This is the mail archive of the mailing list for the Cygwin project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: Cygwin service doesn't get NT shutdown notification

On 26 Apr 2001, at 18:15, the Illustrious Christopher Faylor wrote:

> On Thu, Apr 26, 2001 at 01:23:42PM -0700, Paul Garceau wrote:
> >On 26 Apr 2001, at 11:35, the Illustrious Fred Yankowski wrote:
> >>I'm trying to write a wrapper for the Cygwin port of PostgreSQL so
> >>that it can run as an NT service and stop cleanly when the service is
> >>stopped or shutdown due to system shutdown.  I've got it working well
> >>when the service is stopped manually, but when I shutdown/restart my
> >>NT system the service process seems to get killed without being
> >>notified of the shutdown.
> >>
> >>I've written a simple NT service that can be built for both Cygwin and
> >>native NT (based on code from Corinna Vinschen, Chuck Wilson, and
> >>Jason Tishler).  The native Win32 version *does* catch and handle the
> >>shutdown event, but the same code built for Cygwin does not seem to
> >>get the event.
> >
> >Just guessing, but are you including a specific message handling
> >subroutine/module geared to interface between the Cygwin (bash) shell
> >and the NT4 (comspec) shell?  Without being able to draw from
> >experience (another guess -- hopefully an educated one), there is a
> >possibility that Cygwin registry settings are overwriting PostGreSQL
> >registry settings, thereby corrupting your NT4 registry when it comes
> >to shut down (system shut down/restart).
> I have no idea what registry settings would have to do with this.  Why
> would Cygwin's manipulation of the registry (which should only happen
> when you mount or umount stuff) interfere with anything else?

	Just a wild guess based on the fact that the registry is manipulated 
by Cygwin (mount, umount, setup, etc.).  Not familiar enough with 
Cygwin Architecture or Setup to be definitive.


		Paul G.

> I assume that this problem actually has something to do with the way
> Cygwin handles signals and the way it traps things like "shutdown
> events".
> Inspection of Cygwin code would probably be pretty enlightening in
> that regard.
> It's possible that just setting a SIGHUP handler will be adequate to fix
> this problem.
> cgf
> --
> Want to unsubscribe from this list?
> Check out:

Nothing real can be threatened.
    Nothing unreal exists.

Want to unsubscribe from this list?
Check out:

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]