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


Channel log from 3 September 2014   (all times are UTC)

00:18
<Nemno76>
almost got a working version of xubuntu ltsp now stuck on nbd-client
00:18
Negotiation: Error: INIT_PASSWD badExiting.
00:28
<gdi2k>
Nemno76, sorry never seen that error before. Xubuntu LTSP 14.04 worked out of the box for me (using fat clients)
00:29
<Nemno76>
i'm trying to use it in virtualbox
00:29
and i didn't use the standalone version
00:30
<gdi2k>
I did use the standalone version (although I disabled DHCP as I already have that elsewhere). I have also virtualised my LTSP install, but under KVM not virtualbox
00:30
<Nemno76>
it got a little weird, i got one dnsmasq in my server, virtualbox ...used only for tftp
00:31
i got one dnsmasq outside my virtual box to handle the dhcp stuf
00:32
and now that part is working
00:32
so i get to see a linux kernel boot
00:32
<gdi2k>
I would suspect it has to do with DHCP and dnsmasq
00:32
the client goes through two separate DHCP request cycles
00:33
if it gets sent back to your original tftp server rather than your actual ltsp server during the second DHCP, nbd root fails to mount
00:34
<Nemno76>
without booting a client.. can an working system use nbd-client hostip 9571 /dev/nbd0
00:34
?
00:34
<gdi2k>
I would assume so, but I don't know for sure, never tried it
00:34
but DHCP is most likely your issue
00:34
<Nemno76>
cause that also gives the init_paaswd
00:35
<gdi2k>
you've read this?
00:35
https://help.ubuntu.com/community/UbuntuLTSP/ProxyDHCP
00:35
<Nemno76>
i think so... i've read a lot...
00:36
but i'll check if i missed something
00:36
<gdi2k>
especially the IPAPPEND 3 thing
00:36vagrantc has left IRC (vagrantc!~vagrant@freegeek/vagrantc, Quit: leaving)
00:37
<gdi2k>
you didn't enable encryption for nbd swap?
00:37
<Nemno76>
yes i'v seen that thing... but when i edit /opt/ltsp/i386/etc/ltsp/update_kernels.conf and do a ltsp_update_kernels
00:38
the /var/lib/tftpboot/ltsp/i386/pxelinux.cfg/default is not updated
00:38
<gdi2k>
I had that problem too when I used to use IPAPPEND 3
00:38
had to manually add it each time
00:40
<Nemno76>
yes i checked and its in there the ipappend3
00:40
<gdi2k>
look at: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=127098
00:40
maybe wrong nbd port
00:41
<Nemno76>
what's the standard port for this
00:43
hmmm 10809
00:45
<gdi2k>
yes, seems to be correct. tcp 0 0 0.0.0.0:10809 0.0.0.0:* LISTEN 1047/nbd-server
00:45
<Nemno76>
Negotiation: .Error: It looks like you're trying to connect to a newstyle server with the oldstyle protocol. Try the -N option.
00:45
Exiting.
00:46
<gdi2k>
did the -N option work?
00:47
<Nemno76>
not yet
00:47
checking the manual at the moment it needs an argument
00:50
<gdi2k>
does your nbd config look ok in /etc/nbd-server/ ?
00:55
<Nemno76>
yes i think so
00:56
there is a /etc/nbd-server/config and that uses the /etc/nbd-server/conf.d/ltsp_i386.conf
00:56
<gdi2k>
what xubuntu version are you using? fresh install or upgrade?
00:56
<Nemno76>
fresh install 14.04
01:07Parker955_Away is now known as Parker955
01:09Nemno76 has left IRC (Nemno76!~teknomen@s5375b099.adsl.online.nl)
03:02telex has left IRC (telex!~telex@freeshell.de, Remote host closed the connection)
03:04telex has joined IRC (telex!~telex@freeshell.de)
04:00
<work_alkisg>
gdi2k: you have crashes due to low memory on clients with 2 GB RAM and 8 GB NBD swap?!
04:00work_alkisg is now known as alkisg
04:00
<alkisg>
That doesn't really make sense, are you sure the crashes are related to low memory?
04:00vagrantc has joined IRC (vagrantc!~vagrant@freegeek/vagrantc)
04:04Parker955 is now known as Parker955_Away
04:09
<alkisg>
gdi2k: from http://archive.today/FKlQ:
04:09
"My point is that decreasing the tendency of the kernel to swap stuff out is wrong. You really don't want hundreds of megabytes of BloatyApp's untouched memory floating about in the machine. Get it out on the disk, use the memory for something useful."
04:11
Try with the default swappiness first, verify that no real issues exist, and *then* try to tune it.
04:12
And of course, when RAM is exhausted and swap is used too much, the clients will lag for seconds, even minutes, that's not specific to ltsp
04:13cyberorg has left IRC (cyberorg!~cyberorg@opensuse/member/Cyberorg, Remote host closed the connection)
04:16cyberorg has joined IRC (cyberorg!~cyberorg@opensuse/member/Cyberorg)
04:50Fenuks has joined IRC (Fenuks!~Fenuks@109.202.18.160)
05:00
<gdi2k>
alkisg, sorry was away. yes, seems strange to me too, but the browsers really exhaust the RAM fast with these fancy web applications. I did try with default swapiness initially but tuned it in an attempt to delay swapping (sacrificing cache). I am sure it has to do with swapping, it is very easy to reproduce. Open Firefox or Chrome and open 10 - 15 tabs each. the whole thing will come to a standstill in seconds
05:01
<alkisg>
gdi2k: I don't understand what the problem is
05:01
If you're out of ram, you're out...
05:01
If you're out of ram and play with swappiness and don't see any improvement, it's expected...
