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


Channel log from 24 May 2019   (all times are UTC)

02:24jeblesh has joined IRC (jeblesh!2e78d540@gateway/web/freenode/ip.46.120.213.64)
05:13woernie_ has joined IRC (woernie_!~werner@pD9E8BE6A.dip0.t-ipconnect.de)
05:15jeblesh has left IRC (jeblesh!2e78d540@gateway/web/freenode/ip.46.120.213.64, Ping timeout: 256 seconds)
05:56ricotz has joined IRC (ricotz!~ricotz@ubuntu/member/ricotz)
06:23SYS64738 has joined IRC (SYS64738!~jhonny5@159.213.93.166)
06:30kjackal has joined IRC (kjackal!~quassel@195.235.52.106)
07:01kjackal has left IRC (kjackal!~quassel@195.235.52.106, Ping timeout: 258 seconds)
07:08kjackal has joined IRC (kjackal!~quassel@195.235.52.106)
07:39kjackal has left IRC (kjackal!~quassel@195.235.52.106, Ping timeout: 259 seconds)
08:06Sleaker has left IRC (Sleaker!~quassel@2604:880:a:7::e1b, Ping timeout: 252 seconds)
08:10Sleaker has joined IRC (Sleaker!quasselcor@2604:880:a:7::e1b)
08:33SYS64738 has left IRC (SYS64738!~jhonny5@159.213.93.166, Ping timeout: 252 seconds)
08:42statler has joined IRC (statler!~Georg@gwrz3.lohn24.de)
09:41SYS64738 has joined IRC (SYS64738!~jhonny5@159.213.93.166)
11:42Faith has joined IRC (Faith!~Paty_@unaffiliated/faith)
11:47spectra has left IRC (spectra!~spectra@debian/developer/spectra, Ping timeout: 257 seconds)
11:55spectra has joined IRC (spectra!~spectra@debian/developer/spectra)
12:11spaced0ut has joined IRC (spaced0ut!~spaced0ut@unaffiliated/spaced0ut)
15:05woernie_ has left IRC (woernie_!~werner@pD9E8BE6A.dip0.t-ipconnect.de, Remote host closed the connection)
15:12spaced0ut has left IRC (spaced0ut!~spaced0ut@unaffiliated/spaced0ut, Excess Flood)
15:26kjackal has joined IRC (kjackal!~quassel@195.235.52.106)
15:41SYS64738 has left IRC (SYS64738!~jhonny5@159.213.93.166, Ping timeout: 272 seconds)
15:48kjackal has left IRC (kjackal!~quassel@195.235.52.106, Ping timeout: 268 seconds)
15:55kjackal has joined IRC (kjackal!~quassel@195.235.52.106)
16:12spaced0ut has joined IRC (spaced0ut!~spaced0ut@unaffiliated/spaced0ut)
16:45kjackal has left IRC (kjackal!~quassel@195.235.52.106, Ping timeout: 248 seconds)
17:34statler has left IRC (statler!~Georg@gwrz3.lohn24.de, Remote host closed the connection)
20:01vagrantc has joined IRC (vagrantc!~vagrant@unaffiliated/vagrantc)
20:25
<alkisg>
vagrantc: I'm thinking to change the defaults to NFS+ltsploop, you ok with that?
20:26
<vagrantc>
alkisg: i'm sort of pleasantly surprised :)
20:27
alkisg: i presume you mean for the new code?
20:27
<alkisg>
Eh :D
20:27
Sure for the new code it's already there; I meant for ltsp5
20:27
<vagrantc>
ah
20:27
<alkisg>
...for new installations, of course
20:28
<vagrantc>
it's obviously too late for buster to switch like that
20:28
<alkisg>
Sure
20:28
<vagrantc>
but there are always backports (at least if someone uploads them)
20:29
<alkisg>
So it's not too intrusive if I make it the default for new installations, via the ppa and 20.04, right?
20:29* alkisg will need to read up on conffiles...
20:29
<vagrantc>
your call, really
20:29
<alkisg>
Ty
20:30
<vagrantc>
conffile upgrades are really smooth if they haven't changed anything locally
20:30
they're kind of annoying if people have made local edits, but at least it's a predictible annoyance
20:30
<alkisg>
That's nice... there's also the matter of running ltsp-config nfs before it actually works
20:30
and apt install nfs-kernel-server
20:31
<vagrantc>
well, that should be handled in recommends/depends
20:31
at least installing it
20:31
<alkisg>
Recommends won't work with upgrades... and depends might be too intrusive...
