Topology: ServerA--NE40--provincial network--NMS server B.
Symptom: The NMS server B telnets to the server A, but the connection disrupts after several minutes.
1. Check the mask of the server A, and it is the same as that of gateway.
2. Ping the server A from NMS server B, and no packet is discarded regardless of big or small packet, indicating the link at between is normal.
3. Reset the server A, but the problem persists, so the server runs well.
4. NE40 could telnet to the server A normally; telnet to the server A with the uplink interface address of NE40 as the source address, but the connection disrupts after some minutes, prompting "The connection was closed by the remote host!". The disruption is not regular. According to the analysis on prompts, it is because NE40 has transmitted packets to the server A, but A does not echo to it or it does echo but the link is disconnected, so the router does not receive it.
5. When telnet connection disrupts, check ARP at the server A, and it has the ARP entry to the gateway; also, it could ping to the gateway and the server B.
6. Check the routing table of the server A. The server A is configured with double net cards, and with different gateways, so it has two default routes with the same priority, failing the load balance.
7. Add the host route at the server A to NMS server B for provincial network. It is normal in test, so the problem is solved.