|00:08||alkisg has joined #ltsp|
|00:38||johnny has left #ltsp|
|00:38||johnny has joined #ltsp|
alkisg: you up yet?
Hi vagrantc, *I* should be asking that question :)
It's 8:40 am here :)
just wondering why screen-session.d/XS00-halt-reboot was put into ltsp-trunk rather than ldm-trunk ...
seems like a feature of ldm that itsn't useful without ldm, no?
The reason is in the recent messages sent in the ltsp-developer mailing list,
In short, the correct way to put it was in ldm *script*, and Gadi said that we shouldn't modify it at this point,
Because in other ldm* places it didn't work correctly (either left ghost processes, or it caused flickering)
yes, but you can put it in screen-session.d/XS00-halt-reboot in the ldm-trunk package
No, because that causes flickering (X starts before the shutdown)
ldm-trunk can (and should) ship hooks to ltsp that are only relevent to ldm.
it's just a matter of which source ships the files.
you can put it in the same exact place.
Then we'd need a hook for "after X stops". We currently don't have one, do we?
Now I got it
i think we'd have to add the screen-session.d stuff to ldm, but that's where it belongs.
just like ldm ships the screen script for ldm support on ltsp.
Well that's fine from me, and sorry that I didn't think about it
it's one of those pedantic things that i try to keep a watch on :)
I *did* ping a lot of people to watch over my first commit, though :D
alkisg: no big deal :)
moved it to /var/run, at least :)
I just thought that the debian policy said that one package shouldn't write files in another package's directories....
...so I just went with the flow and put it where all other screen-session.d/* scripts were :)
*.d directories are hooks explicitly designed for other packages to drop files into
vagrantc: in any case, it isn't strictly an ldm thing
they're not supposed to modify other packages files, particularly files in /etc, but *.d directories are made to allow one package to extend the functional.ity of another
alkisg: then why is the file called /var/run/ldm-logout-action ?
Because it was already named as such :D
(from the previous stgraber attempt)
I thought of renaming it, but then I decided not to divert too much
what other things would actually make use of it?
So e.g. one could use it from a kiosk session... or even from rdesktop
ok, well... then the file should stay in ltsp-trunk
though the name leads to confusion, then
You're right on that, but I ldm *is* the main way to use that trick
Sure, if I was writing this from scratch, I'd name it ltsp-logout-action or something...
Or even just ltsp-halt-reboot, because there won't always be a logout involved
seems like one uses /tmp/ldm-logout-action and one uses /var/run/ldm-logout-action
Right, that was because stgraber couldn't write to /var/run (it ran as user, not as root)
so it's two different mechanisms for the same thing?
alkisg, we need this patch to make our build system happy, http://pastebin.com/dc1faff1 GenericName and Categories
renamed it to LTSP Client Reboot as it is confusing if user hits that icon and nothing happens on normal session
vagrantc: yes, and *maybe* stgraber will sometime deprecate his, I don't know...
|00:54||* vagrantc is trying to get it figured out before uploading to debian|
vagrantc: I'm OK with whatever changes you think appropriate :)
i'll poke stgraber to see if it'll be deprecated
and then i'll just make sure it doesn't hit a debian stable release
cyberorg: seems like changing the icon may not be compatible with other distros
alkisg, is it possible to install .desktop files via makeinstall instead of putting it in doc?
though it's just an example anyways
vagrantc, yes that is why i used something generic enough "gdm"
cyberorg: yes, but at least Ubuntu afaik doesn't want to install any ltsp .desktop files...
They're just there as examples
ok np, i do it from rpm spec so will carry on with that :)
And if this gets accepted, they won't be at all useful anymore: https://bugs.launchpad.net/ubuntu/+source/gnome-session/+bug/491940
vagrantc: if you think we should change the xprop name, I'll need to propose a different patch for gnome-session, so please tell me soon enough for Lucid :)
alkisg: i think the name *should* be made generic, yes.
weather it's worth it, i don't know.
|01:02||* alkisg wonders why there's only feedback *after* the commit, and not before :D|
alkisg: sorry ... was caught up with other things and saw a flurry of activity...
alkisg: i tried commenting on irc as much as i could...
