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


Channel log from 24 March 2011   (all times are UTC)

00:41bobby_C (~bobby@85-124-22-227.teleworker.xdsl-line.inode.at) joined #ltsp.
00:47alkisg (~alkisg@ubuntu/member/alkisg) joined #ltsp.
01:01xavier_brochard (~xavier@82.254.253.34) joined #ltsp.
01:04gnunux (~emmanuel@194.167.18.244) joined #ltsp.
01:15dobber (~dobber@89.190.197.77) joined #ltsp.
01:23
<mnemoc>
!screens
01:24
<muppis>
mnemoc, I think our bot has found a better place to be.. :(
01:24
<mnemoc>
oh :(
01:41xavier_brochard (~xavier@82.254.253.34) left irc: Ping timeout: 255 seconds
01:46xavier_brochard (~xavier@82.254.253.34) joined #ltsp.
02:05xavier_brochard (~xavier@82.254.253.34) left irc: Ping timeout: 252 seconds
02:07otavio (~otavio@debian/developer/otavio) left irc: Ping timeout: 255 seconds
02:07xavier_brochard (~xavier@82.254.253.34) joined #ltsp.
02:15ogra_ (~ogra@ubuntu/member/ogra) left irc: Excess Flood
02:15ogra_ (~ogra@p5098ed03.dip0.t-ipconnect.de) joined #ltsp.
02:16ogra_ (~ogra@p5098ed03.dip0.t-ipconnect.de) left irc: Changing host
02:16ogra_ (~ogra@ubuntu/member/ogra) joined #ltsp.
02:25alkisg (~alkisg@ubuntu/member/alkisg) left irc: Quit: Leaving.
02:40Da-Geek (~Da-Geek@nat/redhat/x-snbdpdnohwecyrrm) joined #ltsp.
03:22tobias (~quassel@c-52a8e155.94-9-64736c11.cust.bredbandsbolaget.se) joined #ltsp.
03:23Nick change: tobias -> Guest89595
03:24Nick change: Guest89595 -> tobias
03:24Nick change: tobias -> Guest37872
03:25Nick change: Guest37872 -> tompa82
03:34wwx (~ww@raamat.polva.ee) left irc: Read error: Connection reset by peer
03:35wwx (~ww@raamat.polva.ee) joined #ltsp.
04:18otavio (~otavio@189.114.111.55) joined #ltsp.
04:18otavio (~otavio@189.114.111.55) left irc: Changing host
04:18otavio (~otavio@debian/developer/otavio) joined #ltsp.
04:27MorningSon (~MorningSo@cpe-70-114-21-95.satx.res.rr.com) joined #ltsp.
04:27Nick change: zz_evil_root -> evil_root
04:39tompa82 (~quassel@c-52a8e155.94-9-64736c11.cust.bredbandsbolaget.se) left irc: Remote host closed the connection
05:01xavier_brochard (~xavier@82.254.253.34) left irc: Read error: Operation timed out
05:02xavier_brochard (~xavier@82.254.253.34) joined #ltsp.
05:08
<mgariepy>
good morning everyone
05:13
<knipwim>
hi there
05:14
mgariepy: so ltsp-remoteapps runs on the client together with ltsp-open?
05:14
while ltsp-remoteappsd is listening on the server and providing the mounts to the client?
05:21jtwood (~jtwood@74.81.111.2) joined #ltsp.
05:22
<mgariepy>
knipwim, on the thin clients /usr/share/ldm/rc.d/X01-remoteapps and X51-remoteapps mount the shared directory
05:22
ltsp-remoteappd is running on the thin client
05:23
the programs are started by ssh by ltsp-remoteappsd
05:28
<knipwim>
i got more questions still, but i got to go to a meeting ...
05:28
<mgariepy>
ok no problem, i'll be here all day i guess :)
05:28
talk to you later
05:32knipwim (~wim@ip4da83870.direct-adsl.nl) left irc: Read error: Operation timed out
05:35knipwim (~wim@ip4da83870.direct-adsl.nl) joined #ltsp.
05:51shawnp0wers (~Adium@linuxjournal/staff/shawnp0wers) joined #ltsp.
06:56Trixboxer (~Trixboxer@office.supportdepartment.net) joined #ltsp.
07:15shawnp0wers (~Adium@linuxjournal/staff/shawnp0wers) left irc: Quit: Leaving.
07:19matrix3000 (~matrix@rrcs-70-61-255-226.central.biz.rr.com) joined #ltsp.
07:27Gadi (~romm@ool-18bbe47a.static.optonline.net) joined #ltsp.
07:33alkisg (~alkisg@ubuntu/member/alkisg) joined #ltsp.
07:42alkisg (~alkisg@ubuntu/member/alkisg) left irc: Quit: Leaving.
07:51andygraybeal (~andy.gray@obsidian.casanueva.com) left irc: Quit: Ex-Chat
07:53
<jtwood>
Whats the best way to share printers to the clients? I've got a samba print server...
07:54
<Gadi>
fat or thin?
07:55
<jtwood>
both
07:55
:)
07:55
of course, rigHT!
07:55
I figure the thin clients would be easy actually...
07:56
<Gadi>
yeah, so for the thin clients, you can make the server a client to the print server
07:56
(is it samba and cups print server?)
07:56
<jtwood>
Yeah
07:56
<Gadi>
yeah, so that's easily done with /etc/cups/client.conf
07:57
for the fat clients, set: CUPS_SERVER=<ip> in lts.conf
07:57
(I believe thats the var)
07:57dgroos_ (~dgroos@ip67-152-153-47.z153-152-67.customer.algx.net) joined #ltsp.
07:57
<jtwood>
I can find it..
07:57xavier_brochard (~xavier@82.254.253.34) left irc: Remote host closed the connection
07:58
<Gadi>
oh, y'know what
07:58
<jtwood>
?
