IRC chat logs for #ltsp on irc.libera.chat (webchat)
Channel log from 16 May 2019 (all times are UTC)
| 02:23 | kjackal has joined IRC (kjackal!~quassel@2a02:587:311c:9d00:70ee:5f34:4c95:b67b) | |
| 02:31 | kjackal has left IRC (kjackal!~quassel@2a02:587:311c:9d00:70ee:5f34:4c95:b67b, Ping timeout: 258 seconds) | |
| 04:07 | <alkisg> ...and morning all :D
| |
| 04:51 | Adding a new image now involves these 2 lines in boot.ipxe:
| |
| 04:51 | (in menu): item tara-cinnamon tara-cinnamon
| |
| 04:51 | (and the jump tag): :tara-cinnamon
| |
| 04:51 | The loading command then is the same for any image name: set cmdline root=/dev/nfs nfsroot=${srv}:/srv/ltsp/${img} loop.max_part=9
| |
| 04:52 | Variables make things so much easier! Why were we using syslinux for so long!
| |
| 04:52 | No need to configure/restart nbd, no need to change anything in dnsmasq...!
| |
| 04:55 | ricotz has joined IRC (ricotz!~ricotz@ubuntu/member/ricotz) | |
| 05:20 | <quinox> sweet
| |
| 05:23 | <alkisg> vagrantc: I'm thinking to stop using arch-based names (i386/amd64) and switch to codenames. E.g. if a VM or chroot is named "buster", we'll use that for TFTP too. And for chrootless we can either demand a name parameter or default to the codename as read from /etc/os-release.
| |
| 05:29 | os_a has joined IRC (os_a!~Thunderbi@195.112.116.22) | |
| 05:59 | kjackal has joined IRC (kjackal!~quassel@2a02:587:311c:9d00:70ee:5f34:4c95:b67b) | |
| 06:10 | <alkisg> bcg: hi, could you tell me the kernel cmdline of a fedora/rhel ltsp client? I'm testing "squashfs image over NFS" on dracut, but it doesn't like this much: root=/dev/nfs nfsroot=ip:/path (and then a dracut ltsp hook will mount -o loop /sysroot/image /sysroot)
| |
| 06:15 | <bcg> ltsp/x86_64/pxelinux.cfg/default: APPEND initrd=initramfs-ltsp.img init=/sbin/init-ltsp selinux=0 rw root=dhcp ip=eth0:dhcp net.ifnames=0 rd.net.timeout.carrier=20
| |
| 06:16 | kjackal has left IRC (kjackal!~quassel@2a02:587:311c:9d00:70ee:5f34:4c95:b67b, Ping timeout: 258 seconds) | |
| 06:21 | <alkisg> bcg: thanks; I did manage to boot fedora30 over nfs+squashfs file, but manually, by typing my script in the rd.break=pre-pivot stage
| |
| 06:21 | I used the same cmdline as the other distros: root=/dev/nfs nfsroot=ip:path; but I also had to symlink /dev/nfs to /sysroot manually
| |
| 06:22 | So I think that a dracut ltsp hook can do all that. It'll be great to use the same default cmdline across all distros.
| |
| 06:24 | OK since it's doable, and I'm not familiar with dracut, I'll leave that hook implementation either for you or for after GSoC etc, we'll manage. At this point I basically wanted to verify that this default can work in all distros.
| |
| 06:49 | <bcg> with root=dhcp you can specify nfs path in dhcpd.conf. for my use cases it's better, everything in one file.
| |
| 06:51 | <alkisg> bcg: your case doesn't work for proxydhcp, which is the most common case, AND you still need to modify your pxe configuration to pass kernel parameters,
| |
| 06:51 | while, by putting everything in boot.ipxe, it's again everything in a single file, AND the previous things work
| |
| 06:51 | And it works cross-distro too
| |
| 06:51 | It's a win-win
| |
| 06:53 | (and it supports automatic redirection to grub for loading 32bit kernels under 64bit uefi clients and it supports menus and it easily supports multiple kernels and images a lot of other use cases that dhcp.conf doesn't)
| |
| 06:54 | * vagrantc notices alkisg's promotion to head of marketing for ipxe | |
| 06:54 | <alkisg> :D
| |
| 06:55 | What can I say, I love $variables!
