Thoroughly investigate the problem of intermittent machine interruption in some network segments

Thoroughly investigate the problem of intermittent machine interruption in some network segments

Two servers need to be installed two days ago. As a result, the network that has been faulty in the early stage has come up again. It is clear that the vlan can be pinged, but it is different from that outside the vlan. This phenomenon is like an arp virus, so we conducted a thorough investigation over the weekend, and we must find out the root cause of the problem. In view of the relationship between this network problem and hardware, we also invited network technical engineers from network equipment suppliers.

To reduce interference, shut down the software that was originally running on the Internet, so as to avoid the impact of software control on subsequent judgment.

By disabling or disconnecting the computers in this segment one by one, and retaining only one server installed later, ping the server again.

The network engineer connects to the vswitch to check the status of the vswitch and finds that several machines in this segment cannot learn the mac address, which is abnormal. However, it is strange that the ip addresses shown above are not used by machines in this section.

An OA server used in this section has not found any network interruptions during the test. Later, I found that although the same network segment is used, this server is directly connected to the core switch, this avoids interference with other machines.

After the above checks, the specific cause of the problem was not found. The network engineer said that a server with frequent network disconnection was found to have a small point on the NIC that had been flashing after the remote connection. The network engineer showed that he was searching for the network IP address. He said this was abnormal, the network address is specified and the IP address cannot be found. Because the server has multiple NICs, there is no other problem after checking them one by one. This Nic was previously bound with two ip addresses, and then removed from the other ip address, the system prompts that the Restart service weapon takes effect, but the server does not restart because it needs to provide external services, this is probably the problem. After the computer is restarted, the NIC will find the IP address. After several days of observation, I did not find the network disconnection situation in the previous days again. This network disconnection storm was included in the paragraph.