00:30 | Freejack has left IRC (Freejack!~quassel@unaffiliated/freejack, Ping timeout: 240 seconds) | |
00:36 | Freejack has joined IRC (Freejack!~quassel@unaffiliated/freejack) | |
02:06 | lucascastro has joined IRC (lucascastro!~lucas@186.227.186.18) | |
02:29 | gehidore is now known as man | |
02:57 | man is now known as gehidore | |
03:00 | adrianor1 has joined IRC (adrianor1!~adrianorg@177.134.60.77) | |
03:04 | adrianorg has left IRC (adrianorg!~adrianorg@189.58.230.128.dynamic.adsl.gvt.net.br, Ping timeout: 268 seconds) | |
03:12 | adrianor1 has left IRC (adrianor1!~adrianorg@177.134.60.77, Ping timeout: 240 seconds) | |
03:30 | adrianorg has joined IRC (adrianorg!~adrianorg@187.58.155.6) | |
03:31 | gehidore is now known as man | |
03:32 | man is now known as gehidore | |
05:13 | Statler has joined IRC (Statler!~Georg@p579FE75F.dip0.t-ipconnect.de) | |
05:39 | jgee has left IRC (jgee!~jgee@200.118.140.142, Ping timeout: 245 seconds) | |
06:15 | ricotz has joined IRC (ricotz!~ricotz@p5B2A9583.dip0.t-ipconnect.de) | |
06:15 | ricotz has joined IRC (ricotz!~ricotz@ubuntu/member/ricotz) | |
06:23 | mikkel has joined IRC (mikkel!~mikkel@mail.dlvs.dk) | |
06:29 | lucas_ has joined IRC (lucas_!~lucas@186.227.186.18) | |
06:30 | lucascastro has left IRC (lucascastro!~lucas@186.227.186.18, Read error: Connection reset by peer) | |
06:39 | SYS64738 has joined IRC (SYS64738!~SYS64738@159.213.93.166) | |
06:43 | Statler has left IRC (Statler!~Georg@p579FE75F.dip0.t-ipconnect.de, Remote host closed the connection) | |
07:43 | adrianor1 has joined IRC (adrianor1!~adrianorg@177.18.98.140) | |
07:45 | adrianorg has left IRC (adrianorg!~adrianorg@187.58.155.6, Ping timeout: 240 seconds) | |
07:48 | adrianor1 has left IRC (adrianor1!~adrianorg@177.18.98.140, Ping timeout: 255 seconds) | |
07:53 | forum has joined IRC (forum!~Thunderbi@212-197-177-100.adsl.highway.telekom.at) | |
08:06 | forum has left IRC (forum!~Thunderbi@212-197-177-100.adsl.highway.telekom.at, Quit: forum) | |
08:13 | adrianorg has joined IRC (adrianorg!~adrianorg@177.204.157.203.dynamic.adsl.gvt.net.br) | |
08:22 | Statler has joined IRC (Statler!~Georg@mail.lohn24.de) | |
08:24 | Statler has joined IRC (Statler!~Georg@mail.lohn24.de) | |
08:36 | gdi2k has left IRC (gdi2k!~gdi2k@49.151.73.242, Ping timeout: 255 seconds) | |
08:49 | adrianor1 has joined IRC (adrianor1!~adrianorg@177.132.217.187) | |
08:52 | adrianorg has left IRC (adrianorg!~adrianorg@177.204.157.203.dynamic.adsl.gvt.net.br, Ping timeout: 240 seconds) | |
08:56 | adrianor1 has left IRC (adrianor1!~adrianorg@177.132.217.187, Ping timeout: 260 seconds) | |
09:19 | adrianorg has joined IRC (adrianorg!~adrianorg@186.213.153.156) | |
09:24 | trendzetter has joined IRC (trendzetter!51531c56@gateway/web/freenode/ip.81.83.28.86) | |
09:25 | <trendzetter> hello everyone!
| |
09:26 | I am setting up an webiosk server for taking web based exams
| |
09:28 | I am updating the image. Lets hope it boots now
| |
09:30 | <alkisg> Hi trendzetter
| |
09:30 | Is there any question there?
