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


Channel log from 21 February 2012   (all times are UTC)

00:01
<vagrantc>
alkisg: right, figured that out eventually
00:02
we really should split ltsp-localapps stuff into a separate package...
00:06
maybe i should've pushed that last commit to common ...
00:06
since several of our things depend on that working
00:09
maybe, just maybe i'll upload ltsp to debian today ... i feel like there's always one more thing.
00:11
<alkisg>
vagrantc: I think I'd prefer the symlinks the other way around, from Ubuntu to Debian...
00:12alexqwesa has left IRC (alexqwesa!~alex@109.172.15.11, Quit: Хана X'ам !!!)
00:12* vagrantc really wishes we could just dynamically merge common
00:12
<alkisg>
Who was against it?
00:12
<vagrantc>
i hate having to track every single change that way
00:13
gadi was worried about performance issues
00:13
<alkisg>
First of all, why can't we do that at the packaging stage?
00:13
<vagrantc>
alkisg: that's not a bad idea, really.
00:13
<alkisg>
I.e. not ship any common or Distro/ dirs, just a single init-ltsp.d dir
00:14mischko has left IRC (mischko!~schapman@205.247.25.89, Quit: Ex-Chat)
00:14floorduh has left IRC (floorduh!b85bfd7c@gateway/web/freenode/ip.184.91.253.124)
00:14
<vagrantc>
alkisg: thought not something i want to mess with today :)
00:14
<alkisg>
Hehe, yeah
00:15
<vagrantc>
but yes, that would be the right way to do it.
00:15
in fact, i think we talked about that back in october.
00:16
a little trickier to implement in packaging ... it's not just a simple directory copy
00:16
but totally worth the sanity.
00:17
so far, i've been doing fine just using Ubuntu's, as far as i can tell, with my one workaround
00:18
<alkisg>
vagrantc: I'd also prefer it if we only had a Debian/ dir and we had "ifs" in the code whenever we need to diverge...
00:19
<vagrantc>
alkisg: not sure how much i'd like that
00:19
<alkisg>
It would allow for a common package, like ldm
00:19
No need for 0ubuntux stuff
00:20
<vagrantc>
i do like that it would encourage less divergence :)
00:20
but it also makes it easier for release teams to review code and ignore changes in dirs that essenntially won't affect us
00:20
like importing a new upstream version during a freeze
00:21
to have separate dirs, that is...
00:21
well, actually because of all the symlinks it would drive them crazy.
00:21
hm.
00:21
there are a number of plugins that are totally irrelevent for each distro, or wildly different ideas about what they should look like.
00:22
<alkisg>
I don't think that currently we have wildly different ideas
00:22floorduh has joined IRC (floorduh!b85bfd7c@gateway/web/freenode/ip.184.91.253.124)
00:22
<alkisg>
Examples of irrelevent plugins?
00:22floorduh is now known as chngnme
00:23
<vagrantc>
defaulting to using NBD squashfs images, for one...
00:23
no-recommendss
00:24
locales
00:24chngnme is now known as floorduh
00:24
<vagrantc>
anything having to do with mirror handling
00:24
<alkisg>
I would classify those to "requiring an if "$VENDOR" = "Debian", else...", not as irrelevant plugins
00:25
<vagrantc>
sure, they're where we diverge, but some of which would be really ugly as iff statements.
00:25
artwork
00:25
<alkisg>
I think it'd be easier to pinpoint where there's diversion
00:26
artwork could be handled by packaging
00:26
rules.d or something
00:26
<vagrantc>
in debian i generally add packages to LATE_PACKAGES or EARLY_PACKAGES, whereas the ubuntu plugins generally install the packages directly.
00:26
<alkisg>
Like in ldm now
00:26
There's no reason to not do it the same way there
00:26
<vagrantc>
the artwork in ldm is done totally differently...
00:27
<alkisg>
Same for "no-recommends", I think it's a bad idea for Ubuntu not to include them
00:27
<vagrantc>
in ltsp-build-client, it's package selection.
00:27
in ldm, it's update-alternatives magic
00:27
maybe it's worth a go.
00:28rthomson has left IRC (rthomson!~rthomson@mars.pet.ubc.ca, Quit: Reached EOD)
00:28
<vagrantc>
basically, all the artwork hooks wouldn't do anything because of the package names, i think, so it would essentially be a no-op.
00:29
alkisg: it would also make it possible to install an Ubuntu chroot on Debian and vice versa.
00:30
or at least, easier.
00:30
<alkisg>
:)
00:30
vagrantc: did you get to think about ltsp-client-core stop phase?
00:30
I think that's where `nbd-client -d` belongs, until it goes to the nbd package...
00:30
<vagrantc>
alkisg: no, i've just been plugging away at ltsp on debian
00:31
alkisg: so you wanted to call it from upstart, but not during the sysv stop phase?
00:31
<alkisg>
No I wanted to call it from sysv but we already had upstart jobs handy and we were on a time pressure :D
00:31
And I think that upstart doesn't handle them well, it kills them on timeouts etc
00:32
So I think I'll move them to sysv scripts instead
00:32
But I do think that the ltsp-client-core stop phase is the correct place for them
00:34
Doesn't "service ltsp-client-core stop" mean "shut down the ltsp client"?
00:35
<vagrantc>
suppose it could...
00:36
ok, so now all my menu entries went away :(
00:36
it remembered the login from what i booted as a thin client
00:36
er, fat client
00:36
<alkisg>
Is that lxde? Or gnome-shell?
00:36
<vagrantc>
lxde
00:38
<alkisg>
Tomorrow morning I want to finish the reboot/shutdown problem...
00:38
The best way would be to let the nbd-client script handle it
00:39
I could write a patch for it, and maybe Wouter could accept it, but there's no time to include it in Precise,...
00:39
<vagrantc>
by the time nbd-client runs, it's a bit late, isn't it?