05:02
<gdi2k>
alkisg, the problem is that swap is not behaving as expected - if I look at bmon from the server, it is transferring 3 - 6x total consumed swap back and forth across the network. If 800 mb swap is consumed, why does the server send 3 GB worth of network traffic?
05:03
<alkisg>
Because a single page might get transferred 100 times
05:03
That gets worse when you lower swappiness
05:03
App1 says "I need 10 mb", kernel gets 10 mb for it
05:03
App2 says "I need 5 mb", kernel swaps out app1 to get it
05:04
Then when app1 gets a time share again, it re-swaps
05:04
...many times per second, for all browser processes (=tabs), etc etc
05:04
When I said "strange", I meant the crashes, but you said in the ML message that there were no crashes involved, so I don't see anything strange anymore
05:05
Lags for seconds or minutes are to be expected when you're out of ram
05:05
...and are worse when you force a bad swappiness
05:06
<gdi2k>
alkisg, ok, so I'll put swappiness back to where it was. what is your experience with fat clients and RAM? Is 2GB really not enough any more?
05:06
<alkisg>
1 GB RAM is enough for a single app, single tab
05:06
2 GB RAM is enough for normal use, a few apps, a few tabs
05:07
Power users might need more, depending on what they do
05:07
It's not specific to fat clients or ltsp, of course
05:07
The fat client specs are the same as non-ltsp clients
05:08
When I ordered new PCs for schools here, I asked for 4 GB RAM for the clients, 8 for the servers
05:08
<gdi2k>
alkisg, ok thank you for your input. Many of our users do a lot of things with many open documents, long documents, many open browser tabs, Skype, SIP client, etc. etc. so I guess 4GB is required for those
05:09
<alkisg>
Yup
05:09
<gdi2k>
alkisg, I remember the days when..... lol
05:09
<alkisg>
"640 KB should be enough for everyone" :)
05:10
<gdi2k>
I thought with all the web apps, desktop load would become lighter. instead they load the browser so heavily that a native app would have consumed less in the end!
05:11
<alkisg>
Yes, web browsing in general requires much more ram than native apps
05:11
<gdi2k>
I found Office 365's email app consuming 350 MB the other day - a single tab!
05:11
<alkisg>
Try to re-use apps when possible
05:11
<vagrantc>
well, it's because they're actually not running on the server, they're jus downloading the app from the server and running it locally, in javascript interpreters...
05:12
<alkisg>
E.g. if you have both thunderbird and pidgin open, you might want to try the thunderbird chat tab instead of pidgin, saving 200+MB RAM...
05:12
Or, tell users to use google hangouts inside the browsers instead of a skype separate app
05:12
(just examples)
05:13
<gdi2k>
alkisg, ah, user education! ;) but yes, you're right
05:15vagrantc has left IRC (vagrantc!~vagrant@freegeek/vagrantc, Quit: leaving)
05:20bauerski has joined IRC (bauerski!~witekb@frodo.psp.opole.pl)
05:24skyprox has joined IRC (skyprox!~ident@195.20.195.133)
06:12uXus has left IRC (uXus!~uXus@217.77.222.72, Ping timeout: 245 seconds)
06:12uXus has joined IRC (uXus!~uXus@217.77.222.72)
06:13
<skyprox>
how to start LTSP on Centos 7? google no find this theme
06:22
Last news on ltsp.org have date 31 jan 2013
06:24
<alkisg>
No news is good news :D
06:25
OK seriously, I haven't seen much support for centos here, you might want to try another distribution if it's the first time you try ltsp
06:33
<skyprox>
on centos 6.5 ltsp 5.4.5 starting without problem, but start on centos 7 I have problem
06:35
<alkisg>
What problem?
06:35
The person who last maintained ltsp for rpm-based distributions has been away for more than a year...
06:37map7_ has left IRC (map7_!~map7@teksup41.lnk.telstra.net, Remote host closed the connection)
06:50alkisg is now known as work_alkisg
07:07cyberorg has left IRC (cyberorg!~cyberorg@opensuse/member/Cyberorg, Ping timeout: 276 seconds)
07:46skyprox has left IRC (skyprox!~ident@195.20.195.133, Read error: Connection reset by peer)
07:47work_alkisg is now known as alkisg
07:54
<gdi2k>
alkisg, what do you think of zram? better than nbd swap? http://en.wikipedia.org/wiki/Zram
07:54
and zswap
08:10
<alkisg>
gdi2k: when other swap is available, like NBD_SWAP, I prefer not to waste RAM for compressed pages
08:10
I.e. that compcache, zram etc harm ltsp clients, they don't help them
08:11
They would only be helpful if someone wasn't using NBD or local swap
08:11Grembler has joined IRC (Grembler!~Ben@cpc29-aztw22-2-0-cust128.18-1.cable.virginm.net)
08:11
<gdi2k>
alkisg, but it seems to me that nbd-swap is not acceptable in terms of performance to users, so anything that can reduce swap usage in the first place could be progress
08:12
especially when the client;s CPU is fast
08:13
<alkisg>
gdi2k: you misunderstand the problem
08:13
I'll try to describe it...
08:13
You have 2 GB RAM
08:13
Let's suppose that your apps need 2.5 GB RAM
08:14
And, let's suppose that 600 MB RAM of already running applications is NOT used
08:14
...I mean, actively used, the RAM there does contain useful data for applications, but they won't need it for hours
08:15
At that point swap is useful
08:15
Those 600 MB will be swapped to either NBD or local or zram, whatever
08:15
So, now your apps have enough ram
08:15
All that process is *fast*. It's not slow. It's efficient.
08:15
It's not what you're talking about.