| |
09:30 | <elias_a> alkisg: Come on - he is sharing his enthusiasm. :)
| |
09:30 | <alkisg> Yey!
| |
09:31 | !cookie
| |
09:31 | <ltsp> I do not know about 'cookie', but I do know about these similar topics: 'compiz'
| |
09:31 | <alkisg> Meh we don't have cookies in #ltsp...
| |
09:31 | <elias_a> Only whole grain bread... :)
| |
09:41 | <trendzetter> Yes I am sharing my enthusiasm
| |
09:45 | ||cw has left IRC (||cw!~chrisw@unaffiliated/cw/x-1182934, Ping timeout: 240 seconds) | |
09:58 | ||cw has joined IRC (||cw!~chrisw@unaffiliated/cw/x-1182934) | |
10:01 | markus_e92 has left IRC (markus_e92!~markus_e9@91.115.155.107, Ping timeout: 240 seconds) | |
10:05 | markus_e92 has joined IRC (markus_e92!~markus_e9@91-115-19-99.adsl.highway.telekom.at) | |
10:13 | <trendzetter> I'm back and it's booting the generated image
| |
10:13 | but there are some issues
| |
10:13 | the first is that X is not starting
| |
10:14 | <alkisg> Which distro/version are you using?
| |
10:16 | adrianor1 has joined IRC (adrianor1!~adrianorg@177.156.60.151) | |
10:17 | <trendzetter> ubuntu. I will try to disable X_VIRTUAL_AUTO=false and try again
| |
10:17 | <alkisg> And version?
| |
10:19 | adrianorg has left IRC (adrianorg!~adrianorg@186.213.153.156, Ping timeout: 252 seconds) | |
10:19 | * alkisg gives trendzetter enough time to decide if he wants help or not... :) | |
10:26 | <trendzetter> there is some error saying "switching vt failed". I will probably have to update the image after adding a user to do some debugging
| |
10:27 | <alkisg> OK, if you ever need any help, ping me
| |
10:27 | * alkisg waves until monday... | |
10:28 | <trendzetter> k, thanks
| |
11:14 | adrianorg has joined IRC (adrianorg!~adrianorg@186.215.17.139) | |
11:17 | adrianor1 has left IRC (adrianor1!~adrianorg@177.156.60.151, Ping timeout: 260 seconds) | |
11:21 | GodFather_ has joined IRC (GodFather_!~rcc@47.33.250.142) | |
11:21 | GodFather has left IRC (GodFather!~rcc@47.33.250.142, Read error: No route to host) | |
11:21 | GodFather_ has left IRC (GodFather_!~rcc@47.33.250.142, Remote host closed the connection) | |
11:22 | GodFather has joined IRC (GodFather!~rcc@47.33.250.142) | |
11:25 | adrianorg has left IRC (adrianorg!~adrianorg@186.215.17.139, Ping timeout: 240 seconds) | |
11:44 | adrianorg has joined IRC (adrianorg!~adrianorg@177.132.221.242) | |
11:47 | Michel_ has joined IRC (Michel_!2ef43f32@gateway/web/freenode/ip.46.244.63.50) | |
11:47 | <Michel_> Hello everyone
| |
11:47 | I have a question about booting ltsp
| |
11:48 | When i start a thin client in pxe boot, it shows a lot of log about the boot
| |
11:48 | is it possible to show a picture instead of the log?
| |
11:49 | epoptes_user9 has joined IRC (epoptes_user9!be496121@gateway/web/freenode/ip.190.73.97.33) | |
11:49 | <Michel_> And how do i configure this in LTSP server 5.5.7 on LinuxMint 17.1
| |
11:51 | Can someone help me?
