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


Channel log from 10 August 2016   (all times are UTC)

00:22adrianorg has joined IRC (adrianorg!~adrianorg@186.215.18.89)
00:51adrianorg has left IRC (adrianorg!~adrianorg@186.215.18.89, Quit: leaving)
01:19Freejack has left IRC (Freejack!~Freejack@unaffiliated/freejack, Read error: No route to host)
02:13Freejack has joined IRC (Freejack!~Freejack@unaffiliated/freejack)
02:59GodFather has left IRC (GodFather!~rcc@96-35-101-212.dhcp.bycy.mi.charter.com, Ping timeout: 265 seconds)
03:03Freejack has left IRC (Freejack!~Freejack@unaffiliated/freejack, Ping timeout: 250 seconds)
03:19Freejack has joined IRC (Freejack!~Freejack@unaffiliated/freejack)
05:35ricotz has joined IRC (ricotz!~ricotz@p5B2A9E7C.dip0.t-ipconnect.de)
05:35ricotz has joined IRC (ricotz!~ricotz@ubuntu/member/ricotz)
06:19mikkel has joined IRC (mikkel!~mikkel@mail.dlvs.dk)
06:31Softeisbieger has joined IRC (Softeisbieger!~Softeisbi@ip-62-143-13-166.hsi01.unitymediagroup.de)
06:32forum has joined IRC (forum!~Icedove@93-82-108-242.adsl.highway.telekom.at)
06:42forum has left IRC (forum!~Icedove@93-82-108-242.adsl.highway.telekom.at, Quit: forum)
07:27forum has joined IRC (forum!~Icedove@91-115-230-37.adsl.highway.telekom.at)
07:27forum has joined IRC (forum!~Icedove@91-115-230-37.adsl.highway.telekom.at)
07:27forum has joined IRC (forum!~Icedove@91-115-230-37.adsl.highway.telekom.at)
07:28forum has joined IRC (forum!~Icedove@91-115-230-37.adsl.highway.telekom.at)
07:30forum has joined IRC (forum!~Icedove@91-115-230-37.adsl.highway.telekom.at)
07:31forum has joined IRC (forum!~Icedove@91-115-230-37.adsl.highway.telekom.at)
07:32forum has joined IRC (forum!~Icedove@91-115-230-37.adsl.highway.telekom.at)
07:34forum has left IRC (forum!~Icedove@91-115-230-37.adsl.highway.telekom.at, Client Quit)
07:34forum has joined IRC (forum!~Icedove@91-115-230-37.adsl.highway.telekom.at)
07:36forum has joined IRC (forum!~Icedove@91-115-230-37.adsl.highway.telekom.at)
07:37forum has joined IRC (forum!~Icedove@91-115-230-37.adsl.highway.telekom.at)
07:37forum has joined IRC (forum!~Icedove@91-115-230-37.adsl.highway.telekom.at)
07:37forum has joined IRC (forum!~Icedove@91-115-230-37.adsl.highway.telekom.at)
07:38forum has joined IRC (forum!~Icedove@91-115-230-37.adsl.highway.telekom.at)
07:38forum has joined IRC (forum!~Icedove@91-115-230-37.adsl.highway.telekom.at)
07:40forum has joined IRC (forum!~Icedove@91-115-230-37.adsl.highway.telekom.at)
07:41forum has joined IRC (forum!~Icedove@91-115-230-37.adsl.highway.telekom.at)
07:52forum has left IRC (forum!~Icedove@91-115-230-37.adsl.highway.telekom.at, Remote host closed the connection)
07:52forum has joined IRC (forum!~Icedove@91-115-230-37.adsl.highway.telekom.at)
08:15alkisg has left IRC (alkisg!~alkisg@ubuntu/member/alkisg, Quit: Leaving.)
08:16julienfayad has joined IRC (julienfayad!~julienfay@37.209.252.166)
08:18Gremble has joined IRC (Gremble!~Ben@94.118.222.79)
08:18Gremble is now known as Guest19206
08:19alkisg has joined IRC (alkisg!~alkisg@ubuntu/member/alkisg)
08:21julienfayad has left IRC (julienfayad!~julienfay@37.209.252.166, Ping timeout: 250 seconds)
08:21Guest19206 has left IRC (Guest19206!~Ben@94.118.222.79, Remote host closed the connection)
08:29julienfayad has joined IRC (julienfayad!~julienfay@185.76.178.176)
08:50Grembler has joined IRC (Grembler!~Ben@cpc87179-aztw31-2-0-cust6.18-1.cable.virginm.net)
08:55julienfayad_ has joined IRC (julienfayad_!~julienfay@37.209.252.166)
08:57julienfayad has left IRC (julienfayad!~julienfay@185.76.178.176, Ping timeout: 276 seconds)
08:57julienfayad_ is now known as julienfayad
09:10ORE_ has joined IRC (ORE_!5c0a24b0@gateway/web/freenode/ip.92.10.36.176)
09:11ORE_ has left IRC (ORE_!5c0a24b0@gateway/web/freenode/ip.92.10.36.176)
09:11Russell has joined IRC (Russell!5c0a24b0@gateway/web/freenode/ip.92.10.36.176)
09:12Russell is now known as Guest19635
09:17Guest19635 has left IRC (Guest19635!5c0a24b0@gateway/web/freenode/ip.92.10.36.176, Quit: Page closed)
09:18RussellE has joined IRC (RussellE!5c0a24b0@gateway/web/freenode/ip.92.10.36.176)
09:23
<RussellE>
hi Alkis
09:25
<alkisg>
Hi RussellE
09:25
!vnc-dide
09:25
<ltsp>
vnc-dide: To share your screen with me, run this: sudo apt-get --yes install x11vnc; x11vnc -connect srv1-dide.ioa.sch.gr - this is a reverse connection, it doesn't need port forwarding etc.
09:25
<alkisg>
RussellE: do this from your server ^
09:26
<RussellE>
ok
09:31
<alkisg>
RussellE: in all the cases that you've tried so far, the client failed to boot? And, with what messages or symptoms?
