RFC: a new init system for Arch

fk · 2007-04-07 19:03:27

I think that "upstart" is far better than the current init system of Arch

Can you give some facts? And not thinks like "upstart boots 2s faster !!!"

ekerazha · 2007-04-07 19:23:09

fk wrote:

I think that "upstart" is far better than the current init system of Arch
Can you give some facts? And not thinks like "upstart boots 2s faster !!!"

It boots faster then this: http://bbs.archlinux.org/viewtopic.php?id=29679#p232236 (with this http://www.netsplit.com/blog/articles/2 … n-universe ) and this http://bbs.archlinux.org/viewtopic.php? … =2#p241274 (for the last one... of course you don't need upstart to fix it... but ehy, upstart fixed it ) and it is event-based so it is a clean, intelligent solution.

I wouldn't have reasons to say it is better if I wouldn't think this. I didn't get up one day thinking "well... go to the Arch forum and say upstart is better"

Last edited by ekerazha (2007-04-07 19:32:43)

ekerazha · 2007-04-07 19:39:37

P.S.
If the problem are the SysV-styled scripts... well, maybe we ("we", not "I" ) could write a compatibility layer also for BSD-styled scripts.

broch · 2007-04-07 21:51:42

"It boots faster wink then this"
shrug, it does not, would you like me to post bootchart?
I told you, on much slower box I am getting similar results.

well, maybe we ("we", not "I" big_smile ) could write a compatibility layer also for BSD-styled scripts.

It seems that you are an exception here: this means that you would have to write these scripts if you want.

If you can experimentally prove that upstart will start faster on Arch, then this would be definitely a step in the right direction. I wish that linux would boot as fast as my box with XP or ... BSD to fully functional gui. At this point however I don't think that this is feasible.

deficite · 2007-04-07 23:03:16

Yes, in your little world I am a 2 yr old and init systems can make your computer boot up so fast that it's done before you even press your power button I bow to your great majesty ol' wise one and look forward to the fruits of your efforts which will be beyond my 2 yr old understanding.

PS: Please read my posts before responding to them.

ekerazha · 2007-04-08 09:09:37

broch wrote:

"It boots faster wink then this"
shrug, it does not, would you like me to post bootchart?

Well, I have "upstart" on a system with Ubuntu. If you have an "easy" method to remove "upstart" and restore the old init system, then I'll post some bootcharts. Otherwise, I've no intentions to loose 1/2 day trying to replace "upstart" and risking the system integrity (I use it for work).

However... maybe you could ask these charts to the Frugalware developer (he made some bootcharts) who said "boot and shutdown is now blazingly fast".

I told you, on much slower box I am getting similar results.

I tell you, on a Pentium III 933Mhz with 512MB of RAM and a 7200rpm drive, Ubuntu "powered", after upstart was installed (on an Ubuntu 6.10 alpha release) boot was about 15s faster and no more freezes on shutdown.

It seems that you are an exception here: this means that you would have to write these scripts if you want.

I don't think so, seeing how many "replace the init system" threads there are on this board.

If you can experimentally prove that upstart will start faster on Arch, then this would be definitely a step in the right direction. I wish that linux would boot as fast as my box with XP or ... BSD to fully functional gui. At this point however I don't think that this is feasible.

Well... it is definitely faster (2s on some machines, 15s on others etc. and it will be even faster when packages will take full advantages of the event-based system) but as I've already said, boot speed is not the only reason to adopt upstart and you are ignoring that

@deficite
My time to reply to 2yo trolls is expired.

Last edited by ekerazha (2007-04-08 10:24:14)

dongiovanni · 2007-04-08 09:22:10

I didn't understand what advanteges have the events in comparison with the current init system. I know, what to start befor something other and you can also use a script to check, if network is running or not..the same in other situations. What are the reasons for events?

Last edited by dongiovanni (2007-04-08 09:23:15)

ekerazha · 2007-04-08 09:38:30

dongiovanni wrote:

I didn't understand what advanteges have the events in comparison with the current init system. I know, what to start befor something other and you can also use a script to check, if network is running or not..the same in other situations. What are the reasons for events?

