Symptom

Immediately after an RP failover IPv6 PPPoE sessions are brought down by the router and IPv6 traffic stops flowing on the sessions. This is accompanied by the following error message on the console (where addresses displayed will depend on the router's configuration): %ROUTING-RIB-3-ECMP_ERR_ADD : Path add exceed max number of paths supported by protocol. Table 0xXXXXXXXX, prefix XXXX:X:X::/64, protocol subscriber, intf 0xXXXXXXXX, tunnelid 0, nexthop_table 0xXXXXXXXX, nexthop fe80::XXX:XXXX:XXXX:XXX

Conditions

This issue occurs with IPv6-only or dual stack PPPoE subscriber sessions when the ipv6_rib process is restarted or following an RP failover. This is observed when "ipv6 nd framed-prefix-pool" is used to delegate the prefix.

Workaround

None.

Recovery

The IPv6 subscriber sessions can be bought back up by manualy restarting all instances of the ipv6_nd process. This can be done using the command: process restart ipv6_nd location all Note that router will issue a warning and ask for confirmation before processing this command, for example: router# router# process restart ipv6_nd location all Location all can affect the stability of the System. Proceed? [confirm] On node node0_0_CPU0 ... On node node0_1_CPU0 ... On node node0_2_CPU0 ... On node node0_3_CPU0 ... On node node0_RSP0_CPU0 ... On node node0_RSP1_CPU0 ... complete router# If a large number of IPv6 subscriber sessions are active in the router it may take several minutes for all of them to be bought back up.

Further Problem Description