I read most of topics but I can't find solution with my problem. Neither AI ![]()
I have succesfully implemented IPV6 connection on my DSL (RB5009) connectivity (dhcp-client)... everything is ok on ethernet/LAN
I have succesfully implemented CAPs-MAN.
CAP client/AXs (that run on bridge is ok.
other CAP clients on bridge with VLANs (master/slave) are getting IPV6 route and address but connection is somehow broken and lagged.
IPV6 test fails....
I'm running (possibly) similar setup: IPv6 enabled router running a few subnets over VLANs. CAPs (using local forwarding) connect wireless stations to a few of those VLANs (trictly mapping between VLAN and SSID, no ACL/RADIUS/etc. games with VLAN IDs). APs include AC and AX devices (so either wifi-qcom-ac or wifi-qcom) and hence VLAN setup on some APs is a bit complex in my case.
Both SLAAC (main router is also CAPsMAN server) and stateful DHCPv6 (running on proper linux server) are available. I never experienced any related problems on wireless stations that I could correlate with IPv6.
I'm diggining in configuration and looks like MTU1500 on ND interface has needed to change to 1492. Now it's ok
It is beneficial if MTU is same on all involved interfaces. In your particular case that's between LAN and WAN. If ICMPv6 is not crippled on some hop, different MTUs should not be a problem since there's something like "path MTU discovery" built in IPv6 as well ... but some admins seem to hate ICMP (since IPv4) and are blocking ICMPv6 as well ... and that can break quite a few functions.
And yes, some WAN interface types (e.g. PPPoE) can come with smaller-than-defacto-standard MTU size which can mess with IPv6 connectivity as well. It's not usual to see sub-optimal MTU sizes on IPoIP connections (e.g. when using ether port directly, running ND and DHCPv6) ... unless router's WAN is actually another router (e.g. ISP's device, running PPPoE). In which case RAs (sent out by those devices) should indicate correct routers (IPv6 address of upstream router) and MTU size. However, unfortunately many ISPs are pretty ignorant about proper IPv6 setup so it should not be too big a surprise to see broken setups on ISP side.