Wifi router question

[Saab Dastard]

AO,

A man after my own heart! My Meridian CD transport and DAC are still going strong (admittedly only 20 something years old), although I had to replace my Musical Fidelity amp with a Yamaha device recently. Sadly the Nakamichi cassette deck is just gathering dust these days, as I've largely replaced tapes with CDs.

Sorry for the thread drift!

SD

[mixture]

I don't think that's a requirement though, merely an example of defensive programming.

Agree with you about defensive programming, however "design by committee" appear to have eventually agreed on the word "should" instead of "must" !

When allocating a new address, servers SHOULD check that the offered network address is not already in use; e.g., the server may probe the offered address with an ICMP Echo Request.

-and-

The client SHOULD perform a check on the suggested address to ensure that the address is not already in use.

[Saab Dastard]

I have long maintained that "should" is the most over-worked word in IT.

"it should work..."
"it should have worked..."
"We should have done..."
"We shouldn't have done..."
"We should try..."

etc. etc.

SD

[SpringHeeledJack]

192.168.0.2 holding steady since a few days......Who's the daddy ?



SHJ

[Mike-Bracknell]

Originally Posted by mixture

Agree with you about defensive programming, however "design by committee" appear to have eventually agreed on the word "should" instead of "must" !

It was never in the original RFC. Purely an addition by Microsoft in SP1 of Server 2003, and given how many years DHCP predates 2003 gives you an idea of the amount of implementations that were out there without it. Even now it's not anywhere near the majority, especially in open source packages chosen for size to stuff into router firmware.

I once lobbied a pair of blokes from IBM who had proposed an addition to the RFC for DHCP to specify the DNS suffix search list. Nothing ever came of it either.

[mixture]

Aah.... so I can blame your various lobbying attempts for the many failings of DHCP ...I shall duly procure a DHCP voodoo doll in your name.....

Thanks for the clarification / extra tidbits ! I guess you really have been around since the days of Babbage and the computing dark ages.

[Loose rivets]

I won't hijack this thread about my Seagate Black Armor woes, but upon changing my Motorola U-verse NVG510, the darn 1TB network drive started playing up. I wondered if it was just not seeing the new router.

The router's IP is 192.168.1.254

I have a 10 digit Device Access Code on the same Motorola label that the WEP is on. I was able to look at the devices - it was a huge list. I cleared that list and let it search for my current kit. Rob's computer. (hard wired) Rivetess' computer and the Black Armor backup drive is all I expected to see at that moment. The entire darn list came back with all the stuff OFF except for the named kit above.

Last night I read somewhere it (the router) remembers all this stuff and puts it back anyway. I can't even delete one at a time.

Well, I kind of have hijacked the thread, but a lot of it seems to be germane. Feeling guilty, I'll start a new thread right away.

[llondel]

I once lobbied a pair of blokes from IBM who had proposed an addition to the RFC for DHCP to specify the DNS suffix search list. Nothing ever came of it either.

That would be RFC3397 specifying how DHCP option 119 is encoded, I assume?

As for it being proposed by MS, the cynic in me wonders if that was because they had a hole in their implementation whereby it might try to erroneously re-allocate an address. I'll let them off if so, it is good defensive programming and handles the case where someone's configured a static IP in the middle of the dynamic allocation pool. On the other side of the coin, there's an interesting denial-of-service attack there if you've got something sitting on the network that will respond to such pings from the DHCP server, such that it thinks the entire pool is allocated

[Mike-Bracknell]

Originally Posted by llondel

That would be RFC3397 specifying how DHCP option 119 is encoded, I assume?

It was earlier than that. My work was in 1998-2000.

As for it being proposed by MS, the cynic in me wonders if that was because they had a hole in their implementation whereby it might try to erroneously re-allocate an address. I'll let them off if so, it is good defensive programming and handles the case where someone's configured a static IP in the middle of the dynamic allocation pool. On the other side of the coin, there's an interesting denial-of-service attack there if you've got something sitting on the network that will respond to such pings from the DHCP server, such that it thinks the entire pool is allocated

You'd be right to be cynical. At that time, Microsoft had been kicked hard by a lot of heavyweight for lack of RFC-compliance in a lot of the things they were doing, so it became apparent that someone inside Microsoft had handed down an edict saying they were going to drive compliance through RFCs, which essentially meant them doing what they wanted and proposing it as extensions to existing RFCs.

[SpringHeeledJack]

A quick question for the egg spurts On checking the router settings just now I notice that my MAC address is both IPv4 and IPv6……How is this possible ?



SHJ

[Saab Dastard]

MAC address is layer 2, IP address is layer 3. They are different things.

MAC address is the hardware address of the NIC, IP address is the logical address on the network. IP implementation can be either IPv6 or IPv4 or both.

SD

[SpringHeeledJack]

Thanks SD, however when I look at the attached devices page on the router page, it shows the IPv4 address (192.168.0.2) with my hexadecimal MAC address and again in the IPv6 column same addresses. I know, what little I do know!, that IPv6 is the new kid on the block due to numbering limitations with IPv4. Why would they both be in use at the same time and not just you're either the old v4 or the new v6 ?



SHJ

[Saab Dastard]

Why would they both be in use at the same time

Because they can. The default for many OSs nowadays seems to be for both IPv4 and IPv6 to be enabled at the same time. MS claim that the network stack works best with both enabled.

A network adapter can have multiple network protocols bound to it, and certainly with TCP/IP, a network adapter can have multiple IP addresses assigned to it.

SD