You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Randy Terbush <ra...@zyzzyva.com> on 1997/01/08 16:33:39 UTC

Re: Patch to make 1.2b3 compatible with Tenon's Machten 4.0.3 (fwd)

> Brian Behlendorf wrote:
> > 
> > 
> > Hmm:
> > 
> > > *** DIST/http_main.c	Thu Dec 26 13:14:02 1996
> > > --- http_main.c	Thu Dec 26 13:15:31 1996
> > > ***************
> > > *** 182,188 ****
> > > --- 182,192 ----
> > >       {
> > >       char lock_fname[30];
> > >   
> > > + #ifdef __MACHTEN__
> > > +     strcpy(lock_fname, "/var/tmp/htlock.XXXXXX");
> > > + #else
> > >       strcpy(lock_fname, "/usr/tmp/htlock.XXXXXX");
> > > + #endif
> > >       
> > >       if (mktemp(lock_fname) == NULL || lock_fname[0] == '\0')
> > >       {
> > > 
> > > 
> > 
> > There's gotta be a more generic solution than this, no?
> 
> Yeah. Use /tmp.
> 
> Cheers,
> 
> Ben.

We've been through that problem though....  How about the same
place as the PID file?






Re: Patch to make 1.2b3 compatible with Tenon's Machten 4.0.3 (fwd)

Posted by Brian Behlendorf <br...@organic.com>.
On Wed, 8 Jan 1997, Randy Terbush wrote:
> > > > *** DIST/http_main.c	Thu Dec 26 13:14:02 1996
> > > > --- http_main.c	Thu Dec 26 13:15:31 1996
> > > > ***************
> > > > *** 182,188 ****
> > > > --- 182,192 ----
> > > >       {
> > > >       char lock_fname[30];
> > > >   
> > > > + #ifdef __MACHTEN__
> > > > +     strcpy(lock_fname, "/var/tmp/htlock.XXXXXX");
> > > > + #else
> > > >       strcpy(lock_fname, "/usr/tmp/htlock.XXXXXX");
> > > > + #endif
> > > >       
> > > >       if (mktemp(lock_fname) == NULL || lock_fname[0] == '\0')
> > > >       {
> > > > 
> > > > 
> > > 
> > > There's gotta be a more generic solution than this, no?
> > 
> > Yeah. Use /tmp.
> > 
> > Cheers,
> > 
> > Ben.
> 
> We've been through that problem though....  How about the same
> place as the PID file?

The PID file is created by the server when it's still root, ja?  Whereas the
serialized accept is done between competing children, ja?  If so the children
couldn't create a lock file in the same place where the PID file is created, if
we follow our own advice about making the logfile directory (where the PID file
is created by default, and should probably stay) 700.

	Brian

--=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--
brian@organic.com  www.apache.org  hyperreal.com  http://www.organic.com/JOBS