IRC chat logs for #ltsp on irc.libera.chat (webchat)
Channel log from 7 May 2014 (all times are UTC)
| 00:02 | john_s has joined IRC (john_s!a8d4f382@gateway/web/freenode/ip.168.212.243.130) | |
| 00:04 | <john_s> hello all, having trouble with X on an thin client running ubuntu 14.04, I suspect it doesn't like all of the magical effects, any way to have a basic desktop?
| |
| 00:04 | I tried installing gnome3 which I want to use instead of unity but that didn't improve the situation much
| |
| 00:05 | in the old days I had to disable compiz and use metacity, I don't even know what the current ubuntu window manager's are...
| |
| 00:05 | we're still running ubuntu LTS 10.04
| |
| 00:06 | but trying to figure out 14.04 for a new rollout
| |
| 00:08 | so far I haven';t found any current info for LTSP on 14.04
| |
| 00:12 | john_s has left IRC (john_s!a8d4f382@gateway/web/freenode/ip.168.212.243.130, Ping timeout: 240 seconds) | |
| 00:14 | <vagrantc> and you won't by leaving so quick
| |
| 00:17 | john_s has joined IRC (john_s!a8d4f382@gateway/web/freenode/ip.168.212.243.130) | |
| 00:18 | <john_s> @vagrantc whoops, I got bumped out, I'm using the web client,
| |
| 00:18 | I wouldn't run off that fast :->
| |
| 00:19 | Ark74 has left IRC (Ark74!~Ark74@189.220.248.68.cable.dyn.cableonline.com.mx, Ping timeout: 252 seconds) | |
| 00:20 | <john_s> I really don't even know what the state of the LTSP art is these days, since we're running whatever was current back in ubuntu 10.04 days
| |
| 00:21 | so if anyone has a good pointer to get me back up to speed that would be appreciated
| |
| 00:24 | <vagrantc> you could try with xbuntu or lubuntu ... they're likely lighter
| |
| 00:25 | <john_s> @vagrantc I will try that. The thin clients are atom based with 2 Gig's of ram so I'd expect them to have the chutzpah
| |
| 00:26 | I notice that the DebianEdu project has a note about and lts.conf knob called DEFAULT_DISPLAY_MANAGER
| |
| 00:27 | no man page for lts.conf on Ubuntu, though, perhaps I need to add it via apt
| |
| 00:28 | got it
| |
| 00:30 | hmm, I guess that option doesn't exist on the Ubuntu ver. of LTSP
| |
| 00:32 | john_s_ has joined IRC (john_s_!a8d4f382@gateway/web/freenode/ip.168.212.243.130) | |
| 00:32 | <john_s_> Having trouble staying on here for some reason
| |
| 00:33 | does Debian 7 run a different version of LTSP than Ubuntu 14.04?
| |
| 00:34 | john_s has left IRC (john_s!a8d4f382@gateway/web/freenode/ip.168.212.243.130, Ping timeout: 240 seconds) | |
| 00:35 | <vagrantc> a somewhat different version
| |
| 00:35 | though DEFAULT_DISPLAY_MANAGER is undocumented for a reason... you really don't want to mess with that
| |
| 00:36 | mostly what's different is the versions of software
| |
| 00:37 | <john_s_> thanks, is the DebianEdu site the go to reference for LTSP on Debian?
| |
| 00:38 | <vagrantc> i'm not sure, haven't looked at debian-edu in a while
| |
| 00:38 | they are one of the more sophisticated LTSP setups
| |
| 00:40 | <john_s_> Thank you, do you have a recomendation for good current documentation on configuring LTSP?
| |
| 00:46 | ok, must go, thanks for the help
| |
| 00:47 | john_s_ has left IRC (john_s_!a8d4f382@gateway/web/freenode/ip.168.212.243.130, Quit: Page closed) | |
| 00:47 | * vagrantc wave | |
| 00:48 | Phantomas1 has joined IRC (Phantomas1!~Phantomas@ubuntu/member/phantomas) | |
| 00:50 | Phantomas has left IRC (Phantomas!~Phantomas@ubuntu/member/phantomas, Ping timeout: 252 seconds) | |
| 01:22 | Phantomas1 is now known as Phantomas | |
| 03:45 | Phantomas has left IRC (Phantomas!~Phantomas@ubuntu/member/phantomas, Ping timeout: 250 seconds) | |
| 03:47 | gdi2k has joined IRC (gdi2k!~gdi2k@120.28.224.109) | |
| 04:26 | zama has left IRC (zama!zama@unaffiliated/stryx/x-3871776, Ping timeout: 240 seconds) | |
| 04:27 | zama has joined IRC (zama!zama@unaffiliated/stryx/x-3871776) | |
| 05:04 | cyberorg has left IRC (cyberorg!~cyberorg@opensuse/member/Cyberorg, Remote host closed the connection) | |
| 05:05 | work_alkisg is now known as alkisg | |
| 05:05 | cyberorg has joined IRC (cyberorg!~cyberorg@opensuse/member/Cyberorg) | |
| 05:21 | <cyberorg> alkisg, hi, just completed one ltsp implementation, about 1000 fatclients in single location, any tips for such large setup? :)
| |
| 05:22 | <alkisg> Hi cyberorg, wow, I've never had such a large setup, but I would go for dmcache there
| |
| 05:22 | <cyberorg> using ldap for authentication, nfsserver with ssd disk with 2 node drbd for /home
| |
| 05:22 | <alkisg> A bit of local storage, e.g. a 8 GB SSD disk or stick, and dmcache for automatic local caching
| |
| 05:23 | That should save 90% of the network bandwidth...
