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


Channel log from 28 June 2009   (all times are UTC)

00:02newbie_need_hlp has quit IRC
00:09flokuehn_ has quit IRC
00:09shogunx has quit IRC
00:09ltsppbot has quit IRC
00:11flokuehn has joined #ltsp
00:22shogunx has joined #ltsp
00:28ltsppbot has joined #ltsp
00:40strattog has quit IRC
00:41strattog has joined #ltsp
02:20vagrantc has quit IRC
04:01try2free has joined #ltsp
04:34
<try2free>
hi
04:35
anyone knows how to solve No sound using Intel Corporation 82801EB/ER (ICH5/ICH5R) AC'97 Audio Controller?
04:42Egyptian[Home] has quit IRC
05:04Egyptian[Home] has joined #ltsp
05:24rjune has joined #ltsp
05:35phantom has quit IRC
05:38phantom has joined #ltsp
05:42
<alkisg>
try2free: is the module for this sound card missing?
06:03hanthana has quit IRC
06:43pmatulis has joined #ltsp
06:52lucascoala has quit IRC
06:58try2free has quit IRC
08:21
<sbalneav>
Moning all
08:23
<alkisg>
Hello
08:23
<sbalneav>
Hey alkisg
08:24
<alkisg>
Did you come back from Brazil?
08:24
<sbalneav>
Sitting here in the hotel in Porta Alegre
08:24
No, we don't leave 'till tomorrow
08:24
I've gotten most of the docs updated.
08:25
I just have to go through the xrandr params and get them going.
08:25
<alkisg>
Hooray :)
08:25
<sbalneav>
LTSPManual.xml's up to 92 pages.
08:25
Still a lot of work to be done.
08:25
<alkisg>
If you want, I can look for "undocumented" lts.conf parameters after you're done
08:26
<sbalneav>
But, we've also got a full lts.conf.5 manual page.
08:26
Ah, ok
08:26
<alkisg>
Will the man page be installed as part of ltsp-server?
08:26
<sbalneav>
well, gimme 5 minutes, and I'll post the pdf + html page to the wiki page.
08:27
alkisg: Well, we need to talk about how we'd do that.
08:27
<alkisg>
sbalneav: I think it just goes into the 'manpages' file, for debian/ubuntu at least - I don't know about upstream projects :(
08:27
<sbalneav>
My preference would be to create an ltsp-docs package, that has the xml in it suitable for yelp, and the man page.
08:28
<alkisg>
Yup, that would be best
08:28
<sbalneav>
Well, if we do that, that's ok, but then it's kind of a "manual copy" process from the ltsp-docs-upstream bzr tree into the ltsp-trunk treem
08:29
<alkisg>
I think there are bzr hooks for that
08:29
<sbalneav>
Since we use trunk everywhere else, we should really create a new branch called ltsp-docs-trunk, and just start doing that.
08:29
But my first choice would be to get it packaged, and install a -docs package.
08:31
Simply because that fits in with my edubuntu-handbook strategy, that I want to pull out all the LTSP stuff, referr to the ltsp-docs where needed, and have edubuntu-handbook JUST talk about non-ltsp stuff.
08:31
which will be less intimidating for education people to contribute to.
08:32
<alkisg>
Is the edubuntu-handbook also .xml based, or is it written directly on the wiki?
08:32
<sbalneav>
xml dockbook
08:32
What we need is a really goot system to:
08:32* alkisg needs to have a look at docbook... :)
08:32
<sbalneav>
xml -> wikipage
08:32
then, when people edit the wikil
08:33
xml <- wiki+diffs
08:33
<alkisg>
That's gonna be tricky, isn't it?
08:33
<sbalneav>
Possibly
08:33
as with most things in Linux, it's simply "an excercise in text processing" :)
08:34
<alkisg>
Heh
08:34* alkisg is trying to debug a script he's making and he's stuck in the following not working: eval "MN_ITEM_BASE=\$((\$MN_${1}_ITEM_BASE))" :(
08:35
<sbalneav>
Oh, hold on, I can find you the right kind of syntax for that.
08:36
<alkisg>
Please do, I've been stuck on this for several minutes :)
08:36
<sbalneav>
I think you can just try:
08:37otavio has quit IRC
08:37
<sbalneav>
eval MN_ITEM_BASE=\"\$MN_${1}_ITEM_BASE\"
08:38
If I understand what you're trying to do :)
08:40
<alkisg>
I also wanted to evaluate that as a number with $((x)) at the same time, but I can't get both of them going
08:40
..so that if it isn't defined, 0 is used instead...
