systemd for 11.10 ?
James Hunt
james.hunt at ubuntu.com
Tue May 10 09:50:32 UTC 2011
Patrick,
On 09/05/11 16:58, Patrick Goetz wrote:
>> From: Steve Langasek <steve.langasek at ubuntu.com>
>> Date: Mon, 9 May 2011 11:45:56 +0200
>>
>> And let's not forget that, for anyone tracking the LTS, upstart
>> is*already*
>> the system in use for the previous LTS, 10.04. The fundamentals of how
>> upstart will work in 12.04 LTS are the same as in 10.04 LTS; upstart in
>> 12.04 will include incremental improvements to the usability and
>> instrumentability for system admins, this is not something *new* that
>> admins
>> are being asked to invest their time in learning.
>
>
> Hi -
>
> Your points are well taken, and -- especially if not everyone agrees
> that systemd is superior -- it's fair to keep upstart going for a while
> before jumping to the conclusion that systemd is better.
This comment still implies to me you consider systemd better, but I do
not understand why you have this view?
Everyone
> immediately jumped on the X windows bandwagon many years ago (mostly as
> a foil to Sun's NeWS system), and we haven't been unable to get out from
> underneath using network stacks for locally displayed graphics for the
> 20+ years since.
>
> My point about learning a new system was based on the observation that
> there still seem to be a number of kinks in upstart; in particular,
> we've had timing problems with autofs and statd, both of which
> intermittently refused to start until we tweaked the upstart scripts.
Why does your having to change a configuration file imply we should
throw away a core part of the Ubuntu system? The problem you have
described is a packaging/maintainer configuration file issue for 2
different packages, neither of which are Upstart itself.
> Also, the fact that packages like apache still don't have an upstart
> script indicates the technology isn't fully mature. Under these
> circumstances, if everyone agreed that systemd is better, then it makes
> more sense to switch sooner rather than later.
>
>> http://i.imgur.com/usftZ.png
>
> This is very funny, but there are some points raised in Lennart's
> comparison:
>
> http://0pointer.de/blog/projects/why.html
>
> that merit careful reflection; e.g. shell-free bootup, mount/automount
> handling, and disabling of services without editing files, to name a
> few. There are other points that look interesting, but I don't know
> enough about what he means to comment.
Lots of features does not necessarily equate to a superior system. the
design of Upstart follows "the Unix way": do one job and do it well.
When required, Upstart relies on "helpers" (such as
upstart-udev-bridge). This design is fundamentally sound - *iff* there
were a problem in the udev bridge, it cannot bring down init, which
would of course cause a kernel panic. Adding every possible feature into
a critical system process such as init should only be considered with
extreme caution IMHO.
No one, however, can tell me
> that the accepted method for disabling services in upstart isn't a
> kludge! <:)
The rationale for Upstarts behaviour, which is of course different to
SysV-like systems, is that if you install a service, you'll want it
started on boot: guaranteed. In 9 out of 10 cases, this is the right answer.
For the remaining case, say a database dev who has a handful of db
engines installed, but doesn't want them all auto-started on her
underpowered laptop, we provide the ability to disable that job. Yes,
the interface is simple, but does it really need to be any more complex
than it is?
>
>
>
James.
--
James Hunt
____________________________________
Ubuntu Foundations Team, Canonical.
More information about the ubuntu-devel
mailing list