On 12/16/2009 10:11 AM, Seth Vidal wrote:
On Wed, 16 Dec 2009, Nathanael D. Noblet wrote:
Maybe this is a feature that needs to be addressed in the rpm layer or
something so that upgrades can have multiple effects with regards to
needing a reboot. I'm not sure how PK gets the request to reboot from
a package, but I'm wondering about it.
It doesn't get it from the pkg. It uses the updateinfo.xml metadata that
is generated by our update processing system that is called 'bodhi'.
You can see this data using the yum-security plugin.
Cool.
seems like a package basically has complex upgrade issues, so we
reboot. Are there other tags packages can have other than reboot?
Should there be? etc etc..
No.
The reason for this is that PKs target audience is not someone like me,
and as such no need to provide different messages per package?
I am an advanced user, and manage a handful of servers and
workstations, so yes I don't have to reboot. I'm just wondering about
the reboot 'feature' usage patterns I'm seeing.
And again. PK is not designed for you. The 'reboot often' solution is
not FOR you.
I said this earlier on another subject but you shouldn't be shocked that
camels are slow swimmers.
So basically, PK is designed for the non-experienced users, as such
everything it does is dumbed down, and experienced users should just
ignore it, using other tools to keep their system up to date.
So one last question then, in the case of nfs-utils, (ignoring for now
any nfs specific restart/condrestart issues). The packaging guidlines
will continue to require that a post update script does what is sensible
for an update, and not just depend on the admin rebooting their server?
--
Nathanael
--
fedora-devel-list mailing list
fedora-devel-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/fedora-devel-list
