Re: Port Scan for UDP
Excuse your arrogance, but let me correct you in some points you made! First of all nmap does not scan only the services listed in /etc/services, if you were to have bothered reading the manual before answering you would have read, and I quote: The default is to scan all ports between 1 and 1024 as well as any ports listed in the services file which comes with nmap. NOTE! Comes with nmap! usually located in /usr/local/share/nmap/nmap-services. You could have spared the TCP/UDP diff lecture since the question wasn't directed to that... Although you were correct about UDP and it's difficulty when it comes to remote scanning. jc: If you own the box and *don't* have any reason to assume/think you've been compromised (Just checking) you can check locally using nice tools like: netstat -an --ip for both udp and tcp or netstat -an --udp[--tcp] for either one. lsof -i -n nmap localhost -p 1-[HigherPortNumber] fuser and the list goes on =) -xbud - [EMAIL PROTECTED] I only drink to make other people interesting - On Sunday 21 October 2001 09:45 am, Craig McPherson wrote: I can't believe nobody has answered this correctly yet. UDP is different than TCP in that it is a stateless protocol, and that means you have to understand a few things to interpret UDP port scan results correctly. With TCP scans, you get one of three results: OPEN (meaning that the TCP handshake sequence to open a connection completed), CLOSED (meaning that the target sent a port closed ICMP message), or FILTERED (meaning that no response was received at all: this is also called stealthed by so-called security experts like Steve Gibson but it's a good idea to ignore him just on general principles). UDP is a completely stateless protocol, though. Even if you send a UDP packet to a port that a valid daemon is listening on, the system isn't obligated to send anything back to you at all: there is no handshake sequence to establish a connection. So making a determination is harder than with TCP. If you receive a port closed ICMP message, the port can be safely listed as CLOSED, but if you receive nothing at all, that could mean that the port is either OPEN or FILTERED -- it's pretty much impossible to tell the difference. NMAP assumes that every UDP port that it doesn't receive a responsee from is OPEN, which means that if you have your firewall DROP all UDP connections, every UDP port will appear as open. If you want to fix this, have your firewall REJECT instead of DROP, and the ports will appear correctly as CLOSED to a port scan. DROPing connections without a response is in violation of the RFCs, too, if you care about that sort of thing. Having the local machine portscan itself will also tell you which UDP ports are *actually* open, because I assume you don't have your firewall set to DROP packets from itself. Also, did you know that by default, nmap only scans ports listed in its services file? So although it scans commonly-used ports, it's not scanning the entire system. If you have enough time (this will make the scan very slow, especially over a slow network link), use the -p 1- argument to every scan to force nmap to scan every port from 1 to 65535 instead of just the maybe 400 or 500 ports that it has listed in its services file. That's the only way you can get a complete picture of what your box looks like from the outside. I'm doing portscans on a system I'm working to learn more about securing hosts and setting up iptables. My tcp portscan reported what I expected, only www, ssh and smtp listening. The udp portscan reported a huge list of 'open' ports. I really didn't know what to expect for this scan, so I want to know if this is normal. Just for grins, I removed every udp listing in /etc/services and restarted inetd and the scan came back the same. I figure this is normal, but if someone can confirm this behaviour, I'd really appreciate it. If this isn't secure behaviour, perhaps I can add an iptables entry like: iptables -A INPUT -p udp -j drop However, I don't have any applications running using udp, so the 'open' port doesn't have anywhere to go, as far as I know. Again, if someone can confirm this, I'd really appreciate it. thanks, jc -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Port Scan for UDP