07:58
<Gadi>
scratch that lts.conf var for fat client - that only works for local apps
07:58
<jtwood>
OH YEAH...didn't think about that
07:59
<Gadi>
for fat client, you should drop the same client.conf in the chroot
07:59
<jtwood>
yeah...that makes sense actually...sometime I think too much
07:59
<Gadi>
client.conf just looks like: ServerName <ip>
07:59
btw
08:00
<jtwood>
I am really impressed with the fat client setup...its been very easy once I alkisg and I walked through everything and realized I needed to mod my inetd.conf
08:00
So many configuration files --- ugh...easy to forget one
08:01
<Gadi>
yeah - I think if I were doing fat client, I would use NFS just to avoid the constant chroot updates
08:02
<jtwood>
Excuse me for being completely ignorant...not exactly sure of what is going on between the two...what are the advantages/disadvantages LTSPwise
08:02
<Gadi>
meaning NBD vs NFS?
08:03
<jtwood>
Yeah...but just the LTSP significance
08:03
<Gadi>
well, in our NBD approach, we create a squashfs (compressed) filesystem image out of the chroot and export it as a hard drive over the network
08:03
because it is compressed, there is less network traffic and faster reads
08:04
which means faster booting
08:04dgroos_ (~dgroos@ip67-152-153-47.z153-152-67.customer.algx.net) left irc: Quit: dgroos_
08:04
<Gadi>
but since it is an image *built from* the chroot, when you change the chroot, you have to update it
08:04
hence, ltsp-update-image
08:04
<jtwood>
Oh
08:04
so with NFS
08:04
You run directly from the chroot?
08:05
<Gadi>
yeah, with NFS< you are exporting the chroot itself and mounting it read only
08:05
<jtwood>
Yeah...
08:05
<Gadi>
so, changes you make to the chroot are reflected immediately on the clients
08:05
<jtwood>
I must have taken my idiot pills today -- cause I realize that
08:05
Doh...
08:05
<Gadi>
but, because it is uncompressed and rather large, there is more network traffic and longer booting
08:06
<jtwood>
:_
08:06
<Gadi>
another perk to NBD is it is all contained in one file
08:06dgroos_ (~dgroos@ip67-152-153-47.z153-152-67.customer.algx.net) joined #ltsp.
08:06
<Gadi>
which you can easily back up, move around, etc
08:07
<jtwood>
Yeah, definitely that...too...
08:07
<Gadi>
so, pros and cons to everything
08:07
<jtwood>
Oh, I know...
08:07
I am already catching the cons from users of trying to take away their Windows machines
08:07
<Gadi>
but, once you start running apps in your chroot, the updates increase dramatically over a thin chroot
08:07
<jtwood>
But it seems like everyday we are fighting MS
08:08
These updates...is that just when you actually go into the chroot and update or add packages...right?
08:08
<Gadi>
yeah, thats what I mean
08:09
<jtwood>
Okay...yeah...that is definitely something to think about
08:09
More than likely I will roll out testing with NFS then...think about switching over to NBD after things somewhat stabilize
08:09
<Gadi>
so, if you are the type to update your system as often as updates are available, you might consider it
08:09
of course,doing it with nbd gives you a nice way to backup the last image
08:09
<jtwood>
Yeah
08:09
<Gadi>
so in case an update breaks things, you don't freak
08:09
:)
08:10
<jtwood>
I am paranoid about that...I have a tendency to backup the chroot and the image before adding software, reconfiguring...yadayada
08:10
Amanda has saved my hide a few times too!
08:10
<Gadi>
heh
08:11
<jtwood>
Well, we got lucky and snagged a few Pogo Linux FS on ebay a little while ago...nothing super but they weren't all that old
08:12
packed them full of drives and using one for Samba FS, one for Amanda, and one for LTSP - temporarily
08:13
Well, Gadi...thanks for the suggestions...and the help
08:13
I am going to go mess around with it a bit...I appreciate it as always...
08:13
<Gadi>
cheers
08:13
<jtwood>
Thank you -- good day if I don't catch you later...stay warm
08:24Appiah (~appiah@ip69.hethane.riksnet.nu) joined #ltsp.
08:29shawnp0wers (~Adium@linuxjournal/staff/shawnp0wers) joined #ltsp.
08:29
<shawnp0wers>
Does anyone know offhand if the traditional way (via lts.conf) to set up thin clients as print servers still works if clients are fat clients?
08:36
<Gadi>
it should
08:40
<shawnp0wers>
Thanks Gadi  I realize I could have set up a fat client, etc, etc, but& well I'm sorta lazy. ;)
08:41
<Gadi>
I hear ya - lazy people should have taken over the world by now....
08:44xavierb (~xavier@82.254.253.34) joined #ltsp.
08:48
<xavierb>
@Gadi, you said to jtwood that there is more network traffic with NFS, but that s only true for boot, right?
08:49
<Gadi>
right - or whenever uncached files in the chroot are accessed
08:49garymc (~chatzilla@host81-148-15-59.in-addr.btopenworld.com) joined #ltsp.
08:50
<xavierb>
what sort of uncached files?
08:53dgroos_ (~dgroos@ip67-152-153-47.z153-152-67.customer.algx.net) left irc: Quit: dgroos_
08:56Adriannom (~adrian@cpc10-cwma7-2-0-cust15.7-3.cable.virginmedia.com) joined #ltsp.
08:57
<Gadi>
xavierb: both NFS and NBD have some inherent caching so if the same file is accessed twice it can retrieve it from memory rather than the drive
08:59
so, in our case, you can say that whenever the thin client needs to access a new file from the chroot, with NBD, there will be less bandwidth used than with NFS
08:59
typically, the time when thin clients are accessing the most new files is on boot
09:01dgroos_ (~dgroos@ip67-152-153-47.z153-152-67.customer.algx.net) joined #ltsp.