| |
| 05:23 | <cyberorg> fatclient image is also on ssd
| |
| 05:23 | <alkisg> It would be extremely faster if it was on the clients
| |
| 05:23 | <cyberorg> i've distributed 100 client per virtual server serving fatclient image
| |
| 05:23 | <alkisg> I.e. caching with dmcache
| |
| 05:24 | A single server should be enough then
| |
| 05:24 | <cyberorg> each vm has its own giga lan
| |
| 05:24 | mealstrom has left IRC (mealstrom!~Thunderbi@46.63.63.163, Ping timeout: 264 seconds) | |
| 05:25 | <alkisg> It'd be very nice to add upstream support in ltsp for dmcache, it would benefit all setups, from 10 clients to 10.000...
| |
| 05:25 | <cyberorg> don't think client will go for additional cost of ssds though
| |
| 05:25 | <alkisg> The clients don't have any DOM at all?
| |
| 05:26 | <cyberorg> they have local hard disks, i could use those, but so far performance is quite good
| |
| 05:27 | <alkisg> Current disks are ~= 1 Gbps per client
| |
| 05:27 | <cyberorg> you saw this? http://sourceforge.net/p/cyberorg-home/code/ci/master/tree/dm-cached-nbd/dm-cached-nbd.sh
| |
| 05:27 | <alkisg> So, for 100 clients, it's about 100 Gbps
| |
| 05:27 | You can't match that speed with any networking setup
| |
| 05:27 | For 1.000 clients, 1000 Gbps, etc
| |
| 05:28 | An 8 GB usb 3.0 stick for local storage is 4€, negligible...
| |
| 05:29 | (if you didn't have local disks)
| |
| 05:29 | So I think that local caching is very important and should be on by default, if local storage is detected
| |
| 05:31 | <cyberorg> can you hack up support in ltsp based on that script ^^?
| |
| 05:31 | Ark74 has joined IRC (Ark74!~Ark74@189.214.42.128.cable.dyn.cableonline.com.mx) | |
| 05:31 | <alkisg> Is the script tested + working?
| |
| 05:32 | Did you notice the improvements we're expecting?
| |
| 05:32 | <vagrantc> what's all this talk of dmcache? dm-cache?
| |
| 05:32 | <alkisg> vagrantc: yup, or one of the other caching mechanisms
| |
| 05:32 | <cyberorg> that script is just testing dm-cache setup with nbd/aoe, not tested with ltsp setup
| |
| 05:32 | <vagrantc> just caching the images? filesystems?
| |
| 05:33 | <cyberorg> hi vagrantc :)
| |
| 05:33 | <alkisg> Caching i386.img
| |
| 05:33 | <vagrantc> ah, it's a device-mapper hack
| |
| 05:34 | <cyberorg> vagrantc, dm-cache was created for using ssd as cache for slower disk, in our case network device
| |
| 05:34 | <vagrantc> right
| |
| 05:35 | <cyberorg> as alkisg said it would be nice thing to have, something for debconf? ;)
| |
| 05:35 | <vagrantc> so it dynamically caches at the block device level?
| |
| 05:35 | <alkisg> Yup
| |
| 05:35 | And persists after reboots too
| |
| 05:36 | In an ideal implementation, a bitmap of the cached sectors would be on the client,
| |
| 05:36 | <vagrantc> and how does it handle image updates?
| |
| 05:36 | <alkisg> and, after a few boots, it would be possible to boot the client even with the server powered off
| |
| 05:36 | We would handle image updates in the initramfs
| |
| 05:36 | If the nbd/aoe image is newer, we'd reset the cache
| |
| 05:37 | <MrV> what a charming thing this dm-cache is!
| |
| 05:37 | thank you very much people!
| |
| 05:37 | never new of it
| |
| 05:38 | Knew*
| |
| 05:38 | * vagrantc has been busy testing support for a handful of arm devices with linux 3.15-rc4 all day... | |
| 05:38 | <vagrantc> so then, in the initramfs phase, we'd actually mount the cache device as root?
| |
| 05:38 | <cyberorg> MrV, you can use it on your regular desktop/laptop as well :)
| |
| 05:39 | <vagrantc> backed by nbd/aoe?
| |
| 05:39 | <MrV> cyberorg, yeah, I get it
| |
| 05:39 | <alkisg> vagrantc: no, it's like aufs, we'd mount the combination of nbd+local cache
| |
| 05:40 | <vagrantc> ok.
| |
| 05:40 | <cyberorg> vagrantc, see that script, we would switch root after last step there
| |
| 05:40 | <vagrantc> well, i mean, it creates a virtual device that's composed of the other two devices, no?
| |
| 05:40 | <alkisg> Yes
| |
| 05:41 | <vagrantc> so you'd mount the virtual device as root...
| |
| 05:41 | <MrV> oh, Linux, you never stop surprising me! :)
| |
| 05:41 | <alkisg> We'd need the cow part too, on top of that virtual device
| |
| 05:43 | <vagrantc> ah, right.
| |
| 05:44 | you could use another device for the cow.
