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


Channel log from 11 February 2012   (all times are UTC)

00:01
<vagrantc>
for some reason it's hardcoded in client/initramfs-scripts.d/common/00-getltsconffile
00:02
NBD_ROOT_HOST=server
00:02freedomrun has left IRC (freedomrun!~quassel@89.142.250.116, Read error: Connection reset by peer)
00:05
<vagrantc>
so that's not going to work
00:06
we have sure implemented a maze of scripts. :)
00:15andy__ has left IRC (andy__!~andy@h247.95.31.71.dynamic.ip.windstream.net, Ping timeout: 240 seconds)
00:23mikkel has left IRC (mikkel!~mikkel@80-71-132-15.u.parknet.dk, Quit: Leaving)
00:32
<vagrantc>
looks like i've got some work to do on the debian kiosk plugin
01:08rthomson has left IRC (rthomson!~rthomson@mars.pet.ubc.ca, Quit: Reached EOW)
01:20
<vagrantc>
highvoltage: well, i've almost got a few to poke at ltsp-cluster*
01:22jeffdag has joined IRC (jeffdag!cef8b169@gateway/web/freenode/ip.206.248.177.105)
01:43Parker955_Away is now known as Parker955
01:52bengoa has joined IRC (bengoa!~bengoa@187-27-31-53.3g.claro.net.br)
01:59
<vagrantc>
stgraber: other than the one issue with NBD_ROOT_HOST=server hardcoded in client/initramfs-scripts.d/common/00-getltsconffile, ltsp-trunk is working for me.
02:00
although i need to switch to using kioskSession and screen.d/kiosk for my kiosk implementation
02:00* vagrantc wonders why kioskSession doesn't create the user on the fly
02:01jeffdag has left IRC (jeffdag!cef8b169@gateway/web/freenode/ip.206.248.177.105)
02:02jeffdag has joined IRC (jeffdag!cef8b169@gateway/web/freenode/ip.206.248.177.105)
02:06
<jeffdag>
Hi gang, just discovered ltsp recently, been successfully messing with it (on ubuntu) the last 2 days... one thing remains that doesn't fit too well with the way we need to work. Is it possible for the client to export a usb key as a block device instead of files using fuse? Maybe export the raw /dev/sdx using nbd back to the ltsp-server.
02:07bengoa has left IRC (bengoa!~bengoa@187-27-31-53.3g.claro.net.br)
02:09
<vagrantc>
jeffdag: it's technically possible, sure. it may have consequences like handling unmounts cleanly and such.
02:24
<stgraber>
vagrantc: cool
02:31
<vagrantc>
stgraber: i'm guessing that the NBD_ROOT_HOST=server breaks lts.conf downloading, or rather, doesn't work.
02:58jeffdag has left IRC (jeffdag!cef8b169@gateway/web/freenode/ip.206.248.177.105, Quit: Page closed)
03:17Gershwin has left IRC (Gershwin!~Donkey@unaffiliated/gershwin)
03:38
<vagrantc>
!ltsp-cluster
03:38
<ltsp`>
vagrantc: Error: "ltsp-cluster" is not a valid command.
03:38
<vagrantc>
!ltsp-cluster-agent
03:38
<ltsp`>
vagrantc: Error: "ltsp-cluster-agent" is not a valid command.
03:56adrianorg_ has left IRC (adrianorg_!~adrianorg@186.213.152.248, Ping timeout: 276 seconds)
05:31Parker955 is now known as Parker955_Away
05:40staffencasa_ has left IRC (staffencasa_!~staffenca@128-193-150-248.oregonstate.edu, Ping timeout: 272 seconds)
06:07alkisg has joined IRC (alkisg!~alkisg@ubuntu/member/alkisg)
06:10
<alkisg>
Good morning
06:27
stgraber: I installed the new ltsp-server package, but /etc/ltsp/ltsp-update-image.conf is still there, are all the other changes there, should I go ahead and test, or should I wait for another package?