00:39
<alkisg>
...so, how weird would it be to override the nbd-client initscript with one of ours in init-ltsp.d ? :D
00:39
No, it isn't late
00:39
<vagrantc>
or rather, a bit eaarly.
00:39
too early to stop most services...
00:39
<alkisg>
The way I tried was:
00:40
disconnect swap etc, and leave only the root file system,
00:40
then spawn a shell task
00:40
That task will just wait for the "terminate processes" init phase
00:40
with a shell trap
00:40
At that phase, it'll do nbd-client -d, and echo o > sysrq
00:41bobby_C has left IRC (bobby_C!~bobby@85-124-22-227.teleworker.xdsl-line.inode.at, Ping timeout: 244 seconds)
00:41
<alkisg>
That worked well from sysv, but not so well from upstart, not sure why
00:44* vagrantc dreads writing this lengthy changelog.
00:44
<vagrantc>
i'll mostly steal stgraber's, but still...
00:45alkisg is now known as pumpkin
00:45* pumpkin goes to bed...
00:45pumpkin has left IRC (pumpkin!~alkisg@ubuntu/member/alkisg, Quit: Leaving.)
00:46
<stgraber>
;)
00:50* vagrantc wonders if the menu-cached proess has something to do with the menus being messed up
01:34
<highvoltage>
vagrantc: new bg.png pushed :)
01:37* vagrantc cheers!
01:38
<vagrantc>
with alkisg's new .dmrc processing, i should try and cram out a new ldm tonight too... but not until i get ltsp 5.3.1 solid for debian
01:38Parker955 is now known as Parker955_Away
01:47Parker955_Away is now known as Parker955
02:08adrianorg has left IRC (adrianorg!~adrianorg@186.213.157.3, Ping timeout: 244 seconds)
02:42_UsUrPeR_ has left IRC (_UsUrPeR_!~jsass@c-76-112-192-21.hsd1.mi.comcast.net, Quit: Leaving)
02:56Parker955 is now known as Parker955_Away
03:15
<vagrantc>
still don't know how i want to deal with ltsp-update-image on debian ... simply running it converts your system over to NBD+squashfs, and that's not really desireable just yet.
03:48primeministerp has joined IRC (primeministerp!~ppouliot@static-71-174-244-28.bstnma.fios.verizon.net)
04:03killermike has left IRC (killermike!~killermik@2.26.103.147, Remote host closed the connection)
04:14
<vagrantc>
ah, i'mm subjecting everybody to my /proc workaround....
04:15
needed to run it before the init-ltsp.d stuff is run, and our scripts pretty much assume it's mounted...
04:40risca has left IRC (risca!~risca@wi-secure-8039.cc.umanitoba.ca, Quit: Lämnar)
04:44risca has joined IRC (risca!~risca@wi-secure-8039.cc.umanitoba.ca)
05:07
<vagrantc>
still sometimes get the /var/cache/apt thing ... don't get it.
05:07
just don't get it.
05:08
it's the inconsistancy i don't get...
05:29risca has joined IRC (risca!~risca@wnpgmb0903w-ds01-177-34.dynamic.mtsallstream.net)
05:34
<vagrantc>
well, i'm not totally happy with the way i resolved the discrepancies with ltsp-update-image ... but it's better than where it was and shouldn't break anything for anyone else.
05:36
it breaks ltsp-update-image for debian users switching backk and forth regularly, but only a little broken, rather than breaking it for users just by running a command that seems like it does something vauguely related to what they want.
05:38
now to write the changelog ...
05:47staffencasa has left IRC (staffencasa!~staffenca@128-193-148-241.oregonstate.edu, Ping timeout: 240 seconds)
06:13alexqwesa has joined IRC (alexqwesa!~alex@alexo-veto.broker.freenet6.net)
06:24
<vagrantc>
ok, i'm tagging ltsp 5.3.1
06:43primeministerp has left IRC (primeministerp!~ppouliot@static-71-174-244-28.bstnma.fios.verizon.net, Remote host closed the connection)
06:52risca has left IRC (risca!~risca@wnpgmb0903w-ds01-177-34.dynamic.mtsallstream.net, Quit: Lämnar)
06:54alkisg has joined IRC (alkisg!~alkisg@ubuntu/member/alkisg)
06:55
<vagrantc>
alkisg: hah! just in time.
06:55
<alkisg>
'morning folks
06:55
<vagrantc>
alkisg: uploaded ltsp 5.3.1-1 to debian seconds ago.
06:56
<alkisg>
Coooool!!!
06:56* alkisg checks the trunks...
06:57
<vagrantc>
there's still an intermitant bug with /var/cache/apt/ eating 30MB of space *sometimes* but not always.
06:57
i'm not totally happy with how i've handled ltsp-update-image, but it's better than what we've had in debian before ...
06:57loather has joined IRC (loather!~khudson@wsip-98-175-250-115.sd.sd.cox.net)
06:57
<alkisg>
vagrantc: about r2089, I think the idea was to just copy the template provided by nbd
06:58
<vagrantc>
and configure_localdev should be patched out or removed or not run on debian... it causes two cdpingers to show up, one on cdrom on one on sr0
06:58
<alkisg>
vagrantc: do you have a repository I can use to test in Debian? Or, how long until it's available in unstable?
06:59
We should make that packaging change in ltspfs...
06:59
<vagrantc>
alkisg: i could upload them somewhere... but they should be avilable in the next 1-6 hours at a mirror near you
06:59
<alkisg>
That's soon enough, no need to upload them anywhere then
07:00
And shutdown gives you no problems?
07:00
Or just a stale nbd-server process?
07:00
<vagrantc>
or if you get antsy, you could download from incoming.debian.org even sooner and make a local mirror yourself.
07:00
alkisg: with NBD shutdown gives me stale nbd processes
07:01
but i mostly don't use NBD, so... :)
07:01
<alkisg>
OK... so the most pressing things now is a new ltspfsd package that doesn't contain the udev rules (but they are linked from init-ltsp.d), and the shutdown/reboot services
07:01
<vagrantc>
(and dropping configure_localdev from init-ltsp.d, or significantly changing what it does)
07:02* alkisg hasn't looked at that...
