The following is to share the possible causes of cloud server network packet loss, and the corresponding investigation and solution, I hope to help you!

The possible causes of network packet loss on the cloud server are as follows: TCP packet loss caused by rate limiting triggered, UDP packet loss caused by rate limiting triggered, soft interrupt packet loss triggered, UDP sending buffer full, UDP receiving buffer full, TCP full connection queue full, TCP request overflow, connection number reaches the upper limit, and iptables policy setting rules.

Before troubleshooting, log in to the cloud server and check the instance bandwidth and packet quantity. For linux instances, run the sar -n DEV 2 command to check the bandwidth and packet quantity. In this command, rxpck/s and txpck/s indicate the amount of packets received and sent, and rxkB/s and txkB/s indicate the bandwidth for sending and receiving packets. Use the obtained bandwidth and packet volume data to compare the instance specifications and check whether the performance bottleneck of the instance specifications is reached. If yes, upgrade the instance specifications or adjust the service volume. If no, if the performance bottleneck of the instance specifications is not reached, you can further locate the fault using the online support.

You can use various tools and methods to locate the lost packets. Using the ping command, you can check the packet loss of the target host. If no response is received, you can calculate the packet loss rate. If packet loss is detected, you can increase the Ping packet size and sending frequency to further observe the situation.

The Traceroute command displays the routing path through which the data packet reaches the destination host. Using this tool, you can observe the node where packets are lost and locate the network segment where the fault occurs.

mtr is a network diagnostic tool. Combined with the ping and Traceroute functions, it displays the delay and packet loss of each hop of data packets reaching the target host, helping users to locate the problem network segment more accurately.

A network monitoring tool, such as the Wireshark, can resupply network traffic packets and analyze network traffic packets. By analyzing captured data packets, you can identify the data that does not arrive properly and find out the causes for packet loss.

The SmartPing tool is a network monitoring tool that supports forward and reverse ping plots. It displays the trend of network latency, automatically generates a network topology, and sets the latency and packet loss thresholds. If the latency and packet loss thresholds are exceeded, an alarm is generated.

NQA is a network quality analysis tool. It can simulate different network traffic and measure traffic performance, such as packet loss and delay. Helps locate network quality problems.

kprobe tool for more in-depth network packet loss and delay analysis delay, can provide monitoring linux kernel function execution, help developers quickly locate packet loss and delay problems. It can capture the packet loss point, analyze the specific function packet loss location and packet loss reason.

The above methods can check the packet loss of the halo server network from different angles and take corresponding solutions.