| |
11:52 | <trendzetter> hey michel
| |
11:52 | <epoptes_user9> Hi, I'm installing epopte on debian 7 .. but when I run the epotes-client -c command I get the following error .. error: 140770FC: SSL routines: SSL23_GET_SERVER_HELLO: unknown protocol: s23_clnt.c: 782: Epoptes-client ERROR: Failed to fetch certificate from Can somebody help me
| |
11:52 | <trendzetter> I think the experts will be back on monday or thusday
| |
11:56 | trendzetter_ has joined IRC (trendzetter_!51531c56@gateway/web/freenode/ip.81.83.28.86) | |
11:57 | trendzetter has left IRC (trendzetter!51531c56@gateway/web/freenode/ip.81.83.28.86, Ping timeout: 260 seconds) | |
12:00 | <trendzetter_> I am still stuck on "switching vt failed" when X should start. I created a user in the image to check the log as suggested in the error but the folder (/tmp/.kiosk/.local/..... with the log does not exist - I checked for hidden files)
| |
12:02 | lucas_ has left IRC (lucas_!~lucas@186.227.186.18, Ping timeout: 240 seconds) | |
12:07 | trendzetter_ has left IRC (trendzetter_!51531c56@gateway/web/freenode/ip.81.83.28.86, Ping timeout: 260 seconds) | |
12:18 | lucascastro has joined IRC (lucascastro!~lucas@186.227.185.10) | |
12:23 | AndChat|248436 has joined IRC (AndChat|248436!~AndChat24@37-48-44-206.tmcz.cz) | |
12:29 | epoptes_user9 has left IRC (epoptes_user9!be496121@gateway/web/freenode/ip.190.73.97.33, Ping timeout: 260 seconds) | |
12:31 | donkey_ has joined IRC (donkey_!ad0e3e49@gateway/web/freenode/ip.173.14.62.73) | |
12:31 | <donkey_> my image is reporting that it is out of space
| |
12:32 | but its only 4gb/30
| |
12:36 | other strange things are happening as well
| |
12:36 | it isnt saving the user's profile information. i have to reset up printers and shortcuts every time they log in
| |
12:51 | Trendzetter has joined IRC (Trendzetter!~trendzett@mail.sbat.be) | |
12:52 | <Trendzetter> Hi everyone! I am back
| |
12:54 | <alkisg> donkey_: it sounds like out of ram issues on the client, due to something filling up the cow file system
| |
12:54 | donkey_: distro/version?
| |
12:54 | Also, do you have epoptes installed?
| |
12:56 | Michel_: you should be able to use this guide, even on mint:
| |
12:56 | !ltsp-pnp
| |
12:56 | <ltsp> 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
| |
12:56 | <donkey_> ubuntu server 16.04
| |
12:57 | <alkisg> donkey_: ok, this messages are on the client, or on the server?
| |
12:57 | <donkey_> on the client
| |
12:57 | <alkisg> OK, do you have epoptes installed?
| |
12:57 | !epoptes
| |
12:57 | <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
| |
12:58 | <donkey_> no i dont have epoptes installed
| |
12:58 | <alkisg> Do you have root access to the client?
| |
12:58 | Like this or something?