07:02
<vagrantc>
i also noticed you remove the nbd-client service? i just blindly copied it...
07:03
<alkisg>
Btw I don't know how obvious it was, but I started copying code from ltsp-init-common to init-ltsp.d
07:03
<vagrantc>
but it seemed to leave fewer stale nbd-server processes when i fixed it.
07:03
<alkisg>
It shouldn't, we leave KILLALL=False
07:03
That tells nbd to not kill anything that it doesn't know in its config file
07:03
<vagrantc>
hm.
07:03
<alkisg>
And we don't write anything in its config file
07:04
<vagrantc>
alkisg: what sort of code from ltsp-init-common?
07:04
<alkisg>
...so about ltsp-init-common, I thought that only the code that needs to run from ltsp-client-core could stay there
07:04
E.g. configure_home
07:05
Gadi had put . init-ltsp-common at the start of init-ltsp.d, and then called its functions
07:05
<vagrantc>
well, the idea of ltsp-init-common was that it woudl be a repository of useful functions that could be shared...
07:05
<alkisg>
Yes, but we moved from functions to files
07:05
<vagrantc>
i.e. maybe one distro calls them from one place, but another migght find them better suited elsewhere.
07:05
otherwise we get lots of code duplication
07:06
<alkisg>
If one distro wants to call configure_home at a later stage, it can just create a Distro/xx-home with a bigger number
07:06
It's not like making it a function provides additional flexibility
07:06
<vagrantc>
what if they want to call it after /init ?
07:06
<alkisg>
init-ltsp.d should only have stuff that runs on init
07:07
The rest should stay in ltsp-init-common, which should probably be renamed to ltsp-client-core-functions :D
07:07
<vagrantc>
and so we're presuming what all the other distros want to just dump code in there... maybe we're right... maybe that's always the sanest place to do those things.
07:07
<alkisg>
We do need some code to run after init-ltsp
07:07
<vagrantc>
it's only called ltsp-client-core on debian/ubuntu (we actually rename the upstream version)
07:08
<alkisg>
Yeah, ltsp-core, isn't it?
07:08
And we should remove ltsp-setup as soon as noone else uses it
07:08
<vagrantc>
alkisg: but you're presuming all distros will use the same code from init-ltsp.d rather than at a later phase.
07:08
<alkisg>
Right, we do need a common plan
07:09
<vagrantc>
yes.
07:09
<alkisg>
I don't think that changing the code as we did, and leaving it to ltsp-init-common would help any distro
07:09
<vagrantc>
unfortunately, other than knipwim, i've seen precious little interaction with other distros.
07:09
<alkisg>
It'd just break the current ltsp-init-common for them, forcing them to switch to init-ltsp right now
07:09
I think it's a good time to clean up things then :)
07:10
...and when they're clean, and they want to sync again with upstream, they can follow suit...
07:10
<vagrantc>
fair enough!
07:10
<alkisg>
vagrantc: didn't gadi remove "configure_localdev" completely?
07:10
As it's called from udev now?
07:10
<vagrantc>
he did, and then snuck it back into initramfs-scripts.d for some reason...
07:11
maybe it was a transient issue that's since been fixed?
07:11
<alkisg>
No, the plan was to remove it completely and make it a response to udev rules
07:11
<vagrantc>
it mumbles about it in the init-ltsp.d hook.
07:12
something about coldplug?
07:12
<alkisg>
Udev gets restarted on the real fs, so there's no FIXME there
07:12
We should just remove client/init-ltsp.d/common/50-localdev
07:13* vagrantc wonders if it's hopeless to backport to squeeze now.
07:13
<vagrantc>
alkisg: yes, i was planning on at least removing it from debian before upload, but i'm not going to bother just yet.
07:13
wait till a few more bugs show themselves
07:13
<alkisg>
Yeah
07:14
vagrantc: will you eventually do the ltspfs packaging change?
07:14
I.e. move the rules somewhere that don't apply by default?
07:14
Then we should change client/init-ltsp.d/common/50-localdev to link those rules
07:14
Not remove it completely
07:14
<vagrantc>
alkisg: i'm not goind to be available next weekend, really.
07:15
alkisg: and won't have much time this week...
07:15
<alkisg>
OK, but if stgraber does the change in Ubuntu, should I change the common/50-localdev, or have a local Ubuntu/ copy changed?
07:15
<vagrantc>
alkisg: probably 50-localdev can symlink in /etc/udev/rules.d ...
07:15
because /lib/udev/ may or may not be the location of the week.
07:16
<alkisg>
As long as the rule is in /usr/share/ltspfsd, I don't mind for either dir, /etc/ or /lib/udev :)
07:16
<vagrantc>
ubuntu also has a different numbering scheme from debian, i think.
07:16
<alkisg>
That even makes debian/rules simpler
07:17
<vagrantc>
alkisg: well, i think the /etc one will be respected as a local admin change...
07:17
<alkisg>
....and puts the "if" in common/, or if you want we can have different Distro/50-localdev versions
07:17
vagrantc: but it isn't, it's a packaging change
07:17
:D
07:17
<vagrantc>
yeah, install to /usr/share/ltspfsd, and symlink into /etc/udev/rules.d, i think...
07:17
<alkisg>
(more or less)
07:18
We essentially only "install" the ltspfsd package if init=/sbin/init-ltsp
07:18
<vagrantc>
alkisg: it's not *really* a packaging change.
07:18
if we do that, i think we could kill off ltspfsd-core, too.
07:18
<alkisg>
Yes, that's why I was asking yesterday
07:19
<vagrantc>
i need to look at it more to be sure.