06:49
Looking at the fatclients page... After a few releases, it gets messy, with a whole lot of workaround for each release.
06:51
We should at least remove the comments for unsupported ubuntu versions, but I'd like it even more if we removed all comments for releases before the last released LTS version too. :)
06:52
OK, running ltsp-build-client...
06:54
stgraber: does LDM need synching from Debian?
07:34
<knipwim>
alkisg: if i do a mount --rbind /dev/ $root/dev, the dev/pts and dev/shm get mounted also
07:38
<alkisg>
knipwim: sounds good, but I've no idea if there are any cases where people would want to mount dev and not pts, or vice versa
07:39
I'm ok with always mounting everything... .:D
07:40
stgraber: also the upstart jobs are still there
07:41
<knipwim>
alkisg: me neither
07:42
we could always get it back in if such a use case arises
07:42
<alkisg>
If vagrantc agrees, we could also make it the default to mount anything, and if people don't want that, we can provide a single option to not mount those dirs...
07:42
s/anything/everything
07:43
<knipwim>
a --mount-none :)
07:43
<alkisg>
Yup :)
07:43
Or even "-d" ==> don't mount /dev, etc
07:43
To reverse their meanings to "don't mount those"
07:45
knipwim: if you push the rbind change, then remove the --dev-pts option, I don't think it makes sense then to have both of them. Or at least make the mount-all option not do both of those.
07:47
<knipwim>
check
07:47
will do that later, and also update the man page
07:48
when my server boots again
07:48
<alkisg>
knipwim: what is gentoo using? nfs + bindmoutns?
07:48
<knipwim>
yup
07:48
<alkisg>
If vagrantc switches to nfs + tmpfs + aufs, can you switch there too?
07:49
It'll be easier to assume a completely writeable root
07:50
<knipwim>
is that the new initramfs based boot method?
07:50
<alkisg>
No... it's like that:
07:51
Previously, nfs had problems with tmpfs+aufs
07:51
So that's why bindmounts were used, as the last alternative
07:51
Now it appears that there aren't any problems anymore, as long as the nfs export is indeed read-only
07:51
So, if, at boot time, one creates a tmpfs,
07:51
and uses aufs to bind it with the read-only nfs export,
07:52
then he creates a completely writeable root, with changes stored on the client
07:52
So there's no need to maintain special bintmounts lists, and scripts can go ahead and write wherever they like without thinking about read-only root
07:52
vagrantc is thinking of testing that, and if it works ok, switching debian to default to that
07:53
<knipwim>
but aufs is not in the kernel
07:53
where do you get the module?
07:54
<alkisg>
Ubuntu previously depended on some aufs package, now it uses unionfs
07:55
Ah no, aufs was included in the ubuntu kernel
07:55
Let me see if it's included in the debian kernel too...
07:56
<knipwim>
it's not in the gentoo kernel, but in a package
07:56
install the package and rebuild the kernel
07:56
<alkisg>
Yup, it's there: http://packages.debian.org/squeeze/i386/linux-image-2.6.32-5-486/filelist
07:57
<knipwim>
but i believe it will be doable for gentoo
07:57
<alkisg>
But now Ubuntu started using unionfs, as it's supposed to go mainline in the kernel or something
07:57||cw has left IRC (||cw!~chris@phpgroupware/cw, Ping timeout: 244 seconds)
07:57
<alkisg>
Ah sorry, overlayfs
07:57
I got the name wrong
07:57
<knipwim>
omg, finally
07:57
<alkisg>
I don't know if that works over nfs+tmpfs
07:57
<knipwim>
some kind of overlayfs in the kernel
07:58
<alkisg>
Yeah..
