=== Amaranth [n=travis@ubuntu/member/Amaranth] has joined #upstart | ||
=== Amaranth [n=travis@ubuntu/member/Amaranth] has joined #upstart | ||
=== Md [i=md@freenode/staff/md] has joined #upstart | ||
=== Keybuk [n=scott@12.145.6.2] has joined #upstart | ||
=== juergbi [n=juerg@80-219-18-22.dclient.hispeed.ch] has joined #upstart | ||
=== sadleder [n=sadleder@p508118ad.dip0.t-ipconnect.de] has joined #upstart | ||
=== sadleder [n=sadleder@p508118ad.dip0.t-ipconnect.de] has left #upstart [] | ||
=== wasabi [n=wasabi@ubuntu/member/wasabi] has joined #upstart | ||
=== tale [n=tale@207.235.54.1] has joined #upstart | ||
=== wasabi_ [n=jhaltom@ubuntu/member/wasabi] has joined #upstart | ||
=== mbiebl [n=michael@e180073128.adsl.alicedsl.de] has joined #upstart | ||
=== liquidat [n=liquidat@4560-02bg04.stw-wh.uni-jena.de] has joined #upstart | ||
=== cortana [n=sam@62-31-146-25.cable.ubr12.azte.blueyonder.co.uk] has joined #upstart | ||
mbiebl | Md: Did you have any problem with switching to upstart? | 11:51 |
---|---|---|
Md | mbiebl: actually I have not rebooted yet :-) | 11:58 |
mbiebl | Ah, ok. I've been hacking a bit on native upstart jobs over the weekend. | 11:59 |
mbiebl | The results are pretty promising: http://debs.michaelbiebl.de/upstart/bootchart.png | 11:59 |
mbiebl | Works pretty good so far. | 12:01 |
Md | the next obvious step should be running fsck/mount asyncronously from udev | 12:02 |
Amaranth | mbiebl: 19 seconds from grub to X? | 12:03 |
mbiebl | Amaranth: Yes, including initramfs | 12:03 |
mbiebl | It's a 3 year old laptop. | 12:03 |
Amaranth | _damn_ | 12:03 |
Amaranth | i'm sitting at 35 seconds for my laptop and it's less than a year old :) | 12:03 |
Amaranth | I think I need to do some trimming | 12:03 |
mbiebl | I have to admit, if I switch to gdm instead of entranced, it adds another 8 secs. (gdm is just so slow...) | 12:04 |
Amaranth | heh | 12:05 |
ion_ | entranced, huh? Hadnt heard of that one. | 12:05 |
ion_ | % apt-cache search entrance | 12:05 |
mbiebl | e17 login manager | 12:05 |
ion_ | % | 12:05 |
ion_ | Ah. | 12:05 |
Amaranth | gdm is a key part of a couple nice gnome features | 12:05 |
mbiebl | Still, I think gdm needs some serious profiling ;-) | 12:06 |
mbiebl | Even kdm is a lot faster. | 12:07 |
Amaranth | either way if with gdm you can get a 27 second boot i should be able to get 13 seconds or something :) | 12:07 |
mbiebl | Which fs do you use for / ? | 12:08 |
mbiebl | Reiser3 can take 3-4 secs just for mounting. | 12:08 |
Amaranth | ext3 | 12:08 |
Amaranth | i'm plain and boring :) | 12:08 |
mbiebl | (depending on the size of the partition) | 12:08 |
Amaranth | but i've got a core duo and the fastest laptop HD in existence so... :) | 12:09 |
wasabi_ | I had no idea people still worked on e17 | 12:09 |
ion_ | My boot time probably decreased by a half when i switched from a 5400 RPM HDD with less than MiB of cache to a 7200 RPM one with a 8 MiB cache. :-) | 12:10 |
mbiebl | wasabi_: they have a pretty complete Debian repo at deb http://edevelop.org/debian/ unstable main | 12:10 |
wasabi_ | That stuff usable | 12:10 |
wasabi_ | ? | 12:10 |
Amaranth | ion_: yeah, i've got one of those two | 12:11 |
wasabi_ | Or is it still just tech demos? | 12:11 |
Amaranth | is yours hitachi? | 12:11 |
ion_ | Samsung | 12:11 |
Amaranth | wasabi_: I imagine the login manager, window manager, etc is usable but the rest is still at tech demo level | 12:11 |
ion_ | The older one was an IBM Deahtstar. | 12:11 |
mbiebl | wasabi_: It's definitely usable, but e17 won't become my main desktop though. | 12:12 |
Amaranth | ion_: ah | 12:12 |
ion_ | Death even. | 12:12 |
Amaranth | ion_: sata? | 12:12 |
ion_ | Plain PATA. | 12:12 |
Amaranth | oh | 12:12 |
Amaranth | suppose it doesn't make too much of a difference with these laptop drives | 12:12 |
Amaranth | mbiebl: wow, you disabled most everything to get that time :) | 12:13 |
mbiebl | Amaranth: Really? | 12:14 |
Amaranth | well i've got junk for bluetooth and such here | 12:14 |
Amaranth | looks like all you did was essential "this has to run for my system to boot" stuff plus dbus and X | 12:15 |
mbiebl | Well, not quite: I run syslog-ng, acpid, anacron, cron, dbus, hal, NEtworkManager, dhcdbd, console-kit | 12:15 |
mbiebl | Pretty complete I'd say. | 12:16 |
Amaranth | oh, you made upstart scripts for hal, networkmanager, and dhcdbd? | 12:16 |
Amaranth | because otherwise dbus will start those with it's own init system | 12:16 |
mbiebl | That's actually not true anymore in Debian. | 12:17 |
Amaranth | oh? | 12:18 |
mbiebl | I made all the Dbus init scripts regular sysv init scripts the last week. | 12:18 |
Amaranth | cool | 12:18 |
=== Amaranth should start working on making his boot faster | ||
mbiebl | /etc/dbus-1/event.d/ will soon be gone. | 12:18 |
Amaranth | if i can get it below 20 seconds it's faster than hibernating :) | 12:18 |
wasabi_ | Are we suspecting that upstart will in fact replace dbus' init system? | 12:19 |
mbiebl | wasabi_: sure | 12:19 |
wasabi_ | including activation, or has nobody brought it up? | 12:19 |
mbiebl | Ah, Service activation is a different topic. | 12:20 |
Amaranth | activation is not something upstart should manage | 12:20 |
wasabi_ | Why? | 12:20 |
Amaranth | i know i don't really want upstart managing my rhythmbox ;) | 12:20 |
Amaranth | although i suppose it doesn't matter | 12:20 |
mbiebl | But dhcdbd, network-manager, hal, avahi etc are all native jobs which run as soon dbus is started. | 12:20 |
wasabi_ | Hmm. | 12:20 |
wasabi_ | I'm not sure how I feel about that either. I think I don't like the idea of DBUs launching UI apps etiher. | 12:21 |
wasabi_ | Heh. | 12:21 |
Amaranth | upstart seems a bit heavyweight for something as simple as 'dbus path doesn't exist, $app provides it, launch $app' | 12:21 |
wasabi_ | I think a few people have brought up the idea of using upstart for desktop sessions. | 12:22 |
wasabi_ | I'd like to think about it someday. :0 | 12:22 |
wasabi_ | Basically a non-PID 1 upstart copy. | 12:22 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!