07:19
<alkisg>
OK
07:22
Hmmm yes I prefer /etc to /usr. Maybe if we're careful in the future we'll only need a writeable /run and /etc
07:22
<vagrantc>
alkisg: no, we should ship it in /usr/share/ltspfs, but at runtime symlink it into /etc/
07:23
<alkisg>
Yes, that's what I meant too
07:23
But symlinking into /usr assumes that /usr is writeable
07:23
And maybe, just maybe, in the future we'll only put cows over /usr and /etc
07:24* alkisg is thinking of cases where initramfs's aren't supported
07:24
<vagrantc>
alkisg: symlink from /etc to /usr, /usr doesn't need writeability.
07:25
NBD cow will save us.
07:25
<alkisg>
Sorry meant from /lib/ to /usr/
07:25
<vagrantc>
i think making things writeable everywhere speeded up boot for NFS.
07:25
<alkisg>
I.e. that I prefer the symlink to go to /etc instead of to /lib
07:26
<vagrantc>
because i didn't have to copy up whole directory trees into ram...
07:30
alkisg: by the way, i tested with btrfs, although didn't do any custom options or anything
07:30
<alkisg>
-o compress
07:31
<vagrantc>
alkisg: at filesystem creation time?
07:31
<alkisg>
At every mount time
07:31
<vagrantc>
alkisg: NBD mount as well?
07:31
<alkisg>
No need, it's for writing only
07:32
moment, phone
07:32
<vagrantc>
cool.
07:32
<alkisg>
My btrfs-chroot script: http://paste.ubuntu.com/851017/
07:32
Just a simple wrapper to easily work with multiple btrfs nbd images
07:38
The most crucial time to have -o compress is at ltsp-build-client, or at rsync
07:38
I.e. when all the chroot files are created
07:39
<vagrantc>
using ext3 created an i386.img about 1.3G, using btrfs it was more like 1.8G ... trying with compression now.
07:41
alkisg: if you're really curious, ltsp 5.3.1-1 is already at http://incoming.debian.org
07:41
alkisg: but you'd have to download all the debs and dump them into a repository manually.
07:41
<alkisg>
vagrantc: an 743Mb chroot becomes 281Mb as squashfs and 403MB as btrfs
07:42
I'm very tempted to ditch squashfs completely
07:42
<vagrantc>
having writeability is pretty slick.
07:42
as well as all the other crazy features of btrfs
07:42
<alkisg>
Opensuse has some patches for squahfs writeability... I hope those could go upstream
07:43
Yeah, snapshoting a btrfs loop-based chroot before upgrading..
07:43
...and while in use :D
07:43
<vagrantc>
right.
07:44
<alkisg>
Another thing I want to experiment with is having multiple proprietary drivers in the chroot, and enabling them or not from init-ltsp.d
07:44
<vagrantc>
so it created an image that was 802MB on disk, df -h reports 387MB used, du -smc reports 731MB
07:45
crazy!
07:45
i think a few apt-get upgrades will make one go insane, though.
07:46
unless it has a feature to reconsile identical files and re-merge them again.
07:46
reconcile
07:47
eesh, i've been at this for about 14 hours straight... glad i finally uploaded.
07:47
<alkisg>
Good work though :)
07:47
Heck, it took me hours to do a simple thing as that .dmrc...
07:49
<vagrantc>
the aufs/overlayfs thing really feels good to finally move to ... it was such a bear to maintain the list of what's expected to be writeable.
07:49
(and apparently, lintian grumbles about "writeable" not being a word, it's spelled "writable")
07:49
<alkisg>
Hehe
07:50
Some people in ltsp-developers expressed concerns about ram usage with the init-ltsp writings
07:50
<vagrantc>
if i hadn't tagged and pushed already, i would've fixed that.
07:50
alkisg: on the mailing list?
07:50
<alkisg>
Yes
07:50* vagrantc has been too busy releasing to read ltsp-devel
07:51
<alkisg>
Even thought booting needs less than 1 Mb of cow, I think it's a good idea to allow for mounting an nbd temporary image instead of creating a tmpfs
07:52loather has left IRC (loather!~khudson@wsip-98-175-250-115.sd.sd.cox.net, Quit: This computer has gone to sleep)
07:52
<vagrantc>
alkisg: sure, that's reasonable.
07:52
(unless you get the 30MB on whatever the heck triggers that on Debian)
07:52
<alkisg>
vagrantc: debian doesn't use plymouth, does it? What does it use?
07:53
<vagrantc>
alkisg: some debian installs use plymouth, i'm sure.
07:53
<alkisg>
Ah. But by default? usplash?
07:53
<vagrantc>
alkisg: default LTSP doesn't use any splash, though.
07:53* alkisg hates the fact that plymouth adds 6 Mb to initramfs
07:53
<vagrantc>
none of my machines use splash of any kind... not sure what a default debian install looks like these days.
07:54
<alkisg>
It breaks booting 64mb clients
07:54
Just for a fancy splash screen :(
07:54
<vagrantc>
just because the initramfs is too big
07:54
?
07:54
<alkisg>
Yeah
07:54
It doesn't fit to be uncompressed
07:54
(in RAM)
07:55
<vagrantc>
this sure was the biggest change in Debian's LTSP in ages...
07:55
eesh, i should have probably written a NEWS.Debian about some of the behavior changes...
07:56* alkisg will be happier when we ditch ldm and reimplement the configuration system :)
07:56
<vagrantc>
oh, my computer turns into a pumpkin in a few minutes....
07:56
<alkisg>
Hehe
07:56
'night vagrantc
07:57* vagrantc actually set up a cron job to shutdown computers regardless of weather i'm using them
07:57
<vagrantc>
till another time, have fun reading diffs :)
07:57vagrantc has left IRC (vagrantc!~vagrant@c-76-115-60-19.hsd1.or.comcast.net, Quit: leaving)
08:18alkisg has left IRC (alkisg!~alkisg@ubuntu/member/alkisg, Quit: Leaving.)