08:15
What you're talking about is this:
08:16
after those 600 MB are swapped, your apps still need more ram
08:16
But there are not "infrequently used pages" anymore, that the system can put away for a long time
08:16
So, it swaps *frequently used pages*
08:17
But then those pages are needed again, so it exchanges them with the other frequently used pages
08:17
And it ends up swapping thousand of times per second, causing many GB of traffic
08:17
(moment, phone)
08:19
So, when ram is exchausted, and there are no "infrequently used pages" to swap, *nothing will save you*
08:19
Now, 3rd scenario, suppose you use zswap,
08:20
It might (or might not) make the process of putting pages to swap a bit faster, which *is not significant*,
08:20
but, it will make your clients have less RAM available, which is the most significant thing
08:20
To sum up:
08:20
1) Swap is good, it frees rarely-used RAM for other apps
08:21
2) When you don't have "rarely-used RAM" anymore, swap will make your PCs crawl, they'll be unusuable, and it's just helpful for them not to crash at that point
08:22
3) zram is bad when you have other means of swapping
08:22
...OK, enough time spent on this... :)
08:22
So, 4), just buy more RAM or use less apps :P
08:22
<gdi2k>
alkisg, ok many thanks for the explanation, sounds like a really just have to throw hardware at it
08:26
alkisg, thought there may be just a way to compress swap at the expense of CPU cycles / read write speed, but that's not how it works it seems
08:26
*sorry compress main memory
08:26
<alkisg>
You can compress main memory and that will save you some RAM
08:27
If you don't compress it but put it away to NBD, you'll save *more* RAM
08:27
The problem is that you need more RAM, so zram will make things worse for you.
08:27
zram is useful when you don't have other methods to swap
08:28workingcats has left IRC (workingcats!~workingca@212.122.48.77, Quit: Leaving)
08:28
<gdi2k>
alkisg, understood thank you
08:28
<alkisg>
np
08:29
Ah, a note...
08:29
==> (11:15:40 πμ) alkisg: All that process is *fast*. It's not slow. It's efficient.
08:29
About that:
08:30
zram can make *that* step faster
08:30
But this is *not* the crawl you're experiencing
08:31
(so, zram can help someone task-switch from firefox to libreoffice half a second faster, but when it comes to the "my pc is crawling" issue, it makes things worse)
08:33
<gdi2k>
alkisg, ok thank you, understand.
08:45cyberorg has joined IRC (cyberorg!~cyberorg@opensuse/member/Cyberorg)
08:51F-GT has left IRC (F-GT!~phantom@ppp59-167-136-109.static.internode.on.net, Ping timeout: 245 seconds)
08:52F-GT has joined IRC (F-GT!~phantom@ppp59-167-136-109.static.internode.on.net)
08:58freedomrun has left IRC (freedomrun!~quassel@unaffiliated/freedomrun, Read error: Connection reset by peer)
09:00workingcats has joined IRC (workingcats!~workingca@212.122.48.77)
10:02Nemno76 has joined IRC (Nemno76!~Nemno@s5375b099.adsl.online.nl)
10:10
<gdi2k>
alkisg, I have found that our integrated Intel graphics cards are taking 256M of RAM, which is configurable in the BIOS from 32 - 1024M. I would gain somem valuable RAM if I could safely reduce this. What is the lowest reasonable bios setting for this in your experience, given we don't use any graphics intensive applications (office / browsers etc. only). Screen resolution on the fat clients is 1920x1080
10:19
<alkisg>
gdi2k: do you use 3d applications?
10:19
<gdi2k>
alkisg, no not at all
10:19
<alkisg>
Then put the lowest value possible
10:19
<gdi2k>
alkisg, ok, thank you :)
10:32mealstrom has left IRC (mealstrom!~Thunderbi@46.63.63.163)
10:43bitcheker has joined IRC (bitcheker!~bitchecke@dynamic-adsl-62-10-105-31.clienti.tiscali.it)
10:47schlady has joined IRC (schlady!~schlady@141-53-32-197.ip.uni-greifswald.de)
11:14alkisg has left IRC (alkisg!~alkisg@plinet.ioa.sch.gr, Quit: Leaving.)
11:43Parker955_Away is now known as Parker955
11:58
<Nemno76>
gdi2k: you're right about the problem would be the dhcp config
12:01Fenuks has left IRC (Fenuks!~Fenuks@109.202.18.160, Ping timeout: 252 seconds)
12:01
<Nemno76>
i could connect to the nbd-server from different clients.. and my guess is it will run once the dhcp next-server tag comes true
12:01
but still working on that
12:10Faith has joined IRC (Faith!~paty@unaffiliated/faith)
12:14Phantomas has joined IRC (Phantomas!~phantomas@ubuntu/member/phantomas)
12:44
<gdi2k>
Nemno76, in your /var/lib/tftpboot/ltsp/i386/pxelinux.cfg/default the line should be like this: append ro initrd=initrd.img root=/dev/nbd0 init=/sbin/init-ltsp quiet splash plymouth:force-splash vt.handoff=7 nbdroot=:ltsp_i386 - I think I had that wrong when I was struggling with the same issue
12:47
plus IPAPPEND 3 at the end if you need that
12:47
<Nemno76>
thank you gdi2k i'll look in too that... i know there is something wrong with the dhcp next-server not coming through... i get booted in initramfs and if i do: cat /conf/param.conf the export ROOTSERVER='not_the_ip_i_want'
12:48
<gdi2k>
that's IPAPPEND 3 not being taken
12:49
<Nemno76>
label ltsp-NBD
12:49
menu label LTSP, using NBD
12:49
kernel vmlinuz-3.13.0-35-generic
12:49
append ro initrd=initrd.img-3.13.0-35-generic init=/sbin/init-ltsp root=/dev/nbd0
12:49
ipappend 3
12:51
yess it seems that way... the thing is dnsmasq won't do the dhcp stuf in virtualbox. thats why i'm running it outside the box only as dhcp, proxy
12:51
<gdi2k>
you don't have a nbdroot=:ltsp_i386 on your line
12:52
can you cat /etc/nbd-server/conf.d/ltsp_i386.conf ?