Nah, sorry for my inexperience
cyberorg: I don't remember why, but in iTalc I preferred to use dbus to tell gnome to logout instead of calling gnome-session-save: http://launchpadlibrarian.net/36400397/local_system_ica.diff
alkisg, you get feedback after commit because every tests it after that :)
Heh, well at least for Ubuntu I had packages in my PPA before commiting the patch
|01:08||* vagrantc hasn't even tested it yet|
just spent time building packages
alkisg, building packages locally works, our build service complains if it does not like something, the icon, GenericName and Category was missing, without that packages don't build
vagrantc: the way stgraber did it, left me with ghost processes, that's one of the main reason I tried to do it in a different way (the other one being that xprop would work even if localapps are disabled)
alkisg: yeah, seems better.
cyberorg: I've never even seen those .desktop files, they're in the ltsp source tree?
|01:10||* alkisg looks...|
cyberorg: ah, you are *not* using my patch :)
alkisg, i suspect gnome-session-save is just a wrapper that does dbus call or the other way around
That's stgraber's patch
It was there for some time...
ah, so we have two ways of doing that?
Don't you have ghost processes on the server when you use stgraber's way of doing it?
now that you mention it i'll pay attention :)
But also that way localapps need to be enabled...
(for it to work)
hmm, so what is your way to do it?
for this there was no patching required, just new ltsp packages were enough
Yes, because the patch was there for some time now
there should be DE independent way of doing this, getting gnome-session patched will just get gnome working, not all the other DEs
my way: xprop -root -f LDM_LOGOUT_ACTION 8s -set LDM_LOGOUT_ACTION REBOOT (and call gnome logout)
It's basically the same way, but it's handled by a different code in the scripts
So the "echo" part is done from the client side, not from the .desktop file
then you need to create at least one more patch for KDE too :)
For the .desktop file?
It'd be nice if we could send those patches upstream to kde-session and gnome-session...
This way it'd work even if someone tried to reboot via dbus
no, .desktop way does not require any DE patching
E.g. now with the patched gnome-session, I can properly reboot the client not only from the gnome menus, but also from iTalc etc
need patch for gnome-session, the one that is not a patch to a patch :)
we can discuss with vuntz to get it in gnome 2.28.2 if possible
Ugh, I wouldn't know how to test with vanilla gnome-session... Does debian have a vanilla gnome-session?
|01:30||* alkisg reboots because of a weird X problems he's been experiencing once per week the last months... :(|
|01:30||alkisg has quit IRC|
|01:33||alkisg has joined #ltsp|
|01:43||makghosh has joined #ltsp|
|01:46||* vagrantc hopes other distros start using ltsp-info ...|
alkisg, you are right there are a lot of processes around after shutdown
cyberorg: yup, and because my patch is using the same xprop name as stgraber's patch, the only way to properly reboot is to *remove* stgraber's patch :-/
Maybe that would give us another reason to do as vagrantc said and rename the xprop to something that doesn't have "ldm" in it :D
vagrantc: isn't that upstream now? why wouldn't the other distros include it?
Also btw, the fact that ghost processes remain after the user logs off and are only terminated when X shuts down properly, I think means that we are doing something wrong in the session-closing code...
alkisg, what changes are required in client chroot for xprop -root -f LDM_LOGOUT_ACTION 8s -set LDM_LOGOUT_ACTION HALT to work?
cyberorg: the one and only commit I've made :D
|02:01||* alkisg looks for the link..|
alkisg, that is already in our package, but nothing seem to happen when running that
Do you have the client/screen-session.d/XS00-halt-reboot file?