08:25floorduh has left IRC (floorduh!b85bfd7c@gateway/web/freenode/ip.184.91.253.124, Ping timeout: 245 seconds)
08:47Yet_another_Bill has left IRC (Yet_another_Bill!billy@nat/redhat/x-evqjewzxcuvobyss, Ping timeout: 252 seconds)
08:47Yet_another_Bill has joined IRC (Yet_another_Bill!billy@nat/redhat/x-rolqiqnxsamxvprq)
08:50alkisg has joined IRC (alkisg!~alkisg@ubuntu/member/alkisg)
08:58dobber has joined IRC (dobber!~dobber@213.169.45.222)
09:18alkisg has left IRC (alkisg!~alkisg@ubuntu/member/alkisg, Quit: Leaving.)
09:31dimginis has joined IRC (dimginis!c23fde81@gateway/web/freenode/ip.194.63.222.129)
09:36asmok has joined IRC (asmok!~asmok@adsl-85-157-204-13.regionline.fi)
09:37asmok has left IRC (asmok!~asmok@adsl-85-157-204-13.regionline.fi, Client Quit)
10:19adrianorg has joined IRC (adrianorg!~adrianorg@186.213.157.3)
10:58khildin has joined IRC (khildin!~khildin@ip-80-236-223-18.dsl.scarlet.be)
11:14alkisg_android has joined IRC (alkisg_android!~yaaic@athedsl-156742.home.otenet.gr)
11:21alkisg_android has left IRC (alkisg_android!~yaaic@athedsl-156742.home.otenet.gr, Ping timeout: 276 seconds)
11:49killermike has joined IRC (killermike!~killermik@2.26.103.38)
11:50skumlesen has left IRC (skumlesen!~skumlesen@87-104-113-212-dynamic-customer.profibernet.dk, Quit: always look at the bright side of life.)
11:54loather has joined IRC (loather!~khudson@wsip-98-175-250-115.sd.sd.cox.net)
12:13Trixboxer has joined IRC (Trixboxer!~Trixboxer@115.124.115.71)
12:13adrianorg has left IRC (adrianorg!~adrianorg@186.213.157.3, Ping timeout: 245 seconds)
12:41Parker955_Away is now known as Parker955
12:54Parker955 is now known as Parker955_Away
12:58khildin has left IRC (khildin!~khildin@ip-80-236-223-18.dsl.scarlet.be, Ping timeout: 240 seconds)
13:14brunolambert has joined IRC (brunolambert!bruno@nat/revolutionlinux/x-yyqglvxzuuagcovx)
13:14bmonkj has joined IRC (bmonkj!~skumlesen@mail.kia.dk)
14:12alkisg has joined IRC (alkisg!~alkisg@ubuntu/member/alkisg)
14:14
<cyberorg>
alkisg, hi, can you migrate everything out of ltsp-init-common, i'd then know if anything we need is missing from init-ltsp.d/common/
14:15
what i've discovered so far is configure_network does not seem to know what to do if hostname is 'none' and sound setup seems to be missing
14:15
<alkisg>
knipwim: are you ok with that? ^
14:16
<cyberorg>
apart from those two suse just needs init-ltsp.d/common/ nothing else
14:16
<alkisg>
cyberorg: we've moved configure_network to an init-ltsp.d script too
14:16
Err sorry, let me check...
14:17
<cyberorg>
alkisg, it is, but as i said if hostname is 'none' it does not do anything
14:17
<alkisg>
common/09-hostname
14:17
cyberorg: what do you mean "hostname is none"?
14:18
Hostname is none at the kernel if it's not set
14:18
<cyberorg>
if [ "(none)" = "$hostname" ] || [ "linux" = "$hostname" ] || [ "none" = "$hostname" ]; then
14:18
<alkisg>
We're not looking at the kernel value
14:18
<cyberorg>
we do something like that ^^
14:18
<alkisg>
cyberorg: that's because you're looking at the kernel hostname value
14:18
We don't look at it anymore, there's no point
14:18
We either set the value provided by dhcp, or by lts.conf, or the auto-generated one
14:19
<cyberorg>
http://paste.opensuse.org/41082002
14:19
<alkisg>
There's no need, ever, to look at the kernel value at /proc/wherever/hostname
14:19
cyberorg: yes, we had that code in the past too, I'm saying that it's a broken code
14:19
That's why I replaced it
14:19
Why would we ever want to check what the kernel hostname is?
14:19
We always assign a hostname...
14:20
<cyberorg>
but then the new code does not seem to assign one automatically
14:20
<alkisg>
cyberorg: are you using 09-hostname?
14:20
Or are you still calling configure_networking?
14:21
<cyberorg>
alkisg, using 09-hostname it does not assign hostname automatically, it still remains 'none', using configure_networking autogenerated ws$IP is assigned
14:22
<alkisg>
cyberorg: let's have a look together at 09-hostname
14:22
It unconditionally, without an if, calls hostname "$HOSTNAME"
14:22
Why is that not working for you?
14:22
<cyberorg>
let me boot up client
14:22
<alkisg>
hostname "$HOSTNAME"
14:22
echo "$HOSTNAME" > /etc/hostname
14:23
Maybe you set HOSTNAME=(none) before reaching that point?
14:23
To check, put break=09-hostname, and `echo $HOSTNAME`.
14:23
<cyberorg>
alkisg, exactly, something in initrd sets it to 'none'
14:23
<alkisg>
(as a kernel parameter)
14:23
Then you need to fix your initrd
14:23
Remove stuff, probably
14:24
<cyberorg>
instead of 'if [ -z "$HOSTNAME" ]; then" the above will work for us
14:25
<alkisg>
cyberorg: ok, but why not fix the initrd instead? Isn't it ltsp code that sets HOSTNAME=(none)?
14:25
No need to keep old and currently broken code around...
14:25
<cyberorg>
no, kiwi initrd creator does it
14:26
<alkisg>
Wait. Are you using init=/sbin/init-ltsp ?
14:26
Or are you calling init-ltsp from initramfs?
