01:13 | vagrantc has joined IRC (vagrantc!~vagrant@unaffiliated/vagrantc) | |
02:50 | vagrantc has left IRC (vagrantc!~vagrant@unaffiliated/vagrantc, Ping timeout: 252 seconds) | |
06:43 | ricotz has joined IRC (ricotz!~ricotz@ubuntu/member/ricotz) | |
07:01 | kjackal has joined IRC (kjackal!~quassel@80.107.182.228) | |
10:59 | Faith has joined IRC (Faith!~Paty_@unaffiliated/faith) | |
11:32 | kjackal has left IRC (kjackal!~quassel@80.107.182.228, Read error: Connection reset by peer) | |
11:32 | kjackal has joined IRC (kjackal!~quassel@2a02:587:3101:f300:58d8:9170:d224:a967) | |
12:58 | sutula has left IRC (sutula!~sutula@207-118-149-102.dyn.centurytel.net, Ping timeout: 244 seconds) | |
13:22 | ogra has left IRC (ogra!~ogra_@ubuntu/member/ogra, Ping timeout: 245 seconds) | |
13:36 | nehemiah has left IRC (nehemiah!~nehemiah@hs-user-138.wia.cz, Ping timeout: 272 seconds) | |
13:37 | <Faith> Hello I'm trying to configure LTSP with Ubuntu 18.04, but it's not working. I've followed the instruticions on http://wiki.ltsp.org/wiki/Installation/Ubuntu, with the greek repo, adapting to my case (must be isc-dhcp-server and chroot with nbd). But when the client boot, it stops in this error: "no search or nameservers found in /run/net-ens160.conf /run/net-ens160.conf /run/net6-*.conf Negotiation: ..Error: Unknown error in reply to NBD_OPT_GO; cannot
| |
13:37 | continue"
| |
13:38 | I can connect with the nbd-client command so nbd is working, I don't know if the problem is on nbd-client or dns from the client, because the DNS config is missing on /run/ens-160.conf
| |
13:38 | See here https://pastebin.com/jTw4XRHw
| |
13:43 | ogra has joined IRC (ogra!~ogra_@ubuntu/member/ogra) | |
13:50 | <alkisg> Faith: do you have a grad-amd64 nbd export?
| |
13:50 | <Faith> yes
| |
13:51 | when I run nbd-client 10.20.20.4 -N /opt/ltsp/grad-amd64 /dev/nbd0
| |
13:51 | <alkisg> Why are you sending ROOTPATH='/opt/ltsp/amd64' if you want grad-amd64?
| |
13:51 | <Faith> it works
| |
13:51 | <alkisg> Send /opt/ltsp/grad-amd64
| |
13:51 | Ah ok got it
| |
13:51 | Change IPAPPEND 3 to IPAPPEND 2
| |
13:52 | In /var/lib/tftpboot/ltsp/*/pxelinux.cfg/default
| |
13:52 | !ltsp-discuss
| |
13:52 | <ltsp> ltsp-discuss: at https://lists.sourceforge.net/lists/listinfo/ltsp-discuss
| |
13:52 | <alkisg> If you want more details, search the mailing list, I answered it analytically a month ago or so
| |
13:53 | <Faith> I'll try
| |
13:53 | Thank you :)
| |
13:53 | <alkisg> yw
| |
13:57 | <Faith> alkisg, I works!
| |
13:57 | but every time that I run ltsp-update-image it will overwrite pxe config
| |
13:57 | <alkisg> Yes, read the mailing list for the how-to that I wrote there
| |
13:57 | <Faith> ok
| |
13:58 | thank you again :)
| |
13:58 | <alkisg> np
| |
14:08 | nehemiah has joined IRC (nehemiah!~nehemiah@37-247-80-214.staticip.skydsl.de) | |
14:11 | nehemiah has left IRC (nehemiah!~nehemiah@37-247-80-214.staticip.skydsl.de, Remote host closed the connection) | |
14:53 | spaced0ut has left IRC (spaced0ut!~spaced0ut@unaffiliated/spaced0ut, Ping timeout: 246 seconds) | |
16:00 | spaced0ut has joined IRC (spaced0ut!~spaced0ut@unaffiliated/spaced0ut) | |
16:56 | dominion31489 has joined IRC (dominion31489!d1f9948f@gateway/web/freenode/ip.209.249.148.143) | |
16:57 | <dominion31489> Hello Im looking some info.