| |
| 05:44 | <cyberorg> how do you guys recommend distributing clients to different nbd servers? i've used pxe's files (gethostip) to direct clients to different servers
| |
| 05:45 | <alkisg> We're currently using swap space, wherever that is, it may as well be a local swap partition
| |
| 05:45 | <cyberorg> swap for cow?
| |
| 05:46 | <alkisg> cow = tmpfs, which caches to swap when necessary
| |
| 05:46 | khildin has joined IRC (khildin!~khildin@ip-213-49-84-229.dsl.scarlet.be) | |
| 05:49 | <alkisg> About distributing nbd, I've never done it, but I would do it based on DNS round robin etc, so that if a server gets down, that would somehow update DNS, and the NBD client on reconnection would reconnect to another server
| |
| 05:50 | (not sure if nbd-client in the initramfs accepts dns though)
| |
| 05:53 | cyberorg has left IRC (cyberorg!~cyberorg@opensuse/member/Cyberorg, Ping timeout: 255 seconds) | |
| 05:58 | alkisg has left IRC (alkisg!~alkisg@plinet.ioa.sch.gr, Changing host) | |
| 05:58 | alkisg has joined IRC (alkisg!~alkisg@ubuntu/member/alkisg) | |
| 06:00 | <MrV> alkisg, no, cyberorg was saying about different nbd-servers for different machines/networks
| |
| 06:00 | <alkisg> MrV: ah, I thought his 10 nbd-servers and 1000 clients are on the same network
| |
| 06:01 | Then why "distribute"?
| |
| 06:01 | Ah, to "send the nbd image"?
| |
| 06:01 | <MrV> then maybe I was wrong :)
| |
| 06:01 | <alkisg> No he said "distributing clients"...
| |
| 06:01 | OK
| |
| 06:01 | <MrV> his solution with pxe files seems elegant to me...
| |
| 06:02 | <alkisg> 1000 pxe files?
| |
| 06:02 | I'd use lua scripting in pxelinux instead of going there
| |
| 06:02 | <MrV> but I don't quite get it how did he change pxe.NETWORK to make it boot from different nbd-server?
| |
| 06:02 | <alkisg> And do a `modulo 10` on the client mac address or ip...
| |
| 06:02 | dead_inside1 has left IRC (dead_inside1!~taylor@76.75.3.174, Ping timeout: 276 seconds) | |
| 06:03 | <MrV> in my case - 7 networks, 7 pxe files
| |
| 06:03 | <alkisg> nbdroot=xxx
| |
| 06:04 | NeonLicht has joined IRC (NeonLicht!~NeonLicht@darwin.ugr.es) | |
| 06:08 | <MrV> nbdroot=NBDSERVER_IP:ltsp_i386
| |
| 06:08 | ?
| |
| 06:16 | dead_inside has joined IRC (dead_inside!~taylor@76.75.3.174) | |
| 06:18 | alexxtasi has joined IRC (alexxtasi!~alex@unaffiliated/alexxtasi) | |
| 06:19 | <alkisg> Yup, or nbd-ip:/opt/ltsp/i386 in newer ltsp versions
| |
| 06:19 | Or plain nbd-ip if you're using the correct rootpath
| |
| 06:20 | <MrV> OK. Thank you.
| |
| 06:23 | alkisg is now known as work_alkisg | |
| 06:24 | NeonLicht has left IRC (NeonLicht!~NeonLicht@darwin.ugr.es, Ping timeout: 252 seconds) | |
| 06:25 | cyberorg has joined IRC (cyberorg!~cyberorg@opensuse/member/Cyberorg) | |
| 06:35 | Yaann has joined IRC (Yaann!~yleger@online.vlq16.iliad.fr) | |
| 06:50 | <wouter> work_alkisg: just for reference, recent LTSP uses the nbd-client nbdroot support, right?
| |
| 07:04 | work_alkisg is now known as alkisg | |
| 07:05 | <alkisg> Hi wouter
| |
| 07:06 | If you have 5 mins, I can list all the related issues we have about that... /me checks the code...
| |
| 07:06 | First: ./hooks/nbd:force_load nbd
| |
| 07:06 | I think it shouldn't be forced, it creates nbd nodes even when nbd isn't used for root, swap etc
| |
| 07:07 | I think the scripts that need the nbd module, should ensure it's loaded
| |
| 07:08 | (i.e. use manual_add_modules instead of force_load)
| |
| 07:08 | Second, we have a workaround for https://bugs.launchpad.net/ubuntu/+source/nbd/+bug/696435
| |
| 07:09 | * vagrantc waves to wouter | |
| 07:09 | <alkisg> ...I don't think it should be solved in nbd though, but since it affects all netbooted NBD-based installations, you might want to take a look in case you can help in solving it
| |
| 07:10 | Hmmm, I think that's all, we use the upstream nbd initramfs scripts for everything else
| |
| 07:16 | gdi2k has left IRC (gdi2k!~gdi2k@120.28.224.109, Ping timeout: 255 seconds) | |
| 07:16 | gdi2k has joined IRC (gdi2k!~gdi2k@120.28.224.109) | |
| 07:17 | * wouter waves back | |
| 07:18 | polbosn has joined IRC (polbosn!02552fb2@gateway/web/freenode/ip.2.85.47.178) | |
| 07:19 | <polbosn> Καλήμερα σας. Μήπως υπάρχει τρόπος να κόβει ο 'Επόπτης' το internet απο τα τερματικά;
| |
| 07:20 | <alkisg> !greek | echo polbosn:
| |
| 07:20 | <ltsp> polbosn: greek: Στο παρόν κανάλι μιλάνε μόνο Αγγλικά, για υποστήριξη στα Ελληνικά από την υπηρεσία Τεχνικής Στήριξης ΣΕΠΕΗΥ διαβάστε το http://ts.sch.gr/wiki/IRC και στη συνέχεια πληκτρολογήστε /j #ts.sch.gr
| |
| 07:21 | <alkisg> https://answers.launchpad.net/epoptes/+question/247725
| |
| 07:21 | polbosn: δες την παραπάνω ερώτηση ^
| |
| 07:22 | <wouter> alkisg: just committed a fix for the force-load thing
| |
| 07:22 | alkisg: the wait-for-root issue seems to be a bug in wait-for-root, not in nbd
| |
| 07:22 | I'd recommend filing a bug there :-)
| |
| 07:22 | <polbosn> Sorry. Ok i will see your link. Thank you.