| |
12:58 | !screen_02
| |
12:58 | <ltsp> screen_02: To get a root shell on an Ubuntu thin client: https://help.ubuntu.com/community/UbuntuLTSP/ClientTroubleshooting#Using_a_shell_SCREEN
| |
12:58 | <donkey_> the weird thing is though, i had this exact environment running in a virtualbox vm for testing
| |
12:58 | and the test environment worked fine and had no problems
| |
12:59 | <alkisg> One usual reason is the apt-daily task
| |
12:59 | So, when you create the chroot and test it, apt has no updates
| |
12:59 | <donkey_> but when i built the production environment in vmware using the same software now i'm getting the issues
| |
12:59 | <alkisg> A week later when you've moved to production, apt has security updates
| |
12:59 | And it autoinstalls them, wasting all client ram
| |
12:59 | That's just one of the possible causes that explains what you're seeing
| |
12:59 | To verify, you'll need root access to the client
| |
13:00 | <donkey_> i have root access to the client
| |
13:00 | <alkisg> In general, it's best to follow this guid to setup ltsp:
| |
13:00 | !ltsp-pnp
| |
13:00 | <ltsp> 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
| |
13:00 | <alkisg> OK, what's the output of: df -h | nc termbin.com 9999
| |
13:02 | <donkey_> http://termbin.com/w52w
| |
13:02 | <alkisg> That's from the server, not from the client
| |
13:02 | We want the output of the client
| |
13:03 | <donkey_> can i do it in the chroot or do i have to be at the terminal
| |
13:03 | <alkisg> You have to be on the terminal; unless you're using ssh or epoptes
| |
13:03 | <donkey_> yea two secs then. i gotta run over there to do it
| |
13:03 | <alkisg> And specifically on a terminal that has the issue
| |
13:03 | That's where epoptes helps, you don't have to run over there :)
| |
13:03 | <donkey_> all of them have the same issue. its the same image/user
| |
13:04 | <alkisg> You can't use the same user on many clients
| |
13:04 | Software isn't written with that in mind (gnome, firefox etc etc)
| |
13:04 | <donkey_> well that might be part of the problem then
| |
13:04 | <alkisg> I don't think so
| |
13:04 | I think it's just the apt-daily service
| |
13:05 | <donkey_> yea let me get a terminal here then
| |
13:07 | <Trendzetter> what should I check first if X is not starting on boot but I can get an x screen by running startx?
| |
13:08 | <alkisg> Trendzetter: I don't know, I'm still waiting for the "which distribution version" question I asked first... :)
| |
13:09 | <Trendzetter> Ubuntu 16.04.2 LTS
| |
13:09 | <alkisg> Trendzetter: and which guide are you following to create a kiosk setup?
| |
13:10 | <Trendzetter> http://ternarybit.org/chrome-web-kiosk-guide/
| |
13:11 | <alkisg> Trendzetter: which cpu do your clients have, and how much RAM?
| |
13:12 | <Trendzetter> intel i5 and 4 or 8gb ram or something. I am actually using a brand new Dell optiplex 3040
| |
13:12 | as cleint
| |
13:13 | <alkisg> And why do you want all this extra trouble, instead of running just a normal ltsp diskless fat client setup?
| |
13:13 | <Trendzetter> how normal?
| |
13:13 | <alkisg> !ltsp-pnp
| |
13:13 | <ltsp> 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
| |
13:13 | <alkisg> It's the usual ltsp installation page; it covers 90+% of the cases...
| |
13:14 | <Trendzetter> I will look into it but it seemed a very close match to my needs
| |
13:14 | <alkisg> And if you want autologin, it's easy; if you want guest profiles (cleared on login), it's easy etc etc
| |
13:15 | It's a custom tutorial that noone really supports though
| |
13:15 | While with ltsp-pnp, you won't get to "why don't I have X?" issues
| |
13:15 | <donkey_> ok
| |
13:15 | output from the terminal is http://termbin.com/0wgg
| |
13:16 | <Trendzetter> but I also have older clients too. all of them are optiplex that can run windows vista so probably that classifies as powerful client machines
| |
13:16 | <alkisg> donkey_: overlay 802M 3.1M 799M 1% / ==> this means it's not the apt-daily issue I was saying. What's the exact client error message, and when do you see it?
| |
13:16 | Trendzetter: yes, so just use ltsp-pnp
| |
13:19 | <donkey_> i was installing acrobat in wine
| |
13:19 | and it reported out of space
| |
13:19 | plus the other issues. no saved profiles and cups isnt connecting even though i can see the service running on the server
| |
13:20 | <alkisg> That may be related to "only one user for all clients", or it might be a wine issue
| |
13:20 | <Trendzetter> I am very grateful for your answers but the docs at wiki.ltsp.org seem to be much more complicated and less strait forward than the howto I was following. Is their some step by step guide for a default setup ?
