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


Channel log from 29 December 2019   (all times are UTC)

03:43vagrantc has left IRC (vagrantc!~vagrant@unaffiliated/vagrantc, Quit: leaving)
04:26
<alkisg>
markit, those ltsp.conf options are grouped and documented together
04:26
vagrantc released a new ltsp, which is newer in the repos than the ppa, this isn't an issue
09:43markit has joined IRC (markit!~marco@mail.ammdomus.it)
11:24
<markit>
alkisg: ping, just in case you are not busy (nothing urgent)
12:59gbaman has joined IRC (gbaman!~gbaman@host81-142-46-233.in-addr.btopenworld.com)
13:17alkisg_android has joined IRC (alkisg_android!~yaaic@109-178-252-18.pat.ren.cosmote.net)
13:17
<alkisg_android>
markit, pong
13:23
<markit>
alkisg_android: first a bug report about documentation, don't know if you was able to read my last message of yesterday
13:24
<alkisg_android>
i replied, its not a bug, see irxlogs
13:24
they're just groupped together
13:25
<markit>
I see, ok, but the man look confusing, and LIGHTDM_CONF and SDDM_CONF are not explained and confusing
13:26
I mean, LIGHTDM_CONF seems useful to pass some config to lightdm
13:26
while SDDM_CONF seems useful only to pass the path of the config file...
13:26
<alkisg_android>
Kubuntu uses sddm
13:26
not lightdm
13:26
<markit>
yes, but can I use SDDM_CONF=someconfigparameterhere?
13:27
like I could do with LIGHTDM_CONF?
13:27
<alkisg_android>
sure, the paragraph mentions either string or filename for all of them
13:27
<markit>
i.e. to have the equivalent of "greeter-hide-users=true" would be for me "MinimumUid=6000"
13:28
ok, now I see... it's "obvious" when you already know :P
13:28
<alkisg_android>
Finally, the *_CONF parameters can be either filenames or direct text, and provide a way to write additional content to the generated display manager configuration.
13:29
the dm man pages are separate
13:29
ltsp.conf doesn't include all the possible sddm options etc
13:30
https://www.mankier.com/5/sddm.conf
13:30
<markit>
Now I've tried to broadcast the screen, but seems not to work. You told me once that could be done even with users NOT logged it, right? Doesn't work so and nor if the users are logged (let me test it again to be sure after reboot of the server)
13:32
<alkisg_android>
if it doesn't work, right click from epoptes, open Terminal, root, locally, and run ps faux | nc termbin.com 9999
13:33
<markit>
should also work without users logged in? Or do you prefer having clients (2) logged?
13:34
<alkisg_android>
it should work
13:34
if it doesn't, run the above command
13:34
<markit>
(btw, boot seems much slower than ltsp5, kernel and services takes a lot more time to load)
13:36
"right click from epoptes"?
13:36
do you mean "help -> remote support"?
13:37
in any case I've run the above command in a terminal (konsole), is not related to remote support
13:38
https://termbin.com/lwij
13:38
<alkisg_android>
no i mean right click on the red monitor thumbnail on epoptes,before client login
13:39
boot is faster in ltsp19 compared to 5. but kibuntu 20.04 may be slower than18.04
13:40
<markit>
https://termbin.com/aohyl
13:40
(from one of the two clients)
13:41
<alkisg_android>
at that point there, also run xrandr
13:41
does it work?
13:41
<markit>
yes
13:41
<alkisg_android>
and broadcast doesn't?
13:42
<markit>
no, nothing happens, not even error message
13:42
server and clients all have same resolution, btw
13:42
<alkisg_android>
ok file a bug report in Epoptes
13:43
<markit>
ok, any further info I need to provide?
13:43
<alkisg_android>
no jist kubuntu 20.04
13:46
<markit>
ok, thanks. Btw, with iftop, from boot to a responsive desktop with Kubuntu I've TX 197MB RX 8MB TOTAL 205MB just to reach login (42"), then TX 80MB RX 5MB TOTAL 85MB to reach a responsive desktop (additional 34").