09:08dgroos_ (~dgroos@ip67-152-153-47.z153-152-67.customer.algx.net) left irc: Quit: dgroos_
09:09Gadi (romm@ool-18bbe47a.static.optonline.net) left #ltsp.
09:12drdata (~drdata@tw2.udallas.edu) joined #ltsp.
09:19Roasted_ (~jason@206.82.22.190) joined #ltsp.
09:23
<abeehc>
Roasted_: as it happens someone corrected my misinformation again if you check the logs
09:24
<Roasted_>
abeehc, in regard to what?
09:25
<abeehc>
booting many clients etc
09:26
<Roasted_>
ah
09:26
well I'm trying to bond my two NIC's now
09:26
to see if that would help
09:26
previously I was only using 1
09:26
<abeehc>
word that's a good call
09:26
smarter person than me said your clients would get faster
09:26
i don't thinkg that's strictly booting
09:26
<Roasted_>
I've never bonded NICs before, and I followed the guide I read but my 2nd NIC is still pulling DHCP addys
09:27
If I scatter the boot process 50/50 they all boot. If I run around like a nut and power all 30 on at same time there's always a few that don't boot
09:27
a matter of 3 or 4, but still. enough of a nusance I want to figure out what's up
09:27
<abeehc>
i can't; find it in the damn log page
09:27Lns (~Lns@pdpc/supporter/professional/lns) joined #ltsp.
09:28
<abeehc>
http://pastebin.com/4zBUspbB
09:28Da-Geek (~Da-Geek@nat/redhat/x-snbdpdnohwecyrrm) left irc: Quit: Leaving
09:28
<abeehc>
check that bug it looks popular
09:28
<Roasted_>
ah yeah, I saw that
09:28
I didn't read through the bug though
09:29gnunux (~emmanuel@194.167.18.244) left irc: Quit: Ex-Chat
09:30
<Roasted_>
looks like that bug has a workaround tho
09:30
I'm not sure if it's *my* problem though, but I'll save it for sake of being aware
09:40dobber (~dobber@89.190.197.77) left irc: Remote host closed the connection
09:56xavierb (xavier@82.254.253.34) left #ltsp ("Konversation terminated!").
10:02
<knipwim>
mgariepy: meeting over, and at home :)
10:02dgroos (~dgroos@63.225.132.145) joined #ltsp.
10:02
<vvinet>
!m
10:04ogra_ (~ogra@ubuntu/member/ogra) left irc: Excess Flood
10:04
<mgariepy>
knipwim, what do you want to know?
10:04ogra_ (~ogra@p5098ed03.dip0.t-ipconnect.de) joined #ltsp.
10:06ogra_ (~ogra@p5098ed03.dip0.t-ipconnect.de) left irc: Changing host
10:06ogra_ (~ogra@ubuntu/member/ogra) joined #ltsp.
10:07
<knipwim>
okay, when i'm in a local app, what triggers the opening of a remote app? is this done by ltsp-open?
10:08
<mgariepy>
yeah, ltsp-open is responsible to copy the file somewhere it can be accessed by the applications server then it runs xdg-open on the server for this file.
10:09
<knipwim>
so all (ltsp-remoteapps, ltsp-remoteappsd and ltsp-open) exist on the client
10:10
<mgariepy>
yes exactly
10:10
we use mailcap to make ltsp-open appear in firefox or chromium
10:11
<knipwim>
the entries in the mailcap file determine which filetypes are opened remotely and which locally
10:13
<mgariepy>
yes
10:14
usually local apps have a higher priority then remote apps
10:16
<knipwim>
check, the understanding is beginning to emerge
10:17
now time to try it out
10:17
thanks
10:18ogra_ (~ogra@ubuntu/member/ogra) left irc: Excess Flood
10:18
<mgariepy>
knipwim, what are you trying to do exactly ?
10:19ogra_ (~ogra@p5098ed03.dip0.t-ipconnect.de) joined #ltsp.
10:19
<knipwim>
to include the ltsp-remoteapps functionality in the gentoo ltsp-package
10:19
<mgariepy>
ok cool
10:19
<knipwim>
and add some documentation in the gentoo wiki
10:20
<mgariepy>
this is a nice feature remote app for LTSP, it makes it almost like a normal desktop :)
10:21alkisg (~alkisg@ubuntu/member/alkisg) joined #ltsp.
10:26
<Roasted_>
Is there a guide you guys recommend using with bonding NICs? Each guide I've found either hasn't worked or is for a much older version of Ubuntu.
10:27
<alkisg>
Roasted_: that's the last one I wrote when I tried bonding back in 8.10: https://help.ubuntu.com/community/UbuntuLTSP/Trunking
10:27
haven't tried it since
10:27
<Roasted_>
9.10 vs 10.04 is where it seems things changed :(
10:27
I doubt 8.10 would help
10:28
I really hate to only use one NIc when I have two here
10:28
<alkisg>
jtwood: compressed NBD is about 3-4 times faster than NFS, *not only while booting*, but while running apps as well
10:28
E.g. openning openoffice on a fat client with NFS takes 20 secs and 5 secs with compressed NBD
10:28
But with the default ubuntu setup NBD is no longer compressed
10:28
Roasted_: are they pci or pci-e?
10:29
<Roasted_>
alkisg, I'm not sure. It's built into the server board... I would assume E...
10:29
alkisg, what do fat clients use by default? NFS or NBD?
10:29
<alkisg>
nbd
10:29
<Roasted_>
gotcha
10:29
<alkisg>
If they're pci it won't make a difference as the pci bus speed is about 1 gigabit anyway
10:30
If they're pci-e, you'll double your bandwidth
10:30
<Roasted_>
they';re PCIE
10:30
according to lspci
10:30
<alkisg>
Are your clients and your switch gigabit?
10:30
<Roasted_>
yes
10:30
<alkisg>
Yup, then do that, it'll help a lot
10:30
<Roasted_>
I'm trying.
