ARP problems caused by C2960S tftp

ARP problems caused by C2960S tftp
Environment: 1. management HOST: win7 (ip: 192.168.90.24/24) 2. vswitch: 10.108.0.5/24 model: C2960S (sw_ver: 15.0 (2) SE6) symptom: log on to the vswitch one day for debugging, save the configuration to the local machine tftp, and continue with some other operations, suddenly disconnected from the vswitch, the local machine can no longer Ping the vswitch. Handling process: Suddenly, first check the internal network and find that only the local machine cannot access this switch. perform the following tests: 1. Access the gateway from the local machine is normal; access to other internal network segments is normal. 2. the Local Machine accesses devices in the same network segment of the switch. 3. the IP address of the vswitch accessed by the local machine. Timeout. 4. Exception. when other hosts access the vswitch, it seems that the problem is between the local host and the vswitch. First, process the local host: 1. restart the computer and try to access it. Exception 2. delete useless software related to the local machine and try to access it. Exception 3. disable the local firewall and try to access it. Exception 4. clean the local process and try to access it. Exception 5. the local machine executes ipconfig/flushdns arp-d, and the fault remains unknown: a previous operation on the local machine triggers a security mechanism of the vswitch, resulting in all the connections from the local machine to the vswitch being rejected. Log on to the vswitch for troubleshooting: 1. view the exchange ARP cache, as shown below: It can be obvious that the exchange has a IP-MAC record of the management host, in theory should be able to pass (this falls into the pit ). After searching for the Internet for half a day, I found an introduction to the similar situation. After a hard time, I still had no progress. The switch arp table is cleared manually, and the host can be accessed by suddenly managing it. Is it ARP ?? In order to reproduce the fault, the administrator attempts to re-execute the operation on the Management host, log on to and back up the configuration of tftp, and continue the operation. A few minutes later, the operation suddenly fell offline. Switch Host Logon, delete ARP records, and recover again. Which operation causes ARP records ?? After multiple tests, we found that as long as the backup configuration is performed to tftp, the server where tftp is located cannot access this switch, and the switch will generate an ARP project for the tftp server. Cause analysis: The management host (90.24/24) and the switch (0.5/24) are in two different networks. When the management host uses telnet to access the switch, data is transmitted to the switch through the gateway, the switch also transmits data through the gateway. However, after executing copy running tftp, the switch adds an ARP cache locally and records the ip-mac correspondence between the management host. As a result, when the switch returns data, it directly captures MAC address packets from the ARP table for transmission. Data is transmitted only in this segment without going through the gateway, leading to abnormal disconnection between the management host and the switch, the host that uses this IP address cannot communicate with the vswitch. This should be a bug in the IOS switch. I found a similar situation on the Internet,