14:26
<cyberorg>
if i put if [ "(none)" = "$hostname" ] || [ "linux" = "$hostname" ] || [ "none" = "$hostname" ]; then instead of 'if [ -z "$HOSTNAME" ]; then in 09-hostname it will work fine
14:26
using init=/sbin/init-ltsp
14:26
<alkisg>
OK, please do this:
14:26
Boot a client with init=/bin/bash
14:27
And type: echo $HOSTNAME
14:27
It isn't logical to get (none) as the output
14:30
<cyberorg>
something in this image didn't like init=/bin/bash, stuck after stopping udevd
14:31
<alkisg>
What I'm saying is this
14:31
init starts with an almost clean environment
14:31
No hostname, nothing
14:31
Then we source some env vars that we saved at init-bottom/ltsp
14:32
LTSP saved those, not some other initramfs script
14:32
That's were HOSTNAME=(none) came up in the past
14:32
But it didn't make sense to save a variable with value=(none)
14:32
That's why we removed it from the initramfs script and from 09-hostname too
14:32
I think you need to remove some ltsp code from the initramfs
14:32khildin has joined IRC (khildin!~khildin@ip-80-236-223-18.dsl.scarlet.be)
14:33
<alkisg>
If you really have to do that if above, sure, go ahead and do it, it doesn't hurt, it just doesn't make sense
14:34
<cyberorg>
initrd sets hostname to 'none' as it gets nothing from dhcp
14:34
<alkisg>
Yes, that's what ltsp in ubuntu/debian did in the past too
14:35
Not the other initramfs scripts, but the ltsp code in the initramfs
14:35
And it's broken behavior
14:35
Why set it to (none) and not leave it empty? It doesn't make sense
14:35
What ltsp scripts do you have in the initramfs? Do you have a branch I could look at?
14:36
<cyberorg>
none, we use kiwi for creating initrd
14:36
<alkisg>
OK, if the kiwi code sets HOSTNAME=(none) while at init, then kiwi is broken
14:37
But if you need that, sure, go ahead and commit it, along with a comment to remove it when kiwi is fixed
14:37
<cyberorg>
i can't find anything in kiwi's code that sets it https://github.com/openSUSE/kiwi
14:38
mostly it should be just this file: https://github.com/openSUSE/kiwi/blob/master/modules/KIWILinuxRC.sh
14:38
<alkisg>
That's why I'm insisting, I think you're overlooking some script that needs to be removed
14:38
<cyberorg>
setupNetwork in there
14:39
nothing there handles hostname so not sure where none comes from, could be default distro behaviour
14:41
<alkisg>
Wow that's a very big script. Yeah I don't see HOSTNAME there.
14:41
cyberorg: you can try with any opensuse machine, boot it with init=/bin/bash
14:41
<||cw>
could be in dhclient?
14:41
<cyberorg>
so apart from that and sound just the stuff in init-ltsp.d/common is sufficient for suse
14:41
<alkisg>
Sound should come from the udev rules
14:42xsl has joined IRC (xsl!~silence@unaffiliated/xsl)
14:42
<cyberorg>
pulseaudio stuff?
14:43
see start_sound in ltsp-init-common, we use that
14:43
<alkisg>
Ah, ok
14:44
In Ubuntu we call that from ltsp-core
14:44
initscripts/ltsp-core
14:45
<cyberorg>
can that be moved in init-ltsp.d/common as well?
14:45
<alkisg>
Let me check...
14:46
<cyberorg>
may be do away with ltsp-core and just use init-ltsp.d/common?
14:46
<alkisg>
Hmm not sure. Some services need other services to get launched
14:46
So I think that we'll always need to run some code after most services are loaded
14:46
E.g. pulseaudio may need some alsa service
14:47
So maybe we can't launch it before init has even started
14:47
I think init-ltsp.d should only modify files, and ltsp-core should start services
14:47
<cyberorg>
ok got it
14:48
in that case is everything that could be moved to init-ltsp.d/common is done? i can then start working on packaging it
14:49
<alkisg>
No, it's not done, Gadi left the code at ltsp-init-common for compatibility reasons and I only moved the code that I needed to touch for other reasons
14:49
But if opensuse and gentoo are ready, we can move the other code too
14:49
I.e. not source ltsp-init-common from common/ at all
14:51
<cyberorg>
opensuse is ready, with some stuff duplicated in those two places it is slightly confusing as i don't know what exactly to remove from our ltsp-client.init script
14:51
<alkisg>
knipwim: ?
14:51
cyberorg: feel free to move the code into common/ though
14:52
<cyberorg>
alkisg, this is our current init script http://kiwi-ltsp.svn.sourceforge.net/viewvc/kiwi-ltsp/trunk/kiwi-ltsp/ltsp/suse-12.1/root/etc/init.d/ltsp-client.init?revision=561&view=markup
14:53
not qualified to "move the code into common/" as not sure what other distros use
14:54dead_inside has joined IRC (dead_inside!~dead_insi@76.75.3.174)
14:54
<alkisg>
cyberorg: e.g. we moved configure_swap to 50-swap
14:55
Half of the functions there can be moved to init-ltsp.d scripts
14:55
I.e. the ones that don't depend on other services running
14:55floorduh has joined IRC (floorduh!b85bfd7c@gateway/web/freenode/ip.184.91.253.124)
14:55
<alkisg>
The reason is that e.g. having the swap available before the initscript get loaded allows thin clients with less ram to boot
14:55
Or that we can disable services before init loads them
14:56
<cyberorg>
i know, removed most of it in current testing
14:56
<alkisg>
Nice
14:57
Wow, the kiwi git is more than 300mb?!
14:57
I was hoping to grep for HOSTNAME there...