12:52
<Nemno76>
if got dnsmasq as tftp server set up inside virtualbox... thats true i'lll check that
12:52
<gdi2k>
look at: https://help.ubuntu.com/community/UbuntuLTSP/ProxyDHCP especially the section on Failure: Insufficient Information to set up nbd, quitting - that's what solved it for me
12:53
I had to change the nbdroot= variable to make it work
12:53
you can do nbd_root=[yourserverip]:ltsp_i386
12:55
<Nemno76>
append ro initrd=initrd.img-3.13.0-35-generic init=/sbin/init-ltsp root=/dev/nbd0 nbdroot=192.168.1.200:ltsp_i386
12:56
<gdi2k>
yep, but check the first line in /etc/nbd-server/conf.d/ltsp_i386.conf to make sure it matches "ltsp_i386"
12:56
should be [ltsp_i386]
12:58
<Nemno76>
cat /etc/nbd-server/conf.d/ltsp_i386.conf
12:58
[/opt/ltsp/i386]
12:58
exportname = /opt/ltsp/images/i386.img
12:58
readonly = true
12:58
<gdi2k>
ah
12:59
then you need nbdroot=:192.168.1.200/opt/ltsp/i386
12:59
then you need nbdroot=192.168.1.200:/opt/ltsp/i386
13:03telex has left IRC (telex!~telex@freeshell.de, Remote host closed the connection)
13:04telex has joined IRC (telex!~telex@freeshell.de)
13:23bauerski has left IRC (bauerski!~witekb@frodo.psp.opole.pl, Quit: Leaving.)
13:27
<Nemno76>
still no go
13:28schlady has left IRC (schlady!~schlady@141-53-32-197.ip.uni-greifswald.de, Ping timeout: 250 seconds)
13:37
<gothaggis>
so i am having this problem still and i'm not sure how to troubleshoot it. Ubuntu 14.04 LTSP, idesk + fluxbox. using samba/winbind for AD authentication. for testing purposing i'm running the server in VirtualBox. The thinclient gets the image and shows the login screen - if i type my username and password, a popup box says 'verifying password' for a few seconds, screen goes black, then login screen reloads. On the server I've looked in both a
13:37
uth.log and syslog - nothing at all in auth.log and in syslog it shows the client getting the image but thats it. I have run ltsp-update-sshkeys and ltsp-update-image, still same issue.
13:37
the thin client is a shuttle Xs35V2 with 2gb ram
13:38
it uses intel gma 3150
13:38
and i can login to server using my AD creds so i know that part is working
13:39
(both local login + ssh)
13:53Phantomas has left IRC (Phantomas!~phantomas@ubuntu/member/phantomas, Remote host closed the connection)
13:58
<Nemno76>
gothaggis: if you remove the quiet and splash from the append ro initrd ...etc...etc you can see what happens on the client (oh this is from the file /var/lib/tftpboot/ltsp/i386/pxelinux.cfg/default
14:06
hmmm is this supposed to be this way in /scripts/local-top/nbd from line 47:
14:07
case "$nbdroot" in
14:07
*,*,*)
14:07
nbdsrv="${nbdroot%%,*}"
14:07
nbdport="${nbdroot%,*}"
14:07
nbdport="${nbdport##*,}"
14:08
its sets nbdserv, nbdport? and nbdport again
14:10
<gothaggis>
Nemno76: i removed the splash - it goes by so quickly its difficult to see what is happening. i think i did notice something about the graphics system failing? even though it displays the login screen fine, could there be an issue with the graphics drivers that prevents it from logging in?
14:12
<Nemno76>
gothaggis: i'm a noob at this trying to work my way through... so it's just the way i'm try to figure out my problem.
14:12
did you also remove the quiet ?
14:13
if you got a login screen then you're ahead of my setup
14:21
<gdi2k>
Nemno76, if you're on the graphical login screen then I wouldn't worry about any graphical related errors during the boot. but I have little to no idea about how to integrate with AD, but keep in mind that PAM and co does not work the thin / fat clients
14:22
the other thing is: ltsp-update-sshkeys
14:22
you can do man ltsp-update-sshkeys to learn about it
14:24
<Nemno76>
gdi2k: I was trying to help gothaggis, so your answer | gothaggis :p
14:25
<gdi2k>
Nemno76, oops, you're right
14:25
IRC novice here lol
14:25
regarding you're issue, I'm not sure what else to check. we have moved to using ipxe to boot the clients instead, which cuts out all the DHCP and IPAPPEND confusion and we've not looked back
14:26
*your issue ;)
14:30
<Nemno76>
i think i'll start over using a vbox host configuration so i can use the ltsp-server-standalone... did you see the nbd startup snippet? about the double assignment. maybe i'll make my own initrd.img
14:30
<gothaggis>
gdi2k: ad integration is working fine on the server, i have it working fine on 10.04 so i know it works :) the thing is, i'm not seeing any entries in auth.log, which i am on my 10.04 server when clients log in
14:30
so its not even getting that far it seems
14:31
<gdi2k>
gothaggis, ok, I don't understand the glue between the ssh-based login from the client to the AD on your server, so I'm not going to be helpful to you :(
14:44Fenuks has joined IRC (Fenuks!~Fenuks@176.51.101.232)
15:05
<championofcyrodi>
hey guys! So I got k12linux to boot a client. using "ltsp-build-client --arch x86_64 --fat-client --chroot x86_64 --release epel-6-x86_64" However, the login screen just flashes to a low res. console, then back to the login screen when entering a correct user/pw combo.
