The process of troubleshooting the problem of decreasing the traffic volume of a single word

"Article Summary "

Recently, a ringtones support staff has reflected a decline in the number of orders after a module restart, hoping that developers will help find the cause of the problem.

In this paper, the process of finding the cause of the problem is introduced in detail, which provides a useful reference for the analysis and solution of the related software problems.

First, the problem description

In a ringtones, there is a software system to implement the word order generation function. One day, the ringtones of the field support personnel sent a message, reflecting the scene of a single-word sorting module after the restart of the number of orders generated is significantly lower than before. The meanings expressed in their emails are as follows:

(1) The field personnel just restarted the module and did not modify any configuration.

(2) After restarting the module (in Figure 1, the single sorting module a) generated by the size of each single word file is only a few one-tenth before the restart, the drop is quite severe.

(3) Another identical single-sorting module (in Figure 1, the single-pickup module B) was also deployed on the scene, but the amount of the order generated was not reduced.

second, on-site deployment diagram

After receiving the email, we quickly began to troubleshoot the problem. Before troubleshooting the problem, we learned about the deployment of each module in the field, as shown in 1.

Figure 1 Field deployment diagram

three, the cause of the initial analysis

We asked the field support staff to return to the word conversion module, the word pickup module A, and the message sorting module B's log and related Word file. Before looking for specific reasons, we are very puzzled that the word order Pickup module A and the single pickup module B are exactly the same, why one has a problem and the other is not a problem? We basically ruled out the possibility of a problem with the program of the word order sorting module A, and looked for other reasons instead.

The program processing of the single-word sortation module A is shown in roughly flow 2.

Fig. 2 procedure process of the single-word sorting module A

As you can see from Figure 2, the word pickup module A will filter the records in the files scanned from the source directory, and then write the records that meet the criteria into the resulting file according to the specific algorithm. The final statement A in Figure 1 is the result file.

To pinpoint the cause of the problem, we restored the environment at the scene, using the same program version, configuration, and database environment as the site. Based on the log returned from the field, we select the source of the session file for a specific period of time to run, and find that the resulting statement is exactly the same as the final statement generated in the field. This further excludes the possibility of problems with the program.

So what is the cause of the problem?

iv. positioning of the problem

Since the program is not a problem, it is not the source of the order to meet the sorting conditions of the record has become less? We looked at the configuration file in detail, which involves the configuration of the sorting conditions as follows:

Fieldfilter= (11:4), (11:7), (11:8), (11:9), (11:10), (11:11)

It means that only the value of the 11th field in each record of the source statement is equal to 4,7,8,9,10,11, and the record can be singled out to generate a result order, otherwise it will be filtered out.

We queried the source message file after the reboot and found that the value of the 11th field is equal to 4,7,8,9,10,11 's record is really very small. "Paddle", since the source is less, it is not surprising that the results become less.

We also queried in the source message file before the restart, found that the value of the 11th field is equal to the number of records corresponding to 4,7,9,10,11, but the 11th field has a value equal to 8, and the number of records in the field is significantly larger after the start of the 8. Is this the reason for the decline in the volume of the statement business?

We'll contact the field support immediately. View the production process of a record with a value equal to 8 for the 11th field, and find that the process exception resulted in a reduction in the single record of the source phone that satisfies the filter, resulting in a decrease in the overall traffic volume.

After the corresponding operation in the field, the record of the 11th field is equal to 8, and the volume of the single traffic will gradually reach the level before restarting the module.

v. Summary

Through this troubleshooting, we have summed up the following several experience:

(1) The problem of the system is not limited to the most intuitive display of the phenomenon of the module. When we find that there is no problem with a module that "thinks there is a problem," you should start by troubleshooting its upstream module until the cause of the problem is identified.

(2) software is a collection of programs, documents, and data. In the process of troubleshooting, we not only have to look at the program code, for the application of the configuration file, and so on, also need to look carefully.

(3) In order to troubleshoot the problem, we should try to master more comprehensive information (including the program log, database information, etc.), which is conducive to reproduce the problem, the final positioning problem.

In the light of civilization, Wu said, "There are more ways than problems." Indeed, as long as we are good at summing up, good at analysis, then any procedural problems can be solved.

(I Weibo: Http://weibo.com/zhouzxi?topnav=1&wvr=5, No.: 245924426, welcome attention!) )

The process of troubleshooting the problem of decreasing the traffic volume of a single word