Excuse your arrogance, but let me correct you in some points you made! First of all nmap does not scan only the services listed in /etc/services, if you were to have bothered reading the manual before answering you would have read, and I quote: The default is to scan all ports between 1 and 1024 as well as any ports listed in the services file which comes with nmap. NOTE! Comes with nmap! usually located in /usr/local/share/nmap/nmap-services. You could have spared the TCP/UDP diff lecture since the question wasn't directed to that... Although you were correct about UDP and it's difficulty when it comes to remote scanning. jc: If you own the box and *don't* have any reason to assume/think you've been compromised (Just checking) you can check locally using nice tools like: netstat -an --ip for both udp and tcp or netstat -an --udp[--tcp] for either one. lsof -i -n nmap localhost -p 1-[HigherPortNumber] fuser and the list goes on =) -xbud - [EMAIL PROTECTED] I only drink to make other people interesting - On Sunday 21 October 2001 09:45 am, Craig McPherson wrote: I can't believe nobody has answered this correctly yet. UDP is different than TCP in that it is a stateless protocol, and that means you have to understand a few things to interpret UDP port scan results correctly. With TCP scans, you get one of three results: OPEN (meaning that the TCP handshake sequence to open a connection completed), CLOSED (meaning that the target sent a port closed ICMP message), or FILTERED (meaning that no response was received at all: this is also called stealthed by so-called security experts like Steve Gibson but it's a good idea to ignore him just on general principles). UDP is a completely stateless protocol, though. Even if you send a UDP packet to a port that a valid daemon is listening on, the system isn't obligated to send anything back to you at all: there is no handshake sequence to establish a connection. So making a determination is harder than with TCP. If you receive a port closed ICMP message, the port can be safely listed as CLOSED, but if you receive nothing at all, that could mean that the port is either OPEN or FILTERED -- it's pretty much impossible to tell the difference. NMAP assumes that every UDP port that it doesn't receive a responsee from is OPEN, which means that if you have your firewall DROP all UDP connections, every UDP port will appear as open. If you want to fix this, have your firewall REJECT instead of DROP, and the ports will appear correctly as CLOSED to a port scan. DROPing connections without a response is in violation of the RFCs, too, if you care about that sort of thing. Having the local machine portscan itself will also tell you which UDP ports are *actually* open, because I assume you don't have your firewall set to DROP packets from itself. Also, did you know that by default, nmap only scans ports listed in its services file? So although it scans commonly-used ports, it's not scanning the entire system. If you have enough time (this will make the scan very slow, especially over a slow network link), use the -p 1- argument to every scan to force nmap to scan every port from 1 to 65535 instead of just the maybe 400 or 500 ports that it has listed in its services file. That's the only way you can get a complete picture of what your box looks like from the outside. I'm doing portscans on a system I'm working to learn more about securing hosts and setting up iptables. My tcp portscan reported what I expected, only www, ssh and smtp listening. The udp portscan reported a huge list of 'open' ports. I really didn't know what to expect for this scan, so I want to know if this is normal. Just for grins, I removed every udp listing in /etc/services and restarted inetd and the scan came back the same. I figure this is normal, but if someone can confirm this behaviour, I'd really appreciate it. If this isn't secure behaviour, perhaps I can add an iptables entry like: iptables -A INPUT -p udp -j drop However, I don't have any applications running using udp, so the 'open' port doesn't have anywhere to go, as far as I know. Again, if someone can confirm this, I'd really appreciate it. thanks, jc
Re: Questions regarding the Security Secretary Position
Are they both around 20 years of age and steaming hot ? - like the ones we all hope wish we had as receptionists in our corps ? =) -xbud On Sunday 21 October 2001 04:52 pm, Thomas Bushnell, BSG wrote: Matt Zimmerman [EMAIL PROTECTED] writes: On Sun, Oct 21, 2001 at 09:23:03AM -0700, Thomas Bushnell, BSG wrote: Martin Schulze [EMAIL PROTECTED] writes: Q: Is a requirement being a Debian developer? No. It is my understanding that it would be good to have fresh blood in the team. Working on security can cost a lot of time, thus it could even be helpful not being a Debian developer since that implies active package maintenance as well. However, similar knowledge is very helpful, and may be required when working on issues. I think the security secretary, if we have one, should be a Debian developer. We have two of them, and they are both card-carrying developers. Sorry; I was referring to the QA, not the present incumbents.