|00:02||bengoa has joined IRC (email@example.com)|
|00:10||alexqwesa has left IRC (firstname.lastname@example.org, Quit: Хана X'ам !!!)|
|00:16||mischko has left IRC (mischko!~Scott@ip68-101-108-132.oc.oc.cox.net, Quit: Ex-Chat)|
stgraber: ldm 2.2.5-1 uploaded to debian ... pushed and tagged in bzr repos, too.
vagrantc: cool, thanks. I'll sync it into Ubuntu once I see it.
stgraber: will be happy to see zero diff!
yep, that'll be nice ;) next step, LTSP itself ;) well, this one will be tricky if at all possible :)
at least reducing the delta a bit would be nice for ltsp
|00:56||bengoa has left IRC (email@example.com, Quit: Leaving.)|
i was hoping to make an upload of ltsp to debian today, but didn't really have enough time for testing
oh, some of the NBD variable name changes may require a newer version of ltsp...
hope to get that sorted out by monday.
stgraber: what all does the ltsp-livecd stuff entail?
not much really, it should be split as a separate source package (it's already a separate binary package)
it's just a UI configuring network-manager, installing the minimum set of LTSP packages, created 254 accounts and configuring lts.conf
so it's basically a one click LTSP that's not secure at all but works fine from a live CD and offers guest login sessions for any machine booting on the network
|01:26||18WABIP7U <18WABIP7Ufirstname.lastname@example.org> has quit IRC (Ping timeout: 240 seconds)|
not much but so much, sounds like :)
|01:39||rthomson has left IRC (email@example.com, Quit: Reached EOW)|
|01:40||artista-frustrad has joined IRC (firstname.lastname@example.org)|
|02:18||shogunx has left IRC (email@example.com, Ping timeout: 255 seconds)|
|02:22||shogunx has joined IRC (firstname.lastname@example.org)|
|02:24||alexqwesa has joined IRC (email@example.com)|
|02:51||shogunx has left IRC (firstname.lastname@example.org, Ping timeout: 245 seconds)|
|03:03||adrianorg__ has left IRC (email@example.com, Ping timeout: 260 seconds)|
|03:11||shogunx has joined IRC (firstname.lastname@example.org)|
|03:24||yalu has left IRC (email@example.com, Read error: Operation timed out)|
|03:26||yalu has joined IRC (firstname.lastname@example.org)|
|03:56||VectorX has joined IRC (VectorX!~knight@unaffiliated/vectorx)|
hi need some help, got a thin client that was working fine, plugged it in after about a week and getting this error, http://i30.fastpic.ru/big/2012/0204/aa/2c12035b97d0778a6d18da097d2402aa.jpg
the Xorg.7.log file doesnt seem to exist
|04:02||Damianos_ has joined IRC (Damianos_!~Damianos@adsl-070-145-074-043.sip.cha.bellsouth.net)|
|04:03||Damianos has left IRC (Damianos!~Damianos@adsl-070-145-074-043.sip.cha.bellsouth.net, Read error: Operation timed out)|
|04:03||Damianos_ is now known as Damianos|
|04:48||artista-frustrad has left IRC (email@example.com, Ping timeout: 240 seconds)|
|05:40||alkisg has joined IRC (alkisg!~alkisg@ubuntu/member/alkisg)|
stgraber, highvoltage: instead of generating 254 accounts, we could make an nbd-server preconnect script that sees the client IP (e.g. 192.168.0.123) and creates the accounts on the fly (username: ltsp123, password: ltsp123). Do you want me to provide an initial implementation for that?
VectorX: what are your X_* settings in lts.conf for this client? Do you have the nvidia drivers installed in the chroot?
alkisg the server had a nvidia card, i took it out and am running hd3000 from the mobo, i have not installed any drivers other than that
there are no client specific items that i know of
It's weird that it tries to load the nvidia driver instead of nouveau
Put SCREEN_07=shell for this client
There, run lspci -nn -k | grep -A 2 VGA, to see the graphics card, driver etc
Then try X -configure
well screen is set to rdp as default but ill try
alkisg: that sounds cool!
vagrantc: which one? the dynamic user generation?
alkisg: the preconnect thing
Yeah we could do a lot of things there
alkisg: could also hook it into ldminfod or something
Indeed... we could even provide an "authentication" phase if we wanted
Like joining a machine to active directory, which requires a password, we could be asking on the client side for a password and "register" the machine and create a user account etc
Irrellevant: S15-userLoginCheck should come after S20-restrictUser
There's no point in killing user processes if the user is not allowed to login there
|05:53||staffencasa has left IRC (firstname.lastname@example.org, Ping timeout: 248 seconds)|
Can I replace S15-userLoginCheck with a script similar to the one I pasted yesterday, which saves session information in e.g. ~/.cache/ltsp/session.* ?
