Symptom
Memory leak is detected by the OSPF process at libsrms_api library whenever the interface connected to the
SRMS server is flapped.
How to recreate:
Two SRMS servers and each of them send the DUT entries that are similar in every fashion , except the IP address. see attached. Flap the interface on which the DUT receives the SRMS entries from the SRMS servers.
one server has these SRMS entries
srms server #1
------------------
R None 100.164.2.9 0 100.164.0.0/32 0 256 0
R None 100.164.2.9 0 100.164.4.0/32 1024 256 0
R None 100.164.2.9 0 100.164.5.0/32 1280 256 0
R None 100.164.2.9 0 100.164.6.0/32 1536 256 0
R None 100.164.2.9 0 100.165.1.0/32 2048 256 0
srms server #2
------------------
R None 100.164.2.10 0 100.164.0.0/32 0 256 0
R None 100.164.2.10 0 100.164.4.0/32 1024 256 0
R None 100.164.2.10 0 100.164.5.0/32 1280 256 0
R None 100.164.2.10 0 100.164.6.0/32 1536 256 0
R None 100.164.2.10 0 100.165.1.0/32 2048 256 0
Conditions
OSPF with Segment routing mapping server
When two SRMS servers are used, below is the entries in each.
a1[0] R None 100.164.2.10 0 100.164.0.0/32 0 256 0
a1[1] R None 100.164.2.10 0 100.164.4.0/32 1024 256 0
a1[2] R None 100.164.2.9 0 100.163.1.0/32 0 2048 0
a1[3] R None 100.164.2.9 0 100.163.2.0/32 3096 256 0
a2[0] R None 100.164.2.10 0 100.164.0.0/32 0 256 0
a2[1] R None 100.164.2.9 0 100.164.0.0/32 0 256 0
a2[2] R None 100.164.2.10 0 100.164.4.0/32 1024 256 0
a2[3] R None 100.164.2.9 0 100.164.4.0/32 1024 256 0
Workaround
OSPF process restart
Further Problem Description
more information is @ https://techzone.cisco.com/t5/IOS-XR-PI-OSPFv2-v3-Eng/689575157-OSPF-process-memory-leak/td-p/1552953
