<ldir>
An observation - odhcpd lease times all end at the same time, or they do in my case. I have a short 'valid lifetime' coming from my ISP's prefix delegation, 60 mins. This results in all dhcpv6 leases being valid until the same prefix delegation timeout. And this causes a flurry of dhcpv6 renew activity from all my clients at very similar times.
<ldir>
This seems sub-optimal to me. I think a better solution would be to pseudo randomly reduce the leased preferred time for each lease, say up to 1/4 of the preferred lease time. Discuss :-)
<ldir>
the lease time remaining on dhcpv4 leases is mostly determined by when the client requested and the lan lease duration and there's no 'lock step' with when the ISP's WAN side lease was obtained.
<ldir>
I'm probably being stupid... I'm dependant on the PD being refreshed and it have a later 'valid until' time when the dhcp request comes in. Hmmmm.
rua has joined #openwrt-devel
<Habbie>
i was also wondering what a reduced lease time would do
<Habbie>
because then after that you just get 'the rest'
hitech95 has quit [Read error: Connection reset by peer]
<ldir>
I just don't like this business of the lan side & wan side being in absolute lock step.
<Habbie>
how about ULA?
<ldir>
not a fan. I like globally routable.
* ldir
goes and turns on odhcpd logging again.
<ldir>
I don't think that all my dhcpv6 clients should be poking the router for a renew at 'exactly' the same time. A spread of 60 seconds would be something.