With events, things start, "happen", when they are needed. I don't have a list with things to start at boot time (well... now I've it with a compatibility layer ex. for SysV scripts, until the packages will take full advantages of the events system), I don't have an arbitrary list of things to start, but things are started when they are needed/requested and only needed things start. So... I think you'll understand this is a very clean, very intelligent system. However I'm not the upstart developer, so I suggest you to read this http://www.netsplit.com/blog/articles/2 … n-universe to have a better idea of what upstart is.

Last edited by ekerazha (2007-04-08 09:39:51)

T-Dawg · 2007-04-08 13:09:59

Guys play nice here. I really don't want to have to be captain asshole and lock this thread...

deficite · 2007-04-08 13:52:01

@deficite
My time to reply to 2yo trolls is expired.

patroclo7 · 2007-04-08 14:22:34

The "event based" approach is a form of autodetection, i.e. a system for deciding what is needed when something is plugged in or something else happens in the system. The article is interesting, but is a prominent case of anti-kiss philosophy. Here any form of autodetection is not welcome: the user is free to decide the order of the daemons and is invited to read and learn what is required and when. The actual init system allows this kind of informed freedom, while that article maintains that it is not compatible with modern computing. Since most users are happy with the actual init system and the init system is mentioned in all the reviews of archlinux as one of its main positive peculiarities, arch aims to show that the prophetes of automation, autodetection and systems which do automatically what is needed when it is needed are wrong. Thus these proposals are not welcome and I think you should stop to reiterate your point.

ekerazha · 2007-04-08 15:47:37

patroclo7 wrote:

The "event based" approach is a form of autodetection, i.e. a system for deciding what is needed when something is plugged in or something else happens in the system. The article is interesting, but is a prominent case of anti-kiss philosophy. Here any form of autodetection is not welcome: the user is free to decide the order of the daemons and is invited to read and learn what is required and when. The actual init system allows this kind of informed freedom, while that article maintains that it is not compatible with modern computing. Since most users are happy with the actual init system and the init system is mentioned in all the reviews of archlinux as one of its main positive peculiarities, arch aims to show that the prophetes of automation, autodetection and systems which do automatically what is needed when it is needed are wrong. Thus these proposals are not welcome and I think you should stop to reiterate your point.

Well... KISS = "Keep it simple ...". "Simple" is a subjective concept, there aren't math expressions for "simplicity". Maybe the current init system isn't really "simple" because you should manually run every process : basically computing is automation. Personally, I don't see any "anti-kiss philosophy" into upstart, I see an "intelligence" philosophy: things run when they have to run. I'll make some examples:

Current Arch init system -> I decide initially all the lights that should switch on when I'll come back to home. Maybe it will switch on lights that I won't need that day... maybe it won't switch on lights that I'd need that day.

With Upstart -> I come back to home and when I need a light, it will switch on. Only the needed lights will switch on: this is efficiency and effectiveness, not "anti-kiss".

Finally... the init system is mentioned in all the reviews of archlinux as one of its main positive peculiarities... because it is based on simple BSD-styled scripts, not because it is efficient

P.S.
Ho un amico a bassano del grappa

Last edited by ekerazha (2007-04-08 15:48:50)

TomE · 2007-04-08 16:26:16

Upstart on arch is pointless for 3 reasons:

1) Its NOT faster then the current init.
My second box ( a 1GHz p3 based celeron with 384MB of ram and 5400rmp HD ) boots
to console in 22 seconds, 3 seconds slower than the Frugalware dev whose box
is trice as fast.

2)It tries to solve a problem that does not exist.
The init system does not need to start the deamons on hardware events as udev
can easily start cups if a printer is plugin or any thing else.

3)The arch devs are not go to implement it.

ekerazha · 2007-04-08 16:32:40

TomE wrote:

Upstart on arch is pointless for 3 reasons:
1) Its NOT faster then the current init.
My second box ( a 1GHz p3 based celeron with 384MB of ram and 5400rmp HD ) boots
to console in 22 seconds, 3 seconds slower than the Frugalware dev whose box
is trice as fast.

Well... "Your box" != "All boxes". On my system and other systems it is much faster

2)It tries to solve a problem that does not exist.
The init system does not need to start the deamons on hardware events as udev
can easily start cups if a printer is plugin or any thing else.

You should know there are things outside /dev

3)The arch devs are not go to implement it.