15:05
i think last time i saw this on ubuntu, it was because the SSH connection was being closed after authenticating.
15:05
maybe by PAM...
15:06
*sshd module for PAM
15:13
hmmm.. well i can SSH in to the server from the localhost...
15:13
enabling shell on tty2
15:15freedomrun has joined IRC (freedomrun!~quassel@unaffiliated/freedomrun)
15:17deeville has joined IRC (deeville!~deeville@38.117.108.108)
15:31Phantomas has joined IRC (Phantomas!~phantomas@ubuntu/member/phantomas)
15:37vagrantc has joined IRC (vagrantc!~vagrant@freegeek/vagrantc)
15:46uuuolliw has left IRC (uuuolliw!~william@fs-93-93-43-155.fullsave.info, Quit: Konversation terminated!)
15:54andygraybeal has left IRC (andygraybeal!~andy@h166.205.130.174.dynamic.ip.windstream.net, Remote host closed the connection)
15:54
<championofcyrodi>
hmmm the lts.conf in /var/lib/tftpboot/ltsp/x86_64/lts.conf doesnt seem to be getting used.
15:55
although i have SCREEN_02=shell switching to tty2 only shows a cursor with no repsonse hitting enter/return.
16:04ltsp has joined IRC (ltsp!bot@ltsp.org)
16:05mgariepy_ has joined IRC (mgariepy_!~mgariepy@styx-204.ccs.usherbrooke.ca)
16:05slackish has joined IRC (slackish!amcphall@mcphall.org)
16:05markosu_ has joined IRC (markosu_!marko5@kapsi.fi)
16:06tarzeau_ has joined IRC (tarzeau_!~gurkan@mail.aiei.ch)
16:07FrozenZi1 has joined IRC (FrozenZi1!pbrown@evo.paivola.fi)
16:08stgraber has left IRC (stgraber!~stgraber@ubuntu/member/stgraber, Ping timeout: 260 seconds)
16:08mgariepy has left IRC (mgariepy!~mgariepy@ubuntu/member/mgariepy, Ping timeout: 260 seconds)
16:08ltsp` has left IRC (ltsp`!bot@ltsp.org, Ping timeout: 260 seconds)
16:08tarzeau has left IRC (tarzeau!~gurkan@mail.aiei.ch, Ping timeout: 260 seconds)
16:08slackish_ has left IRC (slackish_!amcphall@mcphall.org, Ping timeout: 260 seconds)
16:08markosu has left IRC (markosu!marko5@kapsi.fi, Ping timeout: 260 seconds)
16:08FrozenZia has left IRC (FrozenZia!pbrown@evo.paivola.fi, Ping timeout: 260 seconds)
16:08dberkholz_ has left IRC (dberkholz_!~dberkholz@gentoo/developer/dberkholz, Ping timeout: 260 seconds)
16:08markosu_ is now known as markosu
16:08dberkholz has joined IRC (dberkholz!~dberkholz@smtp.gentoo.org)
16:08dberkholz has joined IRC (dberkholz!~dberkholz@gentoo/developer/dberkholz)
16:10ncom0pl has left IRC (ncom0pl!~ncom0pl@89-71-65-172.dynamic.chello.pl, Excess Flood)
16:10
<||cw>
championofcyrodi: it uses the one that's in the folder all the PXE happens from
16:12ncom0pl has joined IRC (ncom0pl!~ncom0pl@89-71-65-172.dynamic.chello.pl)
16:13stgraber has joined IRC (stgraber!~stgraber@ubuntu/member/stgraber)
16:14
<championofcyrodi>
IIcw: yea, thats the one i'm editing.
16:14
/var/lib/tftpboot/ltsp/x86_64 ( i can see the pxeboot call server://ltsp/x86_64/pxeblahblah...
16:15
actually in the ltsp/x86_64 folder there is pxelinux.0 which is called at boot...
16:15
but there is also a folder called pxelinux.cfg
16:16
however the lts.conf was generated in the x86_64 folder
16:16
so i'm assuming thats where it goes
16:16
i'm starting to think it's just the home folder not mounting and then being logged back out.
16:16
but w/o a shell it's tough to debug
16:17
<gdi2k>
championofcyrodi, also check the session is correctly set in ldm - for us, if we don't explicitly select the correct session, it just restarts ldm ("no response from server")
16:17
there's a settings button in the lower left corner
16:17
it's a one-time thing for each user
16:18
<championofcyrodi>
i did however just see a segfault appear in tty2 at this boot... ldm[2348]: segfault at 8 ip 000000000000404134 sp 00007fffc15343b0 error 4 in ldm[400000+7000]
16:18
thanks i'll give that a try.
16:18
<gdi2k>
championofcyrodi, uh, no idea what that is, I'll leave it to the gurus here
16:19
<championofcyrodi>
my session manager choices are only 2 options, 'default' and 'failsafe xterm'
16:19
failsafe xterm had the same result
16:19
<gdi2k>
championofcyrodi, hmm, I maybe you didn't include a DE in your fat client image? that would explain things...
16:20
<championofcyrodi>
i used this command to build the image: "ltsp-build-client --arch x86_64 --fat-client --chroot x86_64 --release epel-6-x86_64"
16:20
the result was that it completed successfully
16:20
<gdi2k>
have a look here; https://help.ubuntu.com/community/UbuntuLTSP/FatClients
16:20
--fat-client-desktop: Creates and installs a FAT chroot with the specified desktop environment. Options are ubuntu-desktop, edubuntu-desktop among others. The option kubuntu-desktop is not tested.