07:58
<knipwim>
my first ltsp install was with a custom unionfs kernel
07:58
what a hassle
07:59
<alkisg>
Debating overlayfs - http://lwn.net/Articles/447650/
07:59||cw has joined IRC (||cw!~chris@gateway.wilsonmfg.com)
07:59||cw has joined IRC (||cw!~chris@phpgroupware/cw)
08:01
<knipwim>
i don't think removing any generic bind mounts code will break gentoo
08:01
we use our own bind_mounts function in the initscript
08:01
<alkisg>
And you keep your own list for RW_DIRS etc?
08:03
<knipwim>
yup
08:05
<alkisg>
Cool
08:18khildin has joined IRC (khildin!~khildin@ip-80-236-223-208.dsl.scarlet.be)
08:27
<alkisg>
stgraber: the ltsp package in your PPA still doesn't use the new initramfs-scripts.d system, so I can't test anything... we need to decide soon if the new system goes in for Precise or not. If so, we'll need to remove ltsp_nbd and pull in a new NBD package, or at least to patch ltsp_nbd.
08:37||cw has left IRC (||cw!~chris@phpgroupware/cw, Ping timeout: 240 seconds)
08:38||cw has joined IRC (||cw!~chris@gateway.wilsonmfg.com)
08:39||cw has joined IRC (||cw!~chris@phpgroupware/cw)
08:45komunista has joined IRC (komunista!~slavko@adsl-195-098-005-032.dynamic.nextra.sk)
08:51khildin has left IRC (khildin!~khildin@ip-80-236-223-208.dsl.scarlet.be, Quit: I'm gone, bye bye)
08:58cyberorg has joined IRC (cyberorg!~cyberorg@opensuse/member/Cyberorg)
09:19komunista has left IRC (komunista!~slavko@adsl-195-098-005-032.dynamic.nextra.sk, Ping timeout: 248 seconds)
09:22alkisg has left IRC (alkisg!~alkisg@ubuntu/member/alkisg, Quit: Leaving.)
09:23alkisg has joined IRC (alkisg!~alkisg@ubuntu/member/alkisg)
09:40vagrantc has left IRC (vagrantc!~vagrant@c-76-115-60-19.hsd1.or.comcast.net, Ping timeout: 276 seconds)
10:25adrianorg_ has joined IRC (adrianorg_!~adrianorg@186.213.152.248)
10:37knipwim has left IRC (knipwim!~wim@ip4da83870.direct-adsl.nl, Read error: Operation timed out)
10:38khildin has joined IRC (khildin!~khildin@ip-80-236-223-208.dsl.scarlet.be)
10:42knipwim has joined IRC (knipwim!~wim@ip4da83870.direct-adsl.nl)
11:02khildin has left IRC (khildin!~khildin@ip-80-236-223-208.dsl.scarlet.be, Quit: I'm gone, bye bye)
11:35
<knipwim>
alkisg: i'm going to commit the devpts remove
11:36
unless you're busy with a commit spree :)
11:36
<alkisg>
knipwim: hehe no I'm ok
11:36
I always pull before pushing
11:39
<knipwim>
hmm
11:39
the mount --rbind's won't unmount
11:40
with the umount_marked function i think
11:40
<alkisg>
knipwim: let me test...
11:41
<knipwim>
only the explicit mounts are stored, while dev/pts and dev/shm are mounted implicitly
11:41
might be a way around that
11:41
the gentoo installer uses /proc/mounts to determine what is mounted on the ltsp chroot
11:42
so no need to store anything
11:42
<alkisg>
What if someone runs ltsp-chroot twice, in 2 different terminals?
11:43
Or if someone runs ltsp-chroot while ltsp-build-client is approaching the end of installation...
11:43
<knipwim>
they should exit simultaniously :)
11:43
but true, no easy fix for those cases
11:43
<alkisg>
I don't know but I don't like the idea of unmounting everything below /opt/ltsp/i386 even if we didn't mount it ourselves
11:45
<knipwim>
perhaps a uniq on the proc/mounts output?