09:32
<RussellE>
Yes, the RPI3 hangs at initial boot. I dont think I followed your last email instructions correctly
09:33
<alkisg>
RussellE: there are 2 basic methods, one with raspi2, and one with raspberrypi-bootloader
09:33
Can we try both of them, in that order?
09:33
<RussellE>
I got a copy of Ubuntu Mate from the Flavour Maker site and copied the /boot contents onto the SD
09:33
<alkisg>
Or do you think we should only try the second one?
09:33
<RussellE>
Yes we can try anything you suggest
09:34
<alkisg>
(12:33:43 μμ) RussellE: I got a copy of Ubuntu Mate from the Flavour Maker site and copied the /boot contents onto the SD ==> and then it booted?
09:34
The message when it hangs matters a lot, e.g. if it hangs in initramfs>, it's a completely different story compared to when it hangs with a rainbow screen before the kernel is loaded
09:35
<RussellE>
Yes it booted with the old initrd.img and vmlinuz but when I copied the current ones from the server it hangs at the rainbow screen
09:38
<alkisg>
RussellE: OK, can you insert an SD card to the server in order to prepare it?
09:38
<RussellE>
My server doesnt have an SD slot so i will copy the files from the folder onto my laptop then onto the SD
09:38
<alkisg>
Does your laptop run Ubuntu?
09:39
<RussellE>
no I am on windows
09:39
I normally use Teamviewer for remote support
09:39
if that helps
09:40
<alkisg>
And you don't have a usb to sd adapter or usb stick, right?
09:40
<RussellE>
No
09:40
<alkisg>
OK
09:40
I'll prepare the folder and then you can transfer it as you usually do
09:40
<RussellE>
But I can transfer the files easily, will only take a min
09:40
<alkisg>
Btw, do you want me to install epoptes? It helps maintaining the clients
09:40
!epoptes
09:40
<ltsp>
epoptes: Epoptes is a computer lab administration and monitoring tool. It works on Ubuntu and Debian based labs with LTSP or non-LTSP servers, thin and fat clients, standalone workstations, NX clients etc. More info: http://www.epoptes.org
09:40
<alkisg>
(it only takes 1 minute...)
09:41
<RussellE>
Yes I do but I had problems in the past installing it, was on my todo list once I had sorted this issue out
09:41
<alkisg>
OK, will do it now
09:41
We'll first try with the raspi2 approach
09:42
If that fails, we'll then try with the raspberrypi-bootloader approach
09:51
RussellE: can you copy that folder to the sd card, but before doing so, completely remove all the old files from the card?
09:52
<RussellE>
Yes I have formatted the SD again
09:52
<alkisg>
Nice, copy the ~/Documents/RaspberryPi/sd_card contents there
09:54
<RussellE>
ok I have booted the RPI3 with the updated SD contents and its hanging at the rainbow screen
09:54
<alkisg>
OK then let's try the other approach
09:54
<RussellE>
ok
09:55
<alkisg>
RussellE: your dnsmasq is not properly configured, did you manage to completely boot ltsp clients with it?
09:56
I.e. what is running is the "fake" dnsmasq that network-manager spawns, not the real one. Should I fix this?
10:05
<RussellE>
Yes I have been able to boot clients. I used the DNS fix from the guide
10:06
The alternative method from the ltsp-pnp guide to be precise
10:08
<alkisg>
OK
10:10
RussellE: can you try with this updated folder?
10:11
<RussellE>
ok will do
10:13
<alkisg>
It won't completely boot, but if it passes the rainbow screen, it's a good sign
10:13laercio has joined IRC (laercio!~laercio@177.143.30.194)
10:13
<RussellE>
I notice you ln'd, do I need to copy the files again?
10:13laercio has left IRC (laercio!~laercio@177.143.30.194)
10:14
<alkisg>
No, I didn't; I put a comment there
10:14
You don't need to copy the files again
10:15laercio has joined IRC (laercio!~laercio@177.143.30.194)
10:15
<RussellE>
success it booted :-)
10:15laercio is now known as lbssousa
10:15
<alkisg>
up to which point?
10:16
<RussellE>
current message on screen: random: nonblocking pool is initialized
10:16
<alkisg>
OK, let's try a mix now, with the current ltsp initrd...
10:17
<RussellE>
ok
10:18
So I need to copy the vmlinuz and initrd.img onto the SD?
10:18
<alkisg>
RussellE: and cmdline.txt and config.txt
10:18
4 files
10:19
<RussellE>
ok doing now
10:19
<alkisg>
Do overwrite the existing txt files with the new ones
10:22
<RussellE>
I have booted with the updated files and it is hanging at the rainbow screen
10:22vmlintu has joined IRC (vmlintu!~vmlintu@a91-153-33-91.elisa-laajakaista.fi)
10:23forum has left IRC (forum!~Icedove@91-115-230-37.adsl.highway.telekom.at, Ping timeout: 265 seconds)
10:23
<alkisg>
RussellE: can you try again? I'm now using the new kernel, but the old config.txt
10:24
RussellE: please do a complete sync, with erasing before copy
10:24
As I also updated kernel7.img etc
10:27
<RussellE>
so I have copied the sd_card folder, vmlinuz, initird.img, command.txt but not config.txt?
10:28
<alkisg>
RussellE: config.txt is there, cmdline.txt is not
10:29
There's no vmlinuz either
10:29
<RussellE>
ok I will just copy whats in the sd_card folder
10:29
<alkisg>
Yes please
10:30
<RussellE>
ok I have formatted the SD, copied the files and it is hanging at the rainbow screen
10:31
let me try again just to be sure
10:33
Alkisg: no its still hanging at the rainbow screen
10:33
<alkisg>
RussellE: OK, unfortunately that's an issue, because it seems like the stock kernel can't boot rpi3,
10:33
and the custom kernel that can boot it, lacks an initrd
10:34
Did you mention that you were able to boot it using an rpi3 kernel from a ppa?
10:34
Which one was that?
10:34
From ppisati?