Good point

ekerazha · 2007-04-08 16:46:04

I'd want to make one thing clear... this is only a discussion, I clearly don't want (and I can't ) to enforce anybody to adopt upstart. These are just my cents...

Last edited by ekerazha (2007-04-08 16:46:42)

shining · 2007-04-08 17:47:34

ekerazha wrote:

With Upstart -> I come back to home and when I need a light, it will switch on. Only the needed lights will switch on: this is efficiency and effectiveness, not "anti-kiss".

As long as it's smart and not stupid, it is anti-kiss.

hussam · 2007-04-08 19:37:47

Personally, I don't really care about KISS. I never have and I never will. I only care about what I understand.
I don't know much about Upstart because I haven't tried it.
And since I know and fully understand Arch's current init system, I am quite comfortable with it and it works very well.
Why venture into something new when the current init works very well?

ekerazha · 2007-04-08 19:39:22

Pacman automatically resolves dependencies... is this "anti-kiss"?

ekerazha · 2007-04-08 19:40:28

hussam wrote:

Why venture into something new when the current init works very well?

Because there's something that works better

shining · 2007-04-08 20:52:59

ekerazha wrote:

hussam wrote:
Why venture into something new when the current init works very well?
Because there's something that works better

It isn't that clear that it works better (maybe in theory, but in practice?), and anyway it doesn't even matter. Even if it's better, it's not worth switching, because the current system works perfectly, and its simplicity is one of arch strength.
I'm curious, there are so many distros out there, why don't you use one that uses upstart since you care that much? I already find that most of the linux distribs don't have anything great and particular that justify their existence. But here, we have a particularity, the init system. There are distribs with one init system, other distribs with another, so we can test them, make our choice, and pick the best for our need.
If all distributions made exactly the same choices, then it would be totally retarded to have so many distribs in the first place, wouldn't it?

broch · 2007-04-08 20:56:47

ekerazha wrote:

broch wrote:
"It boots faster wink then this"
shrug, it does not, would you like me to post bootchart?
Well, I have "upstart" on a system with Ubuntu. If you have an "easy" method to remove "upstart" and restore the old init system, then I'll post some bootcharts. Otherwise, I've no intentions to loose 1/2 day trying to replace "upstart" and risking the system integrity (I use it for work).
However... maybe you could ask these charts to the Frugalware developer (he made some bootcharts) who said "boot and shutdown is now blazingly fast".
I told you, on much slower box I am getting similar results.
I tell you, on a Pentium III 933Mhz with 512MB of RAM and a 7200rpm drive, Ubuntu "powered", after upstart was installed (on an Ubuntu 6.10 alpha release) boot was about 15s faster and no more freezes on shutdown.
It seems that you are an exception here: this means that you would have to write these scripts if you want.
I don't think so, seeing how many "replace the init system" threads there are on this board.
If you can experimentally prove that upstart will start faster on Arch, then this would be definitely a step in the right direction. I wish that linux would boot as fast as my box with XP or ... BSD to fully functional gui. At this point however I don't think that this is feasible.
Well... it is definitely faster (2s on some machines, 15s on others etc. and it will be even faster when packages will take full advantages of the event-based system) but as I've already said, boot speed is not the only reason to adopt upstart and you are ignoring that
@deficite
My time to reply to 2yo trolls is expired.

I am not sure what bootchart has to do with removing upstart?

You have provided earlier results for Frugalware boot time
http://www.alex-smith.me.uk/?p=59
I don't see anything unusually fast there.

You are saying 15s shorter...
This only shows how bad default ubuntu setup is. Maybe this is not clear, but once you got everything out your system there is not much to improve in terms of speed, here are physical limitations. Arch does job very well, changing to upstart will not really improve anything.

15s is not really that much. I was able to cut down suse and ubuntu boot time 40-50s. This seems more effective than 15s.

And if you need 1/2 day to cut 15s off, that is really not feasible approach

I am not ignoring anything, I am asking about experimental proof. Simply post bootchart with upstart if you want. Bootchart is not a way of showing how fast something boots, but it shows how bad is boot config (sleep state for example).

Other advantages seems theoretical rather that anything else. Computer for me is only a tool. Arch once set works. Including the way it starts services. I don't see any real advantages of upstart on my laptop. Servers? well these stay on line as long as posible and nobody is interested in the "inteligent" way of starting different services. Nobody is playing with services on the servers.