...and put it to S21?
S15-userLoginCheck now kills all ssh connections, not just ltsp connections, right?
sounds like it could be better.
OK I'll do that and if someone complains we can always revert it :)
Since we have more info now, would you care to rephrase the message we display? E.g. you're alread logged in at $HOSTNAME, $VT etc?
Or that message is translated and I shouldn't touch it?
alkisg can i reboot the clients using the server ?
vagrantc: https://git.ipxe.org/ipxe.git/commit/fcc35bf48776fff9ebfd8db537679583221a9cd4 ==> In git one can have a commit line and an extended description line. Can we do that in bzr? I think it would be a nice place to have the comments about e.g. the S15 => S21 change etc...
VectorX: only if you're using a remote control application like epoptes, or if they're already logged in and you write a small script to do that
cant seem to get SCREEN_07=shell to work, the machine that works ends up in a blank screen and the other gets stuck on the ubuntu init screen with 3 dots white on the second pass
alkisg: quiet-splash: to disable the splash screen in Ubuntu, in order to see any boot error messages, run `sudo gedit /var/lib/tftpboot/ltsp/i386/pxelinux.cfg/default` and remove quiet splash .
That's to avoid the dots and see the actualy problems
Also try with a very simple lts.conf, like
without anything else
yeah thats what i got
VectorX: it reports a problem with the vmware driver there, did you install that manually?
i installed virtualbox but not vmware
You installed the virtualbox additions in the chroot?
vbox on the server not in the chroot
Did you run ltsp-update-kernels when you last ran ltsp-update-image?
Otherwise the tftp kernel won't match the nbd kernel...
just did it like 20 mins ago
and updated keys
Other than that, it doesn't sound related to ltsp at all... maybe ask in #ubuntu-x or similar
vagrantc: similar to the description of .deb packages, we can have descriptions in bzr consisting of a one-line header, an empty line, and a multiline comment. Then bzr log --line only shows the one-line description header.
Should we start using commits like those in ltsp? (of course not all of them, only for the ones where a long comment actually makes sense)
|07:15||killermike has left IRC (email@example.com, Ping timeout: 252 seconds)|
|07:21||freedomrun has joined IRC (firstname.lastname@example.org)|
alkisg: that sounds nice! yes I think you should.
OK, let me reboot to precise where we're using nbd-server instead of inetd, as I don't think that inetd exposes the client IP...
|07:39||alkisg has left IRC (alkisg!~alkisg@ubuntu/member/alkisg, Quit: Leaving.)|
|07:43||alkisg has joined IRC (alkisg!~alkisg@ubuntu/member/alkisg)|
alkisg: i usually use bzr log --short ... but sometimes it can be hard to have a meaningful short commit
too many of our commit messages are too short, in my opiinion... if there's a technical way to improve that, sure, would be nice ... but will people do it? :)
vagrantc: just mentioning in the "coding style readme" that that's what we prefer is enough, I think
That way anyone that feels like commenting his commit may do so
Others that don't want to, can continue with one-line descriptions
bzr log --short --line would be required to skip the multiline comments
does bzr log --line only show the first line anyways?
But up to now we had extremely long lines, not multiline comments,
and at some times I had to send a mail to the developers list and explain the commit, and link to the mail in the commit
log --line overrides --short
Which sounds stupid - it'd be better if the explanation was in the commit itself
yes, i'd like to avoid the long lines ...
i have to pipe all the changelogs through | fmt
i think that's because people typically use -m on the commandline, rather than starting an editor?
highvoltage: Unfortunately I think that neither nbd-server nor inetd provide the client IP to the scripts. So we'd need a small standalone daemon, either in inetd/shell, or in python
vagrantc: you can do bzr commit -m 'one line
but who does?
Who knows that he can do that? :D
|08:02||* vagrantc always uses an editor|
I didn't even know that multiline comments were allowed, much less that it would be a good practise...
If someone told me, I would do that
ok, so we encourage in the codeingstyle stuffs.
