[Ovirt-devel] [PATCH] wrappers for taskomatic.rb and host-status.rb
steve linabery
slinabery at gmail.com
Thu May 1 23:13:40 UTC 2008
On Thu, May 1, 2008 at 6:09 PM, Hugh O. Brock <hbrock at redhat.com> wrote:
> On Thu, May 01, 2008 at 11:56:38PM +0100, Daniel P. Berrange wrote:
> > On Thu, May 01, 2008 at 05:37:04PM -0500, steve linabery wrote:
> >
>
> > This is not really a good idea.
> >
> > The decision about how/if to daemonize belows to the initscripts. In
> > SysV init this is daemon/killproc. In the new Upstart world, it doesn't
> > daemonize at all - it keeps ownership of the process so it can detect
> > death & restart.
> >
> > Xen used to follow the approach of your patch too and we just spent
> > time ripping out all that code and making it used daemon/killproc
> > correctly in the initscripts in F10.
> >
> > The simpler way around this problem would have just been to change the
> > existing initscript to pass '-n' when starting the programs so that
> > the Daemons.daemonize() call was skipped - it was redundant when running
> > the program via the 'daemon' shell command. And figuring out whatever is
> > needed to make killproc work correctly.
> >
> > The existing initscript also really needs to be split up so that there is
> > one initscript per daemon that ovirt has.
> >
> > Dan.
>
> Whups... guess I shouldna acked it then.
>
> Steve, care to try again along the lines above?
>
> --Hugh
>
No problem, I'll fix it tomorrow or over the weekend.
I wasn't thinking about the SysV init conventions, unfortunately. Was
just looking at the ruby docs for Daemons and they suggested the
wrapper scripts. So it's their fault...yeah, that's the ticket! ;)
Steve
More information about the ovirt-devel
mailing list