| |
| 07:22 | <alkisg> wouter: I filed it in the initramfs-tools
| |
| 07:22 | <wouter> alkisg: ah, heh
| |
| 07:22 | missed that, then :-)
| |
| 07:23 | <alkisg> wouter: what fix did you commit? where? link?
| |
| 07:23 | I.e. in the nbd source code?
| |
| 07:23 | * alkisg thinks it should be solved in initramfs, too... | |
| 07:23 | <wouter> http://anonscm.debian.org/gitweb/?p=users/wouter/nbd.git;a=commit;h=60c765cb12c64d0ed666688a6862c3cf80a9a296
| |
| 07:23 | <alkisg> Not as a shell workaround, but in the wait-for-root.c code
| |
| 07:24 | Cool, thanks for that
| |
| 07:24 | <wouter> yes, indeed
| |
| 07:24 | note that as initramfs-tools is developed in Debian, it might make more sense to file the bug there, rather than in ubuntu?
| |
| 07:24 | unless this is an ubuntu-specific bug, of course...
| |
| 07:24 | <alkisg> I think I've done that... let me check...
| |
| 07:28 | vagrantc: do you think we should also request maks to implement a configure_networking hook for us, e.g. to post-process net-eth0.conf for ltsp-specific defaults, before e.g. nbd continues mounting its root etc?
| |
| 07:28 | Or is the hackish shell wrapper that I mentioned good enough?
| |
| 07:29 | (ipconfig wrapper)
| |
| 07:29 | polbosn has left IRC (polbosn!02552fb2@gateway/web/freenode/ip.2.85.47.178, Quit: Page closed) | |
| 07:30 | <alkisg> Hmm it looks like I didn't file a bug for initramfs-tools on debian... maybe it has hard for me to test on debian back then...
| |
| 07:34 | vagrantc, wouter, is it easy for you to test debian+nbd without my /usr/share/initramfs-tools/scripts/local-top/nbd_ltsp hack?
| |
| 07:35 | <vagrantc> alkisg: i don't seem to remember needing it
| |
| 07:35 | <alkisg> vagrantc: isn't packaged by default?
| |
| 07:36 | * alkisg doesn't remember if it's included only in ubuntu... | |
| 07:36 | <vagrantc> alkisg: i think it's packaged, but i haven't seen it trigger cause FSTYPE is always set.
| |
| 07:37 | <alkisg> Hmm how do you check that FSTYPE is always set?
| |
| 07:37 | If it's triggered, then it writes param.conf, so FSTYPE is set after that
| |
| 07:37 | (for all following scripts)
| |
| 07:37 | <vagrantc> well, it worked fine without your hook.
| |
| 07:37 | <alkisg> Cool. So it's ubuntu-specific...
| |
| 07:37 | <vagrantc> maybe
| |
| 07:38 | alkisg: i'm speaking from memory, haven't tried
| |
| 07:38 | recenty at all
| |
| 07:38 | <alkisg> It would work, but delay 30 secs
| |
| 07:38 | * vagrantc sleeps | |
| 07:38 | vagrantc has left IRC (vagrantc!~vagrant@freegeek/vagrantc, Quit: leaving) | |
| 07:45 | * alkisg checks https://bugs.launchpad.net/ubuntu/+source/initramfs-tools/+bug/576429 which sounds very similar... | |
| 07:49 | <wouter> alkisg: I didn't see the 30-second delay in my tests back when I originally wrote the initramfs support, but then I never really tested plain-device booting, always used partitions
| |
| 07:49 | <alkisg> wouter: ah, you don't use squashfs?
| |
| 07:50 | (btw, the "no partition table found" nbd message does confuse quite a few users of ltsp...)
| |
| 07:51 | <wouter> alkisg: no, just ext4 or some such
| |
| 07:51 | alkisg: nothing I can do about the partition table thing. You'd get the same thing if you tried installing something to /dev/sda rather than /dev/sda1
| |
| 07:52 | <alkisg> Isn't that message generated by nbd-client?
