this problem using NFS instead of NDB. As I understand it currently my image is being shared via tftp. Sorry I think I'm in a bit over my head here.
09:27* vagrantc will never use windows anything :)
09:27
<_UsUrPeR_>
lulz
09:27
<vagrantc>
rbla1: the image isn't shared via tftp, just the kernel+initramfs is downloaded via tftp
09:28
rbla1: the OS is mounted via NBD or NFS
09:29
<rbla1>
vagrantc: So does the "Ioctl/1.1a failed: bad file descriptor" error mean I have something wrong with my NBD configuration?
09:29
NBD now.
09:29
I tried NFS, but the system would not even boot.
09:30
I followed a guide (http://doc.ubuntu.com/edubuntu/edubuntu/handbook/C/customizing-thin-client.html) but something tells me that I missed a step because as far as I can see th guide never has you export anything via NFS
09:31
<_UsUrPeR_>
bieb: ok, just finished reading
09:31
<bieb>
ok
09:32
<grantk>
rbla1 I used the same guide, but edited my exports to make it work
09:32
<_UsUrPeR_>
I would like you to get a second opinion on this from someone experienced with Windows DHCP, but that IP address should be your eth1 address. As long as you are running your server without modification.
09:33
I know you can use the same eth card for ltsp, but I have never set mine up like that
09:33
<bieb>
ok.. thats what I was thinking
09:33
<rbla1>
grantk: what exactly did you export to make it work?
09:34
<grantk>
Next you will need to edit the /etc/export file to allow the sharing of the LTSP image. The path that you share must match the option root-path line in the dhcpd.conf file. following is the only line we use in our floyd exports file./opt/ltsp/i386      192.168.1.0/255.255.255.0(no_root_squash,no_subtree_check)
09:34
that was the part I added to my personal documentation on it
09:34
<_UsUrPeR_>
bieb: so assign a static IP address to eth1, and make sure that you point the thinclients toward that IP in the windows dhcp configuration file
09:35
<bieb>
_UsUrPeR_: thanks... in the line 067 Bootfile Name: ltsp/i386/pxelinux.0 # for Edgy and up is that the correct location of the bootfile in 9.04?
09:35
<vagrantc>
grantk: you can actually use /opt/ltsp, rather than /opt/ltsp/i386
09:36
unless /opt/ltsp/i386 is mounted on a different filesystem than /opt/ltsp
09:36
<grantk>
I think that was done because there is a 64bit image in there as well
09:36
<ogra>
and * for the ip if the network is for thin clients only anyway
09:36
<grantk>
sorry, that was for vagrant
09:36
<vagrantc>
as i said earlier, i would recommend: /opt/ltsp *(ro,no_root_squash,async,no_subtree_check)
09:36
<ogra>
yeah
09:37
<vagrantc>
grantk: yes, and /opt/ltsp will handle /opt/ltsp/amd64 as well as /opt/ltsp/i386 as well as /opt/ltsp/SOMEOTHERTHINGY
09:37
so you only need one line
09:37
<grantk>
cool, will give it a shot
09:37
vagrantc : offer any better performance or just make things easier to read?
09:37
<ogra>
rbla1, you dont shut down your server over night, do you ?
09:38
<_UsUrPeR_>
bieb: yeah, you're definitely not using dapper :)
09:38* ogra hopes *nobody* uses dapper anymore :)
09:38
<bieb>
right
09:38
<vagrantc>
grantk: you don't have to restart the nfs server everytime you add something in /opt/ltsp/
09:38
<_UsUrPeR_>
werd
09:38
<bieb>
_UsUrPeR_: I was just checking to make sure it wasn't moved again
09:38
<rbla1>
ogra: no it runs all the time
09:38
<vagrantc>
grantk: and it works in all cases, instead of having it match anything
09:38
<_UsUrPeR_>
naw, it's not going anyplace...
09:39
<vagrantc>
grantk: i.e. you change your network, you change which directory you want mounted, etc.
09:39
<ogra>
rbla1, did you check your server logs for the night ? if any restart of inetd or nbd services occured
09:39mikkel has joined #ltsp
09:40
<grantk>
good to know, thanks vagrantc
09:40
<rbla1>
ogra: i will check that now
09:41
<grantk>
anyone know the name of the nbdswap directories created in /tmp?