| |
13:20 | <donkey_> well according to that link its showing dev/nbd0 as 100% used
| |
13:20 | <alkisg> Start with using a different user per client
| |
13:20 | nbd0 is read only, it's ok
| |
13:20 | <Trendzetter> the official ubuntu guide seems to be quite old
| |
13:20 | <alkisg> It's normal to be 100% in use
| |
13:20 | Trendzetter: I don't know where you saw the wiki.ltsp.org link
| |
13:21 | I gave you this one :
| |
13:21 | !ltsp-pnp
| |
13:21 | <ltsp> 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
| |
13:21 | <alkisg> This is not in wiki.ltsp.org
| |
13:21 | I'm maintaining this myself and it's working fine for Ubuntu 16.04.2
| |
13:22 | <Trendzetter> k
| |
13:22 | <donkey_> yea i'll be installing that later
| |
13:23 | <Trendzetter> can I use a server iso to set up the server or should it be a desktop cd (given as example)
| |
13:23 | <alkisg> It's best to use a desktop cd
| |
13:23 | It will save you a lot of trouble
| |
13:24 | And, ubuntu-mate is more tested than others
| |
13:24 | 32bit, to make sure that older clients can boot as well
| |
13:25 | <Trendzetter> k. I will put the current server at the side and give this setup a chance
| |
13:51 | <donkey_> alkisg creating users worked for the CUPS and profile problems
| |
13:51 | all of the shortcuts are staying, and i dont have to restart the cups service everytime someone logs in/out
| |
13:54 | ben_roose has joined IRC (ben_roose!~roose@roose.cs.wichita.edu) | |
13:55 | <donkey_> i dont know that this is going to solve the profile space issue though
| |
13:55 | is there a config entry that places a hard limit on profile space?
| |
14:06 | ZAJDAN has left IRC (ZAJDAN!4d30954b@gateway/web/freenode/ip.77.48.149.75, Quit: Page closed) | |
14:11 | <alkisg> donkey_: so to sum up, some problems are solved and you only have one left?
| |
14:11 | Whats that 'profile space issue'? The wine error?
| |
14:12 | There's no limit in user profiles in ubuntu, no
| |
14:13 | GodFather has left IRC (GodFather!~rcc@47.33.250.142, Ping timeout: 240 seconds) | |
14:13 | <donkey_> well i'm looking at a mount point on the desktop that shows 188mb
| |
14:14 | and the current user profile is 165mb in size
| |
14:14 | <alkisg> donkey_: put the result to pastebin
| |
14:14 | mikkel has left IRC (mikkel!~mikkel@mail.dlvs.dk, Quit: Leaving) | |
14:14 | <alkisg> Like, df -h | nc termbin.com 9999
| |
14:17 | <donkey_> http://termbin.com/08gr
| |
14:18 | <alkisg> donkey_: no, run this as the user that has the issue, not as local root
| |
14:18 | <donkey_> that is as the user
| |
14:18 | on the terminal
| |
14:18 | <alkisg> ...so you're using fat clients?
| |
14:19 | <donkey_> only if its defaulting
| |
14:19 | which it shouldnt be
| |
14:19 | <alkisg> If you build a fat chroot, and you have enough ram, then yes
| |
14:19 | <donkey_> there's no disk in the terminal
| |
14:19 | <alkisg> fat means diskless but running apps locally with the client ram/cpu
| |
14:19 | I don't see /home/username mounted
| |
14:19 | So that's a big issue there
| |
14:20 | What's your lts.conf like?
| |
14:20 | <donkey_> whatever the defaults are
| |
14:20 | <alkisg> Did you say that's mint?
| |
14:20 | <donkey_> ubuntu 16.04
| |
14:20 | <alkisg> Which desktop environment?
| |
14:20 | <donkey_> xfce
| |
14:20 | <alkisg> What's the output of ls /opt/ltsp/*/usr/share/xsessions ?
| |
14:21 | (on the server)
| |
14:21 | <donkey_> xfce.desktop
| |
14:21 | <alkisg> That means that you have a fat chroot with xfce on it
| |
14:22 | And since the client has 800 MB RAM, it defaults to fat client
| |
14:22 | So, it should have mounted /home/username with sshfs, but it didn't
| |
14:22 | That could mean that it hanged previously, or that you have a misconfiguration somewhere
| |
14:22 | What's the output of `ls -l /home` locally on the client?