Let's also mention about the dots in the end of the commit lines
|08:04||* vagrantc -> sleep|
didn't get a new version of ltsp uploaded today, but at least i got epoptes and ldm :)
'night vagrantc, thanks again
|08:11||vagrantc has left IRC (email@example.com, Ping timeout: 248 seconds)|
alkisg: ah, that's a pity
highvoltage: it's very easy to make a small shell script to create users,
to put it in inetd.conf,
and to call it from ldm/rc.d upon login
Or maybe on client boot, as soon as it knows its hostname
So, RC_FILE in lts.conf
The whole implementation should be less than 10 lines
highvoltage: do you want me to do an implementation, and you can then package it in ltsp-live?
If you don't want to restart inetd, I can also make the daemon part in python
A bit safer way to do it would be:
|08:22||freedomrun has left IRC (firstname.lastname@example.org, Remote host closed the connection)|
1) to put HOSTNAME_EXTRA=mac in lts.conf
2) in either ldminfod or nbd-server (now we don't need the ip), to run `arp -an`, and generate users for all the known mac addresses
That even doesn't require any changes in existing chroots... let me give that a try
alkisg: it sounds reasonable to me, but talk to stgraber before uploading anything
it would be really nice to have ltsp live start up faster
on the other side the ltsp parts are usually the most fragile part of edubuntu so we need to make sure it doesn't introduce any new bugs first
highvoltage: I wasn't going to upload anything, just provide a reference implementation that you guys could package if you think it helps :)
I'll send a mail with it, and if you want you can use it or not
alkisg: ok great
The downside with `arp -n` is that in very large networks it will create even more than 255 users
...so maybe that isn't a good idea either
I'll go with the RC_FILE approach
alkisg whats the best distro for ltsp ?
with probably the most support
VectorX: ubuntu and debian are well supported, gentoo too but I don't know its current state, opensuse has kiwi ltsp, fedora is a bit behind
i still cant figure out this nvidia issue with just this one client
VectorX: did it work sometime in the past?
yeah, the only difference was i took out the nvidia card from the server, and i think it worked after that aswell
coz we sorted the rdesktop thing and i tested on that machine
perhaps i should reinstall ubuntu and see
And when did it break?
well today was when i turned it on again maybe after like a week
You don't need to reinstall anything on the server, at worse you could just run ltsp-build-client again, after moving away your existing /opt/ltsp
well i deketed /opt/ltsp/i386 and did a arch i386 aswell
You ran ltsp-build-client already, and the client still won't boot?
Nope, you didn't run ltsp-build-client?
Or nope the client won't boot?
same problem with the vga
im trying again
mv's i386 to _copy, and running build-client
ok now the machine hacing trouble ends up with a [OK] on the top right corner of the screen and is stuck there, the other is at the login prompt gui
took out the splash and rebooting
its got one fail on, stopping save kernal messages
either that or starting restore sound cards, hehe, too far apart to say
hmm few reboots and everything shows [OK] but its not going to the gui login
Cool, with RCFILE_XX arbitrary commands can be executed on LTSP clients. So we can request accounts for the client hostnames without modifying existing chroots. :)
so i changed the monitor from a brand spanking new led to a older lcd, and wala login promopt
That's unrelated to the problem with the nvidia client that you pasted to pastebin, right?
same problem, aparently that was the fix
although i cant really say coz i rebuilt the client, but, changing the monitor fixed what ever, when this was working i was using this same LCD
It doesn't make sense, X can start without a monitor attached
how do you think i feel
i mean this is really the only thing that changed for the client, last week it was working fine with the old LCD, then last night i rearranged everything and plugged in an LED and it didnt boot
If that's true, then that graphics driver has some serious problems and you should file a bug against it
what kind of gave it a way just now was in the pxelinux.cfg/default i added nomodeset and it worked
then i thought maybe ill try switching monitors
and went right thru, i was also wondering why i wasnt able to get another machine with the LED to boot up, this was last week, but i didnt need it at the time and forgot about it
highvoltage, stgraber: so after a bit more thought, here's my suggestion for dynamic user creation:
1) We launch a very simple python tcpserver on the server, which gets the ip from the clients and creates the corresponding user accounts: http://docs.python.org/library/socketserver.html#socketserver-tcpserver-example
2) We put this in lts.conf, so that the clients ping the server: RCFILE_01=/usr/bin/nc server 9999. That's all, no chroot modifications necessary.
