On Mon, Jul 07, 2014 at 11:28:50AM -0500, Patrick Mahoney <pat_at_polycrystal.org> wrote:
> Hi,
>
> The documentation for s6-svscan says
>
> > SIGTERM : acts as if a s6-svscanctl -t command had been received.
>
> However the source in s6-svscan:107
>
> case SIGTERM : cont = 0 ; wantkill = 1 ; killmode =
> KILLMODE_ACTIVE | KILLMODE_SERVICES ; break ;
>
> Is different from s6-svscan:130
>
> case 't' : cont = 0 ; killmode = KILLMODE_ACTIVE |
> KILLMODE_SERVICES ; return ;
>
> In practice, it seems that "s6-svscanctl -t" results in s6-svscan
> exiting without killing any of its child
> processes, while it does do so after receiving SIGTERM.
Ahh, that explains the head-banging I went through more than a year
ago, as there was always a bunch of processes still around after
executing s6-svscanctl -t. I should be more attentive to get the
bugs fixed, not resort to workarounds and forget..
Thank you for analyzing and reporting.
--
Vallo
Received on Fri Aug 08 2014 - 08:09:11 UTC