20:31* vagrantc wonders if we should assume /etc/exports.d/ is supported
20:32
<vagrantc>
i guess we could just bump the minimum version required to avoid editing /etc/exports
20:32
alkisg: "new" recommends should work with upgrades
20:32
<alkisg>
Ah, those 2 solve the issue then
20:32
<vagrantc>
at least in recent years ... really old installations might have issues
20:33
not sure if other distros support /etc/exports.d ...
20:33
been... maybe 4 years since i abandoned NFS :)
20:34
i think i had ltsp-config nfs just edit /etc/exports as it was too complicated to ... check for redundant entries in /etc/exports and /etc/exports.d/*
20:34
though maybe that check should still happen...
20:34
alkisg: so you must really like this nfs+ltsploop ... you're even trying to "backport" it!
20:35
<alkisg>
It's sooooo much easy to have 20 VMs with it!
20:35
That nbd-server configuration is horrible
20:35
(aside from the instabilities)
20:35
I also wanted to ask you about the authentication part. I think that at the end of gsoc, I'll have ltsp ready without authentication; which means it'll work with passwd/shadow in the initramfs or fetched from local media etc (and of course for guest sessions),
20:35
...while the libpam-sshauth part will be missing; maybe sbalneav can take a look, or someone else, or me the next year etc,
20:36
<vagrantc>
i thought this would be your last GSoC?
20:36
<alkisg>
...so the question there is... work on that, which means we have a real display manager, or ... use ldm or reimplement it in python?
20:36
Next year outside of gsoc
20:37* vagrantc wonders if it would be possible to crowsource the last next-generation feature
20:37
<alkisg>
I can widely test the shadow+real dm part here in schools, as we don't mind much about the security implications there,
20:37
^ right
20:37
So I hope it would attract someone to fill that missing bit, the authentication,
20:38
<vagrantc>
authless would still be very useful in it's own right
20:38
<alkisg>
while if I go for a quick ldm merge or reimplementation , it'll again be a bit messy, we'll have to launch our screen scripts and all
20:38
<vagrantc>
i really want to avoid ldm or reimplementing it
20:38
so i would rather see an incomplete thing that makes it clear what's left
20:38
<alkisg>
Can it be uploaded in debian at that stage though? Can it replace the existing ltsp? Should it?
20:39
Maybe it should temporarily use another name, ltsp-next or something?
20:39
Or at least the old package to still be available like ltsp5...
20:40
We can clearly state, "sending shadow is insecure; use kerberos/nfs4 etc, or help in libpam", but is that enough?
20:40
<vagrantc>
we can certainly make uploads to experimental
20:41
it's in the name, after all :)
20:41
<alkisg>
And I can make it available to the ppa, for those that choose to use it. Works for me.
20:41
<vagrantc>
maybe calling the source package ltsp-next
20:42
or some ltsp-CODENAME ?
20:42
i guess ltsp-next always gives room to reimplement it yet again when we've made new mistakes to learn from
20:42
<alkisg>
OK we'll have enough time to decide the name; I wanted to focus on what the roadmap should be
20:43
I'm having a lot of ideas, many of them are worthy but they do also need time for implementation
20:43
And I wouldn't like to turn down good ideas because of lack of time
20:43
<vagrantc>
i'd be happy to upload to experimental, and you can have a ppa for the ubuntu side ... so that sounds pretty good
20:43
<alkisg>
E.g. I'll stop NN-scripts and start using hooks; it's a much more stable interface
20:43
<vagrantc>
the difference?
20:43
<alkisg>
(thought that one while jogging; it's great for new ideas!)
20:44
Suppose that epoptes needs to run something after 09-hostname and before 10-whatever
20:44
So it ships a file 09-z-epoptes, to go between those
20:44
Then we want to renumber all the scripts because we stuff more things there
20:44
...and we can't, because other packages rely on numbering
20:45
In the new method, epoptes would say "add a hook after the hostname function"
20:45
<vagrantc>
ah, so basically dependency based
20:45
after, before, etc?
