QUOTE(wKkaY @ Oct 18 2013, 10:15 PM)
Well in the case of PPPoE, the MTU of the host's link layer is bigger than the router's.
AFAIK there's two ways to handle this - 1) through RA announcements about the smaller MTU (not sure whether this is honored by popular OSes), or 2) MSS clamping (which breaks end-to-end principle). The fallback is PMTU discovery.
Setting a smaller MTU in the RA's seems the possible using the IPv6 -> ND option in RouterOS
QUOTE(asellus @ Oct 19 2013, 11:16 AM)
The highlighted part is what annoyed me to date with this router. Never encountered the problem before with the HE tunnel maybe because HE gives out static /64 and /48. Do you know how to script this so that there is no need to recycle the Ethernet/wireless interfaces in the computers/tablets in the network when a /64 has been assigned?
My new TP-Link TL-WR1043ND with OpenWRT that I have prepared for my other Streamyx 1Mb account doesn't suffer from that problem.
I just tested using both my Windows/Linux boxes.
i. Manually changing the IPv6 address in ROS or disabling/enabling the PPPoE client (sending a PPPoE terminate request and obtaining a new /64 prefix), the Mikrotik sends out an RA and the clients will have the previous + current /64's bound
ii. A BRAS session reset from the headend results in the same behavior from the client .. so hung sessions or administrative disconnects shouldn't be a problem
My Linux client doesn't seem to be unaffected when having multiple /64 prefixes on the interface. Refreshing the client interface may only be needed if you want to purge the older invalid prefixes but it doesn't seem to affect the ability for the clients to utilize IPv6 as long as one of the prefixes is valid. I've tried getting the Routerboard to force the clients to drop the older prefix but nothing seems to work so far.
Interestingly, during one of the server side session terminates I requested.. I was reassigned the same IPv4 address but my IPv6 prefix was incremented by 1.. so it seems the IPv6 prefix isn't being derived from the IPv4 value but rather the session identifier itself.