10:35
<RussellE>
one sec let me find my notes
10:36
This was the .DEB i got that worked
10:36
https://launchpad.net/~p-pisati/+archive/ubuntu/embedded/+files/raspberrypi3-firmware_4.1.19-a192a05-2_armhf.deb
10:36
So yes ppisati
10:38
<alkisg>
RussellE: OK, that's the firmware, but which kernel did you use with that one?
10:39
<RussellE>
Oh I have been using the kernel generated from when I followed the guide some months back
10:40
<alkisg>
So the raspi2 kernel? That's good news then, lets try that
10:40
RussellE: hmm although I see in your mail that you tried with the new kernel and it failed?
10:41
That's what I'm afraid of, that the problem is in the ubuntu stock kernel, and nothing is wrong with the wiki page currently except this kernel bug that needs to be fixed
10:42
RussellE: so, kernel 4.2.0-1025-raspi2 boots your rpi3, but 4.4.0-1017-raspi2 and 4.4.0-1019-raspi2 can't boot it, right?
10:42
<RussellE>
I suspect you are correct because all the steps from the guide when used with the old kernel give me a working environment
10:43
Yes Kernel 4.2.0.1025 works and the others dont
10:43
<alkisg>
It probably is a regression from 4.2 to 4.4 then, and we need to report it. The bad news is that this may take weeks to be fixed.
10:43
You should use the older kernel in the mean time
10:44
<RussellE>
Ok, I agree with your analysis and I needed to confirm from someone who understood the whole thing better than me ;-)
10:44
So where do we go from here with regards reporting?
10:44
<alkisg>
Do you have an rpi2 to test with?
10:45* alkisg doesn't have an rpi3, but would like to see if it only affects rpi3 or if it also affects rpi2
10:46
<RussellE>
Yes i do, and i think i did try it at one point, but I can try again.
10:46
<alkisg>
And did it boot with the new kernel, or not?
10:47
<RussellE>
The only issue is that I am currently using it as a wifi-lan bridge in my lab which is what is giving me my RPI3 Internet at the moment so I would need to reconfigure things in order to test
10:47
<alkisg>
Maybe you can put the rpi3 in its place
10:49
<RussellE>
Yes I am running minibian on it but for some reason when i take out the SD and put it in the RPI3 eth0 stops working (i suspect it is driver related after isusing an apt-get upgrade command)
10:49
<alkisg>
https://launchpad.net/ubuntu/+source/linux-raspi2
10:49
This is where the bug should be reported
10:50
It would also help a lot if "bisection" was done, i.e. an effort to locate the exact upload when it broke
10:50
There are some kernels in that page that can be downloaded
10:51vmlintu has left IRC (vmlintu!~vmlintu@a91-153-33-91.elisa-laajakaista.fi, Ping timeout: 276 seconds)
10:51
<alkisg>
After the bug is reported there, if days pass and noone responds, it would be good to ping people at #ubuntu-kernel
10:52vmlintu has joined IRC (vmlintu!~vmlintu@a91-153-33-91.elisa-laajakaista.fi)
10:52
<RussellE>
ok. I have never reported a bug before but I guess I can learn.
10:52
<alkisg>
Here's a big list of the kernels that were uploaded: https://launchpad.net/ubuntu/xenial/+source/linux-raspi2
10:53
A big help would be to find the one that caused the breakage, the "regression"
10:53
<RussellE>
But before I report you suggest I try the various Kernels to find out which one works?
10:54
<alkisg>
It would make the bug report more easy to work with
10:54
But you can also do it later on if you want...
10:54
It depends on the time you have, the deadlines etc
10:56
<RussellE>
I am happy to test before reporting. But just to clarify, to test, I download a test kernel, open it up and extract the kernel and copy onto the SD?
10:58vmlintu has left IRC (vmlintu!~vmlintu@a91-153-33-91.elisa-laajakaista.fi, Ping timeout: 276 seconds)
10:59
<alkisg>
RussellE: while that won't make the client boot, copying vmlinuz and passing the rainbow screen would be a very good indication on if the problem exists there or not
11:00
So if you do find out that e.g. kernel 4.4.0-1010.12 gives you a rainbow, and 4.4.0-1010.11 shows some text, it will help the bug report a whole lot
11:01
RussellE: you can use the "sdcard.raspi2" folder to test with, because that's the one that uses normal names like vmlinuz instead of kernel7.img
11:02vmlintu has joined IRC (vmlintu!~vmlintu@a91-153-33-91.elisa-laajakaista.fi)
11:02
<RussellE>
understood. I am a little lost with the list and what to download once there, can you show for instance with 4.3.0-1006.6 which file I should download. Once I get it where is the Kernel located in the structure
11:03
<alkisg>
OK, watch the vnc screen
11:03GodFather has joined IRC (GodFather!~rcc@96.92.43.9)
11:05
<alkisg>
linux-image-4.4.0-1000-raspi2_4.4.0-1000.1_armhf.deb (19.4 MiB)
11:07
RussellE: ok, can you try with that one?
11:07
I.e. download with firefox, use dpkg-deb -X, then cp vmlinuz to sd_card
11:07
So now we'll be trying with the fist 4.4 kernel, this one: vmlinuz-4.4.0-1000-raspi
11:08
<RussellE>
ok
11:10
ok so using that kernel I get the rainbow screen
11:10
<alkisg>
Now let's try with a 4.2 one
11:11
Can you try it yourself, and call me if you get stuck?
11:11
<RussellE>
ok will do thanks
11:13
<alkisg>
https://launchpad.net/ubuntu/+source/linux-raspi2/4.2.0-1013.19/+build/8136687
11:14
<RussellE>
I am working from my laptop in case so you wont see much activity on the server
11:32
<lbssousa>
alkisg: do you still have plans to migrate Epoptes VCS to Git? If yes, would you prefer migrating it out to e.g. GitHub or keeping it in Launchpad (now that it has an initial Git hosting)? I'm interested in packaging latest epoptes code in a personal repo in Fedora Copr, but it only supports git repos.
11:33julienfayad has left IRC (julienfayad!~julienfay@37.209.252.166, Quit: julienfayad)
11:39
<alkisg>
lbssousa: there are some plans for that, yes, I think Phantomas has tried the launchpad git hosting as a test
11:40
But since the project is no longer funded, its development has stalled a bit as well
11:40
Let's hope funding is found within a year or so...