16:21
you created a fat client without a desktop I think
16:21
<championofcyrodi>
i'm using centos, not ubuntu
16:21
<gdi2k>
useful if you want to create diskless servers etc.
16:21
ok, but I bet the variables are the same
16:22
check in the chroot if you have all the desktop stuff that you are expecting
16:23
<vagrantc>
the commandline options for ltsp-build-client are highly distro-dependent
16:31
<championofcyrodi>
vagrantc: that was the impression i got when going through the k12linux variant.
16:32
after you run the command you get a sort of gui built with console characters. (not sure what to call it) which shows the release chosen, etc.
16:32
took a while to realize i had to use --release to specify a 64 bit image rather than just --arch
16:45
gdi2k: i suspect you're right. there is no /usr/bin/gedit in the chroot
16:45
i would have expected that to be there
16:46
<gdi2k>
does ltsp-build-client have a man page? on ubuntu it does and specifies the desktop options
16:48
<championofcyrodi>
yes but the man page is quite short. not finding a copy on the google...
16:48
<gdi2k>
on your server? man ltsp-build-image ?
16:49
<championofcyrodi>
yes. the man page on my server is not explaining much about including LDM or any desktop manager.
16:49
it only talks about specifying a 'dist' 'release' and 'arch'
16:49
<gdi2k>
hmm, ok
16:50
<championofcyrodi>
like: ltsp-build-image --list shows the releases i can choose
16:50
http://www.redhat.com/archives/k12osn/2014-May/msg00002.html
16:50
like that
16:51
also 'fat-client' isnt mentioned in the man page... but it is in the --extra-help
16:51
output
16:51
and says 'run most or all applications locally'
16:52
i can chroot in... maybe i can just installed the desktop manager and update the image manually.
16:53
<gdi2k>
may be worth looking at the code. have a look into /usr/sbin/ltsp-build-client, maybe you can make it out from there
16:53
yes, that seems viable too
16:53
<championofcyrodi>
just got to figure out what yum package is needed...
16:54
maybe yum groupinstall "Desktop" "Desktop Platform" "X Window System" "Fonts"
16:54
<gdi2k>
I'm not any good with CentOS and similar :(
16:54
<championofcyrodi>
yea... we have to use centos on our servers because our customer uses rhel in production.
16:54
<gdi2k>
ok
16:54
<championofcyrodi>
and we arent paying for RHEL updates.
16:55
<gdi2k>
before you do a major install in chroot, make sure you chroot in properly, mount proc etc.
16:55
<championofcyrodi>
everyone likes the ubuntu LTSP w/ Intel NUCs i have set up. but there is another network that ubuntu can't be installed on.
16:55Fenuks has left IRC (Fenuks!~Fenuks@176.51.101.232, Ping timeout: 240 seconds)
16:55Parker955 is now known as Parker955_Away
16:55
<championofcyrodi>
im using 'ltsp-chroot'
16:55
i believe that script is supposed to handle the /proc mount right?
16:56
<gdi2k>
it doesn't on mine
16:56
there are manual steps here though: https://help.ubuntu.com/community/UbuntuLTSP/UpdatingChroot
16:56
export LTSP_HANDLE_DAEMONS=false
16:56
sudo chroot /opt/ltsp/i386
16:56
mount -t proc proc /proc
16:56
I think that should do it
16:56
no idea if the daemons thing works on centos
17:01
<vagrantc>
fedora, redhat and related distros have been falling behind in ltsp support...
17:01
<championofcyrodi>
vagrantc: that is the impression i am getting
17:02
<vagrantc>
recent versions generate the manpage from the output of ltsp-build-client --extra-help
17:02
<championofcyrodi>
yea
17:02
http://wiki.centos.org/HowTos/DisklessClients <-- reading this now
17:03
seems to be a pretty manual process
17:04
ltsp.org seems to be giving me timeouts browsing the site...
17:11
heh... the lts.conf generated by CentOS 'ltsp-server' package says, "If you're using Unity, you might want to install gnome-session fallback... in regard to LDM_SESSION="gnome-fallback"
17:12
however, i thought unity was specific to ubuntu
17:12
(not centos/rhel/fedora)
17:15freedomrun has left IRC (freedomrun!~quassel@unaffiliated/freedomrun, Read error: Connection reset by peer)
17:21
<gothaggis>
I know this isn't the best LTSP question - so i use fluxbox..but man it looks ugly. but i like how easy it is to customize - for example the only thing i have on the desktop are icons for the apps that users need (web browser, email). Is it possible to configure unity the same way, remove the sidebar, only allow users to run what is on the desktop, customize right click menu, etc?
17:22
<championofcyrodi>
I'm not sure removing the side bar would make it 'unity' anymore.
17:22
might as well just run xfce or something similar
17:23
at the end of the day, most of your applications just need "X Windows"
17:24
i've actually had aliases on servers to start an X session and load firefox from the console. which literally gives you a grey screen with only firefox.
17:25
in regard to LTSP... it looks like i should have probably installed the EPEL repository first
17:25
otherwise you get the ltsp stuff from the base repo, which seems to be severely dated.
17:30
weird... just did a reinstall of the OS... now getting all kinds of 404 w/ the yum mirrors..
17:32
looks like there was some kind of network routing issue
17:32
(having two NIC cards)
17:33
disabling the card with the static IP on it's own network seemed to fix the issue.
17:33
well, 'ifdown'ing it.
