Re: Squid problem: "ipcache_create_dnsserver: fork: (35) Operation would block"

From: Stephane Bortzmeyer <[email protected]>
Date: Mon, 16 Sep 96 11:36:58 +0200

On Thursday 12 September 96, at 9 h 24, the keyboard of Stephane
Bortzmeyer <bortzmeyer@pasteur.fr> wrote:

> We'll wait to produce the problem again with the higher debug level, then
> we'll try it. One second between each server seems fine (even if you
> launch 20 of them) and will avoid portability problems with usleep.

Apparently, the following patch (on squid 1.0.9) solved the problem. May
be this is something to include. Does anyone see the same problem?

*** ipcache.c~ Thu Aug 22 20:48:17 MET DST 1996
--- ipcache.c Thu Sep 12 17:09:47 MET DST 1996
***************
*** 268,273 ****
--- 268,278 ----
            return -1;
        }
        comm_set_fd_lifetime(sfd, -1);
+ /* Patch S. Bortzmeyer for the Renater cache
+ Wait a bit before returning to let the child exec.
+ */
+ debug(14, 3, "ipcache_create_dnsserver: PATCHED version. Sleeping
one second after creating a server.\n");
+ sleep(1);
        return sfd;
      }
      /* child */

Do not that, since debug doesn't increment the time, the stamps in
cache.log can surprise you.
Received on Mon Sep 16 1996 - 02:37:27 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:33:01 MST