13:47
Test VM, 1.5Gbps lan performance (from epoptes test), vm on ssd
13:47
alkisg_android: I've to run, will fill the bug report as soon as possible, thanks again for your support
13:48markit has left IRC (markit!~marco@mail.ammdomus.it, )
13:48
<alkisg_android>
np, gotta go too, I'm hiking on vikos gorge ;)
13:48alkisg_android has left IRC (alkisg_android!~yaaic@109-178-252-18.pat.ren.cosmote.net, Quit: Yaaic - Yet another Android IRC client - http://www.yaaic.org)
14:14gbaman has left IRC (gbaman!~gbaman@host81-142-46-233.in-addr.btopenworld.com, Remote host closed the connection)
14:15gbaman has joined IRC (gbaman!~gbaman@host81-142-46-233.in-addr.btopenworld.com)
14:19gbaman has left IRC (gbaman!~gbaman@host81-142-46-233.in-addr.btopenworld.com, Remote host closed the connection)
14:20gbaman has joined IRC (gbaman!~gbaman@host81-142-46-233.in-addr.btopenworld.com)
14:33gbaman_ has joined IRC (gbaman_!~gbaman@host81-142-46-233.in-addr.btopenworld.com)
14:36gbaman has left IRC (gbaman!~gbaman@host81-142-46-233.in-addr.btopenworld.com, Ping timeout: 258 seconds)
15:05gbaman_ has left IRC (gbaman_!~gbaman@host81-142-46-233.in-addr.btopenworld.com, Remote host closed the connection)
15:06gbaman has joined IRC (gbaman!~gbaman@host81-142-46-233.in-addr.btopenworld.com)
15:11markit has joined IRC (markit!~marco@mail.ammdomus.it)
15:12
<markit>
hi alkisg, after any modification of /usr/share/ltsp/common/ltsp/ltsp.conf, I just need to issue # ltsp initrd and reboot the client?
15:16gbaman has left IRC (gbaman!~gbaman@host81-142-46-233.in-addr.btopenworld.com, Ping timeout: 265 seconds)
15:19gbaman has joined IRC (gbaman!~gbaman@host81-142-46-233.in-addr.btopenworld.com)
15:20alkisg_android has joined IRC (alkisg_android!~yaaic@109-178-228-239.pat.ren.cosmote.net)
15:20
<markit>
I'm playing with ltsp.conf and sddm, seems that does not take the settings I put there
15:20
alkisg_android:
15:20
<alkisg_android>
markit, install -m 0660 -g sudo /usr/share/ltsp/common/ltsp/ltsp.conf /etc/ltsp/ltsp.conf
15:21
see the man page
15:21
its in etc, not usr
15:21
<markit>
my bad :(
15:21
<alkisg_android>
https://ltsp.org/man/ltsp.conf/
15:22
<markit>
I thought was the replacement of "lts.conf" just for clients, while /etc/ltsp.conf was for "server side"...
15:22
I should start with a clean mind :(
15:23
so /etc/oltsp.conf sketeton is not automatically created by package? That should be fixed :)
15:24adrianorg has left IRC (adrianorg!~adrianorg@191.32.96.65, Quit: leaving)
15:24
<alkisg_android>
its /etc/ltsp/ltsp.conf
15:25
you need that install command to create it
15:35
<markit>
is it possible in SDDM_CONF= specify a "section" the parameter belongs to? I.e. [Users] and parameter MinimumUid=6000
15:35
(btw, now ltsp.conf works, of course...)
15:36
<alkisg_android>
its just text you can put anything
15:36
its copied to the end of sddm.conf
15:42
<markit>
I mean, in sddm.conf there are sections, how can I tell in what section I want that parameter to be put?
15:42
(if I can, of course...)
15:43
<alkisg_android>
you write the section you wwnt
15:43
want
15:50
<markit>
I'm confused... I don't want (if possible) create a parallel sdm.conf file and set SDDM_CONF="/etc/ltsp/sddm.conf"
15:50
I would like to find a solution like for LightDM, so SDDM_CONF="MinimumUid=6000"
15:51
but I need to specify in what section ltsp script has to "virtually" put it
15:53
also I don't understand, the "image" (chrootless) is produced from /etc/, so clients see the "server" /etc as was when the "image" was produced, right? Anything "new" needs to "override" it in some way / with some trick?
