increase and duplicity of bgp label unicast routes

[maycongoncalves]

Dear, I have a problem in bgp sessions, where there is a constant growth of bgp label unicast routes, I noticed that there is duplicity with each update, increasing the table and reaching the maximum prefix, causing the bgp session to drop. can you help me with any suggestion of this problem?

Log:

RP/0/RSP1/CPU0:Jan 2 01:35:05 : bgp[1058]: %ROUTING-BGP-5-MAXPFX : No. of IPv6 Labeled-unicast prefixes received from has reached 1516964, max 2000000
RP/0/RSP0/CPU0:Jan 2 01:35:05 : bgp[1058]: %ROUTING-BGP-5-MAXPFX : No. of IPv6 Labeled-unicast prefixes received from has reached 1516964, max 2000000
RP/0/RSP1/CPU0:Jan 2 01:35:35 : bgp[1058]: %ROUTING-BGP-5-MAXPFX : No. of IPv6 Labeled-unicast prefixes received from has reached 1516977, max 2000000
RP/0/RSP0/CPU0:Jan 2 01:35:35 : bgp[1058]: %ROUTING-BGP-5-MAXPFX : No. of IPv6 Labeled-unicast prefixes received from has reached 1516977, max 2000000
RP/0/RSP1/CPU0:Jan 2 01:36:05 : bgp[1058]: %ROUTING-BGP-5-MAXPFX : No. of IPv6 Labeled-unicast prefixes received from has reached 1517014, max 2000000
RP/0/RSP0/CPU0:Jan 2 01:36:05 : bgp[1058]: %ROUTING-BGP-5-MAXPFX : No. of IPv6 Labeled-unicast prefixes received from has reached 1517014, max 2000000

[Fredrik Lönnman]

The default value to start nagging you about maximum prefixes is at 75% of the configured value, in your case at 1.5M of your 2M maximum prefixes value.

The reason you are seeing two instance of every log is because you are running a system with dual RSPs and have configured BGP GR/NSR redundancy to sync BGP states between the RSPs, thus they will both send you the same warning.

[maycongoncalves]

thanks for the answer, in this case the duplicity would be several equal addresses like next-hop to the network table, these routes only increase with time and reach the limit, to restore it is necessary to clear the session, I am looking for the reason for this growth , any other suggestions?

[Fredrik Lönnman]

Maybe you need to take a look at the peer, why is it advertising so many prefixes? The "several next-hop without network address" is due to how the output formatting is done, these nexthops will correspond to a prefix further up in your output, se it seems that your peer are advertising many nexthops to the same prefix.

[MHM Cisco World]

duplicate can I see both paths ?

[maycongoncalves]

this is an example shows several next-hop without network address, the routes that show network address appears the same, many entries with many same ips, and the routing table is constantly growing.

[MHM Cisco World]

interesting ?
please do show ip bgp x.x.x.x <<- this network is appear in first line I think, 

[maycongoncalves]

I didn't find this command on cisco asr9010, I'm checking with this command show bgp ipv6 labeled-unicast neighbors x.x.x.x routes, the ipv4 routes are ok, the problem is with ipv6 label unicast, routes grow repeatedly, there are two sessions with route reflectors with internet routes.

[MHM Cisco World]

can I see the config ? and if you can config of RR ?

[Harold Ritter]

Hi @maycongoncalves ,

What version are you running on the ASR9k? What platform is the RR and what version is it running? Can you provide a "show bgp ipv6 label-unicast <ipv6 prefix>" for one of the ipv6 prefixes?

Regards,

[maycongoncalves]

