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


Channel log from 19 March 2016   (all times are UTC)

00:55dtcrshr has left IRC (dtcrshr!~datacrush@unaffiliated/datacrusher, Quit: Saindo)
02:17vagrantc has joined IRC (vagrantc!~vagrant@unaffiliated/vagrantc)
06:15vagrantc has left IRC (vagrantc!~vagrant@unaffiliated/vagrantc, Ping timeout: 264 seconds)
07:34alkisg_away is now known as alkisg
07:46vagrantc has joined IRC (vagrantc!~vagrant@unaffiliated/vagrantc)
07:48
<alkisg>
vagrantc: the initramfs scripts all need the prereqs stanza, otherwise they get to run when update-initramfs is called
07:48
Hopefully they weren't doing much harm so far... :)
07:49
(e.g. if you had ltsploop in the server's kernel cmdline, it would move your root file system when you ran update-initramfs)
07:49
<vagrantc>
ouch
07:52ricotz has joined IRC (ricotz!~ricotz@ubuntu/member/ricotz)
08:08
<alkisg>
Fix committed in http://bazaar.launchpad.net/~ltsp-upstream/ltsp/ltsp-trunk/revision/2721
08:10
!learn alkisg-todo as "Remove the 5 sec initramfs delay due to resume="
08:10
<ltsp>
The operation succeeded.
08:19vagrantc has left IRC (vagrantc!~vagrant@unaffiliated/vagrantc, Ping timeout: 260 seconds)
08:39alkisg has left IRC (alkisg!~alkisg@ubuntu/member/alkisg, Quit: Leaving.)
08:41alkisg has joined IRC (alkisg!~alkisg@ubuntu/member/alkisg)
10:03
<alkisg>
Yey, client boot time in 16.04 dropped from 45 seconds to 27 seconds by bypassing all the initramfs delay bugs due to nbd :)
10:04
There's another 10 second delay but it's inside the ubuntu kernel, can't do much about that... :-/
10:04
Pitty, it would make it boot in under 20 seconds then
10:07
Ouch, precise didn't have that bug and so it booted in 10 seconds
10:25adrianorg has left IRC (adrianorg!~adrianorg@177.132.221.25, Ping timeout: 250 seconds)
10:27adrianorg has joined IRC (adrianorg!~adrianorg@201.22.229.158.dynamic.dialup.gvt.net.br)
10:35kjackal has joined IRC (kjackal!~kjackal@2a02:587:3101:3500:64aa:9e0f:3196:7282)
11:21kjackal has left IRC (kjackal!~kjackal@2a02:587:3101:3500:64aa:9e0f:3196:7282, Ping timeout: 246 seconds)
12:54
<alkisg>
Ah, precise didn't have that bug with kernel=3.2, it did have it with kernel=3.13
13:36
!kvm
13:36
<ltsp>
kvm: Virtual thin client: kvm -vga-vmware -ctrl-grab -no-shutdown -net nic,model=virtio -net user,tftp=/var/lib/tftpboot,bootfile=/ltsp/i386/pxelinux.0
14:50robb_nl has joined IRC (robb_nl!~robb_nl@ip-83-134-23-29.dsl.scarlet.be)
15:22kjackal has joined IRC (kjackal!~kjackal@2a02:587:3101:3500:64aa:9e0f:3196:7282)
15:28robb_nl has left IRC (robb_nl!~robb_nl@ip-83-134-23-29.dsl.scarlet.be, Ping timeout: 248 seconds)
16:40robb_nl has joined IRC (robb_nl!~robb_nl@ip-83-134-23-29.dsl.scarlet.be)
16:58robb_nl has left IRC (robb_nl!~robb_nl@ip-83-134-23-29.dsl.scarlet.be, Ping timeout: 276 seconds)
17:01robb_nl has joined IRC (robb_nl!~robb_nl@ip-83-134-23-29.dsl.scarlet.be)
17:18robb_nl has left IRC (robb_nl!~robb_nl@ip-83-134-23-29.dsl.scarlet.be, Quit: I'm gone, bye bye)
18:14vagrantc has joined IRC (vagrantc!~vagrant@unaffiliated/vagrantc)
18:17vagrantc_ has joined IRC (vagrantc_!~vagrant@unaffiliated/vagrantc)
18:23vagrantc_ has left IRC (vagrantc_!~vagrant@unaffiliated/vagrantc, Quit: leaving)
19:37ricotz has left IRC (ricotz!~ricotz@ubuntu/member/ricotz, Quit: Leaving)
19:41
<alkisg>
stgraber: udhcpc in 16.04 is a symlink to busybox, but it doesn't work in the initramfs because busybox-initramfs doesn't have it while busybox-static does have it
19:42
stgraber: one way around it is to drop udhcpc support and use ipconfig, but I think you wanted udhcpc because it supported lease renewal?
