Today I took a closer look into a intermittent problem that I'm experiencing with IPv6 connections.
I'm using v6.46.6 with a CCR1009
Using permanent IPv6 TCP connections (ssh,rdp), sometimes sporadic connection drops kept occurring.
I then tried to discover where the problem was, by sniffing traffic at the LAN interface and at the WAN interface. I then discovered that when the problem happens, packets from the pc (LAN side) where not being routed to the WAN side and as such the client pc kept re-transmitting until it gave up and started a new connection (SYN). From there everything was back working again until the next time. During this, packets from the WAN to LAN where being normally routed from the server to the client PC (also with re-transmissions because the server was not receiving from the pc, but being routed as normal).
That made me look closer to the conntrack. There I saw something strange: the connection timeout stayed most of the time with "23:59:59", but then from one moment to the next, "00:04:59" appeared, and then back to "23:59:59". Same connection, not a new one. Coincidentally or not, in one of those glitches, that problem occurred.
This seems to be an issue with the IPv6 conntrack, or I'm I looking at this wrong?
I also added a forward allow all rule from LAN to WAN, and lan_interface to wan_interface, both to no avail. Packets still did not route from LAN to WAN when the problem occurs.
Can someone take also a look at this?
Thanks and best regards,