14:57mnevans has joined IRC (mnevans!~mnevans@zero.geol.umd.edu)
14:58
<cyberorg>
nice work btw, everything works as before on suse using just common/
14:58
<alkisg>
cyberorg: but do have a quick look at Ubuntu/, you may see some additional functionality there
14:58
I think that vagrantc mostly symlinked Debian/ to Ubuntu/
14:59
<mnevans>
Hello LTSP world. mnevans again, same problem: no localdev (flashdrive) support on AMD E-350 thin client. LTSP5, ubuntu 10.10. It seemed to break after a recent kernel and security upgrade. I repeated steps 6(a)(2)-(6) in the wiki help page we made at https://help.ubuntu.com/community/UbuntuLTSP/AMDE-350. No luck. Any ideas?
14:59
<knipwim>
alkisg: working now, but i'm here in about an hour
14:59
<alkisg>
knipwim: nice
14:59
cyberorg: I'll be here in about an hour too, if you want we can try to eliminate stuff we don't need then
15:00khildin has left IRC (khildin!~khildin@ip-80-236-223-18.dsl.scarlet.be, Quit: I'm gone, bye bye)
15:00
<cyberorg>
60-screen-sessions 50-set-time 50-saned can be put in common?
15:02
<alkisg>
cyberorg: I didn't put the scripts into common/ only for speed reasons, because I didn't have the time to check if they would work on other distros
15:02
<cyberorg>
may be not, as locale etc comes from different place on suse
15:02
<alkisg>
Whatever you want, move it to common/ please
15:04
<cyberorg>
paths used in those are ubuntu specific can't use them
15:05
got to go now, see you tomorrow, 'night
15:06
<alkisg>
Ah, you can't be here when knipwim will be available... ok, bb
15:12
<Hyperbyte>
When I SSH to my server without X11 forwarding (ssh -x), after entering password, I login practically immediately... when I do it with SSH forwarding (ssh -X) there's a 2 second delay during logon...
15:12
Is this normal?
15:13
<alkisg>
Hyperbyte: I think the ssh delay was some ubuntu specific bug... I don't see what you're saying in Precise.
15:25
<Hyperbyte>
You're giving me more and more reasons to upgrade. :P
15:33
<alkisg>
I've been using it since alpha 1... fortunately LTS'es are indeed more stable than non-LTS releases
15:43
<mnevans>
I have to go; will try again later.
15:43mnevans has left IRC (mnevans!~mnevans@zero.geol.umd.edu)
15:44dead_inside has left IRC (dead_inside!~dead_insi@76.75.3.174, Quit: Leaving...)
16:03
<knipwim>
alkisg: is this about moving some ltsp-init-common code?
16:09
all gentoo specific things for setting up the client are in a gentoo initscript
16:10
i will have to move this to init-ltsp.d
16:10
the generic stuff is already available there in common rigt?
16:13
gentoo will use ltsp-client-5.2.12 for some time, so removing the ltsp-init-common won't break it
16:13staffencasa has joined IRC (staffencasa!~staffenca@128-193-148-241.oregonstate.edu)
16:14
<alkisg>
knipwim: yes, about moving stuff out of ltsp-init-common
16:14dead_inside has joined IRC (dead_inside!~dead_insi@76.75.3.174)
16:14
<alkisg>
Basically whatever can be ran by /sbin/init-ltsp, without services running, should go to init-ltsp.d/
16:15
The rest will be left in ltsp-init-common
16:15
So OK since Ubuntu/Debian/OpenSUSE don't need that code, and you don't mind, we can remove it
16:15
<knipwim>
fine by me
16:15
when migrating gentoo to 5.3, i will put gentoo specifics there as well
16:15
<alkisg>
Nice
16:16
<knipwim>
in init-ltsp.d/Gentoo that is :)
16:16
<alkisg>
Yup, I hope we can have lots of common stuff in common/ too :)
16:17
<knipwim>
it's interesting, you can see the differences in distro's
16:18
mostly packaging, but also other quirks
16:20
<alkisg>
bbiab
16:20alkisg has left IRC (alkisg!~alkisg@ubuntu/member/alkisg, Quit: Leaving.)
16:29alkisg has joined IRC (alkisg!~alkisg@ubuntu/member/alkisg)
16:38dobber has left IRC (dobber!~dobber@213.169.45.222, Remote host closed the connection)
16:43Gremble has joined IRC (Gremble!~Ben@cpc10-aztw24-2-0-cust114.aztw.cable.virginmedia.com)
17:01rthomson has joined IRC (rthomson!~rthomson@mars.pet.ubc.ca)
17:02brunolambert has left IRC (brunolambert!bruno@nat/revolutionlinux/x-yyqglvxzuuagcovx, Ping timeout: 245 seconds)
17:03slackish has joined IRC (slackish!amcphall@mcphall.org)
17:05
<xsl>
is it possible to burn CDs on ltsp ? must the CD-RW be installed on the server - or people can use the Clients CD-RW
17:05floorduh has left IRC (floorduh!b85bfd7c@gateway/web/freenode/ip.184.91.253.124, Ping timeout: 245 seconds)
17:05alexqwesa has left IRC (alexqwesa!~alex@alexo-veto.broker.freenet6.net, Read error: Connection reset by peer)
17:08brunolambert has joined IRC (brunolambert!bruno@nat/revolutionlinux/x-yxxkekqobhvttrys)
17:09alexqwesa_ has joined IRC (alexqwesa_!~alex@alexo-veto.broker.freenet6.net)
17:11mnevans has joined IRC (mnevans!~mnevans@zero.geol.umd.edu)
17:13alexqwesa_ has left IRC (alexqwesa_!~alex@alexo-veto.broker.freenet6.net, Ping timeout: 245 seconds)
17:13
<mnevans>
Hello LTSP World. I lost the usb localdev support for flashdrives that we added a few weeks ago (https://help.ubuntu.com/community/UbuntuLTSP/AMDE-350). I tried repeating steps 6(a)(1-6); no luck. Any ideas? Thanks.
17:14alexqwesa_ has joined IRC (alexqwesa_!~alex@alexo-veto.broker.freenet6.net)
17:21
<mnevans>
To continue: I can see that ltspfs is installed on the server; but ltspfsd-core and ltspfsd are not (both 0.7-0ubuntu1). Seems strange. Should they be?