08:41hanthana has joined #ltsp
08:42
<sbalneav>
ok, so
08:42
eval MN_ITEM_BASE=\"\$MN_${1:-0}_ITEM_BASE\"
08:43
!docs
08:43
<ltspbot>
sbalneav: "docs" is For the most current documentation, see http://wiki.ltsp.org/twiki/bin/view/Ltsp/LtspDocumentationUpstream
08:44
<alkisg>
Thanks man
08:46
This is what I was trying to write, but it's too unreadable to keep it :) : eval MN_ITEM_BASE=\"\(\(\$MN_${1}_ITEM_BASE\)\)\"
08:46
<sbalneav>
heh
08:47
Did my line work?
08:47
check out the docs, they're updated
08:47Patina has quit IRC
08:47
<alkisg>
Yeah, the first line will do just fine, I was too overcautious :P
08:49Patina has joined #ltsp
08:49
<alkisg>
CPU: ... More information on this is available in . ==> missing hyperlink
08:50
About the video card, I was wondering, why is 32Mb a lower limit? Both of my labs have 4Mb S3's and worked with no problems...
08:52
i.e. Alt+F1, through Alt+F10. => In ubuntu it's Alt+Ctrl+F1... :-/
08:53
<sbalneav>
Not sure, I think the 32's for shared memory video cards
08:54
cards that actually HAVE ram on them are slightly different.
08:54
alkisg: thanks, I'll either find the link or remove it.
08:55
<alkisg>
Wow, local apps also :)
08:55
<sbalneav>
Yeah, descriptions need to be beefed up.
08:55
<alkisg>
Isn't LOCAL_APPS default =true?
08:55
<sbalneav>
Not sure.
08:55
<alkisg>
I think I can run ltsp-localapps xterm with no lts.conf at all
08:56
<sbalneav>
ok
08:56
I can fix that.
08:57
<stgraber>
morning
08:58
<alkisg>
Great work sbalneav! Thanks! :)
08:58
Hi stgraber
08:59
<stgraber>
hi alkisg
08:59
sbalneav: did you have time to look at that ldm bug ?
09:00* stgraber is working on the new ltsp-cluster web interface, looks pretty good but lacks tons of features
09:01
<sbalneav>
stgraber: Did some work on it, but I'm hampered by a lack of hardware.
09:01
We're going to pop by propus today, I'm hopingg to test then.
09:01
stgraber: while you're at it, check out the updated docs, see if I'm missing any ltsp parameters.
09:02
<stgraber>
sbalneav: do you have some code for me to test ? (I have two thin clients just next to me :))
09:02
I'm implementing session killing in ltsp-cluster so having to kill X everytime is kind of an issue for testing ;)
09:03
<alkisg>
sbalneav: also I saw LDM_SSHOPTIONS in the docs, but it isn't honored by ldm since some time now... /me hopes it gets fixed in ldm.c, not removed from the docs :)
09:05
<sbalneav>
stgraber: Without putting too fine a point on it, you pushed me to get my patch into the tree, when all I wanted to do was send you a patch to try. I ain't freaking releasig ANYTHING at this point untill I'm sure it works.
09:05marcosmamorim has joined #LTSP
09:05
<sbalneav>
It's very very simple, if you need to move on, back out my eater patch :)
09:06
I'll re-add it *when it's ready* :)
09:07
I'll have some time to test today, I'm thinking. If so, I'll email you what I've got.
09:08
alkisg: I've fixed the link here. Should have pointed to the chapter on LDM.
09:08
I need to get a lot more interlinking into the manual.
09:23bobby_C has joined #ltsp
09:32
<sbalneav>
ltspbot: msg Gadi "I hope you don't mind, but I'm using your post to ltsp-discuss as the underpinning to the xrandr chapter of the manual'
09:32
<ltspbot>
sbalneav: Error: No closing quotation
09:32
<sbalneav>
ltspbot: msg Gadi "I hope you don't mind, but I'm using your post to ltsp-discuss as the underpinning to the xrandr chapter of the manual"
09:32
<ltspbot>
sbalneav: Error: "msg" is not a valid command.
09:33
<sbalneav>
Meh, tell him when I see him :)
09:33
Now to document the xrandr commands.
09:34
alkisg: It's not? (sshoptions I mean)
09:35
alkisg: ah, we get the variable, but we never use it.
09:35
OK, I'll add it to my list of things to work on.
09:36
<alkisg>
sbalneav rocks :)
09:37
<sbalneav>
If you're going to be a true ltsp'er, you need to use the LTSP meme "what means <x> rocks?"
09:37
So
09:37
What means sbalneav rocks?