17:34boospy has joined IRC (boospy!~loma@darkbox.osit.cc)
17:48nick111 has joined IRC (nick111!550bb4ee@gateway/web/freenode/ip.85.11.180.238)
18:02Parker955_Away is now known as Parker955
18:10mgariepy_ is now known as mgariepy
18:11mgariepy has joined IRC (mgariepy!~mgariepy@ubuntu/member/mgariepy)
18:19freedomrun has joined IRC (freedomrun!~quassel@unaffiliated/freedomrun)
18:25Grembler has left IRC (Grembler!~Ben@cpc29-aztw22-2-0-cust128.18-1.cable.virginm.net, Quit: I Leave)
18:32adrianorg has left IRC (adrianorg!~adrianorg@177.204.157.55.dynamic.adsl.gvt.net.br, Ping timeout: 272 seconds)
18:34adrianorg has joined IRC (adrianorg!~adrianorg@177.156.56.81)
18:35telex has left IRC (telex!~telex@freeshell.de, Remote host closed the connection)
18:36telex has joined IRC (telex!~telex@freeshell.de)
18:59Parker955 is now known as Parker955_Away
19:03adrianorg has left IRC (adrianorg!~adrianorg@177.156.56.81, Ping timeout: 264 seconds)
19:04adrianorg has joined IRC (adrianorg!~adrianorg@177.204.77.19.dynamic.adsl.gvt.net.br)
19:21Phantomas has left IRC (Phantomas!~phantomas@ubuntu/member/phantomas, Remote host closed the connection)
19:24
<championofcyrodi>
i really want to give up. but cant
19:25
i mean, the ltsp-server packages exist... so SOMEBODY got it to work at some point.
20:15deeville has left IRC (deeville!~deeville@38.117.108.108, Quit: This computer has gone to sleep)
20:52
<ogra_>
championofcyrodi, traditionally (since warren left) fedora based implementations had people show up, implement it for one release and then they vanish before they have many users so nobody fixes the bugs or cares if it works in the next release
20:54
<championofcyrodi>
ogra_: I understand. Unfortunately my company really likes LTSP, so I'll likely be looking to install all the daemons manually and get something cobbled together using the sources from launchpad i guess.
20:54
well.. Fortunately & unfortunately
20:55
Our Ubuntu implementation is great. So now I have to get a CentOS implementation figured out.
20:55
<ogra_>
yeah, will be some work
20:55
<championofcyrodi>
thats for sure. I'll be sure to publicly post the process somewhere as i document, test and implement.
21:04
<vagrantc>
championofcyrodi: centos clients? feel free to make a branch and request merges...
21:09boospy has left IRC (boospy!~loma@darkbox.osit.cc)
21:09
<championofcyrodi>
vagrantc: I need to do a lot more research before digging into using launchpad. I've only experience working on small projects with git, svn, perforce and cvs. So I'm totally new to the this type of collaboration.
21:11
i see a lot of branches here: https://code.launchpad.net/ltsp/+all-branches?field.lifecycle=CURRENT&field.lifecycle-empty-marker=1&field.sort_by=by+branch+name&field.sort_by-empty-marker=1
21:11
But not sure if i understand the convention. on my projects we create a branch for each 'feature request' or 'bug', then merge the branch with a 'master'
21:11
<vagrantc>
main branches are ltsp(-trunk), ltspfs(-trunk) and ldm(-trunk)
21:12
<championofcyrodi>
gotcha. looks like i need to read up on the ltspfs and the ldm portion, as I don't really understand their role in the fusion of services that makes up the LTSP server
21:12
<vagrantc>
championofcyrodi: this is a much less noisy list of branches: https://code.launchpad.net/~ltsp-upstream/
21:14
championofcyrodi: in general, the same workflow applies ... implement a feature in a branch, request to merge it.
21:15
championofcyrodi: there's also git-remote-bzr if you'd rather use git: https://github.com/felipec/git-remote-bzr
21:17
<championofcyrodi>
So it looks like i would want to start looking in to the ltsp-trunk: /server/k12linux/ portion of the project and get an understand of how it's current implementation is intended to work. Before bothering with creating a new branch
21:18
heh... wow, this stuff is pretty old
21:19
relatively speaking.
21:19
<vagrantc>
championofcyrodi: server/share/ltsp/plugins/ltsp-build-client/
21:19
as upstreams go, ltsp isn't very cleanly implemented.... ldm and ltspfs are more straightforward.
21:20
basically, each distro is largely it's own implementation ... try to share where possible, but it's not always obvious
21:21
<championofcyrodi>
i see
21:22bitcheker has left IRC (bitcheker!~bitchecke@dynamic-adsl-62-10-105-31.clienti.tiscali.it, Quit: {exiting})
21:22
<championofcyrodi>
there is a redhat and fedora implementation of ltsp-build-client, but nothing for centos
21:24
<vagrantc>
none of those were too obvious
21:24
another approach might be to try the ltsp-pnp method...
21:25
<championofcyrodi>
so i would likely branch the ltsp-trunk, and create a folder for Centos, and create the proper scripts to build a client, with sequence numbers for each portion of the build process. like in Ubuntu, 001-set-arch is run before 035-install-debs, and thus are two separate files because they affect two separate subsystems of the ubuntu build process?
21:26
or they are symlinks to shared debian files
21:26
since ubuntu is an offshoot of debian?
21:26
err. Downstream rather?
21:28
heh.. I think this is going to be overwhelming for a while.
21:28
<vagrantc>
championofcyrodi: i think the RedHat dir includes support for CentOS ... but i'm not sure.
21:29
<championofcyrodi>
yea, i was just looking at it, and i see kickstart was used to build the chroot... which is a familiar concept to me... So i'll probably need to study up on the redhat plugin for a bit first
21:29
<vagrantc>
but yeah, debian and ubuntu share a lot of plugins because they are simila
21:29
r
21:29
there are also hooks for mint in the ubuntu plugins as it is closely derived from ubuntu
21:30
<championofcyrodi>
makes sense
21:31
a lot of it is intuitive, however i find it difficult to be confident in my understanding without someone to validate what i *think* something is intended to do. so I'll likely be asking questions over the next few weeks as I dig in. Of course, I will make every effort to read and understand on my own before falling back to the relay for support.