09:41
<ogra>
look in nbdswapd :) its a script
09:42
<rbla1>
ogra: as far as I can tell the services were not restarted overnight.
09:42
<ogra>
well, your connection dying somehow indicates that
09:44
<rbla1>
ogra: are there any logs I should check other than /var/log/syslog?
09:46
<grantk>
ogra: checked out nbdswapd and it did not really say, it started up /sbin/mkswap but can not look at that one it is a binary file
09:46
<ogra>
SWAP=$(mktemp $TEMPFILE_OPTS)
09:47
and TEMPFILE_OPTS="${SWAPDIR}/XXXXXX"
09:47
so whatever mktemp makes out of that
09:47
man mktemp :)
09:49dorimar has joined #ltsp
09:49
<grantk>
ogra: I see, thanks, definatly do not have any dirs created by mktemp in there, appreciate it
09:54
<rbla1>
thanks everyone
09:54rbla1 has left #ltsp
10:04Blinny has quit IRC
10:14staffencasa has joined #ltsp
10:15dorimar has quit IRC
10:16dorimar has joined #ltsp
10:22
<grantk>
is the ltsp localdevice package required for nbd in ltsp5?
10:22
I am seeing that it is for ltsp 4.2
10:23
<johnny>
no
10:23
nbd is integrated into the initital ramdisk
10:25
<grantk>
ok, thanks johnny
10:39l0C0 has quit IRC
11:00pmatulis_ has joined #ltsp
11:05grantk has left #ltsp
11:14vagrantc_ has joined #ltsp
11:18artista_frustrad has quit IRC
11:19artista_frustrad has joined #ltsp
11:21vagrantc has quit IRC
11:22highvoltage has quit IRC
11:22highvoltage has joined #ltsp
11:27RobertBerger has joined #ltsp
11:51grantk has joined #ltsp
11:51vagrantc has joined #ltsp
11:53
<grantk>
wanted to thank you guys again. Got my terminals using nbdswap so I will see if they become more stable. Guy that set it up originally did not disable the nfs mounting that we tried a day or two before going back to nbd, or what I thought was going back to nbd.
11:58
<ogra>
grantk, if your server allows it (i.e. has lots and lots of ram) you can use a tmpfs to create the swapfiles in :)
11:59
that will make swapping actually work at a reasonable speed ... otherwise it will be very slow (but not crashing) if it start swapping
12:01tux_440volt has joined #ltsp
12:05vagrantc_ has quit IRC
12:07
<grantk>
only at 4g now and I am already not happy with its usage, so may have to wait until they get me some more memory
12:07
<ogra>
well, you use kde4
12:08
<grantk>
true
12:08
<ogra>
that will hammer the server ram massively
12:09
<grantk>
probably a good idea to switch, or see if kde 4.3 does any better
12:09lucascoala_ has joined #ltsp
12:10
<ogra>
i doubt you will have much fun with any kde after 3 ...
12:10
same will go for gnome 3 once they switched
12:11atomic007za has joined #ltsp
12:13tux_440volt is now known as subhodip
12:16
<grantk>
ya, I went back a few days ago and hated it
12:17
I like gnome but others round here do not, and it would destroy the users worlds, they would have no idea what to do
12:27
<matttttt>
what do you guys use now grantk?
12:27
for a wm?
12:27
<grantk>
most people round her use kde 3.5 or 3.2
12:28
kind of foggy on that now, got one developer who was using xfce, he is moved over to kde4.2 now, and then about 5 people on our ltsp5 server running kde4.2
12:28
working on getting that ltsp5 server stable enough to make another and eventually move everyone here to an ltsp 5 server
12:29lucascoala has quit IRC
12:31pmatulis_ has quit IRC
12:31pmatulis_ has joined #ltsp
12:33lucascoala has joined #ltsp
12:33GodFather has joined #ltsp
12:49lucascoala_ has quit IRC
12:56
<bieb>
how can I change the internal IP on the ltsp server? I am running ubuntu9.04 currently the internal ip that the clients hit is 192.168.0.254, but that address is used on the student network for a gateway
12:57lucascoala has quit IRC
12:57subhodip has quit IRC
12:58
<grantk>
would it be in your /etc/networking/interfaces file?
12:59
<bieb>
grantk: good question I will check. If I change that IP what other things will I have to fix or adjust??
