My bet is that the sig_handler function in pcs/misc/icl_reada.c should
include a call to "exit(1);", to become:
void
sig_handler( int signo )
{
exit_handler();
exit(1);
return;
}
This handler replaces the default handler, which exits the process, so
needs to do the exit itself. Without the extra "exit(1)", the handler
just restores the terminal characteristics (which is what
"exit_handler" does) but does not actually exit the process. Anyone
familiar with the inwards of PCS care to give an opinion on this
theory?
David
On 22 August 2018 at 09:08, Tom Marsh <[log in to unmask]> wrote:
> Ah yes, we ecountered the bash issue too and I had intended to fix it, but thanks for that Sarah. I somehow thought this was an issue that had been fixed in the past but perhaps not. A not-very-computer-savvy-undergraduate of mine incurred the wrath of our system manager when he managed to rack up > 50 zombie processes on a 2-core machine in ~2005 if I remember right.
>
> tom
>
> ########################################################################
>
> To unsubscribe from the STARDEV list, click the following link:
> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=STARDEV&A=1
########################################################################
To unsubscribe from the STARDEV list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=STARDEV&A=1
|