| |
| 06:55 | <vagrantc> it does appear to fix a lot into one place
| |
| 06:55 | <alkisg> vagrantc: xenial booted fine; stretch has a glitch; do we care for it?
| |
| 06:56 | buster boots fine (still talking about nfs+ext4 image, stock vm without nbd-client or ltsp code)
| |
| 06:56 | <vagrantc> alkisg: what sort of glitch on stretch?
| |
| 06:56 | <alkisg> It might be related to the nfs kernel issues with overlay; maybe ubuntu backported some fix there and debian didn't
| |
| 06:56 | <vagrantc> if it's easy, fix it, if it's hard ... can't backport everything :)
| |
| 06:56 | <alkisg> It complains it can't find /root/sbin/init, while it's there
| |
| 06:57 | <vagrantc> huh
| |
| 06:57 | yeah, i found overlay fs to be not quite workable with NFS
| |
| 06:57 | <alkisg> That's what we'll default to now though!
| |
| 06:57 | <vagrantc> like many things, haven't tried in a while
| |
| 06:57 | i thought you were doing an image over NFS ?
| |
| 06:57 | that's a good deal different.
| |
| 06:58 | <alkisg> And of course an overlay on top of that, how else
| |
| 06:58 | So it's still overlay over nfs; although it's a loop mounted image over nfs, which should be faster due to only needing 1 file over nfs
| |
| 06:58 | <vagrantc> yeah
| |
| 06:58 | <alkisg> Not a lot of metadata involved
| |
| 06:59 | OK at this point I think I'll go on, and come back for stretch + fedora in the end, if I have enough time
| |
| 07:03 | Hehe, all this work about getting the cmdline from the chroots, with update-kernels and pxelinux.cfg, and it was possible to use the same cmdline for all distros all along :D
| |
| 07:05 | <vagrantc> well, it certainly wasn't always possible
| |
| 07:06 | <alkisg> Didn't aufs work with nfs?
| |
| 07:06 | The root=/dev/nfs parameter is documented in the kernel, I think it worked in a cross-distro manner since ages
| |
| 07:08 | <vagrantc> ipxe was first uploaded to Debian in 2011 ... and wasn't a very old project at the time
| |
| 07:08 | <alkisg> Oh this isn't related to ipxe
| |
| 07:08 | <vagrantc> and probably didn't have all those features
| |
| 07:08 | <alkisg> It's doable from any loader
| |
| 07:08 | <vagrantc> root=/dev/nfs didn't work in initramfs-tools all along
| |
| 07:09 | i made a bunch of patches to initramfs-tools to get it to work well
| |
| 07:09 | kjackal has joined IRC (kjackal!~quassel@onopfy.static.otenet.gr) | |
| 07:48 | <alkisg> stretch had this issue: initramfs-tools (0.122) unstable; urgency=medium . * [4fa3fe4] init: Validate the real init after devtmpfs is mounted under the real root (Closes: #811479)
| |
| 07:49 | Hmm maybe not, it's supposed to have 0.130 :/
| |
| 07:52 | SYS64738 has joined IRC (SYS64738!~jhonny5@159.213.93.166) | |
| 07:57 | statler has joined IRC (statler!~Georg@gwrz3.lohn24.de) | |
| 08:01 | <vagrantc> i seem to recall that issue being ping-pong-ed a few times.