| |
16:57 | I've been searching online, but does anyone know if LTSP creates separate sessions for multiple users getting into a VM? If user1 and user2 connect to VM1 at the same time, will they see each others changes or will LTSP create separate desktops?
| |
17:01 | Also for a standard VM running Ubuntu with 5GB of memory, how many LTSP clients can connect to a single VM at once?
| |
17:12 | <alkisg> dominion31489: ltsp doesn't use VMs
| |
17:12 | I think you should describe what you want to do, and IF it is related to ltsp
| |
17:13 | If you just happen to use VMs with LTSP, it's not related to ltsp, it doesn't have any code for VM management at all
| |
17:13 | So whatever happens in normal linux, happens in ltsp too
| |
17:13 | <dominion31489> What do you mean it doesnt use VM's? I was able to put it on Ubuntu via Vsphere
| |
17:23 | Im basically trying to use LTSP for a student lab with about 20 VM's running Ubuntu.
| |
17:23 | I wanted to use the thin client manager to manage user accounts and allow them to connect to different VM's.
| |
17:24 | <mwalters> you can run ltsp on a vm, but ltsp doesn't manage VMs in any way
| |
17:25 | <dominion31489> Right, I dont necessarily need it to manage them but work in such a way that user1 can remote into VM1, VM2, VM3 etc.
| |
17:25 | <mwalters> if VM1 is an ltsp server, you can use thin clients and each client will get a separate session
| |
17:25 | <dominion31489> I need a non-concurrent thin client solution
| |
17:26 | Will I need to install LTSP on each VM?
| |
17:26 | <mwalters> probably... then you need to figure out how to target specific servers for each client
| |
17:26 | alkisg: is ltsp-cluster still a thing?
| |
17:26 | <dominion31489> Could I just target their IP address?
| |
17:27 | In the thin client manager I see it allows you to add a guest based on the IP address. Could I not just create a LTSP login for that specific IP/VM>
| |
17:27 | <mwalters> generally, booting is handled via dhcp, but you could probably have vmlinuz/initramfs stage locally, I think and target a specific server
| |
17:28 | login is just "normal" user accounts
| |
17:28 | ldm logs in to the ltsp server via ssh
| |
17:28 | authenticates, rather
| |
17:29 | are the student's desktops thick clients?
| |
17:29 | e.g., they all have an os/disk/etc already
| |
17:32 | <alkisg> mwalters: no, ltsp-cluster isn't maintained anymore
| |
17:33 | dominion31489: the ltsp code there's no "thin client manager"
| |
17:33 | I've no idea what you're talking about, and I'm the most active ltsp developer...
| |
17:34 | I also don't know how "vsphere" relates to ltsp; if they offer a custom build of ltsp, we don't know about it
| |
17:34 | <mwalters> x2go/ldap is probably a better solution than ltsp based on what I see
| |
17:34 | <dominion31489> From the LTSP server, this dude is using a thin client manager.
| |
17:34 | https://www.youtube.com/watch?v=2aFcDlp9-xw
| |
17:35 | <alkisg> dominion31489: did you notice "2013" there?
| |
17:35 | <mwalters> this is 5 years old
| |
17:35 | <dominion31489> Ah
| |
17:35 | <alkisg> You can easily load balance ltsp servers, or direct users to different servers; but I haven't understood what your actual needs are, unrelated to "how to implement them"
| |
17:38 | So anyway to try to answer your initial question, in ltsp, if user1 and user2 log in at the same time, they're working on the "same pc", so they see each other's changes immediately
| |
17:40 | <dominion31489> Ok thanks for your help everyone
| |
17:40 | <alkisg> np
| |
17:47 | dominion31489 has left IRC (dominion31489!d1f9948f@gateway/web/freenode/ip.209.249.148.143, Quit: Page closed) | |
17:55 | kjackal has left IRC (kjackal!~quassel@2a02:587:3101:f300:58d8:9170:d224:a967, Ping timeout: 252 seconds) | |
17:56 | sutula has joined IRC (sutula!~sutula@207-118-151-61.dyn.centurytel.net) | |
18:06 | vagrantc has joined IRC (vagrantc!~vagrant@unaffiliated/vagrantc) | |
18:26 | Statler has joined IRC (Statler!~Georg@gwrz.lohn24.de) | |
18:28 | Statler has left IRC (Statler!~Georg@gwrz.lohn24.de, Client Quit) | |
18:29 | Statler has joined IRC (Statler!~Georg@gwrz3.lohn24.de) | |
18:29 | josefig has joined IRC (josefig!~jose@unaffiliated/josefig) | |
18:32 | <josefig> hello, i got a problem with my UPS and the server shutdown, when I turn on again with new UPS my LTSP Clients have no internet, they can boot but not Internet at all
| |
18:32 | when I log into the server using ssh it has intenet but not the LTSP Clients
| |
18:35 | <mwalters> is your ltsp server also a dhcp server, or are you doing proxydhcp?
