On Wed, Oct 29, 2025 at 04:04:41PM +0100, Claudio Jeker wrote:
> The RTR and RDE process fatal when the parent closes the imsg connection.
> This is not ideal since that happens when one of the other processes dies
> and the parent exits. So this kind of hides the real reason further up in
> the log.
> 
> The SE already does the right thing so just use that idiom and quit the
> main event loop ASAP and exit out.

ok tb

> 
> -- 
> :wq Claudio
> 
> Index: rde.c
> ===================================================================
> RCS file: /cvs/src/usr.sbin/bgpd/rde.c,v
> diff -u -p -r1.658 rde.c
> --- rde.c	29 Oct 2025 10:34:23 -0000	1.658
> +++ rde.c	29 Oct 2025 14:58:30 -0000
> @@ -272,9 +272,11 @@ rde_main(int debug, int verbose)
>  			fatal("poll error");
>  		}
>  
> -		if (handle_pollfd(&pfd[PFD_PIPE_MAIN], ibuf_main) == -1)
> -			fatalx("Lost connection to parent");
> -		else
> +		if (handle_pollfd(&pfd[PFD_PIPE_MAIN], ibuf_main) == -1) {
> +			log_warnx("RDE: Lost connection to parent");
> +			rde_quit = 1;
> +			continue;
> +		} else
>  			rde_dispatch_imsg_parent(ibuf_main);
>  
>  		if (handle_pollfd(&pfd[PFD_PIPE_SESSION], ibuf_se) == -1) {
> Index: rtr.c
> ===================================================================
> RCS file: /cvs/src/usr.sbin/bgpd/rtr.c,v
> diff -u -p -r1.31 rtr.c
> --- rtr.c	14 Apr 2025 14:50:29 -0000	1.31
> +++ rtr.c	29 Oct 2025 14:58:30 -0000
> @@ -264,9 +264,11 @@ rtr_main(int debug, int verbose)
>  			fatal("poll error");
>  		}
>  
> -		if (handle_pollfd(&pfd[PFD_PIPE_MAIN], ibuf_main) == -1)
> -			fatalx("Lost connection to parent");
> -		else
> +		if (handle_pollfd(&pfd[PFD_PIPE_MAIN], ibuf_main) == -1) {
> +			log_warnx("RTR: Lost connection to parent");
> +			rtr_quit = 1;
> +			continue;
> +		} else
>  			rtr_dispatch_imsg_parent(ibuf_main);
>  
>  		if (handle_pollfd(&pfd[PFD_PIPE_RDE], ibuf_rde) == -1) {
>