Re: [PATCH 1/3] xenstored - postpone fork after initialization

[Keir Fraser <keir.fraser@eu.citrix.com>]

On 26/3/08 11:00, "Bastian Blank" <bastian@waldi.eu.org> wrote:

>  HG changeset patch
> # User Bastian Blank <waldi@debian.org>
> # Date 1206528849 -3600
> # Node ID 5e0412c5f5798b5c0acdd4057c91b3820d6c4afd
> # Parent  f5e6cccfdda5537876d6fc2b87ea1124d6043fc8
> Postpone fork after initialization.
> 
> Signed-off-by: Bastian Blank <waldi@debian.org>

I applied all three patches as c/s 17296 and then reverted them in c/s 17304
because they don't work. I think delaying xenstored daemonisation breaks its
startup. Certainly xend itself came up, but xenstore-ls for example did not
work, so clearly xenstored had not initialised properly.

In any case I think it's bad practice to daemonise after initialisation --
if you're going to daemonise it makes sense to do it as early as possible.
Perhaps a better way to go would be to wait for the stdout pipe to be
half-closed? Or to pass switch -P to xenstored and wait for the daemon pid
to be written to the stdout pipe?

 -- Keir