20:45
<alkisg>
Right, each script is now a function, and we support before/after hooks
20:46
<vagrantc>
those dependecy trees can get surprisingly complicated fast
20:46
<alkisg>
It's not dependencies, it's hooks
20:46
<vagrantc>
calling it something different doesn't change it :P
20:46
<alkisg>
Dracut tries to get dependency/event driven, and it's too complicated,
20:46
while initramfs-tools provides clearly defined hooks, pre-init, premount, etc, and it's a lot cleaner
20:46
It's a different concept
20:47
<vagrantc>
but if you have two different hooks after X, which one runs first?
20:47
<alkisg>
Hooks is just before/after; dependency means resolutions and trees
20:47
<vagrantc>
in initramfs-tools i sometimes need something to go in-between something that isn't defined
20:47
which i have had to get fixed upstream
20:48
<alkisg>
Exactly; that's the best part
20:48
Downstream users or apps that need extra hooks, should ask them
20:48
<vagrantc>
initramfs-tools overall has been pretty flexible over it's many years
20:48
<alkisg>
Now there wasn't much cooperation even between ltsp implementation in different distros
20:49
<vagrantc>
that was both the strength and an eventual weakness
20:49
<alkisg>
By disabling 09-z-numbering there, we force people to contact us and cooperate :)
20:50
<vagrantc>
initramfs-tools implements both, though
20:50
<alkisg>
True, but we don't expect so many apps adding hooks to ltsp
20:51
<vagrantc>
always interesting to find out what people end up actually doing :)
20:51
<alkisg>
And we could easily add a function for that if we have to, example:
20:51
pre_hook configure_sound some_epoptes_function
20:52
let's say that's how epoptes adds a hook, if we need to, we can extend it to,
20:52
pre_hook configure_sound some_epoptes_function 10 (priority), or
20:52
pre_hook configure_sound some_epoptes_function 10 before=that_other_package_function (dependency)
20:52
but I hope we won't have to
20:53
I hope it'll be simpler, not more difficult to understand :D
20:53
<vagrantc>
indeed
20:55
<alkisg>
Ah, vagrantc one last thing: https://github.com/NetworkBlockDevice/nbd/issues/87#issuecomment-464342328
20:56
To use root=/dev/nbd0p1, the initramfs code of nbd needs to be changed
20:56
Wouter fixed it for later versions, but what can we do about older ones?
20:57
A simple sed call fixes it; but it's an awful hack, to sed other people's code
20:57
...should I put such code in ltsp, or just tell anyone affected to do it manuall?
20:58
Also, wouter fixed only half of the issue, the other half still doesn't work; the sed command fixes both of them
21:00
Although now if we switch away from nbd, this isn't that important anymore :D
21:04
<vagrantc>
always nice to fix things if we can even if we don't use them, but i know there's a matter of priorities
21:11
<alkisg>
No I mean, is it OK to sed /scrits/local-top/nbd to fix wouter's code while we boot?
21:11
It's it horrible practise?
21:12
But at that point there, we can't put a wrapper or hook or fix it afterwards, so sed is all we can do
21:12
Anyways.. it's late; ty for all the input; good night :)
21:14
<vagrantc>
yeah, seems like bad form in general
21:14
night!
21:19Faith has left IRC (Faith!~Paty_@unaffiliated/faith, Quit: Leaving)
21:21
<josefig>
hi guys :)
21:37
the RCFILE_01 option on lts.conf keeps working?
21:41
<vagrantc>
there were some bugs in some versions where it didn't
21:42
look on the client and see if /etc/rc.local is modified to include your change when you boot it, then it's working...
21:45
<josefig>
i had to create the directory, i added the RCFILE_01="/etc/rc.local/xscreensaver-log" should work
21:47
<vagrantc>
/etc/rc.local is a file, not a directory
21:47
or should be a file
21:51
<josefig>
yes i see my error, so the script should be included there?
21:51
<vagrantc>
RCFILE_* adds lines to /etc/rc.local
21:52
<josefig>
ahh! i see thanks
21:54
vagrantc, the rc.local file should exist right?
21:55
<vagrantc>
just experiment and find ou
21:55
out
21:55
<josefig>
yes, the image is being created :P
21:56
<vagrantc>
if your version of ltsp is more recent than july 2018, it should create the file if needed
21:56
<josefig>
yes it's more recent than that date
21:59ricotz has left IRC (ricotz!~ricotz@ubuntu/member/ricotz, Remote host closed the connection)
22:17adrianor1 has joined IRC (adrianor1!~adrianorg@177.18.102.108)
22:20adrianorg has left IRC (adrianorg!~adrianorg@177.18.175.67, Ping timeout: 245 seconds)