09:37
:)
09:37
<alkisg>
:)
09:38
<sbalneav>
A brazillian came in one time, after seeing on a web page somewhere "ltsp rocks", and asked
09:38
"what means LTSP rocks?"
09:38
Which caused much merriment.
09:38
Since then, it's been an #ltsp inside joke.
09:39
Welcome to the club. Here's your secret handshake :)
09:39
<alkisg>
Heh... I also wondered myself about that LTSPROCKS the first time i saw it :)
09:39
<sbalneav>
yeah
09:40
LTSP is just full of inside jokes relating to travel, pink floyd, Yes, food, etc.
09:41
Sometime when I see you in person, Jim and I will tell you about the "salmon chanted omlette" joke.
09:41
<alkisg>
I've heard some lobster jokes, but not any Pink Floyd ones
09:42
<sbalneav>
Keep and eye out, you'll see us say things like "wish you were here", etc.
09:43
<alkisg>
So... Pink Floyd rock :)
09:45
In 27 of July there's a Deep Purple concert here in Greece... I've bought tickets already, I hope I'll be able to watch it... A good rock era is about to close soon :(
09:47
<sbalneav>
Oh?
09:47
Is deep purple disbanding?
09:57
<alkisg>
No, but they're 70 y.o., I don't think they'll be able to tour around the world for long...
10:10Faithful has joined #ltsp
10:12bobby_C has quit IRC
10:36bobby_C has joined #ltsp
10:36bencrisford has joined #ltsp
10:46bobby_C has quit IRC
11:05vagrantc has joined #ltsp
11:10bencrisford has quit IRC
11:20litlebuda has joined #ltsp
11:53* stgraber plays a bit more with JS and AJAX, that management interface is going to rock :) Hope to have most of it ready by next hackfest
11:54
<stgraber>
for now I have: server pool list (with hardware spec, server load, number of users, ...), user list, killing user session and listing user process
11:55
all the being extremely fast, no load time over a second with a pool of 40 servers and several thousands users
11:55
<vagrantc>
stgraber: some web interface for ltsp-cluster?
11:55
<stgraber>
information are requested only when needed and filtering is done on the client side with JS
11:56
vagrantc: yeah, our old interface sucked so I'm working on the new one
11:57
I haven't started reimplementing the lts.conf generation part though, that'll be the tricky part as the thin clients is viewed in some kind of tree and you have to support inheriting and funny stuff like that but still be hyper fast
11:57
we some times have to generate over 200 lts.conf a minute so the SQL can't be slow
11:57
<vagrantc>
wow
11:58
<stgraber>
http://www.stgraber.org/download/win-linux.png
11:58
that's a view of two appservers, one on Linux, the other Windows
11:58
the UI is translated with gettext
11:58
depending on the language of the browser
11:58
(in this case it was a test of the french translation)
12:02
<vagrantc>
heh. i've never actually seen the ltsp-cluster stuff, or even really looked into it
12:04
<alkisg>
stgraber: so, how many TCs do you have per server? (=simple division, not taking into account the active/inactive ones)
12:07
<stgraber>
usually between 100 and 150 depending on the hardware
12:07
<alkisg>
Woah! :)
12:08
<stgraber>
well, that's usually 2xXeon quadcore at 3.0Ghz and some 8GB of RAM :)
12:08
<alkisg>
That's a good price/efficiency ratio in any case
12:08
<stgraber>
yeah
12:09
that kind of hardware cost like 3000-4000$ so for 100 users that's really not a lot
12:09
especially as we also use them for remote access during the evening (ltsp-cluster's loadbalancer also does NX loadbalancing for remote access on the same cluster)
12:17* vagrantc just got an openbrick-e via thin client with 3 NICs
12:18
<vagrantc>
now i can see how broken the boot process is with multiple NICs...
12:21* alkisg wishes IPAPPEND 2 or 3 was used by default in LTSP :(
12:22
<vagrantc>
hmmm... two of the NICs fail to pxe boot ...
12:22
and the one that suceeds with PXE, fails on initramfs dhcp ...
12:23
<alkisg>
vagrantc: you may put ip=:::::eth2: in pxelinux.cfg/default to select a nic
12:23
<vagrantc>
alkisg: doesn't one of the ipappend options add that?
12:24
<alkisg>
vagrantc: not exactly. IPAPPEND 2 adds the mac address of the boot interface, but our init scripts don't use it
12:24
<vagrantc>
ah.
12:24
<alkisg>
...and IPAPPEND 3 would also pass the IP, so it wouldn't use dhcp (no big deal since ipconfig isn't a real dhcp client)
12:35
<vagrantc>
alkisg: where does it append the mac address?