| |
| 08:10 | <alkisg> vagrantc: well, if I put a `mount --bind /dev /root/dev` command, it works fine, so it's that issue again
| |
| 08:11 | I don't think it's worth the effort to special-case stretch and apply that work around in the ltsp code though; if someone needs stretch he'd have to solve that via initramfs-tools
| |
| 08:11 | * alkisg resists the temptation to try jessie | |
| 08:12 | <alkisg> So stretch, buster, xenial, bionic etc all work fine, and fedora30 with the commands ran manually. I'll write some code to backport nfs+imagefile to ltsp5, so that people here can test it.
| |
| 08:16 | <vagrantc> alkisg: there already was support for ltsploop and all that in initramfs-tools hooks for ltsp
| |
| 08:18 | <alkisg> vagrantc: nice, so your code there may be able to handle it without ltsp5 users needing any new code. I'll try it in the afternoon.
| |
| 08:19 | # Also needs a dummy /opt/ltsp/images/sbin/init-ltsp to be an executable file # to work around a bug/feature in initramfs-tools's nfs mount checks... ?
| |
| 08:20 | vagrantc: doesn't the loop mounting happen before the initramfs-tools validates init?
| |
| 08:21 | bcg has left IRC (bcg!~b@dfx4btyyyyyyyyyyyyyyt-3.rev.dnainternet.fi, Quit: bcg) | |
| 08:26 | bcg has joined IRC (bcg!~b@dfx4btyyyyyyyyyyyyyyt-3.rev.dnainternet.fi) | |
| 08:34 | <vagrantc> yeah
| |
| 08:35 | alkisg: er, no ... the problem was it mounts the NFS directory as the rootfs, and then we re-mount the loop device later
| |
| 08:35 | it's in the initramfs-tools hooks in ltsp
| |
| 09:02 | GodFather_ has left IRC (GodFather_!~rcc@143.59.184.72, Ping timeout: 264 seconds) | |
| 09:17 | stellasolitaria has joined IRC (stellasolitaria!~jhonny5@159.213.93.166) | |
| 09:20 | SYS64738 has left IRC (SYS64738!~jhonny5@159.213.93.166, Ping timeout: 244 seconds) | |
| 10:21 | <alkisg> vagrantc: if the remounting is done at nfs-bottom, and init validation happens after that, I don't see why a dummy file is needed
| |
| 10:21 | I didn't need it in any of the VMs I tried
| |
| 10:21 | (with my code; I haven't tested ltsploop)
| |
| 10:22 | <vagrantc> your code is clearly better, then :)
| |
| 10:22 | <alkisg> No no I don't think so
| |
| 10:22 | I'm thinking maybe you tested in wheezy or jessie or something,
| |
| 10:22 | and things have been fixed since then
| |
| 10:22 | <vagrantc> it was probably stretch while it was still in development
| |
| 10:23 | it was summer of 2015
| |
| 10:23 | <alkisg> Gotcha. I'll test your code on a 18.04 lab tomorrow and report back.
| |
| 10:23 | (it's getting more and more difficult for me to test ltsp5; I modified my office setup too much!)
| |
| 10:24 | <vagrantc> heh
| |
| 10:24 | and when you get something working cleaner, it's hard to go back...
| |
| 10:24 | <alkisg> Right, even if nfs is a bit slower now, I won't switch my office back to nbd
| |
| 10:25 | This new setup allows me to netboot all my VMs without messing with nbd-server each time
| |
| 10:25 | <vagrantc> yay.