12:59
<grantk>
# The primary network interface
12:59
auto eth0
12:59
iface eth0 inet static
12:59
address 192.168.1.70
12:59
netmask 255.255.255.0
12:59
network 192.168.1.0
12:59
broadcast 192.168.1.255
12:59
gateway 192.168.1.254
12:59
that is from my interfaces file
13:00
<bieb>
my eth0 connects to the internet, the eth1 is the "internal" connection that I need to change
13:01
<grantk>
oh, ok, then you should just change the eth1 places there, you will of course want to change ip, netmask, network, gateway and broadcast if they are different
13:01
<bieb>
ok... thanks
13:02
<grantk>
np
13:03
don't forget to restart networking once changes made
13:03
ifconfig eth1 will of course tell you if they took effect
13:03
<bieb>
cool
13:06cliebow has quit IRC
13:07rjune_wrk has quit IRC
13:09rjune_wrk has joined #ltsp
13:14gate_keeper__ has quit IRC
13:22rjune_wrk has quit IRC
13:22
<bieb>
anyone have some time for troubleshooting? I am getting a TFTP error of "File not found" on my clients, although everything was working prior to moving the server to the server room
13:23fotanus has joined #ltsp
13:24rjune_wrk has joined #ltsp
13:26
<bieb>
grantk:
13:27
grantk: you still here?
13:29
<matttttt>
is it possible your client is getting a reply from a different tftp server?
13:30
have you checked the logs or even tcpdump on the tftp server to make sure the request is making it?
13:31
<bieb>
matttttt: which log on the tftp server?
13:32tux_440volt has joined #ltsp
13:33tux_440volt is now known as subhodip
13:33
<matttttt>
well, to see if the request is even coming in you can look at xinetd.log or equiv
13:33
should show something like this for each connecting client: xinetd.log:09/6/3@11:56:24: START: tftp from=::ffff:192.168.5.53
13:34
or better yet, run something like tcpdump -tni eth0X port 69
13:34
on the tftp server, where X is 0 or 1 and watch that while booting a client.
13:34
if your client is indeed hitting your tftp server, you'll see it, and can move on to making sure the files it should be serving are in the right place.
13:35
<bieb>
ok.. let me go check that..
13:35
<matttttt>
if you don't see the client making a tftp connection, it sounds like your client is hitting a different tftp server on your network, either because that one is responding first, or because your dhcp server is telling it to.
13:35Remaille has joined #ltsp
13:37
<vagrantc>
bieb: did you change the ip, like you were thinking you might?
13:37Lns has joined #ltsp
13:37
<vagrantc>
bieb: when you change the ip of the server, you will likely have to update dhcpd.conf as well
13:37
and restart it
13:41
<matttttt>
hi vagrant, i've got a requirement for one of my ltsp servers to give lan access + internet access and some users only lan access, i'm toying with ideas of how to do this with just the ltsp server (no extra routers, firewalls, etc.). I was hoping iptables had some mechanism to base rules on user/group but it doesn't apear to. My plan b was to force all local http traffic through a local squid proxy and just require auth
13:41
anyone have any tips or better solutions?
13:42
i guess that only helps me with http traffic, if there is legit net traffic on another port i need to allow the 'trusted net' users to have, then everyone will have it.
13:44
<Lns>
Does anyone here have an Ubuntu Jaunty LTSP setup they can test tux4kids apps on with sound? We need to figure out whether the issues in https://bugs.launchpad.net/ubuntu/+source/tuxpaint/+bug/269082 are still present in Jaunty...it would be much appreciated!!
13:45
<Gadi>
matttttt: somethin ala http://www.cyberciti.biz/tips/block-outgoing-network-access-for-a-single-user-from-my-server-using-iptables.html ?
13:45GodFather has joined #ltsp
13:46makghosh has joined #ltsp
13:46
<bieb>
matttttt: here is what shows on the screen from the tcpdump:
13:46
IP 192.168.200.1.2070 > 192.168.10.11.69:39 RRQ "ltsp/i386/pxelinux.0 #" octet tsize0
13:46
IP 192.168.200.1.2071 > 192.168.10.11.69:44 RRQ "ltsp/i386/pxelinux.0 #" octet blksize 1456
13:47
vagrantc: I have removed dhcp from the server since we have a Windows DHCp server that handles the whole network
13:47
<matttttt>
gadi, awesome. i guess i was missing that stuff. i'm going to give it a try now, if it works then i'll be set.
