Netty server is running for a period of time, a certain proportion of the client can't link

CPU and memory usage is very low, the client link server port, large server probability no reaction, but the client shows success,

New start the service, each link server can correct response,

You have met? Do you have any good Suggestions?

CodePudding user response:

Information is too little, in what environment running, operating systems, such as Java version, when many users start wrong, what is the percentage

Server logs for error messages, and then analysis, find pressure measurement, and find the reason for the error

CodePudding user response:

reference 1st floor tianfang response:

information is too little, in what environment running, operating systems, such as Java version, when many users start wrong, what is the percentage

Server logs for error messages, and then analysis, find pressure test, find the reason for the error

Java1.8
Centos6.9
Has links to the client 190, each client about every 30 s a message, the server is not active distributed message,

CodePudding user response:

reference 1st floor tianfang response:

information is too little, in what environment running, operating systems, such as Java version, when many users start wrong, what is the percentage

Server logs for error messages, and then analysis, find pressure test, find the reason for the error

Client link server doesn't have any logs, netty log has been opened, the Handler channelActive () method and no log output,

But the client display link successfully, the client sends the message server also does not handle,

CodePudding user response:

The number of connections, BIO won't be a problem, how much more netty such NIO, receives data post-processing process, the problem might be writing database or file?

To optimize the other server? Server optimization reference:

https://blog.csdn.net/leonpengweicn/article/details/69388886

CodePudding user response:

reference 4 floor tianfang response:

the number of connections, BIO won't be a problem, how much more netty such NIO, receives data post-processing process, the problem might be writing database or file?

To optimize the other server? Server optimization reference:

https://blog.csdn.net/leonpengweicn/article/details/69388886

Reply message is received after the client message, or write the mq

The most strange is the server running for a period of time, will appear this kind of circumstance, new start a service, how many client link is how many log output,

CodePudding user response:

Is we do not have the right to end the handler may caused the obstruction, consider increasing the internal message buffer layer, avoid handler invokes the mq connection

Cumulative problem, it is difficult to debug, write client pressure measurement, otherwise it is difficult to find

CodePudding user response:

refer to 6th floor tianfang response:

is do not have the right to end the handler may caused the obstruction, consider increasing the internal message buffer layer, avoid handler invokes the mq connection

Cumulative problem, it is difficult to debug, write client pressure measurement, otherwise it is difficult to find

Thank you, I also prepared today pressure measure,