15:58
<alkisg>
markit: SDDM_CONF can be a multiline var
15:58
SDDM_CONF="[section]
15:58
parameters"
15:59
If you need to customize /etc, you can do so with POST_INIT_x commands
15:59alkisg_android has left IRC (alkisg_android!~yaaic@109-178-228-239.pat.ren.cosmote.net, Quit: Yaaic - Yet another Android IRC client - http://www.yaaic.org)
16:00
<alkisg>
Whatever you put in SDDM_CONF, it's appended to the end of the one and only sddm.conf
16:00
There's no custom file involved
16:02
See the code, it's just an echo $SDDM_CONF >> /etc/sddm.conf
16:02
https://github.com/ltsp/ltsp/blob/master/ltsp/client/init/55-display-manager.sh#L107
16:02
Nothing complicated
16:09
<markit>
ok, wondering what will happen when the resulting "virtual" sddm.conf is parsed and has duplicated sections, probably nothing wrong, time to try it, thanks
16:09
<alkisg>
I've tried a couple of display managers, they didn't have an issue with that
16:38gbaman has left IRC (gbaman!~gbaman@host81-142-46-233.in-addr.btopenworld.com, Remote host closed the connection)
16:39gbaman has joined IRC (gbaman!~gbaman@host81-142-46-233.in-addr.btopenworld.com)
16:42
<gbaman>
alksig, I am back at fiddling around with the new LTSP and Pis (with Raspbian). Got it to the point of happily booting and reaching a (commandline) login prompt, at which point can't seem to log in to any account... But in some digging, no extra users in /etc/passwd
16:42
(got in via disabling PAM and using root, to poke around)
16:44
Working with the Pi, by booting with kernel/initramfs on the card (as only got a Pi 4 handy). Then catted the ltsp.img onto the end of the kernel initramfs
16:44
Given Pis lack of passing in 2 initramfs's
16:46
<alkisg>
gbaman: put the ltsp.img first, and initrd.img second
16:46
After a .gzip the kernel can't decode an uncompressed initrd (ltsp.img)
16:47
I.e. { cat ltsp.img; cat initrd-kernel-version.img; } > initrd.img
16:47
But if you disable PAM then you won't get any users in /etc/passwd
16:48
<gbaman>
I had been uncompressing the gzipped bit, then merging with the uncompressed ltsp.img, then recompressing :)
16:48
<alkisg>
Nah too much trouble that way. Actually: cat ltsp.img initrd-version.img > initrd.img
16:48
No need for {}
16:48
<gbaman>
That sounds far too easy...
16:49
(may have spent a while figuring out why just catting them the other way wasn't working...)
16:50
<alkisg>
https://github.com/ltsp/ltsp/issues/9
16:50
And then https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1840945
16:51
It took a while to find a solution to that one...
16:52
<gbaman>
How did I not come across that Github issue...
16:52
I have been poking around plenty of them
16:56adrianorg has joined IRC (adrianorg!~adrianorg@191.32.96.65)
16:56
<gbaman>
Ok, so putting them around the other way does work, but still same issue, can't get logged on (with PAM enabled again)
16:56
Getting "read: Connection reset by peer"
16:57
Yet, I am not seeing anything come through in auth.log on the server?
16:59
<alkisg>
gbaman: is the user in /etc/passwd now?
16:59
<gbaman>
Well, now I can't get in :)
16:59
To check that
16:59
<alkisg>
No epoptes to check?
17:00
You can also enable root login in ltsp.conf
17:00
man ltsp.conf, see the last lines
17:00
<gbaman>
Nope, no Epoptes apparently working oddly
17:00
<alkisg>
# Set the root password to "qwer1234" for all clients.
17:00
Did you generate ltsp.img on the server?
17:00
So that it contains epoptes keys and users and all?
17:00
<gbaman>
Although, a whole stack of stuff is broke... I have a console in front of me currently via UART serial
17:01
Will give that a go
17:01
And yeah, generating the ltsp.img via ltsp initrd each time
17:01
<alkisg>
On the server, not on a chroot, right?
17:01
<gbaman>
Yeah?
17:01
<alkisg>
OK
17:02
Set a root password to with the POST_INIT command in man ltsp.conf
17:02
So that you'll be able to login to the client, and check things, and also you'll make sure that ltsp.conf parsing is correct , etc
17:06
"Teacher of Computer Science at Westminster School." ? Hey nice we have the same main job now :D
17:07
<gbaman>
Hahaha, yeah a "real" teacher now
17:07
Got an actual classroom of my own now to run PiNet
17:08* alkisg has been working in a school IT support office for the last 7 years and kinda misses the "ltsp in the classroom" a bit ...
17:08
<alkisg>
On the other hand, I have 1000+ teachers/schools now asking for my help daily; I think I prefer that one better
17:09
<gbaman>
Nah, give me a class of students over other teachers...
17:09
Right, now in and the users do exist in /etc/passwd
17:10
But certainly been quite a step up, coming from a software developer over to an actual real CS teacher, especially in a school like this
17:16
I have also tested, on the server I can mount /home via sshfs
17:16
But odd to not be seeing anything in /var/log/auth.log?
17:16
Would have expected to see it trying?
17:21
<alkisg>
If for example ltsp.img is not correct and doesn't have the server ssh keys, then the client won't auto-accept the keys and won't ever reach the server
17:21
Does ssh user@server work, from the client?