| |
| 10:26 | <alkisg> You were wise to endorse nfs for so long; you just didn't push us enough for the loopback thing :D
| |
| 10:30 | <vagrantc> i tried it out of desperation because i couldn't get NFS to work at the time and it seemed easier to go along with what everyone else was doing... :/
| |
| 10:31 | so it was only a quick hack proof of concept
| |
| 10:42 | kjackal has left IRC (kjackal!~quassel@onopfy.static.otenet.gr, Ping timeout: 245 seconds) | |
| 11:11 | kjackal has joined IRC (kjackal!~quassel@2a02:587:311c:9d00:70ee:5f34:4c95:b67b) | |
| 11:21 | stellasolitaria has left IRC (stellasolitaria!~jhonny5@159.213.93.166, Ping timeout: 250 seconds) | |
| 11:23 | SYS64738 has joined IRC (SYS64738!~jhonny5@159.213.93.166) | |
| 11:45 | GodFather_ has joined IRC (GodFather_!~rcc@96-92-43-9-static.hfc.comcastbusiness.net) | |
| 11:56 | Faith has joined IRC (Faith!~Paty_@unaffiliated/faith) | |
| 12:07 | SYS64738 has left IRC (SYS64738!~jhonny5@159.213.93.166, Ping timeout: 258 seconds) | |
| 12:50 | GodFather_ has left IRC (GodFather_!~rcc@96-92-43-9-static.hfc.comcastbusiness.net, Ping timeout: 268 seconds) | |
| 12:59 | SYS64738 has joined IRC (SYS64738!~jhonny5@159.213.93.166) | |
| 13:44 | <alkisg> vagrantc: your ltsploop code seems to work fine; I didn't need to create a fake init
| |
| 13:45 | mwalters, Hyperbyte, so here's the quick how to, for NFS+ltsploop:
| |
| 13:45 | 1) apt install nfs-kernel-server; ltsp-config nfs ==> exports /opt/ltsp over nfs
| |
| 13:46 | <vagrantc> alkisg: i might have nudged them to fix that code
| |
| 13:46 | <alkisg> 2) pxelinux.cfg/default: change cmdline from nbd to: root=/dev/nfs nfsroot=${srv}:/opt/ltsp/images ltsploop=/${img}.img
| |
| 13:47 | replace ${srv} with your server ip, and ${img} with your image name, e.g. amd64
| |
| 13:47 | 3) Reboot, test, report back; the hope is that it's a bit *faster* and a lot more stable
| |
| 13:47 | (or anyone else that wants to test) ^
| |
| 13:48 | You can also test rebooting the server; as long as the client hasn't logged in yet, and is only using nfs and not sshfs or nbd swap, it should continue normally
| |
| 13:49 | * alkisg loves it when he discovers bugs that vagrantc has solved years ago so no debian-based distro is affected anymore :D | |
| 13:50 | <alkisg> vagrantc: have you ever used netconsole? Wouldn't it be better for debugging ltsp client crashes, than remote syslog?
| |
| 13:52 | <vagrantc> i think you need to have network drivers compiled in
| |
| 13:53 | <alkisg> Hmm, I'll give it a test; I hope it works when they're loaded as modules too
| |
| 13:55 | mgariepy has left IRC (mgariepy!~mgariepy@ubuntu/member/mgariepy, Quit: Leaving) | |
| 14:03 | mgariepy has joined IRC (mgariepy!~mgariepy@ubuntu/member/mgariepy) | |
| 14:03 | jeblesh has left IRC (jeblesh!5dac146f@gateway/web/freenode/ip.93.172.20.111, Ping timeout: 256 seconds) | |
| 14:14 | <mwalters> lets see who is around today ;)
| |
| 14:16 | alright, no one at the one office
| |
| 14:16 | <alkisg> ;)
| |
| 14:17 | <mwalters> been getting called over to the day school we operate
| |
| 14:17 | casting is broken, the internet is broken, everything is broken
| |
| 14:17 | they're trying to watch netflix and its not working
| |
| 14:17 | :|
| |
| 14:18 | <alkisg> I went to a school to fix the internet too, they had problems for a week now. It turned out that a teacher removed an ethernet cable from a pc, put it to his laptop, and when he finished...
| |
| 14:18 | ...he put it to the switch, creating a loop
| |
| 14:18 | <mwalters> gahahahahaha
| |
| 14:18 | I worked for a division here, with a single broadcast domain between 7 buildings/schools
| |
| 14:19 | we had a fiber lan between all of them... at least once a month someone looped a switch and brought down the entire network
| |
| 14:19 | <alkisg> I wonder, if I had ran wireshark or something, would I have seen ethernet messages notifying about the loop?