| |
14:24 | Btw, did you run `adduser` inside the chroot at some point?
| |
14:25 | <donkey_> output:
| |
14:25 | total 0
| |
14:25 | <alkisg> And the user has logged in?
| |
14:25 | <donkey_> drwxr-xr-x 14 room6 room6 300 <date> room6
| |
14:25 | <Trendzetter> @alkisg The tutorial says run "ltsp-config dnsmasq" but I need to run my own dnsmasq on openwrt
| |
14:25 | <donkey_> yes the user is logged in
| |
14:25 | <alkisg> Ah ok there's a "room6" there
| |
14:26 | donkey_: so, it appears that you logged in as room6 at some point, then logged out or hanged, and the mount was not properly cleared,
| |
14:26 | and now ltsp thinks that you have a local home, and it's not using sshfs,
| |
14:26 | and it fills up your RAM thinking it's a disk
| |
14:26 | Reboot the client, login, and see `df -h` again
| |
14:26 | It should have an sshfs mount
| |
14:26 | If it doesn't, after a clean reboot, ping me
| |
14:27 | <donkey_> ok
| |
14:27 | <alkisg> Trendzetter: it's ok, you can run 2 dnsmasqs
| |
14:27 | You can even disable the proxydhcp range if you don't like it, afterwards
| |
14:27 | A single line in ltsp-server-dnsmasq.conf...
| |
14:28 | <Trendzetter> I need to make sure that dnsmasq is not giving any answers to dhcp requests
| |
14:28 | <alkisg> By default, it only does proxydhcp
| |
14:28 | So it doesn't give any leases
| |
14:28 | <Trendzetter> mmm
| |
14:29 | I don't know much about it, I will give it a try
| |
14:29 | <alkisg> Trendzetter: you can also run this:
| |
14:29 | sudo ltsp-config dnsmasq --no-proxy-dhcp
| |
14:29 | This won't use proxydhcp at all, but it will do the rest like tftp, dns if you want...
| |
14:30 | <donkey_> do i need to have a user both on the server and in the chroot?
| |
14:31 | <alkisg> donkey_: only on the server
| |
14:32 | Users in the chroot can cause issues
| |
14:32 | <donkey_> ok
| |
14:46 | SYS64738 has left IRC (SYS64738!~SYS64738@159.213.93.166, Remote host closed the connection) | |
15:01 | <donkey_> @alkisg http://termbin.com/y9n3 no home folder
| |
15:01 | it didnt save the profile after restart either
| |
15:02 | so it looks like its using ram as storage rather than the server
| |
15:05 | it looks like that none of the users home directories got created when i made the users
| |
15:05 | i.e. if i log into the server with the specific user, it says no home directory found
| |
15:33 | trendzetter2 has joined IRC (trendzetter2!~trendzett@mail.sbat.be) | |
15:36 | Trendzetter has left IRC (Trendzetter!~trendzett@mail.sbat.be, Ping timeout: 246 seconds) | |
16:53 | lucascastro has left IRC (lucascastro!~lucas@186.227.185.10, Ping timeout: 260 seconds) | |
17:00 | AndChat|248436 has left IRC (AndChat|248436!~AndChat24@37-48-44-206.tmcz.cz, Ping timeout: 260 seconds) | |
17:10 | vagrantc has joined IRC (vagrantc!~vagrant@unaffiliated/vagrantc) | |
17:35 | <vlt> Hello. I'm using LTSP on Ubuntu to boot my clients that finally run rdesktop on X. How can I set the key-repeat rate or its delay?
| |
17:37 | <vagrantc> alkisg: regarding https://github.com/NetworkBlockDevice/nbd/issues/47
| |
17:38 | alkisg: why use sleep at all? if it keeps the file open, removing it immediately should work
| |
17:38 | alkisg: in the past, it didn't keep the file open, and so removing it didn't work
| |
17:39 | <alkisg> vagrantc: we create the file. Then we exit. Then nbd-server opens it. Then we can delete it.
