Crashed process (core.zebra) on Edge Node

Products

• VMware NSX

Issue/Introduction

A similar core dump file can be observed on the edge node under /var/log/syslog.log

2024-11-19T04:34:43.478Z xxxxxx NSX 1244986 - [nsx@6876 comp="nsx-edge" subcomp="node-mgmt" username="root" level="WARNING"] Core file generated: /var/log/core/core.zebra.xxx.gz

At the same time, in /var/log/frr/frr.log

2024/11/19 04:34:43.174828 PIM: igmp_force_startup_mode_on_intf: PIM not enabled on interface backplane-514
2024/11/19 04:34:43.174915 PIM: VRF Deletion: VRF-15543(155)
2024/11/19 04:34:43.174940 ZEBRA: Zebra Route table free for vrfId 155 tableId 406 afi 1 safi 1
2024/11/19 04:34:43.174946 ZEBRA: Zebra Route table free for vrfId 155 tableId 406 afi 1 safi 2
2024/11/19 04:34:43.174949 ZEBRA: Zebra Route table free for vrfId 155 tableId 406 afi 2 safi 1
2024/11/19 04:34:43.174951 ZEBRA: Zebra Route table free for vrfId 155 tableId 406 afi 2 safi 2
2024/11/19 04:34:43.174955 ZEBRA: VRF VRF-15543 id 155 deleted
2024/11/19 04:34:43.175100 PIM: pim_if_del_vif: vif_index=0 < 1 on interface pimreg406 ifindex=0
2024/11/19 04:34:43.175264 PIM: pimd/pim_mroute_msg.c pim_mroute_set: failure: setsockopt(fd=-1,PIM_IPPROTO, MRT_TABLE=406): errno=9: Bad file descriptor
2024/11/19 04:34:43.175404 PIM: Could not disable mroute on socket fd=-1: errno=9: Bad file descriptor

2024/11/19 04:34:49.343651 STATIC: Terminating on signal
2024/11/19 04:34:49.352904 BGP: Terminating on signal

Environment

VMware NSX

Cause

The issue can be observed when VRF gets deleted while the BGP instance and peers are still configured. On VRF deletion, bgpd receives notification from zebra. BGP session reset is invoked in BGPd iterating through all the configured peers in the VRF. 

It is observed during config clean-up, no run-time functional impact. 

Resolution

The issue will be resolved in VCF 9.0.