| |
| 14:19 | <mwalters> each school came back to a central location, so we'd just unplug the media converters one by one until the network stopped storming and then tell them to go find it ;)
| |
| 14:20 | I was NOT the network admin there ;)
| |
| 14:20 | <alkisg> :)
| |
| 14:20 | <mwalters> I think cisco switches have loop protection at least
| |
| 14:20 | at least our sg300s have a checkbox for it... no idea if it works
| |
| 14:21 | I think I have like... 3 "dumb switches" throughout our entire org, tho
| |
| 14:21 | <vagrantc> there's some protocol to detect loops like that and isolate other switches
| |
| 14:21 | or even the looped ports
| |
| 14:21 | had some switches with that enabled ... but don't chain them together!
| |
| 14:22 | <mwalters> pxelinux.cfg/default will get overwritten by update-kernels?
| |
| 14:23 | <alkisg> vagrantc: how would I install a newer kernel in jessie? 3.16 doesn't have overlay. I think I can do this: apt install linux-image-4.9-686-pae; but in ubuntu I don't specify a version, it's like apt install linux-generic-hwe; does putting 4.9 there make sense?
| |
| 14:23 | mwalters: indeed; a longer process is needed to make sure that the cmdline persists
| |
| 14:24 | <mwalters> I have a bunch of entires, presumably because I haven't been cleaning things up... I guess I want to modify the first one? `label ltsp-NBD`?
| |
| 14:24 | <alkisg> mwalters: yes; ltsp currently creates entries for all kernels, but only the first one is used by default
| |
| 14:25 | Paste the final cmdline here, to make sure it's right
| |
| 14:25 | <mwalters> oh, I use a chroot, is that an issue?
| |
| 14:25 | <alkisg> Nope
| |
| 14:25 | GodFather has joined IRC (GodFather!~rcc@96-92-43-9-static.hfc.comcastbusiness.net) | |
| 14:26 | GodFather_ has joined IRC (GodFather_!~rcc@96-92-43-9-static.hfc.comcastbusiness.net) | |
| 14:26 | <mwalters> append ro initrd=initrd.img-4.15.0-48-generic init=/sbin/init-ltsp forcepae root=/dev/nfs nfsroot=10.0.0.3:/opt/ltsp/images ltsploop=amd64.img
| |
| 14:26 | <alkisg> I think ltsploop=amd64.img works, but to be safe, make it ltsploop=/amd64.img
| |
| 14:29 | <mwalters> read: connection refused
| |
| 14:29 | <alkisg> mwalters: did you install nfs-kernel-server etc?
| |
| 14:29 | <mwalters> yes
| |
| 14:30 | <alkisg> What's the output of this, on the server? showmount -e
| |
| 14:30 | <mwalters> /opt/ltsp *
| |
| 14:30 | <alkisg> Hmm, can I see with vnc?
| |
| 14:30 | !vnc-dide
| |
| 14:30 | <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.
| |
| 14:31 | <mwalters> yah one sec
| |
| 14:31 | <alkisg> ty
| |
| 14:33 | <mwalters> sorry
| |
| 14:33 | <alkisg> mwalters: ok you fix it :P
| |
| 14:33 | <mwalters> I'm a moron
| |
| 14:33 | too many subnets
| |
| 14:33 | <alkisg> (like you say to the teachers ;P)
| |
| 14:33 | <mwalters> wrong site
| |
| 14:33 | lol\
| |
| 14:34 | one day... we'll have enough bandwidth for a single server ;)
| |
| 14:35 | or I'll have the mental bandwidth to keep it all straight
| |
| 14:35 | booted and logged in just fine
| |
| 14:36 | ...when I entered the correct IP
| |
| 14:36 | <alkisg> mwalters: great, so, the idea is to let all clients use that new method
| |
| 14:36 | And check if it's more stable etc, so that we switch the default to that
| |
| 14:36 | <mwalters> yeah, gonna try it here with ap hysical client
| |
| 14:37 | that office only ever has like... 2 people in it
| |
| 14:37 | and zero at the moment
| |
| 14:37 | I'll boot all the local clients off tonight and force them to use it
| |
| 14:40 | <alkisg> Great, ty for testing
| |
| 14:41 | <mwalters> do I need to do anything funky for uefi clients?