|09:57||adrianorg__ has joined IRC (email@example.com)|
|10:35||VectorX has left IRC (VectorX!~knight@unaffiliated/vectorx, Read error: Connection reset by peer)|
|10:44||sepski has joined IRC (firstname.lastname@example.org)|
|10:44||sepski is now known as Guest41868|
|10:46||alkisg has left IRC (alkisg!~alkisg@ubuntu/member/alkisg, Quit: Leaving.)|
|11:20||monteslu has left IRC (email@example.com, Ping timeout: 252 seconds)|
|11:32||monteslu has joined IRC (firstname.lastname@example.org)|
|12:09||adrianorg_ has joined IRC (email@example.com)|
|12:12||adrianorg__ has left IRC (firstname.lastname@example.org, Ping timeout: 245 seconds)|
|12:14||artista_frustrad has joined IRC (email@example.com)|
|12:17||staffencasa has joined IRC (firstname.lastname@example.org)|
|12:31||klausade has left IRC (email@example.com, Quit: leaving)|
|12:40||staffencasa has left IRC (firstname.lastname@example.org, Ping timeout: 244 seconds)|
|12:40||artista_frustrad has left IRC (email@example.com, Ping timeout: 240 seconds)|
|13:27||sybarite has joined IRC (firstname.lastname@example.org)|
Can anyone tell me if its possible to create a ltsp network for mac clients too?
sybarite, I believe you can use powerpc clients for LTSP.
|14:04||komunista has joined IRC (email@example.com)|
|14:09||sybarite has left IRC (firstname.lastname@example.org, Ping timeout: 245 seconds)|
|14:13||Trixboxer has joined IRC (Trixboxer!~Trixboxer@22.214.171.124)|
|14:33||adrianorg__ has joined IRC (email@example.com)|
|14:35||adrianorg_ has left IRC (firstname.lastname@example.org, Read error: Operation timed out)|
|15:16||sutula has left IRC (sutula!sutula@nat/hp/x-aojgsubnygspunng, Quit: Terminated with extreme prejudice - dircproxy 1.0.5)|
|15:18||sutula has joined IRC (sutula!sutula@nat/hp/x-oiokfitkauoygjwn)|
|15:21||ludo402 has joined IRC (email@example.com)|
|15:30||ludo402 has left IRC (firstname.lastname@example.org, Quit: IRC webchat at http://irc2go.com/)|
|15:37||komunista has left IRC (email@example.com, Quit: Leaving.)|
|16:11||alkisg has joined IRC (alkisg!~alkisg@ubuntu/member/alkisg)|
|16:12||sunson has joined IRC (firstname.lastname@example.org)|
|16:21||ogra_ has left IRC (email@example.com, Ping timeout: 252 seconds)|
|16:49||killermike has joined IRC (firstname.lastname@example.org)|
|16:57||alkisg has left IRC (alkisg!~alkisg@ubuntu/member/alkisg, Ping timeout: 260 seconds)|
|16:59||alkisg has joined IRC (alkisg!~alkisg@ubuntu/member/alkisg)|
|17:41||alkisg has left IRC (alkisg!~alkisg@ubuntu/member/alkisg, Ping timeout: 240 seconds)|
|17:58||loather has joined IRC (email@example.com)|
|18:00||alkisg has joined IRC (alkisg!~alkisg@ubuntu/member/alkisg)|
|18:09||tier1 has joined IRC (firstname.lastname@example.org)|
|18:12||moose-tek has left IRC (email@example.com, Ping timeout: 245 seconds)|
|19:06||alkisg1 has joined IRC (alkisg1!~alkisg@ubuntu/member/alkisg)|
|19:08||alkisg has left IRC (alkisg!~alkisg@ubuntu/member/alkisg, Ping timeout: 252 seconds)|
syncpackage: Request succeeded; you should get an e-mail once it is processed.
and now ldm is in sync with Debian ;)
|19:16||* Hyperbyte high fives stgraber|
|19:29||* alkisg1 did that with epoptes a few hours ago :D|
|19:32||alkisg1 has left IRC (alkisg1!~alkisg@ubuntu/member/alkisg, Quit: Leaving.)|
|19:43||loather has left IRC (firstname.lastname@example.org, Quit: This computer has gone to sleep)|
|20:04||Guest41868 has left IRC (Guest41868email@example.com, Quit: Forlater kanalen)|
|20:16||loather has joined IRC (firstname.lastname@example.org)|
|21:31||primeministerp has left IRC (email@example.com, Ping timeout: 245 seconds)|
|22:02||vagrantc has joined IRC (vagrantc!~vagrant@75-150-46-245-Oregon.hfc.comcastbusiness.net)|
|22:36||Damianos has left IRC (Damianos!~Damianos@adsl-070-145-074-043.sip.cha.bellsouth.net, Ping timeout: 244 seconds)|
|22:49||alkisg has joined IRC (alkisg!~alkisg@ubuntu/member/alkisg)|
vagrantc, stgraber: I'm thinking of replacing S15-userLoginCheck with the following S21-userLoginCheck, any objections/comments?
alkisg: it does assume that all logins will be from a version that supports the ~/.cache/ltsp/sessions.* business
i.e. a login from an older LTSP will not get detected
Indeed, but how can you separate old logins from plain ssh sessions?
check the environment in /proc/PID/environ for LTSP_CLIENT and such
which won't catch really old sessions...