12:36
<alkisg>
it appends a kernel parameter something like BOOT_IFACE=ma:ca:dd:re:ss
12:36
<vagrantc>
alkisg: where does it append the mac address?
12:36
pfft.
12:36
<alkisg>
vagrantc: it appends it as a kernel parameter.. ?
12:36
<vagrantc>
yeah, got it.
12:36
<alkisg>
Ah , ok: )
12:37* vagrantc did a double-type
12:37
<vagrantc>
it seems to be loading the wrong module for the network cards...
12:38
the 8139cp drive is complaining to try the 8139too driver
12:44
alkisg: we should add support for the BOOT_IFACE stuff...
12:45
<alkisg>
vagrantc: there was a patch a while ago in the ltsp-discuss (or -dev) list that did just that
12:46
<vagrantc>
alkisg: posted by you, or someone else?
12:46
<alkisg>
Someone else
12:48
<vagrantc>
but thankfully in response to your pointer :)
12:48
<alkisg>
Heh, true :)
12:49
vagrantc: while doing that, it would be really nice to also support an SERVER kernel parameter. It's really really valuable
12:49
<vagrantc>
alkisg: could you elaborate?
12:49
<alkisg>
E.g. ipconfig doesn't support proxydhcp offers, so it can't locate the ltsp server (unless IPAPPEND 3 is used, which isn't really a good implementation but a hack)
12:50
So, if I put in pxelinux.cfg/default this line:
12:50
<vagrantc>
with nfs, you just do: nfsroot=ip.of.server:/opt/ltsp/i386
12:51
<alkisg>
DEFAULT vmlinuz ro initrd=initrd.img SERVER=10.160.31.10
12:51
<stgraber>
nbdroot works with the same syntax too though I'm not sure LDM_SERVER then gets set to the right value
12:51
<alkisg>
then the initscripts should try to contact the ltsp server at 10.160.31.10 (I'm suggesting SERVER because that's also used by lts.conf)
12:52
stgraber: it should also be used as the tftp server, to get lts.conf
12:52
<vagrantc>
i think LDM_SERVER defaults to SERVER, which defaults to wherever your root filesystem is mounted.
12:53
<alkisg>
The problem is that in many configurations ipconfig has problems getting the next-server and the filename dhcp options
12:53hanthana has quit IRC
12:53
<vagrantc>
i think it should parse it from nbd_server, rather than adding a new value, then.
12:53
<alkisg>
So it would be nice if they could be provided as kernel parameters...
12:53
<vagrantc>
since this is only in the nbd scripts
12:53
<alkisg>
OK, that's good enough for me. Accepting patches? :)
12:55
<vagrantc>
don't see why not? :P
12:55
<alkisg>
stgraber? are patches for initscripts easy to push, or they need to be accepted from some other team?
12:56
<stgraber>
is ROOTSERVER an ipconfig thing or is that a regular kernel parameter we can use ?
12:56* alkisg looks at kernel-parameters.txt...
12:56
<stgraber>
alkisg: if that's only changing ltsp_nbd that's fine
12:56
though I still have moving to udhcpc in my todolist, waiting for the MIR to be approved
12:58
<alkisg>
stgraber: if udhcpc is approved, please notify me, I'd like to make/suggest a patch to support IPAPPEND 3 with udhcpc, we use it a lot along with dnsmasq here in greece
12:59* alkisg doesn't see a rootserver option in http://www.kernel.org/doc/Documentation/kernel-parameters.txt :(
12:59
<vagrantc>
i think ROOTSERVER is an ipconfig thing
13:00
<alkisg>
But the ip=:::: parameter that pxelinux appends is formally defined, so we can use that.
13:00
I.e. ip=::<server>::::
13:02
<vagrantc>
- (echo "get ${tftpath:-/ltsp/i386}/lts.conf"|tftp ${ROOTSERVER} >/dev/null 2>&1)
13:02
<alkisg>
http://www.kernel.org/doc/Documentation/filesystems/nfsroot.txt ==> ip=<client-ip>:<server-ip>:<gw-ip>:<netmask>:<hostname>:<device>:<autoconf>
13:02
<vagrantc>
+ (echo "get ${tftpath:-/ltsp/i386}/lts.conf"|tftp ${NBD_ROOT_SERVER} >/dev/null 2>&1)
13:02
alkisg: i think that'll fix your problem ^^
13:03
unless you need a different tftp server from your nbd server
13:03
<alkisg>
vagrantc: I think it'll still need a small patch to actually get the nbd server from the kernel command line - or does it?
13:04
<vagrantc>
alkisg: it sets NBD_ROOT_SERVER by parsing it from the kernel cmdline
13:04* alkisg looks...