13:47
<Gadi>
matttttt: also, see: http://copilotconsulting.com/mail-archives/sdlug.2004/msg00045.html
13:48
<vagrantc>
bieb: well, then you'll have to configure the windows dhcp server to tell it which server to download from if you've changed ips ...
13:48
<Gadi>
bieb: the Windows server needs to provide the equivalent of "next-server"
13:49
<bieb>
vagrantc: https://help.ubuntu.com/community/UbuntuLTSP/LTSPWindowsDHCP I followed that.. and the 066 bootserver IP is the IP that I changed to
13:49
<matttttt>
hopefully there is a way to replace user or owner with group, that'll be a lot of extra rules for all of these users.
13:49* matttttt reads
13:49
<Lumiere>
matttttt: replace user/owner with group for what?
13:50
<Gadi>
matttttt: have a look at my last post
13:50
it uses --gid-owner
13:50
<matttttt>
ah, nice.
13:54
<bieb>
any other ideas or things to check? matttttt vagrantc Gadi
13:55Remaille has quit IRC
13:58dorimar has quit IRC
13:58dorimar has joined #ltsp
14:12GodFather has quit IRC
14:12
<grantk>
Sorry, was out to lunch bieb
14:14
<bieb>
grantk: no prob.. the IP info you sent me earler from your interfaces was that auto configured by ubuntu?
14:14
<grantk>
Good question, I did not set it up, just have set my own static ip boxes up that way before.
14:15
you still not getting the ip to change?
14:21
<bieb>
no.. I got it to change.. was just curious how yours was setup
14:22
<Gadi>
bieb: just out of curiosity, do you have a leading slash in your filename?
14:22
boot filename, that is
14:23
<bieb>
Gadi: :D my boss made a mistype in the filename. He added a # at the end
14:23
<Gadi>
ah
14:23
all better?
14:24
<bieb>
yeah.. except now I cant authenticate agains Active Directory.. "no response from server' so I am looking at the doc that I used to get AD auth working
14:27
Gadi: is there anything in Likewise that would get messed up by an IP change?
14:28
<Gadi>
AD is DNS sensitive
14:28
you may need to rejoin the domain
14:28
evrything else should be fine
14:29
make sure DNS is working well (ie you can resolve the AD server by hostname with a ping)
14:29
<bieb>
ok
14:29
<Gadi>
then unjoin the domain and rejoin the domain
14:30
<bieb>
cool.. let me take a look at those things now..
14:31
<Gadi>
Lns: ping
14:33
<Lns>
Gadi: pong
14:33
(on phone too, sorry if delayed response)
14:34
<Gadi>
Lns: I was able to reproduce your xrestop test on openoffice
14:34
5MB -> 26MB
14:34
or so
14:35
I know its cheesy, but you should be able to work around it by launching openoffice as a single application through NX
14:36
i am guessing (tho haven't tried) that NX would mitigate the pixmap usage
14:37
<Lns>
hmm
14:37
I don't have NX set up at all, i'd hate to do that at all 7 schools :(
14:37
<Gadi>
ah
14:38
nevermind
14:38
:)
14:38
<Lns>
Thank you for the workaround though! I might try piloting that at some point, but i sure would love fix the core issue
14:38
or at least be involved in it ;)
14:39
Gadi: have you seen google's nx server implementation yet?
14:40
<Gadi>
nope - been buried under a rock
14:40
you?
14:40
<Lns>
not yet..just heard about it, not sure if it's even available to install yet
14:52
<bieb>
Gadi: can not ping the hostname.. but can ping the IP
14:52* Lns thinks we need to form an "LTSP action team" that puts pressure on upstream devs to make apps more compliant with thin client environments
14:56
<bieb>
if I disappear.. I will be right back..
14:58
<Gadi>
bieb: "close" only counts in horseshoes, as they say
14:58bieb1 has joined #ltsp
14:59
<rjune_wrk>
Gadi: horseshoes, hand grenades, and atomic bombs
14:59
<bieb1>
Gadi: very true
14:59
<Lns>
and US presidential elections, apparently ;)
15:04pmatulis_ has quit IRC
15:04
<johnny>
Lns, that's not how it works exactly..