| |
| 07:52 | <wouter> no, the kernel
| |
| 07:52 | why would nbd-client care what's on the device? It does a handshake with the server, and once that's done it just passes a socket to the kernel
| |
| 07:52 | <alkisg> Hmm I didn't see that in usb sticks that use the whole /dev/sdb as ext4... ok
| |
| 07:53 | * alkisg doesn't know what parts of nbd are userspace and what in the kernel... haven't read the nbd code | |
| 07:53 | <wouter> that might be because someone special-cased USB sticks? dunno
| |
| 07:53 | alkisg: negotiation is done in userspace. Everything else is done in the kernel.
| |
| 07:54 | if you were to generate partitioned NBD devices, you'd get rid of the message
| |
| 07:54 | but that might be a bit heavy-handed
| |
| 07:54 | <alkisg> Unfortunately mksquashfs doesn't support that
| |
| 07:55 | <wouter> you could dd it into an image... but like I said "heavy-handed"
| |
| 07:55 | NeonLicht has joined IRC (NeonLicht!~NeonLicht@darwin.ugr.es) | |
| 07:55 | <wouter> I'm not suggesting you do it, just pointing out the obvious workaround
| |
| 07:55 | <alkisg> Yeah, some fat client images are 5+ GB, hard to copy around...
| |
| 07:58 | gdi2k_ has joined IRC (gdi2k_!~gdi2k@120.28.219.253) | |
| 08:00 | gdi2k has left IRC (gdi2k!~gdi2k@120.28.224.109, Ping timeout: 258 seconds) | |
| 08:03 | bennabiy has left IRC (bennabiy!~Thunderbi@96-37-209-0.dhcp.leds.al.charter.com, Ping timeout: 240 seconds) | |
| 08:07 | <alkisg> !nbd-client
| |
| 08:07 | <ltsp> nbd-client: To try mounting the NBD image from the client initramfs: nbd-client 192.168.67.1 -N /opt/ltsp/i386 /dev/nbd0
| |
| 08:09 | * alkisg notes that wait-for-root isn't available in wheezy, so vagrantc wouldn't have seen the issue there | |
| 08:13 | <alkisg> I think the problem is that this udev attribute is missing for nbd and sr in initramfs: E: ID_FS_TYPE=iso9660
| |
| 08:13 | For nbd it's also missing in the booted system
| |
| 08:13 | So this issue probably needs to be solved in the udev rules
| |
| 08:14 | root@ltsp159:/run# udevadm info --query=all /dev/nbd0 | grep FS_TYPE
| |
| 08:14 | root@ltsp159:/run# udevadm info --query=all /dev/sr0 | grep FS_TYPE
| |
| 08:14 | E: ID_FS_TYPE=iso9660
| |
| 08:14 | root@ltsp159:/run# /usr/lib/initramfs-tools/bin/wait-for-root /dev/nbd0 1 && echo yes
| |
| 08:14 | root@ltsp159:/run# /usr/lib/initramfs-tools/bin/wait-for-root /dev/sr0 1 && echo yes
| |
| 08:14 | iso9660
| |
| 08:14 | yes
| |
| 09:37 | F-GT has joined IRC (F-GT!~phantom@ppp59-167-136-109.static.internode.on.net) | |
| 10:43 | adrianorg has left IRC (adrianorg!~adrianorg@179.182.73.26, Read error: Connection reset by peer) | |
| 10:44 | bennabiy has joined IRC (bennabiy!~Thunderbi@96-37-209-0.dhcp.leds.al.charter.com) | |
| 10:47 | adrianorg has joined IRC (adrianorg!~adrianorg@179.182.77.177) | |
| 11:27 | Grembler has joined IRC (Grembler!~Ben@cpc29-aztw22-2-0-cust128.18-1.cable.virginm.net) | |
| 11:34 | telex has left IRC (telex!~telex@freeshell.de, Remote host closed the connection) | |
| 11:36 | telex has joined IRC (telex!~telex@freeshell.de) | |
| 11:46 | flo1546796 has joined IRC (flo1546796!~flo154679@2001:660:5001:156:d53:bff9:7f6e:c57a) | |
| 11:46 | flo1546796 has joined IRC (flo1546796!~flo154679@unaffiliated/flo1546796) | |
| 11:51 | willou has left IRC (willou!5d5d2b9b@gateway/web/freenode/ip.93.93.43.155, Ping timeout: 240 seconds) | |
| 11:59 | <alkisg> wouter: https://bugs.freedesktop.org/show_bug.cgi?id=62565
| |
| 11:59 | I think it's nbd-specific after all
| |
| 12:00 | So, the solution should go in nbd, either properly in the kernel module, or as a shell workaround in the initramfs
| |
| 12:04 | Phantomas has joined IRC (Phantomas!~Phantomas@ubuntu/member/phantomas) | |
| 12:07 | Phantomas1 has joined IRC (Phantomas1!~Phantomas@ubuntu/member/phantomas) | |
| 12:07 | Phantomas has left IRC (Phantomas!~Phantomas@ubuntu/member/phantomas, Disconnected by services) | |
| 12:07 | Phantomas1 is now known as Phantomas | |
| 12:13 | <alkisg> If not, then we should file it in initramfs-tools in Debian, so that they try to special-case nbd in wait-for-root.c...
| |
| 12:15 | <ogra_> does debian use wait-for-root ?