Hmmm where's the X99 file diff?
alkisg: it's upstream, but gotta remind the distros to include it
Ah, it's on the ldm-trunk, right...
alkisg, yes it is there in the client chroot
cyberorg: and you also have the ldm patch? http://bazaar.launchpad.net/~ltsp-upstream/ltsp/ldm-trunk/revision/1223
ldm still needs patch? i thought that was committed
Yes, it is
I mean the commited ldm patch :D
If those 2 files LDM/rc.d/X99-zzz-logout-action and LTSP/client/screen-session.d/XS00-halt-reboot contain the new code, then runing `xprop -root -f LDM_LOGOUT_ACTION 8s -set LDM_LOGOUT_ACTION HALT` and then logging off should poweroff the client
i see your changes in X99-zzz-logout-action too
(without leaving ghost processes)
oh you mean log out is still required?
Yup, the xprop only "marks" that a halt is needed after logoff
ok that seems to have worked, log out halted the client
and no ugly leftover processes
So in the desktop files something like this would be needed: sh -c 'prop -root -f LDM_LOGOUT_ACTION 8s -set LDM_LOGOUT_ACTION HALT; logoff'
Exec=sh -c '-root -f LDM_LOGOUT_ACTION 8s -set LDM_LOGOUT_ACTION HALT; gnome-session-save --logout'
damn forgot the xprop
Exec=sh -c 'xprop -root -f LDM_LOGOUT_ACTION 8s -set LDM_LOGOUT_ACTION HALT; gnome-session-save --logout'
Exec=sh -c 'xprop -root -f LDM_LOGOUT_ACTION 8s -set LDM_LOGOUT_ACTION HALT; gnome-session-save --logout'
; is better than && here
you copy paste faster :)
;) We could also change that to an "ltsp-logoff" script, and just call that
(that would automatically detect if KDE/gnome etc is used)
|02:16||* alkisg prefers the gnome-session patch, though :D|
there must be something like XDG_LOGOFF
fixed the package, patched .desktop file to your procedure :)
works perfect now
can you update the .desktop in example to use this?
cyberorg: I think stgraber should do that, as he's the one that provided the .desktop file
Or someone else, it's my first patch, I wouldn't want to get too intrusive :D
|02:33||makghosh has quit IRC|
alkisg, you will only be improving it by putting proper command :)
cyberorg: ok, but I'll need an OK from stgraber first. I'll ping him whenever I find him :)
of course :)
Ah, we don't have anything similar for KDE, do we?
iTalc says that this is for KDE logout: dcop ksmserver ksmserver logout 0 0 0
kde is a mess, kde3 and kde4 may be different
I really wish xdg gets gnome & kde more close...
let me test kde4
That iTalc line should be from the kde3 era, so if it works on KDE4, then I guess it's good enough to provide as an example in ltsp.
|02:49||vagrantc has quit IRC|
i get "ERROR: Couldn't attach to DCOP server"
cyberorg: Is that from within the session? Or from a terminal outside of X?
Maybe KDE4 switched to dbus...
alkisg, kde does not use dcop anymore, it is now dbus
Right, so I guess we'd need to run d-feet to see the proper dbus call
(or just google it :))
Heh, anyway, I never use KDE so I'll just pass on that :)
|03:04||Kicer86 has joined #ltsp|
i wonder ...
if we use xprop for all communication already ...
instead of attaching dbus to ssh we should probably just have communication to the client dbus going through xprop too ...
xprop -root -f LTSP_GREETING 8s -s GREET_OGRA HAI
i wonder if the traffic gets to high for that ... dbus can be noisy, but i imagine its way easier to implement
|03:07||* cyberorg logs out of kde via GUI|
your method doesnt really integrate with the desktop logout buttons, does it ?
ogra, that was just giving up finding kde command to log out
kde should send a dbus message as gnome-session does
found this "qdbus org.kde.ksmserver /KSMServer logout 0 0 0" that works, but the person who suggested has 0 2 0 instead of 0 0 0
probably we should just capture that specific message on the server
right, i guess qdbus just sends a message to the system dbus
similar to gnome
hmm, that was typo from him, it is 0 0 0
ltsp should have a listener that sits on top of the system dbus on the server ... and steals the messages ...