17:21
also, does mkdir /tmp/sshfs; sshfs user@server: /tmp/sshfs, work from the client?
17:27woernie has joined IRC (woernie!~werner@p57A0E783.dip0.t-ipconnect.de)
17:28
<gbaman>
Is "server" gone...? The DNS entry?
17:28
Ohh, it's in the hosts file, but no actual IP address
17:37
<alkisg>
gbaman: did the client do a dhcp request in initramfs?
17:37
nfsroot etc?
17:38
<gbaman>
Yeah, but the server IP came from the cmdline.txt
17:40
And yeah, using nfsroot
17:40
<alkisg>
gbaman: what's the output of cat /proc/cmdline?
17:41
<gbaman>
coherent_pool=1M 8250.nr_uarts=1 cma=64M bcm2708_fb.fbwidth=1824 bcm2708_fb.fbheight=984 bcm2708_fb.fbswap=1 smsc95xx.macaddr=DC:A6:32:03:3B:CA vc_mem.mem_base=0x3ec00000 vc_mem.mem_size=0x40000000 dwc_otg.lpm_enable=0 console=ttyAMA0,115200 kgdboc=ttyAMA0,115200 console=tty1 boot=nfs nfsroot=10.0.12.71:/opt/ltsp/armhf elevator=deadline rootwait
17:42
(got the extra stuff in there for serial console)
17:43
<alkisg>
https://github.com/ltsp/ltsp/blob/master/ltsp/client/init/45-networking.sh#L178
17:43
That's the code that detects SERVER
17:43
Try to run that function now, to see if it works or not
17:44
But it should at least fall back to 192.168.67.1, it shouldn't ever be empty
17:46
gbaman: eeeh /opt/ltsp? That doesn't look right
17:47
Check ltsp.ipxe for an example command line
17:47
<gbaman>
Haha, that's me, all my other code still uses /opt/ltsp, so my copy of Raspbian is in there currently
17:47
<alkisg>
Also, if you've ran `ltsp image`, you'll also need ltsp.image= in the cmdline
17:47
<gbaman>
But will get around to moving it over to /srv eventually
17:49
So, running that code does work
17:50
<alkisg>
Put POST_INIT_DEBUG="ltsp_debug" in ltsp.conf, under [clients]
17:50
That will give you a shell right after `ltsp init` runs
17:50
At that point, you'll be able to check if everything works at that point, not afterwards
17:50
bbiab
17:59
<gbaman>
Haha, "/usr/sbin/ltsp: 1: eval: ltsp_debug: not found"
17:59
And then has got stuck on stopping plymouth
18:02
<alkisg>
gbaman: sorry, make that: debug_shell
18:11
<gbaman>
So now made it to a debug:/$ prompt alkisg, but no actual input is working...
18:11
Including UART serial
18:13
Got to nip off for a bit, but will give it another go when back. Thanks for the help so far
18:32
<alkisg>
Put an additional POST_INIT_OPENVT="openvt bash" or something, in case plymouth isn't properly deactivated in your case
18:32
So that you have bash on vt2
18:43
<gbaman>
Hmm, still same issue alkisg
18:52
<alkisg>
gbaman: do you have a shell on vt2?
18:53
alt+ctrl+f2
18:54
Ah it might need an earlier word, e.g. POST_INIT_AOPENVT
18:54
So that it runs before debug shell
18:55
<gbaman>
Ignore me, got it working now
18:55
Well, working enough, so will see if it can get the $server variable now
19:00
So it is already set it seems by that point alksig, in .etc.hosts
19:00
*/etc/hosts
19:00
And set blank
19:01
Aka " server"
19:11
<alkisg>
Try to run that code at that point, to see what goes wrong
19:18
<gbaman>
This is a surprisingly long line to hand type out when there is no ouput as you type :) https://github.com/ltsp/ltsp/blob/master/ltsp/client/init/45-networking.sh#L193
19:19
<alkisg>
Why no output, openvt not working?
19:19
<gbaman>
Oddly not getting output as I type
19:20
<alkisg>
You can just type:
19:20
set -x
19:20
detect_server
19:20
This will show all the executed lines
19:20
I've seen the 'no output' issue when plymouth was running
19:21
but debug_shell is supposed to stop plymouth
19:21
You can pass plymouth.enable=0 to completely disable it
19:21
<gbaman>
Ahh, it is returning on line 179
19:21
<alkisg>
How so?