15:04
Lns, the action team needs to raise money to hire people to fix the issues
15:05
that's the only way it will work
15:05
or convince canonical that they should hire people :)
15:06
<Lns>
johnny: i'm thinking more distro-agnostic approach
15:06
<johnny>
it's distro agnostic to fix upstream projects
15:06
<Lns>
right, but canonical ...
15:06
<johnny>
just picking somebody with money
15:06
<Lns>
hmm, maybe LNS can be the company t do this =)
15:06
not that i have money
15:07
<johnny>
well.. you gotta raise it then
15:07
if 1000 people have the problem, and they commit $100 each.. then you're talking real money :)
15:07
<Lns>
hmmmm
15:07ab2 has joined #ltsp
15:07* Lns will think about this on the way to the school to troubleshoot SDL audio & pulseaudio
15:08
<Lns>
Food for thought
15:08
thx johnny =)
15:08
bbl
15:08Gadi has left #ltsp
15:15dorimar has quit IRC
15:16bieb has quit IRC
15:25
<brendan`>
has anyone here ever had major problems with i386 clients connecting to an amd64 server? i'm running 8.10 on the server, and i can't get my clients to stop completely freezing, sometimes on completely trivial tasks such as saving a file
15:25
i've been debugging for 3 weeks now and thought i had a solution but just found out it's still happening
15:25
the same clients worked fine on the same server until i upgraded from i386 to amd64 on the server
15:27pimpministerp has joined #ltsp
15:27artista_frustrad has quit IRC
15:27artista_frustrad has joined #ltsp
15:29pmatulis has quit IRC
15:38
<vmlintu>
brendan`: I've seen freezes running i386 clients on amd64 server, but haven't ever looked if that would be a cause
15:39
brendan`: what kind of clients do you have?
15:41
and how often do you see the freezes?
15:42subhodip has quit IRC
15:42
<brendan`>
they aren't thin clients per-se
15:42
most of them are old dell optiplex's
15:43
the one in question the most is a p4 1.7ghz with 2gb ram
15:43grantk has left #ltsp
15:43
<brendan`>
the freezes happen randomly
15:43
last week they were happening very often
15:43
every hour or so
15:43
i've tweaked some settings in the lts.conf file and got them to stay pretty stable
15:44
<vmlintu>
every client froze once an hour?
15:44
<brendan`>
but apparently one of my users yesterday couldn't even save a file in open office without it freezing
15:44
no, that's the weird part too
15:44
some clients will stay on for whole days
15:44
then randomly freeze
15:44
the biggest culprit however though is when my users use vmware-server-console to get to a virtualmachine on another server
15:45
so im thinking it has something to do with pixmap caching
15:45
but, the user that was freezing all day yesterday doesn't use vmware, and was only trying to use openoffice
15:46
<vmlintu>
freeze = mouse doesn't move, keyboard does nothing, screen stops changing, network is dead?
15:46
<brendan`>
only one freeze today though, but when i looked that use had at least 3 firefox windows open, a bunch of other stuff, and was trying to paste a huge image into open office
15:46
well, the freeze today was weird, mouse still moved, just couldn't click anything, could still switch over to a terminal
15:46
but, for the last 3 weeks, they have been total freezes
15:46
no mouse, no terminal, only thing to fix them was power off
15:47
im wondering if it is network related
15:47
i have seen alot of NBD errors in my logs
15:47
<vmlintu>
I've heard rumours of these kind of crashes
15:47
<brendan`>
mostly on thin client startup though
15:48
i did actually have to swap out my gigabit card because i had an intel pro e1000 and it wouldn't work with 8.10amd64
15:48
<ab2>
any pointers on troubleshooting LTSP-5 (Ubuntu 9.04) booting via NFS. The root fs doesn't appear to get mounted rw and I get errors about readonly FS.
15:48
I've been tracing through ltsp-client-setup thus far.
15:48
<vmlintu>
our systems are on ubuntu 8.04, but the two kinds of freezes sound the same
15:49
<brendan`>
ab2: edit your /etc/default/ltsp-client-setup under your ltsp chroot
15:50
change your rw_dirs and add in /root
15:50
vmlintu: are you booting nfs or nbd?
