Unable to establish a connection with the server means that the client obtained the IP through DNS domain name resolution during the process of accessing the server, but a problem occurred during the process of establishing a connection with Host. The analysis of the problem mainly has three directions: client, operator Network, Host
The first is to analyze possible problems on the client:#
1.
Are the hosts of the Probe machine hijacked? Analysis method: If it is hijacked, confirm with the customer whether the Host accessed by Probe belongs to the customer. Because if it is directly hijacked to a ErrorIP, this IP is likely to be invalid (after all, the scammer will not open the website for too long)
2.
Probe has a firewall that blocks access to this website. Analysis method: Instant Testing takes a screenshot. Some information can be seen from the screenshots of many actively blocked websites. You can try adding host Instant Testing to see if you can access it. Some firewalls will block domain names.
3.
ProbeNetwork fluctuates, resulting in a large number of packet losses Analysis method: Look at the Probe error reported and the Network delay in executing Ping to see if it is Critical (needs to be enabled in task settings). If there is a packet capture, you can check the packet capture.
Among them, the first two should be Probe must appear in timely monitoring, and the third type can filter out Probe in the task. If Probe performs tasks multiple times and only one problem occurs, then it is likely to be a temporary fluctuation. You can do Instant Testing once to confirm.
Next, let’s analyze whether there is a problem with the operator:#
1.
You can analyze the operator's equipment by viewing TraceRoute. The principles of ping and trace are the same, except that trace can see the Network delay of each layer of equipment.
2.
If you can see from the trace that the public network device has been reached, it means that there is no problem in the environment of our Probe to the public network.
3.
If suddenly every subsequent layer on the public network is * and the final hop count is 30, it means that the operator's equipment has intercepted it.
4.
If the trace can successfully reach the last Host, it means that the entire Network is very smooth. It should be noted that if the Trace is passed, it can only mean that Network is smooth. Ping is implemented by the ICMP protocol and is the Network layer (IP layer). Page access is through ports 80 and 443, which is implemented by the TCP layer. If the IP layer is passed, the TCP layer will continue to check.
Finally, let’s analyze whether there is a problem on the server side:#
1.
First check, the number of connections that cannot be established is Probe as a proportion of the total. Generally, if the problem is Host, the quantity ratio will not be too low. And there will be a concentrated high incidence in a certain period of time.
2.
Check whether the distribution of these Error Probe is regular. Generally, customers will purchase CDN services, so if there is no problem at the origin site, the problem should be concentrated, or the problem should occur repeatedly in a certain area.
3.
report to customers and suggest checking whether the traffic of Host and the usage of Memory are too high. When the traffic usage of Host is too high or the occupancy of Memory is too high, Host has the practice of actively discarding packets to reduce the pressure on the server.