10:30
<alkisg>
Where did you get stuck?
10:30
<Roasted_>
When I had no more internet on the server :(
10:31
I created the file in /etc/modprobe.d
10:31shawnp0wers (~Adium@linuxjournal/staff/shawnp0wers) left irc: Quit: Leaving.
10:31
<Roasted_>
added the 2 alias lines that it said to add
10:31
using mode=6 or whatever
10:31
I installed ifenslave and then I edited the interface file.
10:32
<alkisg>
While file in modprobe.d?
10:32
*what
10:34
<Roasted_>
bonding.conf
10:34
<alkisg>
What tutorial are you following?
10:36
<Roasted_>
https://help.ubuntu.com/community/UbuntuBonding
10:36
<alkisg>
Never tried that one, don't know
10:37
<jtwood>
The best advice for bonding is in the ifenslave docs. There is an example included...works like a charm
10:39Gremble (~Ben@cpc12-aztw24-2-0-cust146.aztw.cable.virginmedia.com) joined #ltsp.
10:40
<jtwood>
alkisg: Thanks for the advice on NBD vs NFS - It really just takes a minute to update the image - Gadi had a good point, but the update time isn't bad yet... ;)
10:40
<alkisg>
jtwood: a minute? wow your chroot is small :D
10:40
<Lns>
Roasted, I made a howto as well for ethernet bonding, I'm sure it's still applicable to whatever version you have: http://wiki.logicalnetworking.net/doku.php?id=networkbondingubuntu804
10:41
<alkisg>
My *compressed* chroot is 7 Gb
10:41
<muppis>
Huge.
10:42
Just under 700 Mb
10:42
<alkisg>
I tried to use a dedicated btrfs partition for the fat chroot too
10:42
So I could update it and export it while still being compressed, so fast
10:43
It turns out my host os, lucid, isn't ready yet for btrfs :D
10:43
(nbd can export whole partitions)
10:43
(and btrfs supports compression)
10:44
<jtwood>
alkisg: btrfs looks very intereting...I'll have to at least keep an eye on it
10:45shawnp0wers (~Adium@linuxjournal/staff/shawnp0wers) joined #ltsp.
10:45
<Lns>
err Roasted_ ^^
10:46
<Roasted_>
Lns, that's pretty much what I've done... but it sounds like a lot changed from 9.10 to 10.04, so I'm not sure how relevant a 8.04 guide will be
10:46
<Lns>
Roasted_ like what's changed? The tools themselves?
10:47
<Roasted_>
the way the config files are laid out, from what I've read.
10:47
<Lns>
Really? Well that's too bad..
10:48
Either way I'm sure a few minor adjustments and you'd be good to go.
10:48
Can't be that different, it's low-level stuff we're talking about, not a whole lot Ubuntu specific
10:48
<jtwood>
Roasted:/usr/share/doc/ifenslave-2.6/examples/two_hotplug_ethernet
10:48
I followed that exactly for bonding...worked like a charm...just adjust to your situation
10:48ogra_ (~ogra@p5098ed03.dip0.t-ipconnect.de) left irc: Excess Flood
10:49ogra_ (~ogra@p5098ed03.dip0.t-ipconnect.de) joined #ltsp.
10:50garymc (~chatzilla@host81-148-15-59.in-addr.btopenworld.com) left irc: Quit: ChatZilla 0.9.86.1 [Firefox 3.6.16/20110319135224]
10:52ogra_ (~ogra@p5098ed03.dip0.t-ipconnect.de) left irc: Changing host
10:52ogra_ (~ogra@ubuntu/member/ogra) joined #ltsp.
10:52
<Roasted_>
if I bond, and both interfaces get the same MAC address, woudlnt that cause an isue with the switch?
10:58
<jtwood>
Roasted_: Depends on the bonding type and the switch config how its all handled...no pro on it myself
10:58
<Roasted_>
this is getting so old
10:58
this is just not working
10:58
maybe I'll just split the lab on each nic and be done with it
10:59
<alkisg>
That will cause you more headache
11:01
<Roasted_>
seems like a more legit issue at this point
11:08
<Lns>
roasted_, read the docs, they explain.
11:12
<Roasted_>
Lns, I have.
11:12
Many a time.
11:12
It's k. I'll just sling it off one port. This headache needs to go away.
11:13
<Lns>
lol k
11:18* alkisg would love to test fat clients with 3 gbps speed on the server :)
11:19
<Roasted_>
I'd love to test fat clients with bonding :)
11:19
but lol.
11:28Gremble (~Ben@cpc12-aztw24-2-0-cust146.aztw.cable.virginmedia.com) left irc: Quit: I Leave
11:28vagrantc (~vagrant@75-150-46-245-Oregon.hfc.comcastbusiness.net) joined #ltsp.
11:34* Lns waves to vagrantc
11:34
<Lns>
alkisg, with 3Gbps on the server, why would you need fat clients at all? ;)
11:35* alkisg didn't get that one :)
11:35
<alkisg>
To have 1 gbps mean network disk transfer rate for each client
11:35
Ah, you mean as opposed to thin?
11:36
<Lns>
right
11:36
<alkisg>
Nah, even with gigabit for each client, fats are better
11:36
<Lns>
I mean, with 3Gbit/sec, you could even use Flash! =p
11:36
<alkisg>
Nope, not full screen.
11:37
And also no googleearth etc
11:37
<highvoltage>
alkisg: I'd like to see that too, 3gbps on server, and gigabit to thin clients, with the image stored in RAM :)
11:37
<alkisg>
And also no HD video for more than 1 client due to *cpu*!!!
11:38
<Lns>
boo
11:38
how easy is it to enable NX in ltsp? I never really tried before
11:38
<alkisg>
highvoltage: isn't the image in ram anyway, after the first client boots?
11:38* Lns wonders how that would work in a LAN environment
11:38
<alkisg>
thin chroot = 200 Mb, easily cached...