19:43
Any ideas? Currently 16.04 only works with IPAPPEND 3...
19:55ricotz has joined IRC (ricotz!~ricotz@ubuntu/member/ricotz)
20:24
<alkisg>
Ah, in 12.04 udhcpc was also a symlink... something goes wrong with copy_exec then
20:51
<maldridge>
alkisg: in my tests with void, I'm at just under 20s to boot
20:52
and that's for fat clients, I haven't figured out how to do thin ones yet
20:52
<alkisg>
maldridge: can you put your dmesg to pastebin?
20:53
<maldridge>
I can't at the moment, those testbeds are at work
20:53
<alkisg>
I got it down to 8 secs with kvm, which doesn't suffer that issue
20:53
<maldridge>
I'm doing all this on kernel 4.4 btw, that might have something to do with it
20:53
<alkisg>
4.4 here as well
20:54
I think that the ip=xxx issue is ubuntu-specific
20:54
<maldridge>
huh, I don't have the slowdown you mention, so I agree it is ubuntu-specific
20:54
<alkisg>
Your dmesg will help in verifying if it's ubuntu-specific or not
20:54
You wouldn't be seeing it in non-ltsp clients, only in ltsp clients
20:55
<maldridge>
sure, I can grab it later today in theory, they're doing electrical work and so the power is off currently
20:55
<alkisg>
Ah, let me just try it in a fedora vm...
20:58
Haha Debian does some weird clock calibration and goes from 0 to 70 secs
20:58
<maldridge>
I should mention that I'm not necessarily using the "normal" ltsp boot process
20:59
I'm pulling my rootfs into ram over http since I only need 2 apps
20:59
<alkisg>
As long as you have an ip= in the kernel cmdline
20:59
<maldridge>
ah, I do not, that gets handled by my initramfs
21:00
some of this I'm still learning, the initramfs is still kind of hazy as to how its getting access to the proper rootfs
21:00
<alkisg>
If `cat /proc/cmdline` in the client doesn't show an ip=xxx entry, then you could add it to pxelinux just to test, even if it does nothing
21:01
Nope, Debian doesn't have that issue either
21:01
So it's quite probably Ubuntu-specific
21:01
<maldridge>
sure, I'll make note of that and test it when able
21:02
I've noticed a lot of ubuntu specific wierdness, part of the reason I'm migrating away from it
21:02
it was a nice stepping stone, but it seems to be less and less "LTS" with every release
21:03
<alkisg>
I don't like it when it diverts from debian too much
21:04
<maldridge>
yeah that gets old fast
21:04
<alkisg>
But it does have some nice things like various kernels and xorg for the lts releases
21:04
PPAs and recipes etc
21:04
<maldridge>
given that MS Windows is rolling now, we're trying a rolling linux at one of my sites, I'm excited to see how it goes
21:09robb_nl has joined IRC (robb_nl!~robb_nl@ip-83-134-23-29.dsl.scarlet.be)
21:11
<alkisg>
stgraber: confirmed, copy_exec on 12.04 copies udhcpc (=busybox static) to the initramfs, while on 16.04 it symlinks it, breaking it
21:18* alkisg tries: copy_exec /bin/busybox /sbin/udhcpc
21:22
<alkisg>
Yey! 12 secs to LDM! :)
21:23
I hope I can get all those fixes in the proper places, so that 16.04 boots as fast as 12.04.1 did...
21:36robb_nl has left IRC (robb_nl!~robb_nl@ip-83-134-23-29.dsl.scarlet.be, Ping timeout: 244 seconds)
22:47robb_nl has joined IRC (robb_nl!~robb_nl@ip-83-134-23-29.dsl.scarlet.be)
22:49robb_nl has left IRC (robb_nl!~robb_nl@ip-83-134-23-29.dsl.scarlet.be, Remote host closed the connection)
23:06ricotz has left IRC (ricotz!~ricotz@ubuntu/member/ricotz, Remote host closed the connection)