|03:15||* ogra is brainstorming .... i'm on holiday and actually have some time for ltsp work|
|03:18||Kicer86 has quit IRC|
alkisg, here is the script to logoff according to kde/gnome http://pastebin.com/m128745d
cyberorg: 0 2 0 might mean reboot instead of logout, or might mean "force logout" or something...
alkisg, that was a typo
ogra, well, since you're on holidays, what better way to spend them than mergins the d-buses :P :D
|03:59||* alkisg is just kidding...|
cyberorg: why don't you commit a .desktop example file for KDE as well? I imagine KDE users would appreciate it...
alkisg, two different .desktop files shouldn't be required, just replacing log out command with a call to script ^^
cyberorg: also, GNOME_DESKTOP_SESSION_ID=this-is-deprecated
I don't think you should rely on an environment variable to see that, I think it'd be better if you checked if sessionmanager is running
(or even pgrep gnome-session)
alkisg, see detectDE() in /usr/bin/xdg-su
I don't have that file, but: echo $GNOME_DESKTOP_SESSION_ID
I trust gnome itself more than some example :)
hmm, yes it says "this-is-deprecated"
so we use whatever new way turns up in xdg
I think the proper way would be to see if the gnome session manager is running on the system dbus
(or the session dbus)
|04:05||ogra has quit IRC|
E.g. call org.gnome.SessionManager.CanShutdown() and see if the call is available, and what it returns
If it returns true, then we're in a gnome session, and we can shutdown.
cyberorg: dbus-send --session --dest=org.gnome.SessionManager --type=method_call --print-reply --reply-timeout=2000 /org/gnome/SessionManager org.gnome.SessionManager.CanShutdown
method return sender=:1.3 -> dest=:1.684 reply_serial=2
If that returns true, then we're in a gnome session.
But I guess you could also just call both of the DE logout functions - I don't see how it would hurt... => 2 dbus-send calls
You could even put both of the calls in the .desktop file, and avoid the need to ship a logout script
something like this? http://pastebin.com/d20e286
|04:20||ogra has joined #ltsp|
|04:21||makghosh has joined #ltsp|
cyberorg: or like this: Exec=sh -c 'xprop etc; gnome-session-save --logout; qdbus org.kde.ksmserver /KSMServer logout 0 0 0"
alkisg, ah send both commands :)
This way no ltsp-logoff script would be needed...
It's a hack, but well, the proper way here would be for ltsp to listen for dbus events :D
alkisg, it works :)
cyberorg: also, .desktop files have an OnlyShowIn=GNOME (or KDE or XFCE) entry , so one could have multiple desktop files and only one would show up depending on the DE
This may be better, as it can have different icons depending on the DE
|04:38||* cyberorg likes this hack for its simplicity|
So, 2 .desktop files, one for KDE and one for GNOME, with the respective OnlyShowIn entry (and later on 1 .desktop file for XFCE)
And both of the desktop files would be installed, no need to only install only one of them
E.g. try: grep OnlyShowIn /etc/xdg/autostart/*.desktop
That's also the way synaptic uses:
dpkg -L synaptic|grep desktop
we use it for compiz settings tool
It may be better, as it allows for different icons. E.g. KDE might not have the gnome icon...