13:05
<vagrantc>
well, *something* sets nbdroot
13:05
though at some point, initramfs-tools changed how those variables are handled... not just blindly exporting everything from the kernel cmdline
13:06* vagrantc wonders if what the BOOTIF parameter is ... looks like 01-$(echo $MACADDRESS | tr ':' '-')
13:07
<alkisg>
That's the one passed by IPAPPEND 2
13:07
<vagrantc>
right, but what's the 01- mean?
13:07
is it always 01- ?
13:08
<alkisg>
01 = ethernet
13:08
So yeah :)
13:08
(at least *I* don't know of any other protocols used :))
13:08
<vagrantc>
it's been a while since i've done any initramfs-tools hacking...
13:09
<alkisg>
These parameters are defined in the pxelinux site
13:09
<vagrantc>
never did manage to get the DHCPPORT and DHCPVENDOR patches merged :(
13:09
<alkisg>
Where, in pxelinux? I haven't heard about them...
13:10
<vagrantc>
http://bugs.debian.org/509077
13:11
<alkisg>
Damn... setting nbdroot isn't enough, another line is needed to set ROOTSERVER from nbdserver is the former is unset
13:12
<vagrantc>
what else uses ROOTSERVER?
13:13
<alkisg>
the swap server
13:13
<vagrantc>
that should default to NBD_ROOT_SERVER instead
13:13
<alkisg>
But that could also be set with nbdswap... hmmm
13:14
<vagrantc>
yes, but the default if nbdswap isn't set should be NBD_ROOT_SERVER, no?
13:15
<alkisg>
Yes, but NBD_ROOT_SERVER defaults to ROOTSERVER
13:15
<vagrantc>
and?
13:15
<alkisg>
...and that's the router, not the ltsp server :)
13:15
<vagrantc>
that's what nbdroot= is for, no?
13:15
<alkisg>
No, that's nbdswap
13:16
<vagrantc>
are we looking at the same code?
13:16
<alkisg>
Still, your right in that if I define both nbdroot and nbdswap, and I change the line that you said, then I think it'll work
13:16
if [ -n "${nbdswap}" ]; then
13:16
NBD_SWAP_SERVER=$( echo "${nbdswap}" | sed 's/:.*//')
13:16
NBD_SWAP_PORT=$( echo "${nbdswap}" | sed 's/.*://')
13:16
else
13:16
NBD_SWAP_SERVER="${ROOTSERVER}"
13:16
NBD_SWAP_PORT=${nbdswapport:-"9572"}
13:17
<ltsppbot>
"vagrantc" pasted "change most occurrances of ROOTSERVER to NBD_ROOT_SERVER" (22 lines) at http://pastebot.ltsp.org/388
13:19
<alkisg>
Yeah, I'm fine with that :)
13:20
<vagrantc>
i'm also going to fix up some wonky whitespace in that code...
13:21
the disadvantage is that if you're serving up nbdroot on multiple networks, it's annoying to have to hard-code nbdroot in pxelinux.cfg
13:22
<alkisg>
I don't get it, is there some alternative to this?
13:23
<vagrantc>
?
13:23
controlling your own dhcp server? :P
13:23
<alkisg>
Ah, right, sure :)
13:27
Well, if IPAPPEND 3 was used, and we got the :<server>:::: from there, then it would work with multiple networks. :) Nah, too much of a hack.
13:45otavio has joined #ltsp
13:48klausade has quit IRC
13:49
<vagrantc>
alkisg: the patch to support BOOTIF involved messing around with udev rules and reassigning which device was eth0 ... which seems a little hackish ...
13:49bobby_C has joined #ltsp
13:49
<alkisg>
Agh, yes, it is :)
13:49
<vagrantc>
alkisg: if we could tweak it to just reset DEVICE if BOOTIF is set, that would work better.
13:52
<alkisg>
Where's DEVICE mentioned? In configure_networking?
13:52
<vagrantc>
yeah
13:52
and conf/initramfs.conf
13:52
which is hard-coded. so a patch to dynamically assign device would be a good thing.
13:54* alkisg always forgets where the configure_networking function is in :) Looking in /etc...
13:55
<stgraber>
vagrantc: hmm, your change won't work if my nbd and tftp server are different, it'll download lts.conf from the nbd server and not the tftp one ...
14:01klausade has joined #ltsp
14:03Lumiere is now known as Lumi|USA2|BRA0
14:12Faithful has quit IRC
14:15
<vagrantc>
stgraber: true enough ...
14:15
stgraber: i suppose you have such setups, eh?
14:21
i suppose we could add a TFTPSERVER variable...
