Forums before death by AOL, social media and spammers... "We can't have nice things"
|    comp.protocols.tcp-ip    |    TCP and IP network protocols.    |    14,669 messages    |
[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]
|    Message 13,523 of 14,669    |
|    Simon Richter to Albert Manfredi    |
|    Re: DHCP: Server not on subnet, is that     |
|    16 Jun 10 11:32:44    |
      9e4d9402       From: Simon.Richter@hogyros.de              Hi,              On 2010年06月16日 00:39, Albert Manfredi wrote:              > The DHCP server or servers then send back their DHCP Offer message,       > and the client picks the one it prefers. This DHCP Offer message       > includes the Server ID for every responding server. So now the client       > has the Server ID it needs.              > NOW the client issues a DHCP Request, using the appropriate Server ID.              Not yet. The server ID is only used for renewals; if I omit the option        from the ACK packet (assuming the MUST in the table is wrong and the MAY       in the explanatory text is right), then the ISC client will renew via a        broadcast, while uDHCPc will ignore the ACK and time out.              I can see either behaviour being correct, given that the RFC seemingly        contradicts itself.              The server as currently coded works fine for serving the local subnet,        or for serving any subnet as long as a default route option can be        added. This might not always be the case, and I'm wondering whether to        treat this as a configuration error (I'd hand out leases that the        clients cannot renew), or whether the behaviour of the ISC client is        indeed correct.              If I understood correctly, configuration using a relay agent will skip        the agent after the client has received an address, which means the        agent need not be reachable via unicast from the client network, but        only from the side of the actual server.              The actual server code already works fine for all cases except for this        one. The easy way out would be to treat it as a configuration error, and       I'm very inclined to do so, but I'd like to be sure that it indeed is one.               Simon              --- SoupGate-Win32 v1.05        * Origin: you cannot sedate... all the things you hate (1:229/2)    |
[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]
(c) 1994, bbs@darkrealms.ca