11:47
the current umount_marked doesn't unmount a specific mount, just the one stored earlier
11:48
so the second ltsp-chroot on the same chroot holds the same marked_mounts
11:48
<alkisg>
knipwim: err wait I didn't get the idea. So, if we use rbind, then how will we know what it actually mounted, so that we unmount it?
11:49
(01:48:02 μμ) knipwim: so the second ltsp-chroot on the same chroot holds the same marked_mounts ==> yes, and that's the correct thing to do, isn't it? And each one will unmount the dirs it mounted, even if they're duplicates...
11:50
<knipwim>
so a doing a uniq on /proc/mounts will work also i gather
11:51
<alkisg>
knipwim: that's what I don't get
11:51
So if a user manually mounts /opt/ltsp/i386/some/dir, will we unmount it?
11:52
Because it shows up in uniq?
11:52
<knipwim>
each unique mount is available in /proc/mounts, each mount from an ltsp-chroot will show up there seperately
11:52
ahh, check, i understand
11:53
using the proc/mounts only accounts for ltsp-chroot
11:53
not "third-party" mounts
11:53
<alkisg>
Yup
11:53
<knipwim>
never mind then :)
11:53
<alkisg>
:)
11:53
<knipwim>
about the rbind
11:54
the mount man says it bind mounts the dir, and mounted subdirs
11:54
<alkisg>
Yes if there's no way to "umount --rbind", maybe it doesn't help us
11:54
<knipwim>
the mounted subdirs can be checked on the host system
11:55
<alkisg>
But again "third-party" mounts will be a problem
11:55
If mount --rbind reported the dirs it actually mounts, that would help
11:56
<knipwim>
perhaps in the case of dev mounts the third-party mounts won't be much of a problem
11:57
<alkisg>
Maybe it'd be saner to do the opposite, i.e. implement rbind ourselves by grepping /proc/mounts to see what else is mounted under /dev,
11:57
...but I think I still like the simpler approach of manually mounting /dev/pts in a separate step
11:59
<knipwim>
i coming to think as well
11:59
mount -v --rbind doesn't show which submounts are done
12:01
<alkisg>
knipwim, so let's simplify only the part where "--mount-dev" means "in ubuntu, mount devpts too, in gentoo mount shm too" etc?
12:01
I.e. merge the --mount-dev-pts option into --mount-dev?
12:05[GuS] has joined IRC ([GuS]!~MysT@190.55.139.179)
12:05[GuS] has joined IRC ([GuS]!~MysT@unaffiliated/gus/x-663402)
12:07
<knipwim>
alkisg: gentoo doesn't need the shm i think, didn't need it before at least
12:08
i can mount the pts by default
12:08
<alkisg>
So it's the same as ubuntu then?
12:08
Nice
12:08
<knipwim>
within --mount-dev
12:08
<alkisg>
No need for separation between those
12:10
<knipwim>
still, the marked_mounts are umounted in the same order as they're mounted
12:10
<alkisg>
knipwim, no in opposite
12:11
marked_mounts="newmount $oldmounts" or something
12:11
The new mount gets in fornt
12:11
front
12:11
<knipwim>
yes
12:11
i'm seeing it now
12:12
nice
12:29[GuS] has left IRC ([GuS]!~MysT@unaffiliated/gus/x-663402, Remote host closed the connection)
12:36alkisg has left IRC (alkisg!~alkisg@ubuntu/member/alkisg, Quit: Leaving.)