| |
18:37 | <josefig> mwalters: yes, i have a dhcp server using dnsmasq. I did again the ltsp-config dnsmasq --no-proxy-dhcp --overwrite but I got when the pc boot 2 options of PXE boot, Boot from Network both times
| |
18:37 | I'm trying again
| |
18:38 | <mwalters> are you using a 2 NIC config on the ltsp server?
| |
18:38 | <josefig> yes i am
| |
18:38 | <||cw> thin clients or fat?
| |
18:38 | <mwalters> oic, I'm not super familiar with that, I'll defer to someone else
| |
18:39 | <alkisg> !install
| |
18:39 | <ltsp> install: http://wiki.ltsp.org/wiki/Installation/Ubuntu for Ubuntu, or http://wiki.ltsp.org/wiki/Installation for other distributions
| |
18:39 | <||cw> is fat, is your server configured as a router?
| |
18:39 | <alkisg> josefig: you need to download the script for "nat"
| |
18:39 | From that web page
| |
18:40 | <josefig> ||cw: fat clients
| |
18:40 | <||cw> maybe you applied it manually at one point and haven't rebooted?
| |
18:42 | <josefig> alkisg: this line >> wget https://git.launchpad.net/sch-scripts/plain/debian/sch-scripts.if-up -O /etc/network/if-up.d/sch-scripts
| |
18:42 | <alkisg> Yes, read it for instructions
| |
18:44 | <josefig> yes, i did again but nothing changed
| |
18:44 | it's chroot-less image
| |
18:54 | <alkisg> from the clients, can you ping 8.8.8.8?
| |
18:55 | <josefig> mea culpa, i found the problem. It was because of docker a iptables policy on forwarding it said DROP, I just changed to ACCEPT and worked all.
| |
19:00 | <alkisg> meh,who uses docker for ltsp :D
| |
19:03 | Statler has left IRC (Statler!~Georg@gwrz3.lohn24.de, Remote host closed the connection) | |
19:04 | <mwalters> was it docker where the official install isntructions were to pipe curl output into an elevated bash shell?
| |
19:04 | <vagrantc> certainly wouldn't be the only one...
| |
19:04 | <mwalters> lol
| |
19:05 | adrianorg has joined IRC (adrianorg!~adrianorg@187.115.105.0) | |
19:06 | <alkisg> In other news, ipv6 works great. No more dhcp required in the initramfs, so it's done normally later on, from /etc/network/interfaces, with dhcp there, not static/manual
| |
19:08 | adrianor1 has left IRC (adrianor1!~adrianorg@187.113.216.7, Ping timeout: 250 seconds) | |
19:09 | <vagrantc> wow
| |
19:09 | hope there won't be any gotchas if we make that the default behavior, or easy to change back if needed
| |
19:09 | but that sounds really cool
| |
19:09 | <alkisg> And for now, I'm using an nbd copyonwrite overlay with settings and swap and all, so no tftp or http required
| |
19:10 | We'll see if http is needed later on
| |
19:10 | <vagrantc> did you resolve the tftp issue?
| |
19:10 | oh wow
| |
19:11 | so you pass a specific nbd-cow to each client?
| |
19:20 | <josefig> nice =)
| |
19:22 | alkisg: no, I use docker in the server only, i removed the service in ltsp clients. I use for software develompment and works great, I am very happy with LTSP to be honest with you
| |
19:27 | <alkisg> vagrantc: nah, lts.conf is there for all clients
| |
19:28 | The nice thing is that all of ltsp-client code can be there
| |
19:28 | And swap, and /etc/ltsp/per-client directories/overlays
| |
19:29 | <vagrantc> so a swap file on the filesystem?