| |
| 12:16 | <alkisg> I'm not sure, I'm guessing it should be there in jessie
| |
| 12:16 | It's not there in wheezy
| |
| 12:17 | <ogra_> it definitely comes from ubuntu ... not sure debian ever adopted it
| |
| 12:17 | (but so does initramfs-tools ... and nowadays ubuntu is always behind on that one :P )
| |
| 12:22 | mealstrom has joined IRC (mealstrom!~Thunderbi@46.63.71.254) | |
| 12:44 | alkisg is now known as work_alkisg | |
| 12:47 | <wouter> mm, is there a way for me to tell udev "inspect this device again, stuff has changed" from userspace?
| |
| 12:57 | gbaman has joined IRC (gbaman!~gbaman@host81-130-112-2.in-addr.btopenworld.com) | |
| 13:00 | <ogra_> udevadmin trigger
| |
| 13:00 | iirc
| |
| 13:01 | gbaman has left IRC (gbaman!~gbaman@host81-130-112-2.in-addr.btopenworld.com, Client Quit) | |
| 13:02 | gbaman has joined IRC (gbaman!~gbaman@host81-130-112-2.in-addr.btopenworld.com) | |
| 13:50 | mealstrom has left IRC (mealstrom!~Thunderbi@46.63.71.254, Ping timeout: 265 seconds) | |
| 13:55 | <wouter> ogra_: I meant by using some sort of syscall
| |
| 13:55 | calling system() from nbd-client is probably a bad idea :-)
| |
| 13:55 | ah well, I suppose I'll have to prod Paul, then
| |
| 13:56 | <ogra_> oh, that i dont knwo :)
| |
| 14:04 | <||cw> wouter: I bet the source of udevadmin is available for review...
| |
| 14:14 | alexxtasi has left IRC (alexxtasi!~alex@unaffiliated/alexxtasi) | |
| 14:17 | flo1546796 has left IRC (flo1546796!~flo154679@unaffiliated/flo1546796, Quit: Quitte) | |
| 14:22 | mealstrom has joined IRC (mealstrom!~Thunderbi@46.63.21.12) | |
| 14:42 | servercbe has joined IRC (servercbe!~servercbe@190.212.241.252) | |
| 14:43 | <wouter> ||cw: fair point
| |
| 14:43 | but I suspect the interfaces aren't public
| |
| 14:43 | <servercbe> hi, good morning. I am using ltsp with 22 pc.. but some user when they go to internet the system send them to the loggin screen, why?
| |
| 14:43 | thanks in advance
| |
| 14:44 | <||cw> servercbe: running out of ram?
| |
| 14:44 | <servercbe> i am using 16GB
| |
| 14:44 | only 5 gb using
| |
| 14:45 | right now
| |
| 14:45 | <||cw> if you login as that user locally on the server's gui console, doe sit do the same thing?
| |
| 14:48 | <servercbe> i dont know... i will do it now.. i will come back soon.. wait please
| |
| 14:59 | Hi, I came back... I works fine if I access to the internet localy on the server}
| |
| 15:01 | should I erase this account and redo it?
| |
| 15:01 | <||cw> i don't even know where to start on that one then
| |
| 15:02 | <servercbe> neigther do I.
| |
| 15:02 | <||cw> I don't know if account delete would do anything, or if it could a client hardware issue, or what
| |
| 15:03 | though I guess it can't hurt too much to recreate the account
| |
| 15:03 | alkisg and vagrantc have a lot more experience, so ask again if you haven't figured it out when they come back
| |
| 15:04 | <servercbe> thanks!
| |
| 15:06 | khildin has left IRC (khildin!~khildin@ip-213-49-84-229.dsl.scarlet.be, Remote host closed the connection) | |
| 15:07 | servercbe has left IRC (servercbe!~servercbe@190.212.241.252, Quit: Saliendo) | |
| 15:24 | mealstrom has left IRC (mealstrom!~Thunderbi@46.63.21.12, Ping timeout: 258 seconds) | |
| 15:32 | cyberorg has left IRC (cyberorg!~cyberorg@opensuse/member/Cyberorg, Remote host closed the connection) | |
| 15:35 | cyberorg has joined IRC (cyberorg!~cyberorg@opensuse/member/Cyberorg) | |
| 15:38 | mealstrom has joined IRC (mealstrom!~Thunderbi@46.63.63.163) | |
| 16:01 | Grembler has left IRC (Grembler!~Ben@cpc29-aztw22-2-0-cust128.18-1.cable.virginm.net, Quit: I Leave) | |
| 16:07 | mealstrom has left IRC (mealstrom!~Thunderbi@46.63.63.163, Ping timeout: 258 seconds) | |
| 16:08 | mealstrom has joined IRC (mealstrom!~Thunderbi@46.63.63.163) | |
| 16:47 | vagrantc has joined IRC (vagrantc!~vagrant@freegeek/vagrantc) | |
| 17:20 | khildin has joined IRC (khildin!~khildin@ip-213-49-84-229.dsl.scarlet.be) | |
| 17:33 | dead_inside has left IRC (dead_inside!~taylor@76.75.3.174, Ping timeout: 252 seconds) | |
| 17:37 | lifeboy has joined IRC (lifeboy!69ecc7b3@gateway/web/freenode/ip.105.236.199.179) | |
| 17:38 | <lifeboy> 'evenin all
| |
| 17:39 | <vagrantc> mornin'
| |
| 17:39 | <lifeboy> Anyone have an idea why nbd_server starts with system startup and gives me the error: Could not open exported file /opt/ltsp/images/i386.img: Permission denied. When I then restart the service, all works correctly
| |
| 17:39 | dead_inside has joined IRC (dead_inside!~taylor@76.75.3.174) | |
| 17:40 | <vagrantc> that's a new one
| |
| 17:41 | is it on a separate partition and mounted after startup tries to start nbd-server?