|04:42||* alkisg still prefers the gnome-session patch, though :)|
we can ship our own icons though
Right, but it's also more elegant (to not execute 3 commands in a row without checking the DE first )
And "OnlyShowIn" is quite standard: http://standards.freedesktop.org/menu-spec/latest/apb.html
we need xdg-logout script to take care of all possible DEs
Yeah, that'd be nice... or even a standarized dbus function call
|04:57||alkisg has quit IRC|
|05:08||alexqwesa_ has quit IRC|
|05:08||Selveste1 has quit IRC|
|05:11||alexqwesa has joined #ltsp|
|05:34||alkisg has joined #ltsp|
|05:36||rjune has quit IRC|
|05:38||rjune has joined #ltsp|
|05:41||pmatulis has joined #ltsp|
|05:56||Faithful has joined #ltsp|
|06:04||Kicer86 has joined #ltsp|
|06:06||alexqwesa_ has joined #ltsp|
|06:07||rjune has quit IRC|
|06:08||rjune has joined #ltsp|
|06:20||alexqwesa has quit IRC|
|06:20||rjune has quit IRC|
|06:20||rjune has joined #ltsp|
|06:24||Egyptian[Home] has quit IRC|
|06:25||Egyptian[Home] has joined #ltsp|
|06:31||kusznir_ has joined #ltsp|
|06:40||rjune has quit IRC|
|06:40||rjune has joined #ltsp|
|06:44||kusznir has quit IRC|
|06:55||rjune has quit IRC|
|06:56||rjune has joined #ltsp|
|07:06||Roel__ has joined #ltsp|
|07:12||rjune has quit IRC|
|07:13||rjune has joined #ltsp|
|07:24||Roel_ has quit IRC|
|07:50||pmatulis has quit IRC|
|07:53||rjune has quit IRC|
|07:54||rjune has joined #ltsp|
|07:58||F-GT has quit IRC|
|08:01||pmatulis has joined #ltsp|
|08:11||F-GT has joined #ltsp|
|08:15||Selveste1 has joined #ltsp|
|08:15||alkisg has quit IRC|
|08:31||din_os has joined #ltsp|
|08:37||alkisg has joined #ltsp|
|09:11||ehrenpr has quit IRC|
has anyone here used the spice protocol
|09:52||Ahmuck-Jr has quit IRC|
|09:52||makghosh has quit IRC|
|09:53||Ahmuck-Jr has joined #ltsp|
|09:53||prpplague_afk is now known as prpplague|
|09:55||rjune has quit IRC|
|09:56||rjune has joined #ltsp|
|09:56||Ahmuck-Sr has quit IRC|
|10:12||F-GT has quit IRC|
|10:12||nubae__ has quit IRC|
|10:12||shogunx has quit IRC|
|10:12||map7_ has quit IRC|
|10:12||nubae__ has joined #ltsp|
|10:15||F-GT has joined #ltsp|
|10:15||map7_ has joined #ltsp|
|10:15||shogunx has joined #ltsp|
|10:22||ehrenpr has joined #LTSP|
can anyone help me setup opendns with my edubuntu server. i have configured dhcpd.conf with opendns address, but it doesnt seem to be working. I have a two nic setup.
|10:24||yanu has quit IRC|
|10:24||foxbuntu has quit IRC|
|10:24||squarepeg has quit IRC|
|10:24||_USUrPeR_ has quit IRC|
|10:24||korcan has quit IRC|
|10:24||Sarten-X has quit IRC|
|10:24||nubae has quit IRC|
|10:24||nubae has joined #ltsp|
|10:27||foxbuntu has joined #ltsp|
ehrenpr: put DNS_SERVER=<ip> and SEARCH_DOMAIN=local in lts.conf
|10:30||rjune has quit IRC|
Also, you'll need to do NATting or something on the server...
do I need to also chane /etc/resolv.conf
|10:31||rjune has joined #ltsp|
|10:32||Ahmuck-Jr has quit IRC|
There's a bug about resolv.conf in Karmic, so you either have to put the patch, or use stgraber's PPA which has a newer ltsp version with the bug fixed
alkisg: "stgraber-ppa" :: https://launchpad.net/~stgraber/+archive/ppa
|10:33||yanu has joined #ltsp|
ok. let me try that
|10:34||Ahmuck-Jr has joined #ltsp|
ehrenpr: why do you want this DNS? Do you use localapps?