| |
19:29 | <alkisg> Yes, typical /swapfile that ubuntu uses nowadays
| |
19:29 | I guess one can encrypt that, haven't tried
| |
19:30 | And the single cow image can be created in a couple of seconds (1 GB sparse file with less than 1 MB actual data)
| |
19:30 | So I think I'll put settings in /etc, and copy them to the cow image in a similar way that ltsp-update-kernels updates kernels
| |
19:31 | Now on to authentication... I'm between samba and libpam-sshauth...
| |
19:32 | <vagrantc> heh
| |
19:32 | <alkisg> (09:09:37 μμ) vagrantc: hope there won't be any gotchas if we make that the default behavior, or easy to change back if needed ==> those are for the new project, so there won't even be old code to revert to :D
| |
19:45 | <||cw> samba's probably a bit heavy, and ssh can use samba/winbind accounts on the server if the admin wants
| |
19:48 | <alkisg> samba is maintained by others, while we'd have to maintain libpam-sshauth ourselves :D
| |
19:48 | Would plain ldap be more light? I think some ltsp users use it anyway...
| |
19:48 | <||cw> but then you have to setup a samba server and maintain the smbpasswd
| |
19:49 | <alkisg> `ltsp-config samba` to simplify the first setup, while I think smbpasswd is automatically maintained if you use normal unix accounts
| |
19:49 | And the side effect would be that windows users would be able to login to the linux server...
| |
19:50 | <||cw> it's not. but it can update the unix account if you use smbpasswd command
| |
19:50 | <alkisg> I don't know if setting all that up and maintaining it, would be easier than maintaining libpam-sshauth though...
| |
19:50 | Faith has left IRC (Faith!~Paty_@unaffiliated/faith, Quit: Leaving) | |
19:51 | <alkisg> Hmm, I might remember wrong, but I thought that it injected itself to the pam stack, so that when I did "passwd" or "adduser", it automatically updated smbpasswd
| |
19:51 | <||cw> been a while since I did a standalone, I think i add to do the -a manually at least
| |
19:52 | does libpam-sshauth already exist?
| |
19:52 | <alkisg> Yes, an ltsp developer wrote it (sbalneav), but it's never been tested in real use
| |
19:53 | <||cw> hm.
| |
19:53 | <alkisg> So it'll need to be reviewed, at least
| |
19:56 | <mwalters> I have like, 75 users... and I think ldap is overkill ;)
| |
20:00 | <alkisg> (unrelated) The fedora live cd doesn't use overlay for the live/writeable file system, it's using /dev/mapper/live-rw... wth is that :D
| |
20:00 | <vagrantc> it's the way to do it before aufs existed
| |
20:01 | uses device-mapper and freinds
| |
20:02 | <alkisg> How come we never used it in ltsp?
| |
20:05 | Heh, they're using an ext4 sparse upon a tmpfs, for the writeable part... I wonder why, maybe ext4 is more stable wrt application usage?
| |
20:05 | <vagrantc> because ubuntu used it's own overlay kernel stuff early on
| |
20:06 | <alkisg> Hmm, I thought 8.04 and 10.04 were still using aufs... only later they switched to overlayfs, which eventually gave its place to overlay...
| |
20:06 | <vagrantc> for a long time, Debian used the RW_DIRS/RW_FILES and such
| |
20:06 | <alkisg> Ah right
| |
20:06 | <vagrantc> alkisg: i mean union fs type filesystems in general
| |
20:07 | aufs was never in-kernel
| |
20:07 | it was always patched in
| |
20:07 | or out-of-tree
| |
20:07 | <alkisg> overlay is still missing as a module from initramfs's
| |
20:07 | One needs to specifically add it, in order to be available
| |
20:07 | Those 2 things are the only ones needed, the overlay module and nbd-client
| |
20:08 | <vagrantc> cool
| |
20:08 | <alkisg> The "ltsp" initramfs scripts can be included with the additional initrd line
| |
20:21 | Meh, the default ubuntu initramfs has dmsetup and not overlay :D
| |
20:21 | ...debian too
| |
20:24 | <vagrantc> dmsetup is useful for a lot of things
| |
20:25 | <alkisg> If it's already there, and stable, and efficient, why not switch to that instead of overlay then...
| |
20:25 | <vagrantc> you couldn't use squashfs as a backend
| |
20:25 | you'd have to use a writeable filesystem
| |
20:25 | <alkisg> How does the fedora live cd do it?