17:26
Going through the DebugLocalDev (https://wiki.ubuntu.com/DebugLocalDev), ltspfs is installed on the server and ltspfsd is not.
17:27
lsmod | grep fuse gives no response - maybe fuse isn't loading, or else it's no longer an insertable module. But libfuse2 and fuse-utils are installed.
17:27
/dev/fuse exists as a device.
17:27
Users are members of fuse group.
17:30
client recognizes the plug event: http://pastebin.com/c243bUPk
17:31
/etc/fstab on client and server both READ: #UNCONFIGURED FSTAB FOR BASE SYSTEM.
17:34vagrantc has joined IRC (vagrantc!~vagrant@75-150-46-245-Oregon.hfc.comcastbusiness.net)
17:37
<mnevans>
When I try step 4, manually issuing a mount command, I get: "Error /tmp/,mnevans-ltspfs/tmp is not mounted."
17:38
Anyone...? Bueller...? Anyone? :)
17:38
Now I have to go but if anyone has any ideas - I will leave this window open and come back to it later.
17:51
<||cw>
mnevans: so it was working, then it stopped? what changed? I've also noticed from ltspfs chatter recently, but I wasn't paying enough attention to know if it affects you. it does seem lots of things are changing for 12.04. I don't kno how much is in the repo yet though
18:16floorduh has joined IRC (floorduh!b85bfd7c@gateway/web/freenode/ip.184.91.253.124)
18:27adrianorg has joined IRC (adrianorg!~adrianorg@186.213.157.3)
18:31alkisg has left IRC (alkisg!~alkisg@ubuntu/member/alkisg, Remote host closed the connection)
18:31hughessd has joined IRC (hughessd!~steve@173-164-117-109-Oregon.hfc.comcastbusiness.net)
18:35alkisg has joined IRC (alkisg!~alkisg@ubuntu/member/alkisg)
18:35risca has joined IRC (risca!~risca@wi-secure-8039.cc.umanitoba.ca)
18:48vagrantc has left IRC (vagrantc!~vagrant@75-150-46-245-Oregon.hfc.comcastbusiness.net, Ping timeout: 245 seconds)
18:49torbjorn has joined IRC (torbjorn!~torbjorn@2001:840:4051::56)
18:50Gremble has left IRC (Gremble!~Ben@cpc10-aztw24-2-0-cust114.aztw.cable.virginmedia.com, Quit: I Leave)
19:08hughessd is now known as hughess
19:09hughess is now known as hughessd
19:14alkisg has left IRC (alkisg!~alkisg@ubuntu/member/alkisg, Quit: Leaving.)
19:15khildin has joined IRC (khildin!~khildin@ip-80-236-223-18.dsl.scarlet.be)
19:48Parker955_Away is now known as Parker955
19:51rthomson has left IRC (rthomson!~rthomson@mars.pet.ubc.ca, Remote host closed the connection)
19:53loather has left IRC (loather!~khudson@wsip-98-175-250-115.sd.sd.cox.net, Quit: This computer has gone to sleep)
20:09markit has joined IRC (markit!~marco@88-149-177-66.staticnet.ngi.it)
20:30xsl has left IRC (xsl!~silence@unaffiliated/xsl, Quit: Connection reset by fear)
20:35
<mnevans>
@ ||cw and others: Thanks - I am now back. What changed was a routine set of upgrades to ubuntu 10.10. I looked at /var/log/apt/history.log, and there were kernel upgrades, to 2.6.35-31.40, 2.6.35-32.41, 2.6.35-32.42, 2.6.35-32.64, 2.6.35-32.65.
20:46kleewyck has joined IRC (kleewyck!~kleewyck@c-71-239-69-134.hsd1.il.comcast.net)
20:53komunista has joined IRC (komunista!~slavko@adsl-195-168-247-205.dynamic.nextra.sk)
20:56_dogan_ has joined IRC (_dogan_!~dogan@95.15.226.24)
20:59bmonkj has left IRC (bmonkj!~skumlesen@mail.kia.dk, Quit: and what a nap it was)
21:21rthomson has joined IRC (rthomson!~rthomson@mars.pet.ubc.ca)
21:21alkisg has joined IRC (alkisg!~alkisg@ubuntu/member/alkisg)
21:37freedomrun has joined IRC (freedomrun!~quassel@BSN-61-15-166.dial-up.dsl.siol.net)
21:58bmonkj has joined IRC (bmonkj!~skumlesen@87-104-113-212-dynamic-customer.profibernet.dk)
22:00alkisg has left IRC (alkisg!~alkisg@ubuntu/member/alkisg, Quit: Leaving.)
22:03komunista has left IRC (komunista!~slavko@adsl-195-168-247-205.dynamic.nextra.sk, Quit: Leaving.)
22:24kleewyck has left IRC (kleewyck!~kleewyck@c-71-239-69-134.hsd1.il.comcast.net, Quit: Leaving)
22:47markit has left IRC (markit!~marco@88-149-177-66.staticnet.ngi.it, )
22:52_dogan_ has left IRC (_dogan_!~dogan@95.15.226.24, Ping timeout: 245 seconds)
22:52Trixboxer has left IRC (Trixboxer!~Trixboxer@115.124.115.71, Quit: "Achievement is not the end, its the beginning of new journey !!!")
23:01brunolambert has left IRC (brunolambert!bruno@nat/revolutionlinux/x-yxxkekqobhvttrys, Quit: brunolambert)
23:17Parker955 is now known as Parker955_Away
23:45
<torbjorn>
I just followed a guide to set up K12linux ltsp on a centos 6 machine (?) if I understand correctly. Would this be the place to ask questions about that?
23:50dead_inside has left IRC (dead_inside!~dead_insi@76.75.3.174, Quit: Leaving...)
23:51joebaker has joined IRC (joebaker!~joebaker@64.245.0.3)