11:39
(or 500 uncompressed)
11:39
<vagrantc>
booting with NFS it's something more like 20MB
11:39
because you don't read every single file
11:40
i remember someone doing some posts to the debian-edu list comparing
11:40
<alkisg>
I don't think nbd reads every single file
11:40
<highvoltage>
alkisg: well, parts of it will of course end up in system cache
11:40
<alkisg>
Just the needed "sectors"
11:40
<vagrantc>
ah, true enough
11:40* Lns is still very happy that debian ltsp defaults to nfs
11:40* vagrantc is happy that some people are happy with that decision
11:41
<alkisg>
Openoffice with nfs starts in 30 secs, with nbd uncompressed in 20, with nbd compressed in 5
11:41
<vagrantc>
it's been a lot of work to fight against the NBD hegemony :)
11:41
<Lns>
The less complex, and the more stable components you use..well that's the debian way =)
11:41
<vagrantc>
yeah, i've noticed fatclients being a bit on the slow side
11:42
<Lns>
you could always use a hard drive on the client to cache the software... ;)
11:42
<alkisg>
Unfortunately I haven't found any good solution for that
11:43
<vagrantc>
that kind of defeats the simplicity goal
11:43
<alkisg>
The best thing would be an nfs client with a very big persistent local cache
11:43
I've seen a paper about such an implementation
11:43
But they didn't maintain the software after the paper, that tells me a lot :)
11:43
<Lns>
vagrantc, was just joking heh
11:44
I mean, honestly sometimes I think that LTSP fatclients defeat the overall purpose of LTSP.. not to knock it *at all* but sometimes I wonder that the goals are too close to having normal workstations
11:45
<alkisg>
Lns, maintaining normal workstations? Really?
11:45
<Lns>
alkisg, sure..I mean, you could put together some rsync / ssh magic and have centrally administered (but still full-blown) linux workstations
11:45
either way you require bigger and faster and more power hungry hardware
11:45
<alkisg>
Compared to the LTSP way, maintaining 1 chroot, with central authentication, nfs homes, etc, etc, I'd never maintain local installations if I had just a gigabig switch
11:46
<evil_root>
Lns.... your crazy
11:46
<alkisg>
Lns, rsync 30 Gb installations => 2 days to sync 20 workstations
11:46
Wouldn't work
11:46
<Lns>
alkisg, rsync doesn't copy every bit, it copies the changed bits
11:46
<alkisg>
Yup, but it reads every bit
11:47
Both on the source and the target
11:47
(unless you get risky and only check timestamps etc)
11:47
<Lns>
30gb installs??
11:47
<alkisg>
But why rsync, if a gigabit network disk is *faster*? What's the point?
11:48
Good fat client labs boot in 12 seconds over the network, while locally they need 40
11:48
(actual benchmark)
11:49
<Lns>
alkisg, they won't boot in 12 seconds if you turn them all on at the same time ;)
11:49
<highvoltage>
alkisg: I wish people could understand that :)
11:49
<alkisg>
I did. They did boot in 12
11:49
<highvoltage>
Lns: why not?
11:50
<alkisg>
On the server the nbd image gets cached, so it's served from RAM, so the seek times are very very small
11:50
<highvoltage>
Lns: well, maybe not on nfs ;)
11:50
<Lns>
lol
11:50
I'm not really talking about boot-up time here
11:50bobberty (~martian_l@124-197-42-208.callplus.net.nz) joined #ltsp.
11:50
<alkisg>
Opening apps has the same speed benefits
11:50
It's not just boot time
11:51
<Lns>
I'm saying that hardware-wise, you get very few of the advantages of ltsp if you require locally running apps
11:51
<vagrantc>
it takes a huge amount of maintenance overhead just to maintain the hardware...
11:51
let alone the software
11:51
Lns: running "apt-get install FOO" in one place to update an entire network seems like about 90% of the advantage of LTSP
11:52
<Lns>
http://www.linux.com/archive/feature/143893
11:52
vagrantc, i agree that this is probably the killer-feature of ltsp
11:52
that as well as the low-power (or even recycled computer) benefits
11:52ry (~ry@static-71-183-64-28.nycmny.fios.verizon.net) left irc: Remote host closed the connection
11:53
<alkisg>
Lns, can you tell a teacher to learn puppet and apply a gconf setting to all the workstations with it? I wouldn't
11:53
While with ltsp he can do it with gconf-editor
11:53
<Lns>
alkisg, no, and I also wouldn't tell a teacher to learn the ins and outs of LTSP ;)
11:53
a normal teacher isn't going to understand gconf either
11:53
<alkisg>
There are 200 teachers so far using LTSP here, so I guess it's doable
11:54
<Lns>
alkisg, right - using, not administering
11:54
<alkisg>
They installed it themselves
11:54
<highvoltage>
Lns: compared to "using" puppet, it's very easy :)
11:54
<Lns>
highvoltage, =p
11:54
alkisg, you have very open minded and technically savvy teachers! =)
11:54
<highvoltage>
(not that I'm against configuration managers)
11:55
<alkisg>
Automating LTSP is much, much easier than automating standalone installations
11:55
We created GUI wizards for the most common tasks in a few days
11:56
"I want to build a fat chroot" => "I'm in a high school so install the appropriate edu apps" => wait 5 hours => ready
11:56
To do that in 20 workstations it would require 20 hours of working, not waiting
11:57
<Lns>
alkisg, oh come on =p you can use cluster-ssh along with a local repository and do it in 5 minutes.
11:57
<alkisg>
You can install ubuntu with cluster ssh?
11:57komunista (~slavko@adsl-195-098-015-248.dynamic.nextra.sk) joined #ltsp.
