> >
> > > > > in this status, I tried to start a new httpd, but the daemon would
> > > > > fail with an error message: "couldn not bind to port" (cern-httpd).
> > > > >
> > > > > I had to reboot the machine in order to start httpd again.
> >
> > Its supposed to wait until all connections have cleared (can be 2 hours),
> > ask the cern folks why they don't set SO_REUSEADDR, or better still use
> > a decent httpd
> >
>
> cern-httpd does use SO_REUSEADDR.
>
> $ pwd
> /usr/www/WWW/Daemon/Implementation
> $ grep SO_REU *
> grep: vms: Is a directory
> HTDaemon.c: if ((setsockopt(master_soc,SOL_SOCKET,SO_REUSEADDR,
>
> I remember that I never before had problems with just killing and
> restarting httpd. Seems to be introduced somewhere between 2.0.16?18? - 2.0.20
Same with Roxen, it started with one of the last patches:
ulpc/src/modules/files/socket.c:
if(setsockopt(fd, SOL_SOCKET, SO_REUSEADDR, (char *)&o, sizeof(int)) < 0)
ulpc/src/modules/spider/spider.c:
setsockopt(fd, SOL_SOCKET, SO_REUSEADDR, (char *)&true, sizeof(int));
Mike
Michael Neuffer i-Connect.Net, a Division of iConnect Corp.
mike@i-Connect.Net 14355 SW Allen Blvd., Suite 140
503.677.2900 Beaverton, OR 97008