I just lately seen that my IPv6 multicast visitors was being broadcasted to all hosts related to a swap, even people who had not subscribed to the visitors.
To repair this, I enabled MLD snooping on the swap, and this fastened the difficulty – now solely these hosts which have subscribed to a given multicast group obtain visitors for it, nice.
Nonetheless this broke fundamental IPv6 performance on the community, and I may now not hook up with (and even ping) hosts related to the identical swap.
The issue turned out to be that the IPv6 neighbour solicitation packets are despatched to a multicast tackle of the shape ff02::1:xxxx:xxxx which the swap was not forwarding because it appeared to resolve that no one had subscribed to these multicast teams.
My query is, is that this a bug within the swap’s IPv6 firmware, is it as a result of I have never configured MLD snooping correctly, or is it as a result of I must configure every host to ship MLD subscriptions for these explicit multicast teams?
I’m not positive what the right behaviour is for MLD snooping and these IPv6 messages. Are hosts meant to subscribe to these teams, or are switches meant to ahead on these requests regardless? If they’re meant to be forwarded all the time, are all packets addressed to ff02::/16 meant to be forwarded with out subscription, or solely ICMP6 visitors? The swap does ahead on packets for ff02::1 and comparable addresses with none subscription which leads me to assume you are not alleged to subscribe to this visitors, however then I discover it laborious to consider that one thing that nearly instantly breaks all IPv6 visitors would go unnoticed for thus lengthy.
The swap sadly is lengthy out of assist and now not getting firmware updates (Cisco 2975), however earlier than I’m going to the hassle of changing it I would like to verify it is truly a firmware bug and never one thing I’ve simply misconfigured that may stay even with a brand new swap!