15:50
<ab2>
brendan: thanks, giving it a try now
15:51
<brendan`>
remember to rebuild your image after
15:51
<vmlintu>
brendan`: nbd
15:51
<brendan`>
Jul 28 10:02:09 ltspserver nbd_server[25346]: Read failed: Inappropriate ioctl for device
15:51
i get lots of errors like that
15:51
none at the time of freezing though
15:52
only when the clients boot up
15:53
<vmlintu>
I've been suspecting that at least some of the hangs could be irq related, but haven't been able to verify that at all..
15:53
<brendan`>
yeah
15:53
i've been wondering, last time i asked in here someone said it could be network related
15:54
im wondering if for some reason packets are getting dropped, the nbd connection is dying, and therefore the client goes down
15:54
but, i tried reverted back to nfs, and was still getting the same issues
15:54
<vmlintu>
my understanding is that if nbd goes down, mouse still moves
15:55
<brendan`>
ahhh
15:55
so in my case it has to be the x server maxing out either ram or videoram
15:55
and the x session crashes
15:55
<vmlintu>
have you tried pinging the clients when this happens?
15:56
<brendan`>
that's why i thought i had everything solved this weekend, i tried hardcoding my video ram in the lts.conf
15:56
and everything seemed good
15:56
i stress tested one of the clients for a good hour, running several instances of OO, firefox, and loaded the vmware server console
15:56
but then yesterday one of my users was freezing saving a file on that same machine
15:56
<vmlintu>
do all the clients have 2 gigs of memory?
15:56
<brendan`>
when the clients freeze, there is nothing
15:57
no ping, no ssh, no terminal
15:57
no, but none have less than 512
15:57bieb1 has left #ltsp
15:58
<vmlintu>
have you seen the freezes ever happen without someone moving a mouse or typing on keyboard?
15:58
<ab2>
adding /root to rw_dirs didn't help yet. I am dumping to a bash shell to pke around and noticed that I cannot cd /root/ even when 'id' says i'm uid=0. /root is 700
15:59
<brendan`>
no, that was another question of mine, alot of times it seemed to happen explicitly when someone was moving the mouse
15:59
mainly over a window that tries to take focus
15:59
like a toolbar in openoffice
15:59
or the vmware server console tries to grab control
15:59
<vmlintu>
I've seeing exactly the same behaviour on my test thin client every 2-3 days now
16:00
<brendan`>
ab2: did you update your image?
16:00
<ab2>
im not using nbd so it was immed. after a reboot.
16:00
<vmlintu>
usually is freezes exactly at the moment when I try to click on icon or something else
16:00
it's got 1Ghz via processor with 512 megs of memory
16:00
<brendan`>
hmmm
16:01
that sounds very close to what i have seen happen
16:01
<vmlintu>
icon being for example icon on evolution's toolbar
16:01
<brendan`>
ab2: if you're using nfs you have to rebuild everytime you do anything
16:01
right, and it tries to take focus of the mouse
16:02
to update a status bar or pop a help balloon
16:02
<vmlintu>
probably
16:02
<brendan`>
that sounds very similar to issues i've seen
16:02pmatulis has joined #ltsp
16:03
<brendan`>
ab2: did you also set BOOT=nfs in your /etc/initramfs-tools/conf.d/ltsp in your ltsp chroot?
16:03
when you change that, you have to update your initramfs
16:03
<vmlintu>
I've been following the memory usage on the thin client by running top, but I've never caught it exactly at the moment when it freezes
16:03
<ab2>
brendan: I did set that. By update do u mean run ltsp-update-image?
16:04
<vmlintu>
But there's always plenty (>200 megs) of memory free when I check it
16:04
<brendan`>
you have to update-initramfs -u in your chroot
16:04
<vmlintu>
And I don't run openoffice
16:04
<brendan`>
then sudo ltsp-update-kernels outside of chroot
16:04
yeah, i've been watching xrestop, top, htop, iftop, iotop, etc nothing shows anything funky
16:04
i have all my clients remote logging to syslog
16:04
nothing indicates anything wrong
16:05
<vmlintu>
that sounds the same
16:05
and that's why I've been beginning to suspect irq related problems
16:05
<brendan`>
ab2: and yes, it doesn't hurt to ltsp-update-image as well
16:06elisboa has quit IRC
16:06ace_suares has quit IRC
16:06
<brendan`>
how do you suppose we get a dev involved in this?