| |
17:39 | If we delete it before we exit, then nbd-server won't even find it to open it...
| |
17:39 | <vagrantc> ah
| |
17:40 | would be nice if there were a hook "after nbd-server connects"
| |
17:40 | so as long as negotiation takes less than 10 seconds...
| |
17:40 | <alkisg> No, negotiation is before prerun
| |
17:40 | We get called after negotiation
| |
17:41 | So it should be only a few msec...
| |
17:41 | The big downside is this:
| |
17:41 | if ever nbd-client properly supports reconnections,
| |
17:41 | then ideally we would like to assign the same nbd swap file, without reformatting it,
| |
17:41 | so that the client can continue from when it got disconnected
| |
17:42 | That's why I had selected a static file name and not a random one
| |
17:42 | This isn't supported currently by nbd-client, so we don't lose anything now
| |
17:42 | And, we can even make the tmp file name random, so that we don't hardcode /tmp/nbd-swap etc
| |
17:42 | * vagrantc doesn't like the idea of swapfiles hanging around on-disk | |
17:42 | <vagrantc> could contain passwords and who knows what else
| |
17:43 | <alkisg> A reconnection might happen within a second
| |
17:43 | The same swap file should be used then
| |
17:43 | <vagrantc> but not within 10 seconds?
| |
17:43 | <alkisg> The server swap partition can also contain passwords etc...
| |
17:44 | The tcp timeout is 2 hours 11 minutes
| |
17:44 | <vagrantc> alkisg: i'll never buy the argument because some things are imperfect it means we shouldn't bother to do anything
| |
17:44 | <alkisg> So a client supposedly can disconnect and reconnect within that time without the connection being dropped
| |
17:44 | We trust the disk; that's not a compromise
| |
17:44 | We trust that /etc/shadow isn't readable
| |
17:45 | That /tmp/nbd-swap isn't readable
| |
17:45 | That /dev/swap isn't readable etc
| |
17:45 | <vagrantc> but random users of clients may not
| |
17:45 | <alkisg> If the client has encrypted the swap file, and another client requests it, it won't be possible to decode it
| |
17:46 | <vagrantc> except our current encryption implementation is broken
| |
17:46 | <alkisg> If the sysadmin doesn't use encrypted swap files, and another requests the same swap file, ok, it's an issue
| |
17:46 | The nbd-client reconnection is broken too, so no harm done :D
| |
17:46 | <vagrantc> at any rate, i don't really see the advantage of having stable names
| |
17:47 | <alkisg> You don't see the advantage of nbd-client --persistent?
| |
17:47 | <vagrantc> ideally, i think the swapfiles should be randomized and deleted as soon as technically possible
| |
17:47 | <alkisg> That's what we're using currently, except it's broken...
| |
17:47 | <vagrantc> alkisg: in theory that would handle transient network failures?
| |
17:47 | <alkisg> Yes
| |
17:48 | <vagrantc> that would be nice, sure.
| |
17:48 | <alkisg> Also note that if someone wants to request a part of a swap file... it would be easy
| |
17:48 | They wouldn't have to rely on persistent names or anything
| |
17:48 | <vagrantc> harder if it's only the inode
| |
17:48 | <alkisg> The nbd protocol doesn't care about inodes
| |
17:49 | It's a network protocol...
| |
17:49 | It can be done at the tcp/ip level, without caring about the server file system at all
| |
17:49 | <vagrantc> alkisg: wait, with swap persistance, any user on the client could connect and access the swap?
| |
17:50 | <alkisg> From what I imagine, currently, with or without swap persistence, any client can break the nbd protocol, yes
| |
17:50 | I don't think they implemented anything about securing client requests
| |
17:50 | Encryption would solve that issue, of course
| |
17:50 | <vagrantc> out current implementation is ip address restricted
| |
17:50 | right
| |
17:51 | really ought to fix encryption
| |
17:51 | <alkisg> IP addresses can be spoofed etc...