we use huawei as route reflector

RP/0/RSP1/CPU0:-asr9010-1#show bgp ipv6 labeled-unicast 2001:480:10::/48
Mon Jan 2 17:10:52.811 UTC
BGP routing table entry for 2001:480:10::/48
Versions:
Process bRIB/RIB SendTblVer
Speaker 3143077313 3143077313
Last Modified: Jan 2 10:42:33.898 for 06:28:18
Paths: (9 available, best #1)
Advertised to update-groups (with more than one peer):
0.5
Advertised to peers (in unique update groups):
2804:15b4:80::e
2804:15b4:0:3::6
Path #1: Received by speaker 0
Advertised to update-groups (with more than one peer):
0.5
Advertised to peers (in unique update groups):
2804:15b4:80::e
2804:15b4:0:3::6
28258 6939 668 22
z.z.z.z (metric 1) from c.c.c.c (z.z.z.z)
Received Label 173904
Origin IGP, metric 0, localpref 700, valid, internal, best, group-best, labeled-unicast
Received Path ID 0, Local Path ID 0, version 3143077313
Community: 28258:21381
Originator: z.z.z.z, Cluster list: c.c.c.c
Path #2: Received by speaker 0
Not advertised to any peer
28258 6939 668 22
x.x.x.x (metric 1) from y.y.y.y (x.x.x.x)
Received Label 124110
Origin IGP, metric 0, localpref 700, valid, internal
Received Path ID 0, Local Path ID 0, version 0
Community: 28258:11481
Originator: x.x.x.x, Cluster list: y.y.y.y
Path #3: Received by speaker 0
Not advertised to any peer
28258 6762 701 668 22
a.a.a.a (metric 1) from y.y.y.y (a.a.a.a)
Received Label 135624
Origin incomplete, metric 100, localpref 700, valid, internal
Received Path ID 0, Local Path ID 0, version 0
Community: 0:21481
Originator: a.a.a.a, Cluster list: y.y.y.y
Path #4: Received by speaker 0
Not advertised to any peer
28258 6939 668 22
x.x.x.x (metric 1) from y.y.y.y (x.x.x.x)
Received Label 226897
Origin IGP, metric 0, localpref 700, valid, internal
Received Path ID 0, Local Path ID 0, version 0
Community: 28258:11481
Originator: x.x.x.x, Cluster list: y.y.y.y
Path #5: Received by speaker 0
Not advertised to any peer
28258 6939 668 22
b.b.b.b (metric 1) from y.y.y.y (b.b.b.b)
Received Label 94577
Origin IGP, metric 0, localpref 700, valid, internal
Received Path ID 0, Local Path ID 0, version 0
Community: 28258:1281
Originator: b.b.b.b, Cluster list: y.y.y.y
Path #6: Received by speaker 0
Not advertised to any peer
28258 13786 2914 668 22
x.x.x.x (metric 1) from y.y.y.y (x.x.x.x)
Received Label 114739
Origin incomplete, localpref 700, valid, internal
Received Path ID 0, Local Path ID 0, version 0
Community: 28258:11581
Originator: x.x.x.x, Cluster list: y.y.y.y
Path #7: Received by speaker 0
Not advertised to any peer
28258 16735 6939 668 22
x.x.x.x (metric 1) from y.y.y.y (x.x.x.x)
Received Label 114739
Origin IGP, localpref 700, valid, internal
Received Path ID 0, Local Path ID 0, version 0
Community: 16735:5 16735:6102
Originator: x.x.x.x, Cluster list: y.y.y.y
Path #8: Received by speaker 0
Not advertised to any peer
28258 6939 668 22
x.x.x.x (metric 1) from y.y.y.y (x.x.x.x)
Received Label 124110
Origin IGP, metric 0, localpref 700, valid, internal, labeled-unicast
Received Path ID 0, Local Path ID 0, version 0
Community: 28258:11481
Originator: x.x.x.x, Cluster list: y.y.y.y
Path #9: Received by speaker 0
Not advertised to any peer
23106 23106 6939 668 22, (received-only)
2804:238:0:1::2 from 2804:238:0:1::2 (d.d.d.d)
Origin IGP, localpref 100, valid, external
Received Path ID 0, Local Path ID 0, version 0
Community: 23106:1001
Origin-AS validity: not-found

in this example we have 3 equal routes for x.x.x.x, with different labels, but we also have cases like this one with equal labels, and these routes accumulate until reaching 2 million until the bgp session crashes

thanks

[MHM Cisco World]

I think this is RR issue not yours 

Origin IGP, metric 0, localpref 700, valid, internal, best, group-best, labeled-unicast <<- group-best make RR send all path of same network (as group) to client.

https://www.cisco.com/c/en/us/td/docs/ios-xml/ios/iproute_bgp/configuration/xe-16/irg-xe-16-book/bgp-additional-paths.html

[maycongoncalves]

Thanks for the answer. I didn't find any option within address-family ipv6 labeled-unicast that would solve this problem or anything on the route reflector side, everything is ok with the settings and the labels continue to multiply inexplicably and the route table grows.

[Harold Ritter]

Hi @maycongoncalves ,

You need to find out which device(s) generates these BGP paths with same the NH, but different labels? In the the sample you provided, 172.24.254.9 seems to be the culprit. What devices is that? Can you check its configuration to see what could be wrong?

Regards,