And won't work in kfreebsd, will it?
well, i'm the only one crazy enough to even think about kfreebsd... right?!
How common will it be for a sysadmin to have one updated chroot and and another older one, loggin in both at the same server?
alkisg: well, i've got multiple concurrent versions of chroots running
I know I used to manage a few networks running up to 3 different version of Ubuntu as thin clients connecting to maybe 2 different version of Ubuntu on the server
alkisg: i'm just thinking about a few extra checks, it doesn't have to work perfectly everywhere, but just a few more safety mechanisms.
(depending on the thin client hardware, on whether they are fat clients or thin clients and depending on what kind of environment you want on the server)
I agree on the concept though, but we just need to make sure it covers most use cases
vagrantc: btw, we're in sync for ldm!
i.e. do not assume homogenous networks :)
stgraber: been working on that delta "since years"!
the biggest thing that improved the delta was moving the themes into their own packages.
yeah, we were mostly back in sync a year or so ago I believe (looking at changelog), but that's the first time we're fully in sync ever, according to https://launchpad.net/ubuntu/+source/ldm/+publishinghistory
it's the first time LP published a package without a -XubuntuY in the version :)
I actually had to force it to do it as the package was still blacklisted in a few places (merge/sync scripts, LP itself and the publisher)
all that's left is LTSP ... :)
so now, that's ltsp-docs, ltspfs and ldm all in sync, I guess my goal will be to have a minimal delta on ltsp by the next LTS, sounds reasonable I guess
ltsp-docs has always been in sync
I doubt we'll ever be in sync for ltsp as we need it to depend on the themes at least and probably some more stuff that we don't share between Debian or Ubuntu or that we share with a different name
though if we really want to be in sync, we could have a control.in doing that bit of the magic for us
true, but if that's the entire difference...
yeah, that too
just checked and indeed ltsp-docs was always imported from Debian, we never touched it in Ubuntu (there really is no reason for Ubuntu specific changes to be honnest)
i actually want to implement some changes in dependencies because a lot of packages (ltsp-client/ltsp-client-core) have the various dependends/recommends/suggests that should all be bumped one direction or another.
|23:20||monteslu has left IRC (firstname.lastname@example.org, Read error: Operation timed out)|
ltspfs was out of sync from time to time because I wanted new versions of it but was resynced from Debian fairly often
i.e. ltsp-client-core suggests or recommends end up as depends for ltsp-client
we used to actually do that with some sort of control.in magic, but somewhere along the way i gave up on it.
Thanks for the comments + good night guys :)
|23:29||alkisg has left IRC (alkisg!~alkisg@ubuntu/member/alkisg, Quit: Leaving.)|
stgraber: if we get to the point where the only packaging differences are dependencies, it would be great to try that again
|23:32||monteslu has joined IRC (email@example.com)|
|23:33||sviesusisalus has joined IRC (firstname.lastname@example.org)|
|23:33||Trixboxer has left IRC (Trixboxer!~Trixboxer@126.96.36.199, Remote host closed the connection)|
vagrantc: yeah, I guess next cycle I'll move ltsp-live to its own separate source package, then try to rebase the existing LTSP packaging on yours. I doubt I'm even using dh7 for it at the moment, so likely quite a bit behind on what you ahve.
|23:43||shamino_ has joined IRC (email@example.com)|
|23:44||sndwrx1 has joined IRC (sndwrx1!~sndwrx@unaffiliated/sndwrx)|
|23:45||zamba has left IRC (firstname.lastname@example.org, Ping timeout: 255 seconds)|
|23:45||shamino has left IRC (email@example.com, Ping timeout: 255 seconds)|
|23:45||marjus has joined IRC (firstname.lastname@example.org)|
|23:45||sndwrx has left IRC (sndwrx!~sndwrx@unaffiliated/sndwrx, Ping timeout: 255 seconds)|