Last edited by broch (2007-04-08 21:00:22)

anykey · 2007-04-08 22:47:00

I just wonder, why the hell are boot times even "important" to anyone here? The only occasion to reboot is when kernel changes? There are suspend-to-ram and suspend2 plenty!

Even if you absolutely *had* to reboot daily, would it really matter if it were 15, 20 or 30 seconds? The important thing is "how well can you work with it", not "it starts some seconds faster".

Please, I beg you, someone explain what boot times are really a "benchmark" for (please, not something stupid. A real reason to nitpick those maybe ten seconds).

broch · 2007-04-08 23:30:51

"I just wonder, why the hell are boot times even "important" to anyone here?"
looks like home user only.

I need my laptop to boot fast to start presentation. Nobody will wait 5 min for my to boot linux.
Unfortunately linux suspend to RAM is not faster than cold boot (plus is not 100% reliable and as I mentioned: I don't have time to fix system and boot again) and suspend to disk is not an option.

AlexExtreme · 2007-04-09 06:25:42

Well, as you all know probably, I've been messing with Upstart. I'll just point out a few things here: I may seem to be ranting on about the boot time, but boot time is simply an added bonus from switching to a more reliable init system. The main benefits of Upstart are it's ability to order startup based on dependencies rather than just defining a static order, and the ability to respond to events such as hardware being added, removed, etc. These events can then be used, for example, to reliably know when a device in fstab has been added to the system rather than just attempting to mount it no matter what.

However, having used Arch, I do think that replacing the init system would be a little stupid in Arch since it's rc.conf system is very well suited to the type of distro that Arch is. I think the Arch initscripts are well designed and they are fairly quick. We're switching to Upstart in Frugalware because it's better suited to the type of distro Frugalware is, a distro for new/intermediate users.

Hope that helps

Last edited by AlexExtreme (2007-04-09 06:27:35)

Mr Green · 2007-04-09 07:01:02

Arch is one of the fastest booting distros around, if its not broke do not fix it ;-)

Last edited by Mr Green (2007-04-09 07:01:16)

Arch Linux

#51 2007-04-07 19:03:27

Re: RFC: a new init system for Arch

#52 2007-04-07 19:23:09

Re: RFC: a new init system for Arch

#53 2007-04-07 19:39:37

Re: RFC: a new init system for Arch

#54 2007-04-07 21:51:42

Re: RFC: a new init system for Arch

#55 2007-04-07 23:03:16

Re: RFC: a new init system for Arch

#56 2007-04-08 09:09:37

Re: RFC: a new init system for Arch

#57 2007-04-08 09:22:10

Re: RFC: a new init system for Arch

#58 2007-04-08 09:38:30

Re: RFC: a new init system for Arch

#59 2007-04-08 13:09:59

Re: RFC: a new init system for Arch

#60 2007-04-08 13:52:01

Re: RFC: a new init system for Arch

#61 2007-04-08 14:22:34

Re: RFC: a new init system for Arch

#62 2007-04-08 15:47:37

Re: RFC: a new init system for Arch

#63 2007-04-08 16:26:16

Re: RFC: a new init system for Arch

#64 2007-04-08 16:32:40

Re: RFC: a new init system for Arch

#65 2007-04-08 16:46:04

Re: RFC: a new init system for Arch

#66 2007-04-08 17:47:34

Re: RFC: a new init system for Arch

#67 2007-04-08 19:37:47

Re: RFC: a new init system for Arch

#68 2007-04-08 19:39:22

Re: RFC: a new init system for Arch

#69 2007-04-08 19:40:28

Re: RFC: a new init system for Arch

#70 2007-04-08 20:52:59

Re: RFC: a new init system for Arch

#71 2007-04-08 20:56:47

Re: RFC: a new init system for Arch

#72 2007-04-08 22:47:00

Re: RFC: a new init system for Arch

#73 2007-04-08 23:30:51

Re: RFC: a new init system for Arch

#74 2007-04-09 06:25:42

Re: RFC: a new init system for Arch

#75 2007-04-09 07:01:02

Re: RFC: a new init system for Arch

Board footer