11:57
<alkisg>
Because the fat chroot saves you from installing the os locally
11:58
<Lns>
alkisg, I was referring to "I'm in a high school so install the appropriate edu apps"
11:58
<alkisg>
On the fat chroot
11:58
<highvoltage>
you shouldn't ever use clusterssh, really
11:58
<Lns>
highvoltage, ?
11:59
<mgariepy>
alkisg, preseed + preseed/late_command
11:59
<highvoltage>
Lns: stgraber has this philipsophy of keeping everything in configuration managers and avoiding undocumented local changes if you have to administer many machines. it works. clusterssh encourage you to do things manually and implement local hacks that you'll lose if you have to re-install
12:00
<Lns>
highvoltage, what local hacks?
12:00
<alkisg>
mgariepy: even if I gave them a ghost-like disk which would be installed without asking anything at all, I'd still need to update the result. And more significantly, I fail to see the benefits.
12:00
<Lns>
And you could always have a standard image
12:01Q-FUNK (~q-funk@ubuntu/member/q-funk) joined #ltsp.
12:01
<highvoltage>
Lns: anything that's not in your configuration manager
12:02
<Lns>
Personally I was drawn to LTSP initially because of 1) central administration and 2) low power terminals. I guess I just never got used to configuring localapps enough to be able to switch between them and remote apps - for instance, firefox always complained when I'd run it as a localapp, then a remote app
12:02
Maybe I'm just missing something dumb
12:03drdata (drdata@tw2.udallas.edu) left #ltsp.
12:03
<Lns>
but it really put a damper on the central-administration aspect of it because there was a big problem when you go back and forth
12:03shawnp0wers (~Adium@linuxjournal/staff/shawnp0wers) left irc: Quit: Leaving.
12:03* alkisg never uses localapps, too complicated to automate for the teachers
12:04
<Roasted_>
so if I back out of trying to set up bonding, and I'm back to 1 NIC and I can't ping at all....
12:04
<highvoltage>
local apps is useful for stuff like gcompris where each instance of the program can easily push up a cpu core usage to 100% and keep it there for the whole session
12:05
also for multimedia
12:05
<alkisg>
Localapps are really fine when there's a sysadmin nearby.
12:06
You can get the most out of your hardware with them
12:07
<abeehc>
word i agree
12:12
<Lns>
alkisg, there is also the headache of maintaining multiple chroots
12:13
<alkisg>
Lns: why?
12:13
A fat chroot can boot thin and fat clients,
12:13
nvidia clients, non-nvidia clients,
12:13
...you name it
12:14
<Lns>
but not cross-arch
12:14
(for obvious reasons)
12:14
<alkisg>
That's true for thin clients too
12:14
<highvoltage>
cross-arch is in my experience pretty rare anyway
12:14
<alkisg>
But yup, we always use 32bit chroots here
12:14
No other arch
12:14
<Lns>
see, 64-bit is just going to saturate more and more, then what?
12:14
you can't hold out forever
12:15
<alkisg>
There will be some time period where some clients will have e.g. 8 Gb RAM and some others will be 32 bit (non amd64 capable)
12:15
At that time we'll still be using i386 chroots, maybe with a pae kernel
12:15
After that period we'll just switch to amd64 chroots
12:15
Again no multiple arches
12:16
<vagrantc>
unfortunately, a lot of thin client hardware can't even run a pae kernel ... requires some syslinux magic to get it to select the correct kernel
12:17
<highvoltage>
alkisg: I guess the more likely scenario for multiple arches is the proliferation of cheap (and capable) ARM based devices
12:17
<vagrantc>
and it's fairly easy to cross-build arm.
12:17
<alkisg>
highvoltage: I agree. In any case there's no difference between thin+fat there, 2 chroots will be needed in both cases
12:17
<highvoltage>
alkisg: *nod*
12:17
<alkisg>
I hope by that time we'll be able to manage the chroots with vbox
12:17
(instead of console)
12:17
<highvoltage>
vagrantc: it is expensive though
12:18
<vagrantc>
highvoltage: expensive?
12:18
<highvoltage>
vagrantc: sorry, my head is short-circuiting, I was thinking of someone cross-compiling a kernel earlier this week
12:19
alkisg: virtualbox? why?
12:19
<vagrantc>
highvoltage: on debian, you can just: apt-get install qemu-user-static ; ltsp-build-client --arch armel
12:19
a little slow to build, but way easier than having a dedicated arm server
12:20
<highvoltage>
vagrantc: yes that's quite cool
12:20
<alkisg>
highvoltage: the chroot + console concept is difficult for the teachers
12:20* Lns can't wait for gnome3 to be released..tried it the other night and it is soo beautiful
12:21
<alkisg>
Now we provide them with wizards to select apps while creating the chroot
12:21
But to add an app, or some gconf setting on the chroot, it's difficult for them
12:21
So in the future I want to try to make the LTSP chroot bootable with vbox,
12:21
so that they can configure it graphically
12:21
Or, bootable directly on a client (e.g. with an external usb disk)
12:22
<highvoltage>
alkisg: hmm... perhaps some software-center integration could fix that better
12:22
<alkisg>
Example - I boot an arm client with an external usb disk with btrfs + compression
12:22
Then I get that usb disk, put it on the server, and directly export the whole disk as an nbd cow image
12:22
Done, arm lab ready
12:22
With no console whatsoever
12:23
And this way each teacher can even maintain his own nbd image, like if it was a normal installation. No additional knowledge needed.
12:28
<Lns>
i want to make the shell "cool" again ;) I hate hearing about all this extra work because people are afraid to learn it
12:29bobberty (martian_l@124-197-42-208.callplus.net.nz) left #ltsp.
12:29
<Roasted_>
If I remove bonding, is there anything specific to do besides changing my interface file?
12:29
<Lns>
roasted_ just reverse the steps you took to (try to) enable it
12:30
unload modules, etc.
12:30
<Roasted_>
unload modules?