19:21
<gbaman>
(well, 180)
19:21
<alkisg>
You said that root=/dev/nfs was there
19:22
Ah you use boot=nfs, instead of root=/dev/nfs?
19:22
<gbaman>
Yeah
19:22
<alkisg>
That's a debianism
19:22
<gbaman>
Haha
19:23
<alkisg>
The correct one is root=/dev/nfs, it's a kernel parameter, documented upstream
19:25
<gbaman>
Ok, so now that is working
19:26
Will give the whole thing a test now
19:26Ark74 has left IRC (Ark74!~Luis@177.238.145.140, Quit: Leaving)
19:28
<gbaman>
Bingo
19:28
Ok, so I can no actually log in to network users :)
19:37
<alkisg>
now or not?
19:37
<gbaman>
*now
19:37
That might help
19:37
Still got a blank screen, but at least can log in (via UART)
19:45
Either lightdm or xorg is unhappy, so will go dig into the logs and see if I can figure out which is annoyed
20:33
<alkisg>
gbaman: ps faux | nc termbin.com 9999
20:33
and: systemctl status | nc termbin.com 9999
20:34
and: systemctl --failed | nc termbin.com 9999
20:35woernie has left IRC (woernie!~werner@p57A0E783.dip0.t-ipconnect.de, Remote host closed the connection)
20:35
<gbaman>
https://termbin.com/549k
20:35
https://termbin.com/8cnc
20:36
https://termbin.com/vf42
20:36
<alkisg>
systemctl status lightdm
20:37
A normal boot shouldn't have failed services
20:37
All of those need to either be fixed, or masked
20:40
<gbaman>
Dec 29 20:23:36 ltsp73 systemd[1]: lightdm.service: Service RestartSec=100ms expired, scheduling restart.
20:40
Dec 29 20:23:36 ltsp73 systemd[1]: lightdm.service: Scheduled restart job, restart counter is at 6.
20:40
Dec 29 20:23:36 ltsp73 systemd[1]: Stopped Light Display Manager.
20:40
Dec 29 20:23:36 ltsp73 systemd[1]: lightdm.service: Start request repeated too quickly.
20:40
Dec 29 20:23:36 ltsp73 systemd[1]: lightdm.service: Failed with result 'exit-code'.
20:40
Dec 29 20:23:36 ltsp73 systemd[1]: Failed to start Light Display Manager.
20:40
Dec 29 20:23:36 ltsp73 systemd[1]: lightdm.service: Triggering OnFailure= dependencies.
20:40
<alkisg>
better use termbin
20:41
pastebin lightdm log
20:41
Or try to restart it, and then use systemctl status, to see why it fails
20:41
<gbaman>
Haha, realised that after hitting paste :)
20:41
https://termbin.com/s3e0
20:42
<alkisg>
This is a "failed to restart too quickly"
20:42
it's not the original message, it doesn't help
20:42
<gbaman>
Getting same thing with service lightdm restart
20:42
<alkisg>
Try stop, then start, then status
20:43
Finally, pastebin lightdm log
20:45
<gbaman>
https://termbin.com/0fw3
20:45
And getting same thing on status after stopping and starting
20:46
( and Xorg.0.log for good measure - https://termbin.com/iznj )
20:47
<alkisg>
console=tty1 breaks systemd's getty service (in cmdline)
20:47
I see some accountsservice warnings, maybe they're related
20:48
But gotta go for now... we may continue another day
20:48
<gbaman>
Thanks for the help, I will keep poking around. Have to keep myself busy not I have these long holidays :D
20:48
*now
20:51
And now Epoptes is fixed, I don't really need my UART console
22:01dsjii has joined IRC (dsjii!~david@047-134-241-234.res.spectrum.com)
22:03highvoltage has left IRC (highvoltage!~highvolta@ubuntu/member/highvoltage, Ping timeout: 268 seconds)
22:03highvoltage has joined IRC (highvoltage!~highvolta@ubuntu/member/highvoltage)
23:09
<gbaman>
Typical alkisg... It was a specific Pi 4 thing, apparently you need to specifically tell it to load the vc4-fkms-v3d device tree overlay for Pi 4, otherwise nothing graphical works. Still plenty else broke, but we now have a login screen and can log into the (very broken) desktop
23:18
<markit>
alkisg: I've upgraded ltsp from 1.0.1-2+t201911250838~ubuntu20.04.1 to 1.0.1-2+t201912251509~ubuntu20.04.1 available today, and broadcast WORKS!
23:21
btw, if I select only some client for broadcast, I've no visual clue then about which ones have currently broadcast going on (clients are not logged, so can't understand based on their thumbail)