The Layer 2 loop always has bad impact for the whole network since it can cause huge ARP broadcast packets in the network, and further it will affect normal ARP learning. If Gateway couldn’t learn ARP in time, the service will be down.
As we know, even the port on access switch is just 1G, but 1G can include huge ARP packets, the core switch couldn’t handle so many ARP packets in time, and most of them are dropped by cpu-defend and ARP anti-attack.
so, for the service going down issue, we can always check the ARP packets with bellow commands:
display cpu-defend statistics all ///to check if there are quite a lot of ARP packets were dropped by cpu-defend.
display arp anti-attack record ///to check if there are quite a lot of ARP packets were dropped by anti-attack