Re: Socket troubles: Too many open files

[starlight.2012q3 () binnacle cx]

FYI I did eventually get that scan of the
China evilhost completed.  It took three
days and in the end was fairly boring
--almost seems like it was specifically
designed as an 'nmap' tarpit.

The socket count went up to 200k by the
time the scan was complete.

Seems to me this might be a socket leak
as each script run appears to never
clean up the socket allocated for it.

If the sockets are kept for valid reasons,
then 'nmap' should advise users to
increase the 'ulimit' for file descriptors
on exhaustion events.

I'm also still wondering if --send-eth
activity should not all occur on a single
raw/packet socket with response traffic
dispatched internally by 'nmap'.

_______________________________________________
Sent through the nmap-dev mailing list
http://cgi.insecure.org/mailman/listinfo/nmap-dev
Archived at http://seclists.org/nmap-dev/