12:30
all I did was create /etc/modprobe.d/bonding.conf and edit my interface file, oh, and install ifenslave
12:30
Ive reversed that yet I cant get any network connectivity via static anymore
12:31* vagrantc recommends etckeeper
12:31
<highvoltage>
Roasted_: do you see your interface in 'ifconfig'? does it have the right address? what does 'route' give you? is your gateway configured correctly? what's in /etc/resolv.conf? can you resolve hostnames?
12:31
<vagrantc>
not that it will fix your problems now ... but in the future ... amazing to be able to see what is actually different
12:31
<highvoltage>
Roasted_: if you describe the symptoms of "what's not working" then it would be easier to pinpoint what went wrong
12:32
<Roasted_>
highvoltage, yes, I see it in ifconfig. I have the static IP I assigned it. I can't ping hostnames or IP addresses. I can only ping myself.
12:32
highvoltage, the info I have in my interface file is the backup I made of it when it worked BEFORE I even began tinkering with bonding.
12:33
is it possible rebooting twice could have fixed it?
12:33
I rebooted once, nothin. rebooted again just because, and I pinged??
12:33
<alkisg>
Roasted_: if you have a dhcp server available, try: sudo dhclient eth0
12:33
This will tell you if it's a simple "/etc/network/interfaces" configuration problem, or something more
12:34
<Roasted_>
alkisg, well, this box has 2 nics, and I let 1 interface be dhcp. I was able to get an IP via dhcp.
12:34
give me a second tho, I want to reboot a few more times and see if it continually works
12:34
that just strikes me odd a 2nd reboot was fine
12:35
haa. 3rd reboot, not fine.
12:35
why did I ever try bonding...
12:36
nevermind. seems as if eth0 didnt come up automatically. a reboot of the network service brought it back.
12:44shawnp0wers (~Adium@linuxjournal/staff/shawnp0wers) joined #ltsp.
12:45matrix3000 (matrix@rrcs-70-61-255-226.central.biz.rr.com) left #ltsp.
12:49xavierb (~xavier@82.254.253.34) joined #ltsp.
12:52
<Lns>
sounds like someone got in a bit over their head ;)
12:53cliebow (~cliebow@WatchGuard.ellsworth-hs.ellsworth.k12.me.us) left irc: Quit: Leaving
12:54Kicer86 (~Kicer86@host-5db0eeee.sileman.net.pl) left irc: Remote host closed the connection
12:55Kicer86 (~Kicer86@host-5db0eeee.sileman.net.pl) joined #ltsp.
12:55Nick change: evil_root -> zz_evil_root
13:07shawnp0wers (~Adium@linuxjournal/staff/shawnp0wers) left irc: Quit: Leaving.
13:09Kicer86 (~Kicer86@host-5db0eeee.sileman.net.pl) left irc: Read error: Connection reset by peer
13:12Roasted_ (~jason@206.82.22.190) left irc: Ping timeout: 255 seconds
13:13xavierb (~xavier@82.254.253.34) left irc: Ping timeout: 260 seconds
13:15xavierb (~xavier@82.254.253.34) joined #ltsp.
13:16Kicer86 (~Kicer86@host-5db0eeee.sileman.net.pl) joined #ltsp.
13:21alkisg (~alkisg@ubuntu/member/alkisg) left irc: Quit: Leaving.
13:54Kicer86 (~Kicer86@host-5db0eeee.sileman.net.pl) left irc: Quit: KVIrc Insomnia 4.0.2, revision: 4740, sources date: 20100627, built on: 2010-08-08 18:29:00 UTC http://www.kvirc.net/
14:00ogra_ (~ogra@ubuntu/member/ogra) left irc: Ping timeout: 252 seconds
14:01ogra_ (~ogra@p5098ed03.dip0.t-ipconnect.de) joined #ltsp.
14:22xavier_brochard (~xavier@lns-bzn-32-82-254-16-28.adsl.proxad.net) joined #ltsp.
14:26Trixboxer (~Trixboxer@office.supportdepartment.net) left irc: Quit: "Achievement is not the end, its the beginning of new journey !!!"
14:26xavierb (~xavier@82.254.253.34) left irc: Ping timeout: 260 seconds
14:27dan_l (~chatzilla@63.193.16.168) joined #ltsp.
14:33
<dan_l>
Is it possible to execute a script on a fatclient from a remote host?
14:34
hello good people! Is it possible to execute a script on a fatclient from a remote host?
14:35* vagrantc heard you the first time
14:35
<vagrantc>
dan_l: i gotta run though. be patient, and someone will eventually respond
14:35vagrantc (~vagrant@75-150-46-245-Oregon.hfc.comcastbusiness.net) left irc: Quit: leaving
14:36
<dan_l>
thnx
14:37mistik1 (mistik1@unaffiliated/mistik1) left irc: Read error: Connection reset by peer
14:37mistik1 (mistik1@unaffiliated/mistik1) joined #ltsp.
14:44Nick change: xavier_brochard -> xavierb
14:49
<xavierb>
dan_l: what do you want to do?
14:53
<dan_l>
Somewhat complex. We have a web based authentication app which tests a users library card and upon authentication we want to execute a script to copy an appropriate set of files for the user.
14:53
The scripst must be executed in the fatclient space
14:56
<xavierb>
You want a web browser to launch a script on the localhost? sorry, I didn't even know it was feasible
14:59
<dan_l>
Not a web browser.
14:59
We connect to a webeserver and launch a login screen on the client using ltsp-remotehosts.
15:00
When the login is accepted we can launch a script on the ltsp server successfully but
15:00
we need to lauch it on the client
15:03
<xavierb>
this wouldn't be different from the script to shutdown a client from the server
15:04
<dan_l>
Sounds correct - I don't know about this script. Can you put me in that direction?
15:07
<xavierb>
I remember about some discussion about that sort of script on the ltsp-discuss mailing list. Nothing more sorry.
