Paul Stimpson wrote:
> SIGKILL - That doesn't sound a very nice way to handle a child process
> when its parent gets SIGINTed (Die with no chance of doing a tidy exit.)
> Would not a more logical thing be to send be a SIGHUP? (Your logical
> control terminal has gone away.)
The application is being halted at the OS level and cannot interpret or
read SIGKILL.
> Just wondering why they did it that way.
>
>From the description of the problem there is a window of only a few
seconds to send a signal. Perhaps no signal was sent and the application
did something to make the kernel baulk and SIGKILL the process.
Understandably the application and test environment have been taken away
for further analysis.
-- Damian
--
Damian L Brasher
http://www.diap.org.uk - Advanced backup volume management.
