Symptom:
Multiple npumgr crashes. 

Fatal Signal 11: Segmentation fault
  PC: [0010f060/A] npumgr_delete_forwarding_rec()
  Faulty address: 0x96769178
  Signal from: kernel
  Signal detail: address not mapped to object
  Process: card=12 cpu=1 arch=A pid=863 cpu=~100% argv0=npumgr
  Crash time: 2014-Jan-23+17:47:24 UTC
  Recent errno: 11 Resource temporarily unavailable
  Stack (23588@0x7effa000):
    [0010f060/A] npumgr_delete_forwarding_rec() sp=0x7effa740
    [000fe5a8/A] sn_npumgr_db_process_forwarding_list() sp=0x7effb754
    [001092a0/A] sn_npumgr_forwarding_add_del_mod_func() sp=0x7effc860
    [0010a23c/A] sn_npumgr_forwarding_add_del_mod_handler() sp=0x7effcd40
    [011bb0d8/A] sn_msg_arriving_handle() sp=0x7effd194
    [01196e04/A] sn_loop_run() sp=0x7efff694
    [01097de8/A] main() sp=0x7efffc0c

Conditions:
15.0 (52633)
The npumgr crash seems to happen sometimes after vpnmgr restart.

Issue happens only if number of  ECMP group in system exceeds 512 and we have combination of IP ECMP and MPLS ECMP routes. Once we have this case, vpnmgr restart triggers this crash.