12:44Trixboxer has joined IRC (Trixboxer!~Trixboxer@115.124.115.71)
13:06toscalix has joined IRC (toscalix!~toscalix@85.137.146.26.dyn.user.ono.com)
13:08toscalix has left IRC (toscalix!~toscalix@85.137.146.26.dyn.user.ono.com, Remote host closed the connection)
13:10toscalix has joined IRC (toscalix!~toscalix@85.137.146.26.dyn.user.ono.com)
13:12toscalix has left IRC (toscalix!~toscalix@85.137.146.26.dyn.user.ono.com, Remote host closed the connection)
13:19toscalix has joined IRC (toscalix!~toscalix@85.137.146.26.dyn.user.ono.com)
13:30toscalix has left IRC (toscalix!~toscalix@85.137.146.26.dyn.user.ono.com, Read error: Connection reset by peer)
13:33toscalix has joined IRC (toscalix!~toscalix@85.137.146.26.dyn.user.ono.com)
13:35toscalix has left IRC (toscalix!~toscalix@85.137.146.26.dyn.user.ono.com, Remote host closed the connection)
13:45toscalix has joined IRC (toscalix!~toscalix@85.137.146.26.dyn.user.ono.com)
14:53killermike has left IRC (killermike!~killermik@2.26.103.208, Remote host closed the connection)
15:07alkisg has joined IRC (alkisg!~alkisg@ubuntu/member/alkisg)
15:56alkisg has left IRC (alkisg!~alkisg@ubuntu/member/alkisg, Quit: Leaving.)
16:23markit has joined IRC (markit!~insegnant@dynamic-adsl-84-223-84-219.clienti.tiscali.it)
16:24markit has left IRC (markit!~insegnant@dynamic-adsl-84-223-84-219.clienti.tiscali.it, Remote host closed the connection)
16:26freedomrun has joined IRC (freedomrun!~quassel@89.142.248.83)
16:27Parker955_Away is now known as Parker955
16:33komunista has joined IRC (komunista!~slavko@adsl-195-098-005-032.dynamic.nextra.sk)
16:40ogra_ has left IRC (ogra_!~ogra@p5098ed03.dip0.t-ipconnect.de, Ping timeout: 276 seconds)
16:40markit has joined IRC (markit!~insegnant@dynamic-adsl-84-223-84-219.clienti.tiscali.it)
16:43toscalix has left IRC (toscalix!~toscalix@85.137.146.26.dyn.user.ono.com, Remote host closed the connection)
16:43
<markit>
hi, anyone expert of epoptes here? server does not start
16:55
mm solved, I had to purge first
16:55markit has left IRC (markit!~insegnant@dynamic-adsl-84-223-84-219.clienti.tiscali.it, Remote host closed the connection)
17:19freedomrun_ has joined IRC (freedomrun_!~quassel@89.142.248.83)
17:20freedomrun has left IRC (freedomrun!~quassel@89.142.248.83, Ping timeout: 245 seconds)
17:22komunista has left IRC (komunista!~slavko@adsl-195-098-005-032.dynamic.nextra.sk, Ping timeout: 260 seconds)
17:30alkisg has joined IRC (alkisg!~alkisg@ubuntu/member/alkisg)
17:37
<alkisg>
What if we used init=/usr/sbin/boot-ltsp instead of using initramfs-scripts.d? We could do a regular NFS or NBD boot, without anything ltsp related, and then create the tmpfs+aufs+everything else there... isn't that much simpler?
17:37
We could even do that *without* initramfs... I'm guessing that will help in arm
17:39
<stgraber>
alkisg: hey
17:39
<alkisg>
Hi stgraber
17:39
<stgraber>
alkisg: sorry, didn't see you e-mail to ltsp-dev before I replied to you in private ;)
17:39
<alkisg>
You're right, I wasn't using your ppa, I saw that a few minutes after I send you the mail
17:39
<stgraber>
alkisg: you'll have a new nbd in Ubuntu in 5min or so
17:40
<alkisg>
I rebuilded the chroot, and booted the client. It booted, but some services delayed the boot for 1-2 minutes
17:40
(configure networking... waiting 60 seconds... etc)
17:40
<stgraber>
alkisg: uploaded
17:40
<alkisg>
So we need to grow the RM_SYSTEM_SERVICES list
17:41
<stgraber>
so new ldm and new nbd in a few minutes in Ubuntu
17:41
<alkisg>
stgraber: thanks!