16:06
my users are beginning to get real pissed at me
16:07
<vmlintu>
when you updated to amd64, did you update the ubuntu version also?
16:07
and did you ever see this before the upgrade?
16:08
<brendan`>
no, we were running an 8.10 i386 setup on the same hardware for almost a year
16:08
then 2 of our 4 hard drives crashed within 2 weeks (luckily they were RAID!!)
16:08
so i did an overhaul of the whole thing
16:08
and figured amd64 was up to speed nowadays so why not take advantage of having an amd64 chip
16:09
4 new hard drives, upped the memory to 4gigs and installed a fresh copy of 8.10
16:09
<vmlintu>
how many users do you have on it?
16:10
<brendan`>
when all my users are here, 10
16:10
it will freeze at times when there are only 1 or 2 though
16:10lucascoala has joined #ltsp
16:10
<brendan`>
and oddly enough, i have one machine that is a crappy old dell, maybe 700mhz, 256ram, integrated video, and it hasn't crashed yet
16:11
but my three optiplex's have
16:11
and the whitebox 1.7ghz p4 with 2gb ram and 32mb nvidia tnt card is the one that crashes the most
16:13
<vmlintu>
the only thing that comes into my mind right now is to try booting the kernel with irqpoll parameter
16:14
<brendan`>
hmm, i'll take note of that
16:15
can't change anything right now, going to NYC for a meeting tomorrow, don't want to lock every one out somehow
16:15
<vmlintu>
I've seen some thin clients behave differently with that
16:16
but I have not tried them for these freezes as this via box doesn't like the parameter
16:19
btw, do you use ps/2 or usb mice and keyboards?
16:20atomic007za has quit IRC
16:27
<brendan`>
combination of both
16:27
mostly usb mice though
16:27
a few ps/2 keyboards
16:28
<vmlintu>
http://ubuntuforums.org/archive/index.php/t-189572.html
16:29pimpministerp has quit IRC
16:29
<vmlintu>
it's old, but it talks about turning off legacy usb support in bios
16:29
maybe you could try that also
16:29CAN-o-SPAM has quit IRC
16:30
<brendan`>
hmmm
16:30
i wonder if noapic or apic in the kernel would have any effect on this
16:31
<ab2>
brendan: updated /etc/default/ltsp-client-setup to include "/root" in rw_dirs, in chroot ran update-initramfs -u, then outside chroot ran ltsp-update-kernels. /root still not readable.
16:31
<brendan`>
i did you add BOOT=nfs in /etc/initramfs-tools/conf.d/ltsp ?
16:32
<ab2>
brendan: yep, just double checked
16:32
<brendan`>
hmmm
16:33
wait
16:33
or writeable?
16:33
./root is not readable?
16:33
<ab2>
boot is now stuck as Server Authorization directory perms error msg.
16:33
yep is not READable
16:33
mode is 0700
16:33
<brendan`>
oh
16:33
<ab2>
`id` returns i am 0
16:34
<brendan`>
hmmm
16:34
you got something else going on
16:34
what are you running?
16:34
ubuntu/debian/fedora?
16:35F-GT has quit IRC
16:35
<ab2>
the ltsp is from Ubuntu 9.04 it is being hosted by HPUX11.23 NFS and ISC DHCP
16:35
<brendan`>
oh right
16:35
hmmm, im not sure what's going on
16:36F-GT has joined #ltsp
16:36
<ab2>
from bash I have these "/" mounted: rootfs on / type rootfs (rw) AND IP:/opt/ltsp/i386 on / type nfs (lots of options here)
16:38
does that sound right? This is before the execution of /etc/init.d/ltsp-client-setup
16:38
<vmlintu>
brendan`: if you try the kernel options or bios settings, I'd be really interested in knowing the results..
16:39
<brendan`>
vmlintu: i'll note that, i'll let you know
16:40
i won't be able to get to it until maybe this weekend
16:40
<vmlintu>
other thing to do would be to find out how many others are seeing these freezes
16:40
<brendan`>
i've been scouring the internet, searching through bug reports, forums, archives, etc
16:40
i have not found anything all that similar
16:41
and anything i have found, haven't improved my situation
16:41
<vmlintu>
I copy that..