| |
| 14:41 | <alkisg> If you already had them working, they should continue to
| |
| 14:42 | <mwalters> but they'll continue to use nbd, won't they?
| |
| 14:42 | <alkisg> Well it depends on how you're booting them
| |
| 14:42 | <mwalters> I need to modify the cmd line there
| |
| 14:42 | <alkisg> If you're booting them normally, they'll use nfs, like all of them
| |
| 14:42 | <mwalters> they're using grub
| |
| 14:42 | <alkisg> If you have a local kernel + grub installed, then yeah you'd need to update it there
| |
| 14:43 | <mwalters> time for more testing
| |
| 14:43 | lets see how badly I can break this ;)
| |
| 14:44 | so, I have my "root" param on the kernel entry, not the initrd entry... does it matter?
| |
| 14:44 | (for grub)
| |
| 14:46 | <alkisg> mwalters: I don't think the cmdline (root etc) ever goes to the initrd entry
| |
| 14:46 | I think it always goes with the kernel entry; or in an additional append line in pxelinux
| |
| 14:46 | <vagrantc> alkisg: you'd need to have a source of packages for jessie, and then you would probably install linux-image-X.Y.Z-N-VARIANT or just install the appropriate linux-image-VARIANT/CODENAME
| |
| 14:47 | the backport kernels get really long names
| |
| 14:47 | <mwalters> that makes sense
| |
| 14:47 | <alkisg> vagrantc: I went ahead and installed it; it's almost working; but chocking with ext4 read errors; possibly hit nfs/overlay issues
| |
| 14:48 | <mwalters> grub client booted fine
| |
| 14:49 | https://termbin.com/iy2y <--- a known good grub.cfg for reference
| |
| 14:50 | it definitely "feels" faster
| |
| 14:51 | <alkisg> I think it's cached better on the client side
| |
| 14:51 | <mwalters> desktop load, application time to launch, etc etc
| |
| 14:51 | hard to tell, this client is sitting on the same switch as an HD camera DVR too =/
| |
| 14:52 | but yeah, I'll leave it up here and let you know if it blows up ;)
| |
| 14:52 | <alkisg> Great
| |
| 14:52 | <mwalters> We have a mix a uefi/grub and pxeclients as well
| |
| 14:53 | the majority of our workload is in chrome, though... so not sure we'll be the best stress testers
| |
| 14:53 | <alkisg> web surfing is one of the most demanding tasks nowadays, so it's fine
| |
| 14:54 | <mwalters> I can't wait for web assembly
| |
| 14:54 | to be widely adopted ;)
| |
| 14:55 | <alkisg> And then CPUs will get hardware acceleration for that
| |
| 14:56 | And then new CPUs will only use web assembly instruction set
| |
| 14:56 | And we'll rediscover the 1970's :D
| |
| 14:59 | <mwalters> I mean, we've already rediscovered lisp!
| |
| 15:00 | <alkisg> and now ltsp!
| |
| 15:00 | <mwalters> lolol
| |
| 15:00 | I really like the idea of lisp/clojure...
| |
| 15:00 | ...but... I need to like... get stuff done today ;)
| |
| 15:05 | * vagrantc skipped sleeping in order to package a chain of guile dependencies today | |
| 15:07 | <vagrantc> not really a *good* idea....
