infinite adding of ipv6 routes to kernel table which were recieved from ebgp peer and piped into kernel protocol

[Arnautov Artem]

Hi
we have problems on one of our router with bird6 and ebgp neighbors
when we recieve ipv6 full view from ebgp neighbor and put it into
separate table everything is OK
when we pipe it into master table everything is OK
when we enable export all ipv6 routes into kernel protocol strange
things happens

while trying to find out how many ipv6 routes were added
trying to run command like
ip -6 ro sh | wc -l
we may sit and wait for a long time without any success

if we looking for route count from bird6 console everything is ok
but if we enable debug off route protocol we see messages like

26-06-2013 12:58:03 <TRACE> kernel: Scanning routing table
26-06-2013 12:58:03 <TRACE> kernel: 2001::/32: seen
26-06-2013 12:58:03 <TRACE> kernel: 2001:200:900::/40: seen
26-06-2013 12:58:03 <TRACE> kernel: 2001:200:c00::/40: seen
......

26-06-2013 12:58:03 <TRACE> kernel: 2001:2b8:11::/48: already seen
26-06-2013 12:58:03 <TRACE> kernel: 2001:2b8:15::/48: already seen
26-06-2013 12:58:03 <TRACE> kernel: 2001:2b8:16::/48: already seen
26-06-2013 12:58:03 <TRACE> kernel: 2001:2b8:17::/48: already seen
26-06-2013 12:58:03 <TRACE> kernel: 2001:2b8:19::/48: already seen
26-06-2013 12:58:03 <TRACE> kernel: 2001:2b8:20::/48: already seen
26-06-2013 12:58:03 <TRACE> kernel: 2001:2b8:21::/48: already seen
26-06-2013 12:58:03 <TRACE> kernel: 2001:2b8:22::/48: already seen
26-06-2013 12:58:03 <TRACE> kernel: 2001:2b8:25::/48: already seen

.....

and then again and again and again same lines of data with "already seen"

even if we define scan time 600 and restarting kernel protocol nothing
changes


with routes recieved from ibgp  there are no such problems


here is sample configuration

router id xx.xx.xx.xx;
define myas = ZZZZZ;


protocol device device {
        scan time 10;
}

protocol kernel kernel {
        export all;
        import none;
        scan time 15;
}

protocol direct direct {
        export none;
        import none;
}


table T_ebgp;

filter pipe_ebgp_in
{
        preference = 500;
        accept;
}

filter pipe_ebgp_out
{
        if (myas,client) ~ bgp_community then accept;
	reject;
}

protocol pipe P_ebgp_peer {
        description "EBGP peer Pipe to master table";
        table master;
        mode transparent;
        peer table T_ebgp;
        import filter pipe_ebgp_in;
        export filter pipe_ebgp_out;
}

protocol bgp R_ebgp_peer {
        description "ebgp IPv6";
        local as myas;
        neighbor 2001:xxx:x:xx::1 as XXX;
        import all;
        export all;
        next hop self;
        table T_ebgp;
        gateway direct;
        #disabled;
}

-- 
---
Арнаутов Артем
ISP NetAssist
+380 44 2398989