11:44Grembler has left IRC (Grembler!~Ben@cpc87179-aztw31-2-0-cust6.18-1.cable.virginm.net, Quit: I Leave)
11:44
<alkisg>
RussellE: need any help? did it boot with 4.2?
11:48
<RussellE>
Alkisg: I had an interesting result. When I used the files from the sd_card folder with the first kernel in the list 4.2.0.1013 (which is older than my currently working Kernel) it hung at the rainbow screen.
11:49
<alkisg>
RussellE: sometimes a mismatch between kernel and firmware is to blame
11:49
<RussellE>
So I formatted the sd card, put my working kernel and firmware files on, confirmed it booted ok, then replaced vmlinux with the older 4.2.0-1013 copy and guess what it booted
11:50
<alkisg>
Right, so that old firmware works with the 4.2 kernel, but the new firmware doesn't
11:51
What we want to report is that the current stock ubuntu kernel and firmware don't work
11:51
<RussellE>
I used the firmware from https://launchpad.net/~p-pisati/+archive/ubuntu/embedded/+files/raspberrypi3-firmware_4.1.19-a192a05-2_armhf.deb
11:51
<alkisg>
So try with the new firmware, and find the kernels which work with that
11:51
Unfortnately noone will care about a firmware from some ppa, other than ppisati himself, which does participate in the ubuntu kernel maintenance...
11:52
It's best to try with the stock firmware, everyone will care about that
11:52
<RussellE>
ok so the newest firmware is the one in the guide right?
11:52
<alkisg>
Yes
11:52
<RussellE>
ok I will use that
11:52
<alkisg>
RussellE: but, we did verify that the ppisati firmware doesn't work with 4.4, correct?
11:53
If it does work with the latest kernel, then it's significant information that only the current firmware is to blame
11:54
<RussellE>
Yes we did verify that ppisati doesnt work with 4.4
11:54
<alkisg>
OK
12:02
<RussellE>
alkisg: using the firmware from the wiki and my old kernel i get the rainbow screen
12:04
but to confirm using the firmware from ppisati and the old kernel I can boot successfully
12:04
So also the firmware is an issue?
12:41
alkisg: As a further test I was able to successfully boot the RPI3 using the Ubuntu Pi Flavour Maker firmware (https://launchpad.net/~ubuntu-pi-flavour-makers/+archive/ubuntu/ppa/+build/9426137/+files/raspberrypi-bootloader_1.20160315-1~xenial1.0_armhf.deb) and my old Kernel
12:41
<alkisg>
RussellE: there are 3 options: (1) the kernel is broken, (2) the firmware is broken, (3) the firmware doesn't match the kernel version
12:42
(03:02:42 μμ) RussellE: alkisg: using the firmware from the wiki and my old kernel i get the rainbow screen ==> I think this one is "3", but it can also be "2"
12:43
Is there any kernel that can be booted with that firmware?
12:45
If the new kernel can't be booted with any firmware, and the new firmware can't boot any kernel, it's possible that both are broken then
12:45
<RussellE>
so far no but i have only tested a few but i have tested the oldest kernel (which is older than my working copy) and the newest and both have failed with that firmware
12:46
<alkisg>
OK in that case there's no need to check any additional kernels
12:46
We can just report that they don't boot at all, period :D
12:46
What we could do is check older kernels with older firmwares
12:46
Let me find the link for the official older firmwares...
12:47
<RussellE>
ok
12:47
<alkisg>
Main page: https://launchpad.net/ubuntu/+source/linux-firmware-raspi2
12:48
...they didn't have many older releases there... https://launchpad.net/ubuntu/+source/linux-firmware-raspi2/+publishinghistory
12:49
...so no, not much to test there
12:49
So maybe it's enough testing; do file the bug report and let's see what their answers are
12:51
...those are its older versions, with the older "raspberrypi2-firmware" name: https://launchpad.net/ubuntu/+source/raspberrypi2-firmware/+publishinghistory
12:52
RussellE: can you try with this one? https://launchpad.net/ubuntu/+source/raspberrypi2-firmware/4.1.15-b70b451-0ubuntu1/+build/9001776
12:54
<RussellE>
will do now
12:55
<alkisg>
Try with your "working" kernel first, if that fails, no need to bother
12:55
Just file the bug report then
12:55
<RussellE>
ok
12:56
should i try with my RPI2 before reporting the bug?
13:00adrianorg has joined IRC (adrianorg!~adrianorg@189.58.230.109.dynamic.adsl.gvt.net.br)
13:06
<RussellE>
alkisg: Looking at that file it looks identical to the one in the wiki
13:08
alkisg: I am getting the same result, rainbow screen. i guess its time to report the bug
13:49rrn has joined IRC (rrn!~rrn@unaffiliated/rrn)
13:51
<rrn>
Hi, is it possible to realize all the functionality of an existing workstation (webcam, scanner, keyboard, mouse, multiple high-resolution monitors, etc.) with LTSP on appropriate hardware?
13:51
From what I've read on the documentation, my impression is that LTSP offers limited functionality.
13:54ben_roose has joined IRC (ben_roose!~roose@roose.cs.wichita.edu)
14:02
<||cw>
rrn: with "fat client" mode, yes
14:02
<rrn>
||cw: excellent :)
14:02
<||cw>
it's still PXE boot and diskless, but everything runs locally, so all that works the same as a normal install
14:03
<rrn>
||cw: Oh, so it would be merely diskless? That's not interesting.
14:03
<||cw>
ideally you have a gigabit LAN for this, but you can have a smaller server
14:03
rrn: how so? it's still using the server's OS image
14:04
<rrn>
||cw: Yes, I'll have dual gigabit links for each server and workstation to a high-end managed switch.
14:04
<||cw>
the only drawback is that you can't just power off the client and resume your session later
14:04
<rrn>
But computation is performed locally, not on the server.
14:04
<||cw>
right
14:05
<RussellE>
.
14:05
<rrn>
My goal is to imitate Plan 9 by having a separate file server, a diskless computation server, and diskless workstations.