17:42
<stgraber>
alkisg: and I also uploaded a new snapshot of trunk to the PPA, just going to bump the build score now so it builds quickly
17:42
<alkisg>
stgraber: since any of the initramfs-scripts.d stuff hasn't landed yet, I'm really thinking we should rename it to initscripts.d,
17:43
and use init=/usr/sbin/boot-ltsp
17:43
That will do the cow stuff, call the initscripts, and then exec /sbin/init
17:43loather has joined IRC (loather!~khudson@wsip-98-175-250-115.sd.sd.cox.net)
17:43
<alkisg>
No special "ltsp" kernel parameter needed, no boot=ltsp_nbd, nothing
17:43
We can even do that with the old nbd, afaik
17:44
I can try it and maybe have it up and working by monday
17:44
It's not a big change
17:44
What do you think?
17:44
<stgraber>
right, so the only difference from a regular boot to a LTSP thin client boot would be the added init=/usr/sbin/boot-ltsp?
17:44
<alkisg>
Yup
17:45
I just tried init=/bin/bash with a regular system, and then exec /sbin/init, and it worked fine
17:45
<stgraber>
that should work, as long as make really sure not to polute the environment or keep anything open, otherwise the exec will likely fail
17:45
<alkisg>
Everything is already mounted at that point
17:45
So we don't even need to mount --bind /proc etc
17:45
We just modify/delete a few files, and exec /sbin/init
17:46
We can even leave the cow mounting in the initramfs, so that it's a smaller change
17:46
<stgraber>
sounds good
17:47
<alkisg>
OK, thanks, I'll do some tests, commit innocent stuff, and talk with you and vagrantc about the init= tests on monday
17:48
<stgraber>
sounds good. Having vagrant's opinion would be nice indeed
17:49
alkisg: btw, any idea what's going on with the ldm theme? (assuming you also have the issue where you get the ltsp theme and not the ubuntu one)?
17:49
<alkisg>
stgraber: I got the ubuntu theme...
17:49
...both on my fat chroot and my thin chroot
17:49
Both builded today
17:49
<stgraber>
lucky you ;)
17:49
<alkisg>
Hehe
17:50
<stgraber>
ok, I guess I'll test again then, maybe it got fixed at some point in trunk and I didn't see it
17:51
I'll wait for the new nbd and ldm to be in the archive and for ltsp to be built in the PPA and will get a clean chroot building
18:08bengoa has joined IRC (bengoa!~bengoa@2001:1291:229:2:216:cbff:feab:6cc9)
18:08bengoa has left IRC (bengoa!~bengoa@2001:1291:229:2:216:cbff:feab:6cc9)
18:12vedranm has left IRC (vedranm!~vedranm@inf2.ffri.hr, *.net *.split)
18:12vedranm has joined IRC (vedranm!~vedranm@inf2.ffri.hr)
18:22
<alkisg>
Yup, init=/sbin/ltsp-init, with contents==`#!/bin/sh; mount -o remount,rw /; rm /etc/init/tty3.conf; mount -o remount,ro /; exec /sbin/init "$@"`, works fine on standalone systems. So the idea seems to be valid.
18:31
I'm not seeing the ubuntu logo though, I don't know if it affects plymouth somehow.
18:51markit has joined IRC (markit!~marco@88-149-177-66.staticnet.ngi.it)
19:24komunista has joined IRC (komunista!~slavko@adsl-195-098-005-032.dynamic.nextra.sk)
19:56priyank has joined IRC (priyank!~priyank@122.177.175.7)
19:57
<priyank>
Hi i am getting TFTP File not found error
19:58
and I am not able to boot my thin client
19:58
<alkisg>
Distro/version?
19:58
<priyank>
ubuntu
19:58
11.10
19:58
<alkisg>
Version?
19:58
OK
19:58
!tftp
19:58
<ltsp`>
alkisg: tftp: Here's a page to help you troubleshoot TFTP problems in Ubuntu: https://help.ubuntu.com/community/UbuntuLTSP/Troubleshooting/TFTP
20:03
<priyank>
hi I am not able to guess the problem mind remote desktop?