14:21
<alkisg>
or an ltspserver kernel parameter
14:22
<vagrantc>
alkisg: yes, but what does ltspserver mean if you have tftp on one server and nbd on another?
14:22
<alkisg>
ltsp==tftp AFAIK
14:22
<vagrantc>
apparently, stgraber raised a concern that they will be on different servers.
14:23
<alkisg>
Eh, I mean where ldm is going to connect to
14:23
ltspserver == override rootserver. That's all.
14:23
(or we could even name it rootserver to be clear)
14:23
<vagrantc>
i think that's unclear, as LTSP servers can be the tftp part, the NFS/NBD part, and the application server. so i object to calling it ltspserver
14:24
with rootserver we're back to square 1, as rootserver is clearly the nbd server, no?
14:24
<alkisg>
Nope. Rootserver is the dhcp server that ipconfig contacted
14:25
<vagrantc>
sure, but in the context of LTSP, the thing that serves up the root filesystem is the rootserver, no?
14:26
it's merely a poor assumption on ipconfig's part to assume the next-server is the same as the root server.
14:26
well, it's a reasonable assumption, actually :)
14:26
<alkisg>
Is it?
14:26
It could keep the rfc name for it
14:26
<vagrantc>
well, an LTSP server consists of tftp and nfs or nbd, basically.
14:27
look at the differences between the ltsp-server and ltsp-server-standalone packages ...
14:27
<alkisg>
...and the application server?
14:27
<vagrantc>
application server is *frequently* on the same server, but not always.
14:28
<alkisg>
Well, stgraber just said that nbd could also be on a different server
14:28
<vagrantc>
that's not the *core* of what LTSP is... LTSP is really the network booting OS.
14:28
alkisg: yes, it's possible.
14:28
but that's not the typical setup...
14:28
so it's the sort of thing that needs an override.
14:28
<alkisg>
Well, ltspfs, audio handling etc... isn't that also ltsp? Isn't that part of the application server?
14:29
Hrm... ok, not really
14:29
<vagrantc>
alkisg: sure, but i wouldn't say that's the core of LTSP ... those are additional features, but they wouldn't be much use without the network booting part.... or it wouldn't be LTSP without the network booting part.
14:30
maybe we're just talking semantics here...
14:30
<alkisg>
Anyway. The "rootserver" name is defined by ipconfig, so it isn't "ours" to change... so what if we just re-used it as a kernel parameter?
14:30
<vagrantc>
i bet it will override it.
14:30
<alkisg>
Yes, what I'm saying is that our scripts should check the kernel command line and override ROOTSERVER with rootserver
14:33* alkisg wonders where is nbdroot read from... can't find it in init
14:34
<vagrantc>
it might be relying on old behavior of initramfs-tools just exporting all kernel commandline options as variables.
14:34
which it doesn't do anymore.
14:34
<alkisg>
If so, it won't even work anymore :(
14:34
So how's stgraber operating his nbd server?
14:35
<vagrantc>
or at least, i don't think the version in debian did anymore...
14:37* alkisg also doesn't see init accepting *) kernel parameters in jaunty
14:38
<vagrantc>
ah, it does seem to work.
14:38
well, at least, from break=bottom
14:39
now i just need to figure out how to re-write the configure_networking function to support $BOOTIF
14:43
<alkisg>
something with ip ... show, cutting the dev name and the mac address, and greping for $BOOTIF, I guess...
14:44
<vagrantc>
something to that effect, sure...
14:45
<alkisg>
Hm.. I can't get a client to boot with nbdroot and break=init
14:54
OK, it seems that it accepts any kernel parameter as an env var.
14:54
So, this line would be enough to allow passing rootserver:
14:54
ROOTSERVER=${rootserver:-$ROOTSERVER}
14:55
<vagrantc>
alkisg: does "IPAPPEND 3" also set BOOTIF ?
14:55
<alkisg>
Yes
14:55
But it prevents dhcp from happening
14:57
<vagrantc>
sure
14:57
but ipconfig is still used to configure the networking
15:01
ok, i'm going to modify update-kernels to support IPAPPEND... a quick hack for what should be much more flexible...
15:04
<alkisg>
vagrantc, stgraber: could you also put this line right after "configure_networking" ?
15:04
ROOTSERVER=${rootserver:- $ROOTSERVER} # There's an extra space before the $ to prevent smilies from interfering
15:06
This allows for a rootserver kernel parameter. If it's unset, nothing is changed; if it's set, then one can specify the location of the ltsp server.
15:06
I've tried it, works fine for me
15:08
<vagrantc>
alkisg: don't quite get the extra space bit?