14:05
<||cw>
but any pentium dual core better will do that just fine
14:05
<Hyperbyte>
rrn, if you use thin clients, then all hardware has to be connected to the server, or LTSP has to support the hardware remotely via the thin clients. With USB sticks, printers, etc, this is all implemented, because those are common use cases. Two screens will also work with LTSP, but I doubt it can transmit USB webcams over the network.
14:06
<||cw>
but even with thin clients, you can still have localapps for the things that cna't be easily tunneled back to the server
14:06
<rrn>
Hyperbyte: I see, that's what I thought.
14:06
<Hyperbyte>
rrn, with fat clients, you get regular Linux experience, but without harddisk, so anything that would normally work, will work.
14:06
<||cw>
storage and printers can, I'm not sure any work has been done on webcams and scanners, though SANE does have a network mode that might do it
14:07
but IMO, network scanners are the way to go, either standalone or MFP
14:08
but webcams are beastly amount of data, which would push limits even on a gigabit lan with more than 1 going
14:08
<RussellE>
alkisg: Final updates. I have confirmed that the Kernel is broken on both RPI2 and RPI3 from Xenial (4.3.0-1006.6) onwards. The stock firmware works on RPI2 but is broken on RPI3. I will now raise a bug alert
14:08
<||cw>
so you really do want to process them locally
14:08
<Hyperbyte>
rrn, also, there is no such thing as a "diskless computation server". With a thin client, all software is started on the server. With a fat client, it's all started on the client. Server only does dhcp for PXE boot, NBD for providing the image, and SSH for logging in. No more.
14:09
<rrn>
||cw: I'm simply trying to centralize computing as much as possible---I really love this concept.
14:09
<Hyperbyte>
rrn, technically, if your router does DHCP, your NAS supports TFTP and NBD and you have another server that clients can SSH to, you don't even need an LTSP server.
14:09
<rrn>
Hyperbyte: I mean, the computation server (a.k.a. thin client server) is itself diskless.
14:10
<Hyperbyte>
rrn, for fat clients, that is. It wouldn't surprise me if you can use a raspberry pi as LTSP fat client server. They have almost no load.
14:10
<||cw>
rrn: so where does it boot from?
14:10
<rrn>
||cw: A separate file server.
14:10
<||cw>
how
14:10
<rrn>
NFS
14:10
<||cw>
basically as a ltsp fat client?
14:10
<rrn>
and the file server will be connected to a SAN
14:11
<||cw>
will you have a 10G link between that server and the file server?
14:11
or FC?
14:11
<Hyperbyte>
rrn, are you trying to centralize the software or hardware as well?
14:11
<rrn>
Hyperbyte: I wish to centralize as much as possible---I'm really interested in Plan 9's philosophy.
14:11
<||cw>
I guess NFS means 10G, if 1G you're going to have bandwidth issues
14:12
<rrn>
||cw: Yes, I'll have >= 2 10 gigabit links.
14:12
<||cw>
bell labs plan 9? that's distributed
14:13
rrn: well, NFS is only going to use 1 link so...
14:13
<Hyperbyte>
rrn, thing is - if you centralize hardware, this is going to be more expensive on your infrastructure and not make much difference. I've done it for one my clients, because they have a proper gigabit network, not too many clients and don't want computers making noise in their offices. But even for 20 workstations, it requires a beast of a server if you want performance.
14:14
They have a server with Dual Opteron processors, total 16 cores at 3.2ghz, with 64 GB ram, and sometimes still complain it's slow, if people are doing resource intensive things.
14:15
rrn, also, things like playing video fullscreen, will not be possible on HD screens, because gigabit network simply can't handle a raw HD videostream.
14:15
<rrn>
Hyperbyte: This is a home network (just a microwave RF lab and home offices) so there will be relatively few clients.
14:15
<Hyperbyte>
rrn, if you're willing to make those investments and concessions, then yeah, sure - centralize hardware. Otherwise, don't do it.
14:16
rrn, regardless... calculate how large a bitmap image is at 1920x1080, 32bit and then see how many frames/second you can dump over a gigabit link.
14:16
<rrn>
Hmm I see
14:17
10 gigabit costs nothing *as long as* the distance is short
14:17
<Hyperbyte>
rrn, LTSP thin clients are great for limited, well-defined business use (browsing web to a degree, e-mail, crm's, office, etc). For personal use, I wouldn't go that way.
14:17
<||cw>
I think ltsp has some tools to get where you want, but you're ging to have to combine some other tools in too to reach anything like plan9. but fat clients would be the closes model to a distributed system without rewriting the whole OS
14:19
I don't think plan9 imagined things like 1080 webcams. but if it did, you'd have some local processing on the device the webcam is plugged into in order to optimize it for network consumption.
14:19
like an IP cam does.
14:19
which if you want to go full plan9, you'd use IP cams, not usb webcams
14:20
<rrn>
||cw: Oh right, I forgot about IP cams!
14:21
Hmm too bad that distributed, centralized computing is rare---otherwise we'd have optimized hardware and faster networking at lower costs.
14:22
<alkisg>
!ltsp-pnp | echo rrn:
14:22
<ltsp>
rrn: ltsp-pnp: ltsp-pnp is an alternative (upstream) method to maintain LTSP installations for thin and fat clients that doesn't involve chroots: https://help.ubuntu.com/community/UbuntuLTSP/ltsp-pnp
14:22
<alkisg>
..that's the fastest/easiest way to test ltsp thin and fat clients
14:22
<||cw>
I wouldn't be surprised if there's a daemon you could install into your thin image that would take the usb video stream and make it a compressed network stream. in fact I've be surprised if there isn't
14:23
rrn: distributed and centralized are conflicting terms
14:23
<rrn>
||cw: Not in the semantics that I use.
14:23
<alkisg>
RussellE: nice! please send me the bug link when you do so.
14:24
<rrn>
||cw: ``centralized'' meaning that general classes of nontrivial operations (disk storage and processing, computation) are centralized to dedicated servers with hardware and software optimized for the purpose
14:25
||cw: ``distributed'' meaning that subsystems, ordinarily combined into a general-purpose PC or workstation, are separated geographically over a network
14:25GodFather has left IRC (GodFather!~rcc@96.92.43.9, Ping timeout: 258 seconds)
14:26
<||cw>
I'm not sure centralized is the best word for that.