20:09
also how to bridge server and thin client
20:12
alkisg: Please reply
20:16
<alkisg>
priyank: no, I don't have time right now to help you with remote desktop, but I spend a lot of time writing that wiki page to help people with tftp problems, so it'd be nice if you followed that and replied here with what step you're having problems with
20:17
<priyank>
ok thanks
20:32Trixboxer has left IRC (Trixboxer!~Trixboxer@115.124.115.71, Remote host closed the connection)
21:01priyank has left IRC (priyank!~priyank@122.177.175.7, Remote host closed the connection)
21:21alexqwesa__ has left IRC (alexqwesa__!~alex@2001:5c0:1400:b::320f, Ping timeout: 245 seconds)
21:21alexqwesa__ has joined IRC (alexqwesa__!~alex@109.172.15.11)
21:43vagrantc has joined IRC (vagrantc!~vagrant@75-150-46-245-Oregon.hfc.comcastbusiness.net)
21:54komunista has left IRC (komunista!~slavko@adsl-195-098-005-032.dynamic.nextra.sk, Quit: Leaving.)
21:54
<alkisg>
Hey vagrantc, got a minute?
21:55
I was thinking, instead of calling the initramfs-scripts.d from the initramfs with chroot, to put init=/sbin/ltsp-init as a kernel parameter, run the scripts from there, and then exec /sbin/init to continue
21:56
I tried it and it seems like it's working, except for plymouth disappearing, maybe because it doesn't like "echo"es and prefers log_begin_msg etc instead
21:57
So, to boot an ltsp chroot as an ltsp chroot, we wouldn't put "ltsp" or other magic kernel parameters, just init=/sbin/ltsp-init
21:57
And to boot it as a normal system, we'd just remove that
21:58
If you agree, I'll rename initramfs-scripts to ltsp-init, and initramfs-scripts.d to ltsp-init.d, and will put an `exec /sbin/init` line at the bottom of ltsp-init
21:58
stgraber thought it sounds good...
22:00
<vagrantc>
oalknot a lot of a minute
22:00
but messing with init sounds scarey.
22:02
<alkisg>
Everywhere I tried it it worked
22:02
live cds, normal systems etc
22:03
What we need is to run some stuff in the real root, before the initscripts start running... isn't that the point if the init= kernel parameter? :D
22:04
<stgraber>
vagrantc: it's basically like chaining a second initramfs, not anymore scary than that ;)
22:04freedomrun_ has left IRC (freedomrun_!~quassel@89.142.248.83, Remote host closed the connection)
22:05
<vagrantc>
except that the initramfs software we're usinng is thouroughly and widely tested
22:05
<stgraber>
indeed :)
22:05
<alkisg>
...though the initramfs-scripts.d is widely broken right now... :)
22:05
<stgraber>
in theory there's no risk of execing another script at the end of initramfs (instead of /sbin/init), the exec /sbin/init
22:05
so long as we make sure not to make any more of a mess than we do in the iniramfs
22:05
<alkisg>
...so if we're to do extensive tests, I think it's a good time to test both of them
22:06
<stgraber>
mostly thinking of fds/processes/environment
22:06
<alkisg>
The initramfs-scripts.d scripts only mess with files... they don't run daemons or other weird stuff
22:06
<vagrantc>
i got initramfs-scripts.d stuff working great yesterday, just had to patch out the hard-coded NBD_ROOT_HOST
22:06
<stgraber>
alkisg: the fact that this breaks plymouth is a bit concerning though, it shows that "something" is wrong as plymouth shouldn't die when doing another exec
22:07
<alkisg>
In Ubuntu the main breakage come from removing RC_WHITELIST and introducing RM_SYSTEM_SERVICES, I'll try to remove more services, now it waits 60 secs to configure the network etc - at the end it boots
22:08
stgraber: I did put an "echo" though at the test script... I wonder if plymouth doesn't like that and it prefers the log_begin_msg set of functions...
