standalone-nnrpd "dies" when hitting ressource limits

[Sven Paulus]

Hi,

the standalone nnrpd parent process dies if there are temporary problems
when it is trying to fork(). This gives evil netizens the possibility to
"shut down" the NNTP port remotely: Just open enough connections in
parallel, so that the process hits the ressource limits (max. number of 
file descriptors or processes) and the parent nnrpd exit()s.
No, I don't think wrapper shellscripts are a real solution ...

How about:

--- nnrpd.c.old	Fri Jul  7 17:36:05 2000
+++ nnrpd.c	Fri Jul  7 17:36:32 2000
@@ -964,8 +964,8 @@
 	    
 	    for (i = 0; (pid = fork()) < 0; i++) {
 		if (i == MAX_FORKS) {
-		    syslog(L_FATAL, "cant fork %m -- giving up");
-		    exit(1);
+		    syslog(L_FATAL, "cant fork %m -- giving up for this connection");
+		    continue;
 		}
 		syslog(L_NOTICE, "cant fork %m -- waiting");
 		(void)sleep(1);

?

This closes the connection when fork()ing has failed MAX_FORKS times and
goes on accepting the next connection (maybe some children died in
between, so there is room for new children). 

Sven