14:26
even with a thin client you have some processing on the client
14:27
<rrn>
oh of course, processing for the purpose of interfacing with the user
14:27
<||cw>
storage and processing are just nodes in the distributed model
14:28
<rrn>
Ok, scanners are no problem thanks to saned.
14:29
What about surround audio, GPU acceleration, and removable storage devices (mass storage, PTP, etc.)?
14:29
<||cw>
alkisg: anyone tried webcams on thin clients that you know of?
14:30GodFather has joined IRC (GodFather!~rcc@75-145-237-204-Michigan.hfc.comcastbusiness.net)
14:30mikkel has left IRC (mikkel!~mikkel@mail.dlvs.dk, Quit: Leaving)
14:31
<||cw>
rrn: ltspfs for storage, things that can use xv can do some gpu offloading, but it's limited until you switch up to PCoIP or some similar graphics offloading
14:32julienfayad has joined IRC (julienfayad!~julienfay@37.209.252.166)
14:32
<||cw>
audio is low bandwidth and works well
14:33
<rrn>
||cw, Hyperbyte: Ok, excellent, I'll proceed with actual experimentation of LTSP!
14:51GodFather has left IRC (GodFather!~rcc@75-145-237-204-Michigan.hfc.comcastbusiness.net, Ping timeout: 258 seconds)
15:04
<alkisg>
||cw: they work fine with localapps, and they don't work by default on normal, remote apps
15:04
I think someone tried with usb over ip, but I don't know if he succeeded
15:04
...usb webcams, I assume, not ip webcams
15:04
<||cw>
right
15:05* alkisg didn't see all the chat, why is rrn considering thin clients instead of fat ones?
15:05
<alkisg>
Like vagrantc says, "they're so 90s..." :D
15:08GodFather has joined IRC (GodFather!~rcc@75-145-237-204-Michigan.hfc.comcastbusiness.net)
15:08
<rrn>
alkisg: Fashion trends are meaningless; I want a highly-distributed network with centralized operations.
15:09
<alkisg>
rrn: no they aren't; if everyone supports one thing, and you're looking for another, you're in for a lot of work :)
15:09
LTSP has been used for distributed services
15:09
And by nature it's centralized
15:11
<rrn>
alkisg: I don't use mainstream equipment, and if something isn't available, I custom-build it or obtain it from academia or R&D (e.g., I'm in the process of building an atomic clock ensemble fusor b/c there's only 1 Russian company in the entire world that manufactures such a piece of equipment).
15:13
<alkisg>
I would love to have 1.000.000 years in order to be able to build my own OS... but unfortunately I don't
15:14
Anyway, LTSP has been used to create clusters for distributed computing, so we know it works there
15:14
<rrn>
webcams should be no problem, as long as I can get clients (e.g., Skype) to reach them
15:16
Yeah, I really abhor a lot of the technology that's out there---I consider it to be designed and engineered incorrectly.
15:17
I generally like space, military, and academic design and engineering philosophies though.
15:24RussellE has left IRC (RussellE!5c0a24b0@gateway/web/freenode/ip.92.10.36.176, Ping timeout: 250 seconds)
15:24GodFather has left IRC (GodFather!~rcc@75-145-237-204-Michigan.hfc.comcastbusiness.net, Quit: Ex-Chat)
15:25GodFather has joined IRC (GodFather!~rcc@75-145-237-204-Michigan.hfc.comcastbusiness.net)
15:26kjackal_ has joined IRC (kjackal_!~quassel@2a02:582:ccb:6a00:6081:e8d2:35cb:6965)
15:29kjackal has left IRC (kjackal!~quassel@2a02:582:ccb:6a00:54bd:509b:9561:e2d1, Ping timeout: 260 seconds)
15:39bennabiy has left IRC (bennabiy!~bennabiy@unaffiliated/bennabiy, Remote host closed the connection)
15:41sutula has left IRC (sutula!~sutula@207-118-159-111.dyn.centurytel.net, Ping timeout: 265 seconds)
15:42sutula has joined IRC (sutula!~sutula@207-118-159-111.dyn.centurytel.net)
15:45GodFather has left IRC (GodFather!~rcc@75-145-237-204-Michigan.hfc.comcastbusiness.net, Ping timeout: 258 seconds)
15:49GodFather has joined IRC (GodFather!~rcc@75-145-237-204-Michigan.hfc.comcastbusiness.net)
16:12GodFather has left IRC (GodFather!~rcc@75-145-237-204-Michigan.hfc.comcastbusiness.net, Ping timeout: 258 seconds)
16:32julienfayad has left IRC (julienfayad!~julienfay@37.209.252.166, Quit: julienfayad)
16:36julienfayad has joined IRC (julienfayad!~julienfay@37.209.252.166)
16:41Statler has joined IRC (Statler!~Georg@p5DD6E2C1.dip0.t-ipconnect.de)
16:56boj has joined IRC (boj!3d0c2373@gateway/web/freenode/ip.61.12.35.115)
16:56
<boj>
Hi
16:57
Can someone guide me fat or thin clients suits my environment ?