| |
| 17:41 | er, before
| |
| 17:41 | <lifeboy> No, just one partition I think. Let me make sure.
| |
| 17:41 | <vagrantc> what are the permissions on /opt/ltsp/images/i386.img ?
| |
| 17:42 | <lifeboy> Yes, just one partition.
| |
| 17:42 | <vagrantc> is it running as a different user during startup than when you restart the service?
| |
| 17:43 | <lifeboy> -rw-r----- 1 root root 332587008 Mar 18 08:55 /opt/ltsp/images/i386.img
| |
| 17:43 | I restart using sudo, so I would say it's root both times
| |
| 17:44 | Is it possible to let nbd_server do more verbose logging maybe?
| |
| 17:45 | <vagrantc> on my debian system, nbd-server is running as user "nbd"
| |
| 17:45 | <lifeboy> hmmm
| |
| 17:46 | <vagrantc> your permissions on i386.img seem too restrictive
| |
| 17:46 | are you trying to hide something in there?
| |
| 17:46 | (and then exporting it over the network anyways?)
| |
| 17:46 | sudo chmod +r /opt/ltsp/images/i386.img
| |
| 17:46 | then i bet it'll work correctly
| |
| 17:47 | <lifeboy> No particular reason it has those permissions that I can recall
| |
| 17:47 | nbd_server runs as user "nbd" on my side too
| |
| 17:47 | <vagrantc> maybe you have a very unusual umask?
| |
| 17:47 | <lifeboy> Actually I do, now that I think about it.
| |
| 17:48 | <vagrantc> well, make it world-readable, and then it should work.
| |
| 17:48 | <lifeboy> ok, will do
| |
| 17:48 | <vagrantc> although having a umask that restrictive will break lots of things...
| |
| 17:49 | <lifeboy> I was "forced" by a policy to ensure that all new files are not readable my anyone except the user that created it (for security reasons)
| |
| 17:50 | Put of course this should really apply to user's home directories, not system directories
| |
| 17:50 | dead_inside has left IRC (dead_inside!~taylor@76.75.3.174, Ping timeout: 250 seconds) | |
| 17:50 | <lifeboy> s/Put/But/
| |
| 17:51 | I see /opt has 0007 umask...
| |
| 17:52 | <vagrantc> eeyk.
| |
| 17:52 | so any random user can rewrite all of /opt ... nice :P
| |
| 17:55 | <lifeboy> no, 007 is the umask, not the file permissionss
| |
| 18:02 | dead_inside has joined IRC (dead_inside!~taylor@76.75.3.174) | |
| 18:03 | <vagrantc> umask is the default file permissions on created files... so i don't get what you mean, then.
| |
| 18:03 | files do not have a umask...
| |
| 18:04 | or maybe i misunderstand the term...
| |
| 18:05 | <lifeboy> I se the system default umask to 007 instead of 022 in the user's .profile
| |
| 18:05 | I set...
| |
| 18:14 | <vagrantc> so /opt has some file permissions resulting from a 007 umask.
| |
| 18:18 | telex has left IRC (telex!~telex@freeshell.de, Ping timeout: 240 seconds) | |
| 18:23 | telex has joined IRC (telex!~telex@freeshell.de) | |
| 18:34 | <lifeboy> Yes, which means for sure "others" can't read or list the file, but owner and group can if the program that creates the file sets the permissions to allow that.
| |
| 18:34 | I have no set the the permissions for i386.img to 0755 and will restart the server now and see if the problem is resolved.
| |
| 18:35 | I hove now set...
| |
| 18:39 | <vagrantc> you might also need to adjust the permissions of /opt, /opt/ltsp and /opt/ltsp/images as well
| |
| 18:40 | also, if you created the chroot with these permissions, the image might have bunk permissions as well
| |
| 18:41 | so i'd "mount /opt/ltsp/images/i386.img /mnt" and make sure the permissions look sane
| |
| 18:41 | lifeboy: sounds likely you'll need to set up some exceptions for your policy...
| |
| 19:22 | <lifeboy> I can exclude "root" from the policy, so only the actual user's new files will have the restrictive umask
| |
| 19:26 | After the restart, nbd_server is fine now and I didn't have to manually restart it. I'll work on the umask for root, so a future build of i386.img will not have this problem again.
| |
| 19:26 | Thanks, vagrantc!