| |
| 15:07 | but so close
| |
| 15:11 | * alkisg was wondering if vagrant was in a different timezone than usual | |
| 15:11 | <alkisg> OK, jessie booted too; it needed an initramfs-tools hack as well as kernel 4.9
| |
| 15:13 | <vagrantc> alkisg: i stepped outside of time
| |
| 15:13 | had to get into the right mindspace to churn through build dependencies on a language with a few too many parens
| |
| 15:14 | * alkisg is using hot tsipouro + honey for that | |
| 15:14 | <alkisg> https://en.wikipedia.org/wiki/Rakomelo
| |
| 15:18 | <mwalters> oh this sounds delicious
| |
| 15:19 | I haven't been able to find much Eastern European/Mediteranian foods/drinks since moving away from the North Eastern US =/
| |
| 15:20 | We had one "Greek" restaraunt close by, but it closed :(
| |
| 15:24 | The only saving grace is my Romanian friend who makes homemade plum wine ;)
| |
| 15:25 | tested a couple clients that aren't getting bombarded by this IP camera traffic... they definitely seem faster
| |
| 15:26 | alkisg: so this just mounts the root fs over nfs, right? Home dirs still use nbd?
| |
| 15:26 | <alkisg> mwalters: originally we have: nbd for /, and sshfs for /home
| |
| 15:26 | and nbd for swap
| |
| 15:27 | With the change we have: nfs for /, sshfs for /home, and nbd for swap
| |
| 15:27 | <mwalters> eer, yeah, that ;)
| |
| 15:27 | ok cool... yeah, "insecure" nfs seems perfectly fine for /
| |
| 15:27 | <alkisg> nbd is insecure too; the root file system is not encrypted in any of the ltsp cases in all those years
| |
| 15:28 | Only /home is encrypted with sshfs, and that doesn't change now
| |
| 15:28 | <mwalters> good to know, thanks
| |
| 15:31 | client recovers after server reboot too, this is nice
| |
| 15:31 | it was hard frozen while the server restarted, but came right back to life once the server was back up
| |
| 15:31 | <alkisg> But note that sshfs might be disconnected after server reboot, so that will only work on login screen, not on session
| |
| 15:31 | <mwalters> looks like it survived in this case
| |
| 15:32 | oh, nope
| |
| 15:32 | just cached lists
| |
| 15:32 | <alkisg> If you switch /home to *insecure* nfs, then this will work too
| |
| 15:33 | I.e. from sshfs to nfs
| |
| 15:33 | But that's not what we want to test now :)
| |
| 15:33 | <mwalters> yeah, I don't think that'll fly here ;)
| |
| 16:26 | SYS64738 has left IRC (SYS64738!~jhonny5@159.213.93.166, Ping timeout: 246 seconds) | |
| 16:33 | <alkisg> vagrantc: with NFS, what happens if someone runs ltsp-update-image and there are clients connected?
| |
| 16:35 | <vagrantc> depends on the nature of the image ...
| |
| 16:35 | adrianor1 has left IRC (adrianor1!~adrianorg@177.18.180.185, Ping timeout: 246 seconds) | |
| 16:35 | <vagrantc> if it's a delete-and-replace, it should work fine, as the inode should be held open
| |
| 16:35 | if it changes the live filesystem (e.g. ext4) ... that is not good.