16:57
I have 2GB RAM with dual core processor for all my clients
16:58
And the server is I5 with 16 GB RAM
17:01Softeisbieger has left IRC (Softeisbieger!~Softeisbi@ip-62-143-13-166.hsi01.unitymediagroup.de, Remote host closed the connection)
17:15
<julienfayad>
hi boj, depends on the use you’ll make out of your clients
17:15
but with this config. I would say go with FAT
17:16
I actually think LTSP will boot as FAT by default since it’s over 400 or 800 MB of ram which is set as a trigger to boot as FAT client
17:16alkisg_web has joined IRC (alkisg_web!6db20f55@gateway/web/freenode/ip.109.178.15.85)
17:23
<boj>
Can you pls share me a good document for configuring and understanding FAT client
17:24
Because i am confused by seeing this configuration https://help.ubuntu.com/community/UbuntuLTSP/FatClients
17:25
I am actually looking for a document like this "https://www.google.co.in/url?sa=t&rct=j&q=&esrc=s&source=web&cd=3&cad=rja&uact=8&ved=0ahUKEwiEoPDurrfOAhWKp48KHRQQCOwQFggmMAI&url=https%3A%2F%2Fwww.thefanclub.co.za%2Fhow-to%2Fhow-create-ubuntu-1104-x64-ltsp-server-32bit-thin-clients&usg=AFQjCNHaaV4t0JfCQRO9tjZQttHax_qOCA&sig2=v1hVodh71JNn8rnxBx24WA"
17:25
<julienfayad>
well I used ltsp-pnp
17:25Statler has left IRC (Statler!~Georg@p5DD6E2C1.dip0.t-ipconnect.de, Quit: Leaving)
17:25
<julienfayad>
so I don’t have a chroot
17:25
and didn’t need to do any extra configuration (I use FAT clients)
17:26
<boj>
You have that document ?
17:29
<julienfayad>
here you go boj: https://help.ubuntu.com/community/UbuntuLTSP/ltsp-pnp
17:33
<boj>
julienfayad:Thank you..Can we use ubuntu 16.0 ?
17:33alkisg_web has left IRC (alkisg_web!6db20f55@gateway/web/freenode/ip.109.178.15.85, Quit: Page closed)
17:35
<julienfayad>
you are welcome boj
17:36
I think there is no such version (16.0) it’s 16.04
17:36
if this is what you mean yeah that’s the version I’m using
17:37
<boj>
julienfayad :Great and thank you..Lemme try and ping the forum if i need any assistance
17:38
<julienfayad>
sure, good luck
17:38boj has left IRC (boj!3d0c2373@gateway/web/freenode/ip.61.12.35.115, Quit: Page closed)
18:17kjackal has joined IRC (kjackal!~quassel@2a02:582:ccb:a00:6081:e8d2:35cb:6965)
18:17kjackal_ has left IRC (kjackal_!~quassel@2a02:582:ccb:6a00:6081:e8d2:35cb:6965, Ping timeout: 260 seconds)
18:32GodFather has joined IRC (GodFather!~rcc@50.245.123.216)
19:44
<julienfayad>
Hi there, alkisg sorry yesterday with my manipulation I managed to update kvno for one of my principal which led to a total inability to mount with kerberos
19:44
I fixed the pb with the help of the guys at #kerberos
19:48
<alkisg>
julienfayad: any issues left? is it working fine in both the ltsp server and the clients now?
19:49
<julienfayad>
actually I’m back to the initial issue
19:49
mounting is working on all stations BUT on the ltsp clients the mapping is not correct
19:51
<alkisg>
How is the mounting done? Via an FSTAB_x line in lts.conf?
19:51
<julienfayad>
nop not now
19:51keyboardKat has joined IRC (keyboardKat!ad0bffe9@gateway/web/freenode/ip.173.11.255.233)
19:51
<julienfayad>
it’s done with mount -t nfs -o sec=krb5 {share} {mountpoint} command
19:52
<alkisg>
https://wiki.debian.org/NFS/Kerberos
19:53
Each host should have a copy of its own key inside /etc/krb5.keytab. For each host, locally run kadmin -p adminuser/admin (adminuser/admin is an admin principal) with the commands:
19:53
...
19:53
Is this done for all ltsp clients?
19:53
<keyboardKat>
Hello, I'm attempting to setup PiNet for the first time and it's going great except for 1 hickup: Trying to ssh into it doesn't work (using root as suggested on website)
19:54
<julienfayad>
acutally nop, the keytab contains the principal of the ltsp server now
19:55
<alkisg>
julienfayad: then I imagine that kerberos doesn't work for ltsp clients, so that explains the anonymous user
19:55
!raspberrypi | echo keyboardKat:
19:55
<ltsp>
keyboardKat: raspberrypi: (#1) Ubuntu/LTSP on Pi 2: https://help.ubuntu.com/community/UbuntuLTSP/RaspberryPi, or (#2) Debian/LTSP (with raspbian chroot) on Pi: http://cascadia.debian.net/trenza/Documentation/raspberrypi-ltsp-howto/, or (#3) unofficial Ubuntu/LTSP (with raspbian chroot) on Pi: http://pinet.org.uk/
19:55
<julienfayad>
I’m not sure
19:55
because the mount actually works
19:55
<alkisg>
keyboardKat: #3 is not supported here, but only with a mail to its author
19:56
<julienfayad>
and if I chmod the share with 777 I can get access to the mount on the ltsp client
19:56
<keyboardKat>
Okay I'll give #2 a shot
19:56
<alkisg>
If you have pinet, you have an ubuntu server, don't you? why not #1 then?
19:56
old pis?
19:57
julienfayad: are you sure that kerberos is required for the mount command?
19:57
<keyboardKat>
yes I have a ubuntu vm setup, hmm I'll look at #1 as well since I have the pinet generally working
19:58
<julienfayad>
kerberos is not required and without kerberos it works well but I want to enable kerberos on some shares that needs extra security
19:59
<alkisg>
(10:55:32 μμ) julienfayad: because the mount actually works ==> I was replying to that part, meaning that if the problem is in kerberos, the mount command might work, but not the authentication... dunno
20:01
<keyboardKat>
Both of those links (#1 and #2) explain how to setup the LTSP environment. I kind of already have that with PiNet. The only missing piece is being able to SSH into it (I enabled it as specified here: http://pinet.org.uk/articles/advanced/ssh-information.html).
20:02GodFather has left IRC (GodFather!~rcc@50.245.123.216, Ping timeout: 258 seconds)
20:02
<julienfayad>
actually when the auth on kerberos doesn’t work it usually issue a permission denied at mount
20:03
<keyboardKat>
I'll email the author of pinet for some pointers
20:03
<alkisg>
keyboardKat: they explain how to build a chroot with the stock ltsp procedure, without relying on other scripts that we don't know and we can't support
20:03
<julienfayad>
alkisg here is one question from the #kerberos room if you mount with kerberos, the identity it sees is the kerberos principal. I don't know what the default mapping from principal name to uid is (I would have expected it to try the first component as a user name and use getpwnam(), but apparently this is not happening on the ltsp hosts?)