15:08
<alkisg>
My pidgin turns : - $ into a smily when they're together
15:09
<vagrantc>
ah.
15:09
<alkisg>
So I put an extra space just for the irc clients, it should be removed if it's put in the script
15:09
<vagrantc>
sure
15:12
alkisg: so, basically, this is to override what the dhcp server thinks is the next-server...
15:12
<alkisg>
Yes.
15:12
Useful in the dnsmasq/proxydhcp scenario, because ipconfig doesn't support getting proxy dhcp offers
15:13
opensuse has something similar - it names it kiwiserver : http://en.opensuse.org/LTSP/Tips_and_Tricks#Using_another_DHCP_server.2C_not_on_LTSP_server
15:15
It's also useful when the ltsp dhcp3-server coexists with an external dhcp server. Not a recommented setup, sure, but hacks like these are necessary in some configurations.
15:17
<vagrantc>
well, i've got a patch that supports BOOTIF
15:18
might have to re-work it a little to get it into initramfs-tools, but it's a proof of concept, at least.
15:22rjune has quit IRC
15:27
<alkisg>
vagrantc, could you pastebin it?
15:29
<ltsppbot>
"vagrantc" pasted "patch to initramfs-tools to support BOOTIF" (24 lines) at http://pastebot.ltsp.org/389
15:29
<vagrantc>
filing a wishlist bug in debian
15:31
<alkisg>
vagrantc: is ifconfig available in the initramfs?
15:31
<vagrantc>
it is in debian...
15:35
<alkisg>
Yup, in Ubuntu also. I didn't see any other script using it, so I mistakenly thought it wasn't available.
15:36
<vagrantc>
i'm sure i'll get complaints about using so many binaries, and possibly ifconfig complaining too, but it at least gives something to get reviewed :)
15:37
it's a lot harder to get the mac address with ip, as it doesn't list the interface on the same line as the device
15:38
<alkisg>
ip -o link show
15:38
<vagrantc>
ah, much better :)
15:38
it's hard to use ip because it's got so many options :)
15:38
<alkisg>
But it has some trouble with my wireless, it lists both wlan0 and wmaster with the same hwaddress
15:39
So leave ifconfig there until (and if) something better is found :)
15:41bobby_C has quit IRC
15:41
<vagrantc>
alkisg: same here for wlan0 and wmaster ... but one is link/ether and one is link/ieee802.11
15:42
<alkisg>
vagrantc: and I think you can get sed out with ${BOOTIF#*-}
15:43
<vagrantc>
but i want to change - to :
15:44
<alkisg>
That would even work if the first 01- is some other protocol
15:44
That's what tr is for
15:44
and you already use tr
15:44
<vagrantc>
then i'm changing out sed for tr ...
15:44
<alkisg>
You already have tr, though
15:44
<vagrantc>
could probably do some other magic with shell expansion ...
15:45
<alkisg>
You could also get rid of awk
15:45
<vagrantc>
tell me more...
15:45
<alkisg>
Gimme a BOOTIF example
15:46
01-A1-B2-C3-D4-E5-F6 ?
15:47
ah, lowercase
15:47
<vagrantc>
yeah, though i think ip uses uppercase
15:48
er, lowercase
15:48
<alkisg>
ok, trying...
15:49
<vagrantc>
always wondered if this stuff was available in /proc
15:51
unreliable to rely on proc, though... changes too often.
15:52
<alkisg>
normal_mac="$(echo ${BOOTIF#*-} | tr [A-Z\-] [a-z:])"
15:53
now for the hard part... :(
15:54
<vagrantc>
ip -o link show doesn't work from the initramfs
15:55
<alkisg>
ah
15:56
it works for me! :-/
15:57
<vagrantc>
this is using busybox ip"
15:58
<alkisg>
Is that with break=top / bottom / init ?
15:58
<vagrantc>
break=bottom
15:58
<alkisg>
(it shouldn't make any difference, but...)
16:00
vagrantc: ip -onelink link show
16:00
<vagrantc>
is yours using the "real" ip command, or busybox's?
16:00
<alkisg>
I think it's a bug
16:01
I used "init=/bin/bash" to break, so maybe it was the real ip command
16:01
vagrantc: -oneline I meant
16:02
`ip -oneline link show` works with both busybox and real ip
16:02
<vagrantc>
alkisg: "ip -oneline link show" works for me
16:02
in the initramfs
16:03
<alkisg>
It says "options = ... -o[neline] " - I guess -o should be supported but it isn't declared in the getopts...
16:05
<vagrantc>
how i'm going to get that mac address out of there without awk is beyond me, though :)
16:06
<alkisg>
You could use read
16:06
<vagrantc>
egads...