21:34
<vagrantc>
good luck
21:34
<championofcyrodi>
Everyone i work with uses linux but develops web applications with scala/java/python/js stacks. Nobody I work with is involved with linux this much, so i'm going to be on my own for the most part.
21:34
thanks.
21:34
<vagrantc>
championofcyrodi: i might actually focus on getting the client-side support working first, and making sure the ltsp-pnp method works
21:34
which doesn't use ltsp-build-client at all
21:34
<championofcyrodi>
okay. I'll start there.
21:35andygraybeal has joined IRC (andygraybeal!~andy@h66.205.130.174.dynamic.ip.windstream.net)
21:35
<vagrantc>
and then once that's working well enough, maybe work on ltsp-build-client plugins
21:37
<championofcyrodi>
looking here: https://help.ubuntu.com/community/UbuntuLTSP/ltsp-pnp I'm not seeing what "pnp" stands for
21:38
<vagrantc>
plug-and-play
21:38
all it really is is building a client image out of the server's filesystem.
21:39
<championofcyrodi>
Ah, essentially excluding /proc and /sys...
21:39
that seems a lot faster than building via package repo
21:39
<vagrantc>
excluding a bunch of stuff, but sure.
21:40
<championofcyrodi>
(was thinking "Plug-n-Play"???)
21:40
<vagrantc>
sure
21:40
i've occasionally thought we should come up with a better name for that method
21:41
but nothing springs to mind
21:41
<championofcyrodi>
in docker world they call it 'creating a base image'
21:41
<vagrantc>
i think also the implementation used dnsmasq with proxydhcp for single-nic configurations with a separate dhcp server
21:41
which is maybe where the plug-n-play idea came from
21:43
<championofcyrodi>
yea, thats what it says. cool, i'll do some testing with pnp via virtual machine(s) and network to get a better understanding of it, then go from there. thanks for your time, hopefully I can contribute back a little bit and get some centos support rolling. It's a very stable distro compared to fedora, and doesnt require subscription fee for updates like rhel.
21:45
<vagrantc>
right
21:46
i gave a clueless test of the rhel support on centos last time someone was working on it
21:46
clueless, in that i don't know that family of distros ...
21:48
<championofcyrodi>
ah. i see it used a lot when deploying clusters of systems to support distributed stuff like Apache Hadoop or Storm.
21:49
mainly centos in the development enviornments (because its free), and rhel in the production environments, because you want expert/paid support when you have real users and 24/7 uptime requirements.
21:50
and fedora generally used as a workstation since it has the newest kernel and can do fancy hardware things with extra kernel modules
21:50
like the Consumer IR sensor for using a remote control with your desktop
21:51
(which you would never need for a server)
21:52
okay, i'm going to go read and relax. thanks again.
21:56spectra has left IRC (spectra!~spectra@debian/developer/spectra, Quit: ZNC - http://znc.sourceforge.net)
22:01cyberorg has left IRC (cyberorg!~cyberorg@opensuse/member/Cyberorg, Ping timeout: 246 seconds)
22:01spectra has joined IRC (spectra!~spectra@debian/developer/spectra)
22:02cyberorg has joined IRC (cyberorg!~cyberorg@opensuse/member/Cyberorg)
22:14
<Nemno76>
gdi2k: Thanks for you support.. I did get it the xubuntu working in virtualbox. Only with a new install and then the ltsp-server-standalone. Host vbox with 2 nics one for internal and one for the world
22:17
<gdi2k>
Nemno76, ok at least you got something working ;) I don't know if virtualbox complicated anything. I use standalone too, but with local dhcp disabled
22:17
also virtualized, but on kvm
22:18
<Nemno76>
gdi2k: may be i'll look into the other setup i had where i used a vbox host with bridged lan, to get it working... it must be the next-server dhcp thing... some script will use that ip to do the filemount stuff..
22:19
<gdi2k>
Nemno76, I assumed you were using a bridged lan - how is the networking set up with virtualbox?
22:21
<Nemno76>
gdi2k, Yes i was using bridged lan and one eth adapter.. There is nat setup internal network setup, host setup... al for each adapter. (http://www.virtualbox.org/manual/ch06.html)
22:24
gdi2k, The thing that didn't go as i expected was the dnsmasq dhcp stuff in the vbox host setup with bridged eth. So when i used the dnsmasq outside the virtualbox, it did handle the dhcp (all but next-server (tftp address))
22:25
so i'm glad i got it working
22:26
<gdi2k>
Nemno76, that's the main thing - we use bridged networking on our setup and it works as if it were a metal box
22:27
but it was not easy to get working in the beginning either with external dhcp etc.
22:27
as I said, ipxe ftw!
22:28
especially as it allows failover - if booting from one server fails, it sends the client to the next one
22:33
<Nemno76>
Yes, it would be nice to get it working. For me this is/was to learn something, and may be to use it in a future job. I do not have practical use for it at the moment.
22:59Faith has left IRC (Faith!~paty@unaffiliated/faith, Quit: Saindo)
23:02vagrantc has left IRC (vagrantc!~vagrant@freegeek/vagrantc, Quit: leaving)
23:21mmetzger has left IRC (mmetzger!~mmetzger@99-71-214-107.lightspeed.mdldtx.sbcglobal.net, Ping timeout: 250 seconds)
23:44Nemno76 has left IRC (Nemno76!~Nemno@s5375b099.adsl.online.nl, Remote host closed the connection)