OVN ndb container restarted

Issue

• We saw the OVN ndb container restarted in the same time of the issue.

• We were able to recover coredumps on some computes but nothing in the controllers and the following stack is seen:

#0  0x000055851cf5c271 in get_local_datapath (tunnel_key=<error reading variable: Cannot access memory at address 0x100>, local_datapaths=0x55851f157ab0) at controller/local_data.c:241
#1  remove_local_datapath_peer_port (pb=<optimized out>, ld=<optimized out>, local_datapaths=0x55851f157ab0) at controller/local_data.c:242
#2  0x000055851cf2a4a0 in remove_pb_from_local_datapath (pb=<optimized out>, b_ctx_out=b_ctx_out@entry=0x7ffe997f59a0, ld=<optimized out>) at controller/binding.c:1752
#3  0x000055851cf2ad7a in handle_deleted_lport (b_ctx_in=<optimized out>, b_ctx_out=<optimized out>, pb=<optimized out>) at controller/binding.c:2103
#4  handle_deleted_lport (pb=<optimized out>, b_ctx_out=b_ctx_out@entry=0x7ffe997f59a0, b_ctx_in=<optimized out>) at controller/binding.c:2081
#5  0x000055851cf2e739 in binding_handle_port_binding_changes (b_ctx_in=0x7ffe997f5a00, b_ctx_out=0x7ffe997f59a0) at controller/binding.c:2363
#6  0x000055851cf563a8 in runtime_data_sb_port_binding_handler (node=0x7ffe997f6dc0, data=0x55851f157ab0) at controller/ovn-controller.c:1367
#7  0x000055851cf72472 in engine_compute (recompute_allowed=<optimized out>, node=<optimized out>) at lib/inc-proc-eng.c:369
#8  engine_run_node (recompute_allowed=true, node=0x7ffe997f6dc0) at lib/inc-proc-eng.c:424
#9  engine_run (recompute_allowed=true) at lib/inc-proc-eng.c:449
#10 0x000055851cf27446 in main (argc=<optimized out>, argv=<optimized out>) at controller/ovn-controller.c:3629

• In addition, a user brought us up that instances on an internal network also has a network outage without FIP and unbound port.