| |
| 16:36 | <alkisg> vagrantc: well, ltsp-update-image completely deletes the old ext4 or squashfs image there and replaces it
| |
| 16:36 | It doesn't open it and change its contents
| |
| 16:37 | Ehm, sure, if someone has vbox running with the .vmdk file open, AND the clients running, that 's not good
| |
| 16:38 | but generally people will be exporting to squashfs, not using vms directly
| |
| 16:38 | So if I understood you correctly, we're good :)
| |
| 16:38 | <mwalters> I can tell you in a second ;)
| |
| 16:39 | <vagrantc> right
| |
| 16:39 | i definitely experiemented with it a bit back when i was using ext2 images
| |
| 16:40 | <alkisg> mwalters: try to e.g. change a letter in /etc/adduser.conf in the chroot, run ltsp-update-image, THEN cat /etc/adduser.conf on the client without rebooting it, it should still show the old one
| |
| 16:40 | Don't cat before, to avoid caching
| |
| 16:41 | <mwalters> getting rabbitholed... no dns from chroot anymore
| |
| 16:42 | probably more to do with my IPA deployment
| |
| 16:45 | I took a hammer to resolv.conf and I'm updating now ;)
| |
| 16:47 | the client seemed to survive the image update process
| |
| 16:48 | however ldm is not prompting me to reboot
| |
| 16:48 | (after logging out)
| |
| 16:49 | <alkisg> that ldm prompt is no longer valid
| |
| 16:53 | GodFather_ has left IRC (GodFather_!~rcc@96-92-43-9-static.hfc.comcastbusiness.net, Ping timeout: 255 seconds) | |
| 16:53 | GodFather has left IRC (GodFather!~rcc@96-92-43-9-static.hfc.comcastbusiness.net, Ping timeout: 252 seconds) | |
| 17:02 | adrianorg has joined IRC (adrianorg!~adrianorg@177.156.56.54) | |
| 18:46 | statler has left IRC (statler!~Georg@gwrz3.lohn24.de, Remote host closed the connection) | |
| 19:05 | xdrazen has joined IRC (xdrazen!b57393a4@gateway/web/freenode/ip.181.115.147.164) | |
| 19:06 | <xdrazen> hi there, how is the development right now? i am back from kinda 3 years out
| |
| 19:08 | <vagrantc> alkisg has been busy reinventing LTSP almost from scratch
| |
| 19:08 | if you had asked just 3 months ago, i would say development was looking kind of bleak
| |
| 19:08 | now i'm excited about the possibilities :)
| |
| 19:09 | <xdrazen> alkisg sounds me... really sounds me
| |
| 19:11 | ***before trying to do implements LTSP... right now the office where i work, uses windows and that, and i do not know why the boss wants to "spy" the pcs of the employees, right now, we are using tightvnc, but the screen goes blank, so it is a bit strange; we will use ultravnc and there it does not feel so bad
| |
| 19:12 | *** what i am asking is if, with LTSP and diskless workstations, is possible to that other user or even the admin could see what the clients are doing
| |
| 19:23 | <alkisg> !epoptes
| |
| 19:23 | <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
| |
| 19:23 | <alkisg> xdrazen: we've developed that one for ltsp ^
| |
| 19:26 | <xdrazen> <alkisg> your name... i remember it
| |
| 19:26 | **nick
| |
| 19:27 | OMFG then epoptes is what i was looking for, thank you very much. Now i have to make work office 2010 in wine
| |
| 19:28 | <mwalters> I always struggle with the boss who wants to spy... if you don't trust your employees, why did you hire them :(
| |
| 19:31 | <xdrazen> i am not even agree, but in this $#%#$ country, if you say that you will not do something, always is anyelse more that will do the same for less money... ; anyways i think it deppends more about paranoia or kinda
| |
| 19:35 | well, see you later, nice to see you again. Long live LTSP!
| |
| 19:39 | xdrazen has left IRC (xdrazen!b57393a4@gateway/web/freenode/ip.181.115.147.164, Ping timeout: 256 seconds) | |
| 20:56 | kjackal has left IRC (kjackal!~quassel@2a02:587:311c:9d00:70ee:5f34:4c95:b67b, Ping timeout: 258 seconds) | |
| 21:00 | ricotz has left IRC (ricotz!~ricotz@ubuntu/member/ricotz, Remote host closed the connection) | |
| 21:02 | Faith has left IRC (Faith!~Paty_@unaffiliated/faith, Quit: Leaving) | |
| 23:38 | kjackal has joined IRC (kjackal!~quassel@2a02:587:311c:9d00:70ee:5f34:4c95:b67b) | |
| 23:42 | kjackal has left IRC (kjackal!~quassel@2a02:587:311c:9d00:70ee:5f34:4c95:b67b, Ping timeout: 258 seconds) | |