Re: [BProc] Re: bpsh bug (fwd)

[Erik A. H. <er...@he...>]

On Wed, Feb 06, 2002 at 03:47:25PM -0500, Nicholas Henke wrote:
> 3.1.6

Hrm.. Ok.  Here's something quick and easy to try with bpsh.c:

        /* need timeouts for dead nodes and the like */
        if (late_connections) {
-           tmo.tv_sec  = 0;
-           tmo.tv_usec = 50000;
+           tmo.tv_sec  = 5;
+           tmo.tv_usec = 0;
        } else {
            tmo.tv_sec  = 300;  /* completely arbitrary */
            tmo.tv_usec = 0;
        }

That's just cranking one of the connection timeout for processes that
I've received SIGCHLD for.  I don't *think* it should make a big
difference but I've seen too much weirdness out of TCP recently to not
try it.

An strace trace of it screwing up might be useful but strace on my
alphas here seems to get confused my all the SIGCHLDs showing up at
once.

- Erik
-- 
Erik Arjan Hendriks          Printed On 100 Percent Recycled Electrons
er...@he...                   Contents may settle during shipment