20:05
<alkisg>
julienfayad: I haven't used kerberos... is there an ltsp part in this question?
20:05DomiX has joined IRC (DomiX!~mod@nly93-h03-87-88-87-168.dsl.sta.abo.bbox.fr)
20:05DomiX has left IRC (DomiX!~mod@nly93-h03-87-88-87-168.dsl.sta.abo.bbox.fr)
20:05
<julienfayad>
hehe I’m not sure
20:14
my doubts go toward a service that handles the mapping which isn’t running properly on ltsp clients…
20:14
but I have no idea which one is doint this work
20:14
all I can find is about idmapd
20:14
but this isn’t running on neither the ltsp server nor the clients (yet on the ltsp server mounting with kerberos works like a charm)
20:17lbssousa has left IRC (lbssousa!~laercio@177.143.30.194, Quit: lbssousa)
20:23
<alkisg>
julienfayad: try to put this line in lts.conf: FSTAB_0="# dummy entry to disable deletion of nfs services"
20:23
This will prevent the ltsp clients from auto-removing nfs related services
20:23
<julienfayad>
ohh I see what you are trying to do
20:23vmlintu has left IRC (vmlintu!~vmlintu@a91-153-33-91.elisa-laajakaista.fi, Ping timeout: 265 seconds)
20:23
<julienfayad>
but I have on fstab line that does an NFS mount on a non kerberized share
20:24
<alkisg>
But I imagine that you won't be able to make everything work correctly until you declare all ltsp clients to kerberos
20:24
Then the nfs services are not deleted (and you don't need that FSTAB line)
20:24
<julienfayad>
well I just temporary cleaned the keytab file on that ltsp client
20:24
but it didn’t change much
20:24
<alkisg>
Did you add the client to kerberos?
20:24
<julienfayad>
I’ll try to reboot it and redo the cleaning jusr for the sake of it
20:25
<alkisg>
Clearing is enough? Doesn't the client need to be added?
20:25
<julienfayad>
by clearing I mean:
20:25
<alkisg>
addpriv -randkey nfs/hostnamename@REALM ktadd nfs/clientname@REALM
20:25
<julienfayad>
1. remove the wrong keytab
20:25
2. Add the client as a host principal in the KDC
20:25
3. get a new keytab with the new principal
20:26
4. make sure host is FQDN and IP are correctly binded in between the KDC and the ltps client
20:32
well even after a reboot this doesn’t work
20:33
and people at #kerberos seem to be a bit lost by the fact that it’s an ltsp client
20:38
<alkisg>
When you're starting to clean up and reconfigure things, I don't know why it would matter that it is an ltsp client
20:38
It's just a live booted machine, it doesn't matter if you used usb, live cd, or the network... you should be able to dynamically reconfigure things
20:39
<julienfayad>
well yeah I am able to reconfigure things
20:39
but this isn’t changin much
20:39
still same behavior, able to mount but wrong mapping
20:39
and since the mapping works well when I mount a share that isn’t kerberized
20:40
this means that my uid/gid mapping between server and client are all correct (which makes sense given I use ldap)
20:40
but when it’s through kerberos that fails
20:40
<alkisg>
It sounds like misconfigured kerberos to me
20:40
(to the client)
20:41
But I haven't used kerberos, so I can't comment much there..
20:41
<julienfayad>
as if kerberos is trying to use a service that it can’t find and thus maps uid to nobody and gid to 4294967294
20:41
well from my little experience with kerberos
20:41
when something is misconfigured
20:41
the whol thing fails
20:41
so in the case of an nfs mount
20:41
the whole mount would fail
20:41
not just the mapping
20:42
but nobody at the kerberos room seems to know how is the mapping happening when idmapd isn’t used
20:44
my basic understanding is that since uid and gid are the same on the server and the client I shouldn’t need idmapd
20:44
that’s why it works out of the box on the ltsp server
20:44
but why it doesn’t work on the ltsp clients is what confuses me
20:45
I’m making a last tentative, if that doesn’t work I’ll get back at it tomorrow (since #kerberos went suddenly silent)
20:47
by the way, off the kerberos topic, do you have an idea of what could cause this undesired behavior: I created two folders /media/shared_all and /media/shared_marketing in the ltps server
20:47
I rebuilt the image
20:47
the /media/shared_all shows but the /media/shared_makreting doesn’t
20:47
any idea why ?
20:50
OMG I made it work!
20:50
and indeed your intuition was correct
20:53
rechecking all configs on the client and make sure all is correct after a fresh reboot and before the first mount tentative will make it work
20:53
if you miss one detail
20:53
issue a mount with a wrong value
20:53
even if you fix it and remount that won’t make it (a full reboot is necessary)
21:07
<alkisg>
Nice :)
21:07
<julienfayad>
yeah partially nice
21:07
the mapping shows correctly
21:07
<alkisg>
The /media folder is not included, check /etc/ltsp/ltsp-update-image.excludes
21:08
<julienfayad>
but I’m getting a permission denied when trying to access the share with a user that has belongs to the correct gid
21:09
I’ll check it but what is weird is one folder got picked up but not the other one...
21:10* alkisg waves, 'night all...
21:20julienfayad has left IRC (julienfayad!~julienfay@37.209.252.166, Quit: julienfayad)
22:13vagrantc has joined IRC (vagrantc!~vagrant@unaffiliated/vagrantc)
22:14ricotz has left IRC (ricotz!~ricotz@ubuntu/member/ricotz, Quit: Leaving)
22:45ben_roose has left IRC (ben_roose!~roose@roose.cs.wichita.edu, Remote host closed the connection)
23:05GodFather has joined IRC (GodFather!~rcc@96-35-101-212.dhcp.bycy.mi.charter.com)
23:15fnurl has left IRC (fnurl!~paul@36-229-138-234.dynamic-ip.hinet.net, Read error: Connection reset by peer)
23:32vagrantc has left IRC (vagrantc!~vagrant@unaffiliated/vagrantc, Quit: leaving)