Is this a Solaris machine, by chance? (ie. I've seen the same thing on
my box(es) and have heard that others have seen the same, but no posted
solutions... yet) It's a problem somewhere in hostmon, I think (from
memory, so don't shoot me)
On Mon, 14 Dec 1998 ttsg@ttsg.com wrote:
> Date: Mon, 14 Dec 1998 17:48:40 -0500 (EST)
> From: ttsg@ttsg.com
> To: nocol-users@navya.com
> Subject: [nocol-users] Need to know more
>
> Hi,
>
> We just had a major fiasco here because we missed a nocol event.
>
> We are getting "(libdebug) eventlog syswrite error: " and don't know
> why. The bit of perlnocol that is doing this is in eventlog:
> -------------------------------------------------------------------------------
> ## Here only if socket has been opened and have to write out EVENT.
> $bytesleft = length ($event) ;
> while ($bytesleft)
> {
> local($written);
> local($offset) = 0;
> $written = syswrite($nlogfd, $event, $bytesleft, $offset);
> if ($written <= 0)
> {
> print STDERR ("(libdebug) eventlog syswrite error: $!\n");
> close($nlogfd);
> $nlogfd = -1; # reset
> return(-1);
> }
>
> $bytesleft -= $written;
> $offset += $written
> } # end: while($bytesleft)
>
> -------------------------------------------------------------------------------
>
> The machine creating the events is the same machine receiving
> the events. Why would the open be ok, then the write fail? Why should it
> fail when its on the same machine? Why isn't $! getting set? Why isn't
> eventlog ever checked for a return code? How do I investigate this more?
>
>
> (The event that happened was something in hostmon went from Warning
> to Error and there is no log of it and our paging program wasn't run)
>
> Thanks, Tuc/TTSG
>
--
Russell M. Van Tassell Commercial Systems Corp.
russell@cscorp.com http://www.cscorp.com/
Ph: 1-888/COMM-SYS
|