| |
20:25 | <vagrantc> you said they use ext4
| |
20:25 | <alkisg> It has a squashfs image
| |
20:25 | For the overlay, not for the rofs
| |
20:26 | <vagrantc> hmmm... maybe there are more complicated things going on than i thought
| |
20:26 | <alkisg> squashfs rofs + tmpfs + sparse file on that tmpfs, formatted as ext4
| |
20:26 | But I still can't understand the dmsetup magic to do that
| |
20:26 | <vagrantc> i thought they make a virtual block device, which wouldn't work well with differing filesystems ... but clearly that's not the case
| |
20:26 | <alkisg> The squashfs rofs is there on the cd though, so it's definately doable
| |
20:28 | <vagrantc> i do recall fedora folks being envious of the unionfs/aufs/overlay(fs) approach being simpler
| |
20:28 | *ages* ago
| |
20:28 | may have improved since then
| |
20:29 | <alkisg> Heh. And now overlay is just a modprobe away. Oh well since we do need a modified initramfs anyway, for nbd-client, I guess it's not worth it
| |
20:29 | I'd love it if nbd was a bit like aoe. Modprobe and mount, no nbd-client...
| |
20:29 | <vagrantc> aoe has issues you didn't like?
| |
20:30 | <alkisg> It doesn't have any way to avoid flow contol
| |
20:30 | So, if someone has even one 100mbps client, it tells the server to transmit at 100 mbps for all clients
| |
20:30 | (total, i.e. not gigabit anymore)
| |
20:32 | Maybe now with ipv6 and link-local addresses, someone could do the same thing that aoe does, over ipv6
| |
20:32 | Only for the local subnet, and with multicasting for advertising the shares, etc
| |
20:37 | <vagrantc> so ipv6 link-local addresses wouldn't work across routers, which i'd had use-cases for in the past
| |
20:38 | but that's a pretty sub-optimal setup for a client anyways
| |
20:38 | but relying on it would make that basically impossible
| |
20:39 | surely might be worth it
| |
20:39 | kjackal has joined IRC (kjackal!~quassel@2a02:587:3101:f300:58d8:9170:d224:a967) | |
20:40 | <vagrantc> an ltsp sever on multiple networks with multiple interfaces would have a different link-local address for each network
| |
20:43 | <alkisg> Virtualbox manages to send packets to the network for VMs with bridged NICs, even if the host doesn't have an IP
| |
20:44 | <vagrantc> that's not a router, though
| |
20:44 | <alkisg> If we could do that, i.e. setup our own virtual interface, we wouldn't bother with the rest of the networking
| |
20:44 | It supports ethernet, ipv4, ipv6, everything
| |
20:44 | <vagrantc> it just depends on existing network topology and all that
| |
20:46 | hard-coding network topology without any way to change it will mean it can't be used in some environments
| |
20:46 | <alkisg> About virtualbox? I think it's using https://en.wikipedia.org/wiki/TUN/TAP
| |
20:46 | <vagrantc> sure
| |
20:46 | <alkisg> It doesn't hardcode anything. I don't know if we could use that ourselves
| |
20:47 | <vagrantc> well, you're talking about only supporting ipv6 link-local addresses
| |
20:47 | <alkisg> I.e. to have an "ltsp-eth0" virtual card, that we could use only for the nbd/aoe connection, over any protocol, while the os wouldn't touch it
| |
20:47 | Oh I'm talking about 2 different approaches here
| |
20:47 | virtualbox/tun/tap is a completely different approach than ipv6 link local
| |
20:48 | But both of them allow the OS to set an ipv4 IP without breaking our nbd connection
| |
20:48 | <vagrantc> right
| |
20:57 | <quinox> (I have 12 users and are really happy with LDAP)
| |
21:19 | spaced0ut has left IRC (spaced0ut!~spaced0ut@unaffiliated/spaced0ut, Quit: Leaving) | |
21:31 | adrianor1 has joined IRC (adrianor1!~adrianorg@177.18.175.43) | |
21:34 | adrianorg has left IRC (adrianorg!~adrianorg@187.115.105.0, Ping timeout: 250 seconds) | |
21:57 | ricotz has left IRC (ricotz!~ricotz@ubuntu/member/ricotz, Remote host closed the connection) | |
23:13 | vagrantc has left IRC (vagrantc!~vagrant@unaffiliated/vagrantc, Quit: leaving) | |
23:30 | nikoh77 has joined IRC (nikoh77!~nikoh77@host88-180-dynamic.53-79-r.retail.telecomitalia.it) | |