yes. Also I need some filtering for my school. and figured this would be the simplest way to go
do you recommend something else
|10:39||lucascoala has joined #ltsp|
|10:41||Sarten-X has joined #ltsp|
|10:41||korcan has joined #ltsp|
|10:41||_USUrPeR_ has joined #ltsp|
|10:41||squarepeg has joined #ltsp|
|10:42||rjune has quit IRC|
|10:42||rjune has joined #ltsp|
and dans guardian
there are some firefox add ons that do filtering as well
ehrenpr: u from the us?
yes western ny. I have used opendns filtering at home and find it a good start, but Working at a school I want to be extra cautios and setup a couple levels
|10:52||rjune has quit IRC|
|10:52||rjune has joined #ltsp|
|11:26||Faithful has quit IRC|
|11:40||alincoln has quit IRC|
|11:42||alincoln has joined #ltsp|
|12:45||ehrenpr has quit IRC|
|13:46||laprag has joined #ltsp|
|14:31||Ahmuck has quit IRC|
|14:31||Ahmuck has joined #ltsp|
|14:48||pmatulis has quit IRC|
|14:51||Kicer86 has quit IRC|
|15:04||Faithful has joined #ltsp|
|15:06||vagrantc has joined #ltsp|
|15:52||Ahmuck has quit IRC|
|15:53||Ahmuck has joined #ltsp|
|16:05||din_os has quit IRC|
|16:15||yanu_ has joined #ltsp|
|16:15||yanu has quit IRC|
|16:24||johnny has left #ltsp|
|16:30||jhutchins has joined #ltsp|
|16:34||yanu has joined #ltsp|
|16:39||alkisg has quit IRC|
|16:45||yanu_ has quit IRC|
|16:58||rjune has quit IRC|
|16:58||rjune has joined #ltsp|
|17:00||sene has joined #ltsp|
|17:11||Faithful has quit IRC|
|17:11||Faithful has joined #ltsp|
|17:12||SalsaDoom has joined #ltsp|
Hi fellas. Having a problem with logging into an ubuntu 9.10 ltsp server. It boots ok, and loads up an x11 login screen just fine -- but any user account says "authentification failed". /var/log/auth.log doesn't show anything about this..
|17:13||Selveste1 has quit IRC|
Now, I've read what appears to be the relevant documentation, but there really isn't a lot. Google hasn't helped me either so far...
|17:13||Selveste1 has joined #ltsp|
|17:20||Ahmuck-Jr has quit IRC|
|17:21||johnny has joined #ltsp|
|17:28||SalsaDoom has quit IRC|
|18:00||pmatulis has joined #ltsp|
|18:24||alexqwesa_ has quit IRC|
|18:26||alexqwesa_ has joined #ltsp|
|18:30||Ahmuck-Sr has joined #ltsp|
|18:36||Ahmuck-Sr has quit IRC|
|18:43||Ahmuck-Sr has joined #ltsp|
|18:47||alexqwesa_ has quit IRC|
|19:08||Sarten-X has quit IRC|
|19:09||Sarten-X2 has joined #ltsp|
|19:09||alexqwesa_ has joined #ltsp|
|19:15||avena has joined #ltsp|
|19:57||pmatulis has quit IRC|
|20:27||rjune has quit IRC|
|20:28||rjune has joined #ltsp|
|20:47||Sarten-X2 has quit IRC|
|20:56||Sarten-X has joined #ltsp|
|21:10||rjune has quit IRC|
|21:11||rjune has joined #ltsp|
|21:33||johnny has left #ltsp|
|21:43||rjune has quit IRC|
|22:21||sene has quit IRC|
|22:25||vagrantc has quit IRC|
|22:36||avena has quit IRC|
|23:08||Ahmuck-Sr_ has joined #ltsp|
|23:23||Ahmuck-Sr has quit IRC|
|23:39||Sarten-X has quit IRC|
|23:40||Sarten-X has joined #ltsp|
|23:41||johnny has joined #ltsp|