| |
17:51 | <vagrantc> well, not only that, but the client wouldn't need to spoof it ... they're already running locally
| |
17:51 | they're already running from the correct ip
| |
17:51 | <alkisg> The client would require admin rights
| |
17:51 | <vagrantc> to run nbd-client?
| |
17:52 | <alkisg> To use the same source port as the root nbd-client...
| |
17:52 | <vagrantc> but if the file was still there, they could make a new connection, no?
| |
17:52 | <alkisg> We can detect reconnections vs new connections
| |
17:52 | (at least in theory)
| |
17:53 | <vagrantc> but in practice, do we do anything about it?
| |
17:53 | <alkisg> In practise nbd reconnections don't work at all
| |
17:53 | <vagrantc> network swap is kind of a mess
| |
17:53 | <alkisg> So we don't care about it, and it's fine :)
| |
17:53 | Right now we don't have any security issue related to reconnections or persistence
| |
17:54 | We do have one that related to non-encryption or non-authentication etc
| |
17:54 | *relates
| |
17:55 | vagrantc: this is what I'm thinking to commit: http://termbin.com/vgzl
| |
17:56 | <vagrantc> alkisg: as long as the "The kernel won't remove it from disk while nbd-server is still using it." is true
| |
17:56 | alkisg: seems fine to me
| |
17:57 | <alkisg> Cool, I'll commit it in a few days, I have it running on production now to test it
| |
17:58 | vagrantc: should I also randomize the name?
| |
17:58 | Like, /tmp/nbd-swap-XXXXX?
| |
17:58 | <vagrantc> alkisg: seems better to me
| |
17:58 | <alkisg> OK
| |
17:59 | (06:01:38 μμ) donkey_: @alkisg http://termbin.com/y9n3 no home folder ==> you have an issue, it needs troubleshooting. Install epoptes to make it easier, see epoptes.org
| |
18:01 | (08:35:18 μμ) vlt: Hello. I'm using LTSP on Ubuntu to boot my clients that finally run rdesktop on X. How can I set the key-repeat rate or its delay? ==> https://askubuntu.com/questions/140255/how-to-override-the-new-limited-keyboard-repeat-rate-limit ==> xset r rate, probably
| |
18:09 | <vlt> alkisg: Thanks, I'll try that.
| |
19:02 | Statler_ has joined IRC (Statler_!~Georg@p579FE75F.dip0.t-ipconnect.de) | |
20:05 | quinox has left IRC (quinox!~quinox@ghost.qtea.nl, Quit: :wq) | |
20:06 | quinox has joined IRC (quinox!~quinox@ghost.qtea.nl) | |
20:29 | * alkisg notes down a test for tuesday... to run nbd-client -persistent, and then to pull out the server network cable, and see if nbd-server immediately dies with SIGPIPE, which it shouldn't... | |
20:55 | ben_roose has left IRC (ben_roose!~roose@roose.cs.wichita.edu, Remote host closed the connection) | |
21:10 | <Hyperbyte> mmmm
| |
21:23 | GodFather has joined IRC (GodFather!~rcc@47.33.250.142) | |
21:29 | fnurl has left IRC (fnurl!~url@36-227-13-126.dynamic-ip.hinet.net, Read error: Connection reset by peer) | |
21:29 | fnurl has joined IRC (fnurl!~url@36-227-38-174.dynamic-ip.hinet.net) | |
21:38 | GodFather has left IRC (GodFather!~rcc@47.33.250.142, Ping timeout: 240 seconds) | |
21:42 | jgee has joined IRC (jgee!~jgee@200.118.140.142) | |
21:45 | ricotz has left IRC (ricotz!~ricotz@ubuntu/member/ricotz, Quit: Leaving) | |
21:54 | Statler has left IRC (Statler!~Georg@mail.lohn24.de, Remote host closed the connection) | |
22:01 | lucascastro has joined IRC (lucascastro!~lucas@186.227.185.10) | |
22:02 | lucascastro has left IRC (lucascastro!~lucas@186.227.185.10, Remote host closed the connection) | |