15:08
<dan_l>
thnx
15:14
<xavierb>
dan_l: I'v found something from august 2007! look for "I need to remote reboot a LTSP 5 client" subject in the archives
15:19jtwood (~jtwood@74.81.111.2) left irc: Ping timeout: 252 seconds
15:31
<dan_l>
xavierb: Thanks for that - definitely some clueful stuff
15:37
<xavierb>
dan_l: There was something more recent and more interesting, but I can't find it
15:38xavierb (xavier@lns-bzn-32-82-254-16-28.adsl.proxad.net) left #ltsp ("Konversation terminated!").
16:00komunista (~slavko@adsl-195-098-015-248.dynamic.nextra.sk) left irc: Quit: Leaving.
16:22vagrantc (~vagrant@75-150-46-245-Oregon.hfc.comcastbusiness.net) joined #ltsp.
16:39staffencasa (~staffenca@128-193-144-246.oregonstate.edu) left irc: Quit: Leaving
16:50Q-FUNK (~q-funk@ubuntu/member/q-funk) left irc: Quit: Leaving.
16:52Selveste1 (~Selveste1@88.83.86.20.static.dong.customer.smilecontent.dk) joined #ltsp.
17:07ry (~ry@static-71-183-64-28.nycmny.fios.verizon.net) joined #ltsp.
17:13Selveste1 (~Selveste1@88.83.86.20.static.dong.customer.smilecontent.dk) left irc: Ping timeout: 240 seconds
17:28vagrantc (~vagrant@75-150-46-245-Oregon.hfc.comcastbusiness.net) left irc: Quit: leaving
17:40bobby_C (~bobby@85-124-22-227.teleworker.xdsl-line.inode.at) left irc: Ping timeout: 252 seconds
17:58artista_frustrad (~artista_f@189.75.139.149) left irc: Ping timeout: 240 seconds
18:15chupacabra (~chupacabr@cpe-70-112-14-99.austin.res.rr.com) left irc: Ping timeout: 260 seconds
18:27chupacabra (~chupacabr@cpe-70-112-12-155.austin.res.rr.com) joined #ltsp.
18:32chupacabra (~chupacabr@cpe-70-112-12-155.austin.res.rr.com) left irc: Ping timeout: 240 seconds
18:33Lns (~Lns@pdpc/supporter/professional/lns) left irc: Quit: Leaving
18:42otavio (~otavio@debian/developer/otavio) left irc: Quit: leaving
19:15alkisg (~alkisg@ubuntu/member/alkisg) joined #ltsp.
19:20
<alkisg>
[14:59] *** [dan_l] We connect to a webeserver and launch a login screen on the client using ltsp-remotehosts.
19:20
===> what is ltsp-remotehosts?
19:22
Also, those files that you're copying go to the user home directory? Then you can also run the script on the server, as it does has access to the user home dir...
19:54leio (~leio@gentoo/developer/leio) got netsplit.
19:54jbrett (brett@nat/hp/x-eannvbbirycdpfap) got netsplit.
19:54HrdwrBoB (~hrdwrbob@94.75.233.172) got netsplit.
19:54sep (~sep@40.211.jostedal.no) got netsplit.
19:54ogra_ (~ogra@p5098ed03.dip0.t-ipconnect.de) got netsplit.
19:54MorningSon (~MorningSo@cpe-70-114-21-95.satx.res.rr.com) got netsplit.
19:54alexqwesa (~alex@alexo-veto.broker.freenet6.net) got netsplit.
19:54mgariepy (~mgariepy@ubuntu/member/mgariepy) got netsplit.
19:54shamino (~shamino@a91-156-185-177.elisa-laajakaista.fi) got netsplit.
19:54mnemoc (~amery@shell.opensde.net) got netsplit.
19:57leio (~leio@gentoo/developer/leio) returned to #ltsp.
19:57jbrett (brett@nat/hp/x-eannvbbirycdpfap) returned to #ltsp.
19:57HrdwrBoB (~hrdwrbob@94.75.233.172) returned to #ltsp.
19:57sep (~sep@40.211.jostedal.no) returned to #ltsp.
20:00ogra_ (~ogra@p5098ed03.dip0.t-ipconnect.de) returned to #ltsp.
20:00MorningSon (~MorningSo@cpe-70-114-21-95.satx.res.rr.com) returned to #ltsp.
20:00alexqwesa (~alex@alexo-veto.broker.freenet6.net) returned to #ltsp.
20:00mgariepy (~mgariepy@ubuntu/member/mgariepy) returned to #ltsp.
20:00shamino (~shamino@a91-156-185-177.elisa-laajakaista.fi) returned to #ltsp.
20:00mnemoc (~amery@shell.opensde.net) returned to #ltsp.
20:04MorningSon (~MorningSo@cpe-70-114-21-95.satx.res.rr.com) left irc: Quit: WeeChat 0.3.0
20:17drdata (~drdata@pool-173-57-189-158.dllstx.fios.verizon.net) joined #ltsp.
20:23alkisg (~alkisg@ubuntu/member/alkisg) left irc: Quit: Leaving.
20:23artista_frustrad (~artista_f@189.75.139.149) joined #ltsp.
20:41dgroos (~dgroos@63.225.132.145) left irc: Quit: dgroos
21:18cyberorg (~cyberorg@opensuse/member/Cyberorg) left irc: Ping timeout: 255 seconds
21:19cyberorg (~cyberorg@opensuse/member/Cyberorg) joined #ltsp.
21:52loathing (~khudson@69.43.168.134) left irc: Read error: Operation timed out
21:53loathing (~khudson@internal-nat.djnetworks.net) joined #ltsp.
23:19Nick change: zz_evil_root -> evil_root
23:39Nick change: evil_root -> zz_evil_root
23:57Kicer86 (~Kicer86@host-5db0eeee.sileman.net.pl) joined #ltsp.
00:00--- Fri Mar 25 2011