16:41
<brendan`>
although, i shouldn't say that
16:41
it has improved
16:41alkisg has joined #ltsp
16:41
<brendan`>
i used to be able to get them to freeze pretty consistently
16:41
but i tweaked a few things
16:42
and now i stress tested one last weekend, and i couldn't get it to freeze no matter what i did
16:42
i had firefox open with many tabs, open office going in several instances, and vmware server console open
16:42
nothing
16:42
but then on this same machine, one of my users has been freezing just saving a file in open office, nothing else open
16:42
i'm almost wondering if there are settings in their /home that are screwing with things
16:42
<matttttt>
so i know i could setup a sniffer or network monitor and figure this out, just wondering if anyone knows offhand, haven't had time to check yet, about how much data comes down over the wire when a client pxeboots (everything up to the ldm screen prompt)
16:43
<brendan`>
tons
16:43
<matttttt>
that's not helpful :)
16:43
<brendan`>
haha
16:43
well, the size of your image at least, i would imagine
16:43
<alkisg>
No, it's smaller
16:43
Not all the image is transfered
16:44
<brendan`>
how much of it?
16:44
<vmlintu>
maybe I should start dumping all network traffic to disk to see what the client sends as the last thing before it freezes..
16:45
<alkisg>
brendan`: Well, you could e.g. have 15 Gb image or 150 Mb image; it wouldn't make much difference while booting the client
16:45
Only the files needed while booting would get transferred in both cases
16:45
<brendan`>
interesting
16:46
do people really have 15gb images?
16:46
mine is 241mb
16:46
<alkisg>
Nah, I don't think so :)
16:46
<brendan`>
haha
16:46
<matttttt>
well if my image is 150m? how much is likely being transferred?
16:46
<alkisg>
But I imagine 1 Gb is common with extensive localapps usage
16:46
<brendan`>
right
16:48
<alkisg>
matttttt: I don't remember, but you can easily measure it yourself using e.g. iptraf, iptop, wireshark...
16:52[1]ab has joined #ltsp
16:55
<vmlintu>
I just realized that my test box has a serial port.. I wonder how difficult it would be use it for kernel debugging..
16:57
<matttttt>
hm, that can't be right
16:57
only 1.5m to the ldm screen?
16:57
from a 150m image
16:57ogra_ has quit IRC
16:58
<matttttt>
oh well, i'll find out tomorrow. quitting time
16:58D`aZ has joined #ltsp
17:02lucascoala has quit IRC
17:02fotanus has quit IRC
17:03lucascoala has joined #ltsp
17:03DaZ has quit IRC
17:04D`aZ is now known as DaZ
17:10ab2 has quit IRC
17:14mikkel has quit IRC
17:26[1]ab has quit IRC
17:40alkisg has quit IRC
17:41alkisg has joined #ltsp
17:52epsas is now known as TronCarter
17:59
<staffencasa>
has anyone setup GRUB to PXEBOOT?
18:00
<alkisg>
I read that grub uses an ancient gpxe version, so when I need grub + pxe I just load gpxe.krn with grub...
18:14ace_suares has joined #ltsp
18:40staffencasa has quit IRC
18:44Lns has quit IRC
18:52ace_suares has quit IRC
18:55alkisg has quit IRC
19:03RobertBerger1 has joined #ltsp
19:06pmatulis has quit IRC
19:17GodFather has joined #ltsp
19:19artista_frustrad has quit IRC
19:20artista_frustrad has joined #ltsp
19:21RobertBerger has quit IRC
19:51vagrantc has quit IRC
20:00GodFather has quit IRC
20:22try2free has joined #ltsp
20:27try2free has left #ltsp
20:58makghosh has quit IRC
21:48F-GT has quit IRC
21:51F-GT has joined #ltsp
22:04map7|away is now known as map7
22:17ace_suares has joined #ltsp
22:39F-GT has quit IRC
22:41F-GT has joined #ltsp
22:43RobertBerger has joined #ltsp
22:50map7 is now known as map7|away
22:55Ahmuck-Jr has joined #ltsp
23:01RobertBerger1 has quit IRC
23:08RobertBerger has quit IRC
23:12Ahmuck-Jr has quit IRC
23:20artista_frustrad has quit IRC
23:20artista_frustrad has joined #ltsp
23:30yanu has quit IRC
23:31yanu has joined #ltsp