On Apr 7, 2015, at 2:07 PM, Yuri <y...@rawbw.com> wrote: > On 04/07/2015 07:53, Brooks Davis wrote: >> I suppose that since dhclient has been killed and restarted it can't >> know it's on the same network, but in practice you want to try to get >> the same lease again and fall back if it turns out you've moved or your dhcp >> server is broken and lost state. I don't see how this would hurt anything. > > Let's say dhclient is restarted after a while (ex. after the reboot), when > some other host already has that same IP address. dhclient sends the > broadcast with it, and the response will be sent to another host, which > currently has that address, and that other host will discard this response. > dhclient keeps trying for many seconds, doesn't get any response. Then it > falls back to sending from 0.0.0.0->255.255.255.255 (as it should have done > in the first place), and immediately gets the valid response. The problem > delays DHCP handshake, this is how this can hurt.
In point of fact, the IP used for the source address doesn't matter too much for DHREQUESTs because they are subnet local (by definition), and the replies will reach the original sender because they are addressed to the layer-2 MAC address of that host. DHCP operates as much on layer-2 as layer-3 and dhclient, ISC dhcpd, and other DHCP software should handle such cases. There is a specific protocol coming from Zeroconf defined here: http://www.ietf.org/rfc/rfc5227.txt <http://www.ietf.org/rfc/rfc5227.txt> ...which talks about how to handle potential IP conflicts via ARP probing. Regards, -- -Chuck _______________________________________________ freebsd-net@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-net To unsubscribe, send any mail to "freebsd-net-unsubscr...@freebsd.org"