16:07
<alkisg>
read index iface .... hwaddress
16:07
if [ $hwaddress = $normal_mac ] then iface .... something like that
16:07
But I don't know if it's worth it :)
16:09
(all that in a `while read`, of course... - nah, not worth it)
16:11
How do you use the alternate dhcp port patch with PXE clients?
16:15
<vagrantc>
alkisg: i think it was used with custom etherboot images ... probably support for it in gPXE, too.
16:15
<alkisg>
No, it was dropped in gpxe. Anyway I think proxy dhcp is better than custom port
16:15
It also works with real PXE clients
16:15
<vagrantc>
possibly.
16:16
right, that is nice.
16:17
<alkisg>
I hope someone gets proxy dhcp support to dhcp3-server as well...
16:17
<vagrantc>
what about dhcp-helper, dhcp3-relay ?
16:17
<alkisg>
OK, it's late here, goodnight vagrantc... thanks for your work
16:18
No, I don't think those can help
16:18
They're useful when the dhcp server is on another subnet
16:18
But not with an external dhcp server in the same subnet
16:18
<vagrantc>
ah, got it.
16:19
<alkisg>
g'night :)
16:19
<vagrantc>
alkisg: good night!
16:19alkisg has quit IRC
16:51
<vagrantc>
alas, a bash-ism i actually want ... ${parameter//-:/} could replace tr '\-' ':'
16:55
<panthera>
vagrantc: are you comming to debconf?
16:55
<vagrantc>
panthera: yes!
16:55
panthera: see you there?
16:57
<panthera>
vagrantc: of course ;)
16:58
vagrantc: will be nice to see you again.
16:58
<vagrantc>
panthera: what are you working on these days?
16:59* vagrantc didn't notice panthera lurking in #ltsp before
16:59* panthera wants to do a new target in the image build script for debian-live soon.
17:00
<panthera>
(one that doesn't do all the, for us, useless thin client hacks)
17:00
that's why.
17:00
<vagrantc>
i long lurked in debian-live, and never took the time to dive into it
17:02
<panthera>
indeed, quite some time passed by..
17:23sene has joined #ltsp
17:44GodFather has joined #ltsp
17:46GodFather has quit IRC
17:47joao has joined #ltsp
17:50hanthana has joined #ltsp
17:58Faithful has joined #ltsp
17:58Faithful1 has joined #ltsp
17:58Faithful11 has joined #ltsp
17:58Faithful has quit IRC
17:58Faithful11 has quit IRC
17:58Faithful1 has quit IRC
17:58Faithful has joined #ltsp
17:59cliebow has joined #ltsp
18:06litlebuda has quit IRC
18:08sene has quit IRC
18:08GGD is now known as GGD_away
18:09sene has joined #ltsp
18:10joao has quit IRC
18:14sene has quit IRC
18:14hanthana_ has joined #ltsp
18:14sene has joined #ltsp
18:16sene has joined #ltsp
18:21J45p3r__ has joined #ltsp
18:21J45p3r__ has left #ltsp
18:29hanthana has quit IRC
18:32cliebow has quit IRC
18:34cliebow has joined #ltsp
18:38lucascoala has joined #ltsp
18:38sene has joined #ltsp
18:39sene has joined #ltsp
18:40sene has quit IRC
18:52hanthana_ has quit IRC
19:13fotanus has joined #ltsp
19:49fotanus has quit IRC
19:50fotanus has joined #ltsp
19:59pmatulis has quit IRC
20:02pmatulis has joined #ltsp
20:02pmatulis has joined #ltsp
20:06pmatulis has quit IRC
20:07pmatulis has joined #ltsp
20:10try2free has joined #ltsp
20:19
<try2free>
hi
20:21
How to solve No sound using Intel Corporation 82801EB/ER? please help.
20:21
i am using ubuntu 8.04
20:26vagrantc has quit IRC
20:56cliebow has quit IRC
21:00try2free has quit IRC
21:00
<sbalneav>
:)/win 4
21:35lucascoala_ has joined #ltsp
21:43try2free has joined #ltsp
21:48lucascoala has quit IRC
22:04try2free has quit IRC
22:35try2free has joined #ltsp
22:35
<try2free>
hi
22:35
sorry, my connection drop...
22:35
How to solve No sound using Intel Corporation 82801EB/ER? please help.
22:58Egyptian[Home]1 has joined #ltsp
23:00alkisg has joined #ltsp
23:03Egyptian[Home] has quit IRC
23:04ogra has quit IRC
23:05ogra has joined #ltsp
23:18lucascoala_ has quit IRC
23:29try2free has quit IRC
23:49alkisg has quit IRC