22:08
<stgraber>
alkisg: you'll actually need to add something instead of removing something to avoid the 60s wait
22:08
alkisg: initctl emit static-network-up
22:08
and that'll avoid the 60s
22:09
<alkisg>
OK, I'll give that a try tomorrow
22:09
<stgraber>
needs to be done after upstart is started though
22:09
<alkisg>
The initramfs-scripts.d vs ltsp-init isn't a big change... if I get the rest working smoothly tomorrow, you two can decide if you want to switch to that or not
22:10
The scripts will be the same in any case
22:10
stgraber: to minimize the initramfs changes, I'll just put a chroot $rootmnt /usr/share/ltsp/initramfs-scripts at the end of ltsp_nbd
22:10
<stgraber>
sounds good
22:10
<alkisg>
...and we can explore the new NBD script in 12.10 or later
22:11
<stgraber>
yeah, that way nbd-proxy should still work too ;)
22:11
(for these who want it)
22:11
<alkisg>
It would work with the new nbd script too,
22:11
I asked wouter and he put support for an NBDCLIENT env var
22:11
So with NBDCLIENT=nbd-proxy-wrapper, it would be possible
22:12
<stgraber>
ah cool, yeah, we'd need to write a small wrapper but that'd work
22:12
<alkisg>
it would launch that instead of nbd-client
22:12
stgraber: since we're doing changes though, and since you pulled the new nbd script, I think we should switch to that new nfs-like syntax that you proposed
22:12
I.e. nbdroot=server:port/path
22:13
or nbdroot=[ipv6]:port/path
22:13
I can do that tomorrow too, if you agree
22:13
<stgraber>
yeah, that'd make sense
22:13
<alkisg>
OK, lots of things for tomorrow then :)
22:14
Thanks guys, /me goes to bed...
22:15alkisg has left IRC (alkisg!~alkisg@ubuntu/member/alkisg, Quit: Leaving.)
22:17* stgraber thinks we'll need quite a few bugfix releases for all that stuff before we can release 12.04 in April ;)
22:18
<vagrantc>
well, i'm happy with initramfs-scripts.d now.
22:33vagrantc has left IRC (vagrantc!~vagrant@75-150-46-245-Oregon.hfc.comcastbusiness.net, Quit: Changing server)
22:35markit has left IRC (markit!~marco@88-149-177-66.staticnet.ngi.it, )
22:36vagrantc has joined IRC (vagrantc!~vagrant@75-150-46-245-Oregon.hfc.comcastbusiness.net)
22:38vagrantc has joined IRC (vagrantc!~vagrant@75-150-46-245-Oregon.hfc.comcastbusiness.net)
23:00vagrantc has left IRC (vagrantc!~vagrant@75-150-46-245-Oregon.hfc.comcastbusiness.net, Quit: leaving)
23:00vagrantc has joined IRC (vagrantc!~vagrant@75-150-46-245-Oregon.hfc.comcastbusiness.net)
23:03epaphus has left IRC (epaphus!~epaphus@200.122.149.9, Disconnected by services)
23:03epaphus has joined IRC (epaphus!~epaphus@200.122.149.9)
23:05epaphus2 has joined IRC (epaphus2!~epaphus@200.122.149.9)
23:07andy__ has joined IRC (andy__!~andy@h247.95.31.71.dynamic.ip.windstream.net)
23:08Parker955 is now known as Parker955_Away
23:11Steve_The_Pirate has joined IRC (Steve_The_Pirate!~Gary@cpc16-haye17-2-0-cust169.haye.cable.virginmedia.com)
23:42freedomrun has joined IRC (freedomrun!~quassel@89.142.248.83)
23:48alexqwesa__ has left IRC (alexqwesa__!~alex@109.172.15.11, Quit: Хана X'ам !!!)