| |
| 19:27 | <vagrantc> glad to hear it's working
| |
| 19:33 | lifeboy has left IRC (lifeboy!69ecc7b3@gateway/web/freenode/ip.105.236.199.179, ) | |
| 20:18 | khildin has left IRC (khildin!~khildin@ip-213-49-84-229.dsl.scarlet.be, Quit: I'm gone, bye bye) | |
| 20:50 | Ark74 has left IRC (Ark74!~Ark74@189.214.42.128.cable.dyn.cableonline.com.mx, Ping timeout: 255 seconds) | |
| 21:00 | Ark74 has joined IRC (Ark74!~Ark74@189.214.42.128.cable.dyn.cableonline.com.mx) | |
| 21:05 | professor_ has joined IRC (professor_!b1027b42@gateway/web/freenode/ip.177.2.123.66) | |
| 21:06 | Ark74 has left IRC (Ark74!~Ark74@189.214.42.128.cable.dyn.cableonline.com.mx, Ping timeout: 250 seconds) | |
| 21:12 | gbaman has left IRC (gbaman!~gbaman@host81-130-112-2.in-addr.btopenworld.com, Remote host closed the connection) | |
| 21:18 | Ark74 has joined IRC (Ark74!~Ark74@189.214.42.128.cable.dyn.cableonline.com.mx) | |
| 21:23 | gbaman has joined IRC (gbaman!~gbaman@host81-130-112-2.in-addr.btopenworld.com) | |
| 21:32 | imox has joined IRC (imox!~imox@p57A1F229.dip0.t-ipconnect.de) | |
| 21:43 | lmds_ has left IRC (lmds_!~lmds@tui.pi-et-ro.net, Ping timeout: 240 seconds) | |
| 21:44 | lmds_ has joined IRC (lmds_!~lmds@tui.pi-et-ro.net) | |
| 22:37 | calisto has joined IRC (calisto!~calisto@unaffiliated/calisto) | |
| 22:40 | * vagrantc wonders if it would make sense to support drbd as a root device | |
| 22:40 | <vagrantc> then you could actually have multiple servers, and reboot one of them without taking the clients down
| |
| 22:41 | <calisto> hi, I have a problem I install ltsp, and cant login has a user
| |
| 22:41 | I create users in the machine
| |
| 22:42 | I testing on raspberry and works, but
| |
| 22:42 | a pc booting from the net
| |
| 22:42 | Ark74 has left IRC (Ark74!~Ark74@189.214.42.128.cable.dyn.cableonline.com.mx, Ping timeout: 258 seconds) | |
| 22:43 | <calisto> ldm ask the user and password, and tellme unable to run /usr/bin/startlxde
| |
| 22:43 | lxde are installed on the real machine
| |
| 22:44 | but not on the chroot, can help me
| |
| 22:45 | <vagrantc> if you've configured them as thin clients, lxde would be installed on the server
| |
| 22:45 | if you've configured them as fat clients, you'd install lxde in the chroot ...
| |
| 22:47 | calisto: what exactly says it's "unable to run /usr/bin/startlxde"?
| |
| 22:47 | <calisto> vagrantc: I have a amdPC booting from the net and a raspberry pi booting from SD
| |
| 22:48 | the amdPC is a thin client?
| |
| 22:49 | vagrantc: startlxde is the message tellme ldm when I try to login
| |
| 22:50 | <vagrantc> i've never seen LDM report something like that ... maybe something in Xsession.d hooks
| |
| 22:50 | calisto: could you select a different session at login?
| |
| 22:51 | <calisto> only can loguin as xterm session
| |
| 22:51 | <vagrantc> but it works?
| |
| 22:51 | <calisto> yes
| |
| 22:52 | <vagrantc> does the xterm say "user@SERVER" or "user@ltspXXX" ?
| |
| 22:53 | <calisto> second option
| |
| 22:53 | user@ltsp83
| |
| 22:53 | <vagrantc> looks like it's configured as a fat client
| |
| 22:54 | you'll either need to set LTSP_FATCLIENT=false or install LXDE into the chroot
| |
| 22:55 | <calisto> ok vagrantc, but where set LTSP_FATCLIENT=false?
| |
| 22:55 | <vagrantc> !lts.conf | echo calisto
| |
| 22:55 | <ltsp> calisto lts.conf: http://manpages.ubuntu.com/lts.conf
| |
| 22:55 | <vagrantc> calisto: what distro?
| |
| 22:56 | <calisto> debian
| |
| 22:57 | <vagrantc> then you should be able to add it directly to /opt/ltsp/*/etc/lts.conf
| |
| 22:58 | Phantomas has left IRC (Phantomas!~Phantomas@ubuntu/member/phantomas, Ping timeout: 265 seconds) | |
| 22:58 | <vagrantc> unless you're using NBD, then you'll want to edit /var/lib/tftpboot/ltsp/*/lts.conf or /srv/tftp/ltsp/*/lts.conf
| |
| 22:58 | Phantomas has joined IRC (Phantomas!~Phantomas@ubuntu/member/phantomas) | |
| 23:02 | <calisto> yes....... works 8-)
| |
| 23:04 | thanks vagrantc
| |
| 23:15 | telex has left IRC (telex!~telex@freeshell.de, Remote host closed the connection) | |
| 23:16 | telex has joined IRC (telex!~telex@freeshell.de) | |
| 23:22 | gbaman has left IRC (gbaman!~gbaman@host81-130-112-2.in-addr.btopenworld.com, Remote host closed the connection) | |
| 23:48 | imox has left IRC (imox!~imox@p57A1F229.dip0.t-ipconnect.de, Quit: imox) | |
| 23:53 | gbaman has joined IRC (gbaman!~gbaman@host81-130-112-2.in-addr.btopenworld.com) | |
| 23:59 | gbaman has left IRC (gbaman!~gbaman@host81-130-112-2.in-addr.btopenworld.com, Ping timeout: 245 seconds) | |