initscripts-2011.09.2-1
 

Hi guys,

I pushed a release candidate of initscripts to [testing]. I'd still
like to include some man pages that Seblu is working on, but I'll be
busy this weekend and next week, so I thought I'd get some feedback on
this as soon as possible, even if it will not be the final release.

Look out for:

The timeouts for SIGTERM/SIGKILL at shutdown have been increased and
the error reporting in case the timeouts are reached is improved. This
should hopefully make it simpler to spot any problems with processes
not being properly killed before shutdown. Such problems have been
reported, but I have not managed to reproduce them.

CHANGES
-------------

Deprecation:
* We'll complain if /usr is on a separate mount point and has not been
mounted by the initramfs (mkinitcpio does not yet support this, but
there are patches available)

Features:
* We load fragments from sysctl.d in addition to the sysctl.conf file
* rc.d is can now filter the output in a nice way
* We pivot into a shutdown ramfs if it exists in /run/initramfs (this
is mostly for testing purposes as mkinitcpio still does not support
this)

Towards read only /etc:
* /etc/profile.d/locale.sh is now a static file, rather than being
generated on every boot (this means we need to use -f when updating)
* /etc/localtime may now be a symlink, and is not written to unless it
is necessary.

Cheers,

Tom

initscripts-2011.09.2-1
 

On 30/09/11 00:00, Tom Gundersen wrote:

Hi guys,

I pushed a release candidate of initscripts to [testing]. I'd still
like to include some man pages that Seblu is working on, but I'll be
busy this weekend and next week, so I thought I'd get some feedback on
this as soon as possible, even if it will not be the final release.

Look out for:

The timeouts for SIGTERM/SIGKILL at shutdown have been increased and
the error reporting in case the timeouts are reached is improved. This
should hopefully make it simpler to spot any problems with processes
not being properly killed before shutdown. Such problems have been
reported, but I have not managed to reproduce them.

CHANGES
-------------

Deprecation:
* We'll complain if /usr is on a separate mount point and has not been
mounted by the initramfs (mkinitcpio does not yet support this, but
there are patches available)

Features:
* We load fragments from sysctl.d in addition to the sysctl.conf file
* rc.d is can now filter the output in a nice way
* We pivot into a shutdown ramfs if it exists in /run/initramfs (this
is mostly for testing purposes as mkinitcpio still does not support
this)

Towards read only /etc:
* /etc/profile.d/locale.sh is now a static file, rather than being
generated on every boot (this means we need to use -f when updating)



Please do not encourage using -f when updating (people will use
"pacman -Syuf", which is evil...). Instead, "mv
/etc/profile.d/locale.sh{,.bak}" and then update.




* /etc/localtime may now be a symlink, and is not written to unless it
is necessary.

Cheers,

Tom

initscripts-2011.09.2-1
 

On Fri, Sep 30, 2011 at 8:36 AM, Allan McRae <allan@archlinux.org> wrote:
> On 30/09/11 00:00, Tom Gundersen wrote:
>> Towards read only /etc:
>> * /etc/profile.d/locale.sh is now a static file, rather than being
>> generated on every boot (this means we need to use -f when updating)
> Please do not encourage using -f when updating * (people will use "pacman
> -Syuf", which is evil...). * Instead, "mv /etc/profile.d/locale.sh{,.bak}"
> and then update.

I presume the new local.sh will be generated in exactly the same way,
and hence there's no real need for *.bak?

initscripts-2011.09.2-1
 

On Thu, Sep 29, 2011 at 16:00:52 +0200, Tom Gundersen wrote:
> * /etc/localtime may now be a symlink, and is not written to unless it is
> necessary.


Hi,

why not always make it a symlink, so updates to tzdata package are applied
immediatly and don't require a reboot? Is it for scenario's where /usr is
not available?


Geert


--
geert.hendrickx.be :: geert@hendrickx.be :: PGP: 0xC4BB9E9F
This e-mail was composed using 100% recycled spam messages!

initscripts-2011.09.2-1
 

On Thu, Oct 6, 2011 at 2:10 AM, Geert Hendrickx <geert@hendrickx.be> wrote:
> On Thu, Sep 29, 2011 at 16:00:52 +0200, Tom Gundersen wrote:
>> * /etc/localtime may now be a symlink, and is not written to unless it is
>> necessary.
>
> why not always make it a symlink, so updates to tzdata package are applied
> immediatly and don't require a reboot? *Is it for scenario's where /usr is
> not available?

Yes. We have sort of given up on supporting booting with an
unavailable /usr, but we don't want to break people's setups
unnecessarily, so we'll keep the backwards compatibility for the time
being.

Cheers,

Tom