Customer is using MX Virtual-Chassis for CGNAT service. If [service-set <> flow-log] is configured, every time when the routing-engine is rebooted online, NSD daemon core dump is seen.

"show system coredump" can observe NSD core dump file after RE is rebooted.

Due to a JUNOS bug with MX virtual-chassis, when jflow-log serivce configuration is parsed, the NDS daemon accesses the uninitialised zone pointer. This causes the crash due to SIGSEGV.
This problem is seen only on MX virtual-chassis rebooting and it will recover by itself, hence won't cause any service impacts.
Customer can either ignore the core dump, or can upgrade to a JUNOS release that having PR1793765 fix.