The company tested the service without problems, and there will be some large and small problems after on-site deployment. The summary is as follows:
1. Check the jdk version (the jdk version is different once)
2. Junk data: Through checking the log, it was found that some test data on the spot was not deleted in time, and finally checked out when the service was in normal use, causing some puzzled bugs. (3 times)
3. Query optimization: When some timing tasks are polling, if the query speed is slow, it will cause stalls. There are both front and backends. It is necessary to optimize the query speed. Store commonly used data in redis to reduce interaction with the database (2 times ).
4. Server cpu and its disk: the front desk is stuck. Through investigation, it is found that the problem is that the drive letter space for storing the log is full, and the implementation personnel did not clean it up, which caused the front desk to call the interface when writing the log stuck. (1 time)
Five. Packaging errors
1. The package of the general model and service is not in the same package as the subproject. If the general model or service code is changed, the parent project needs to be packaged (more time-consuming). If only the subproject is packaged, it will cause changes. The code did not compile (1 time)
6. Configuration error
Check the configuration of the site project, database, message middleware, redis connection address; missing fields in the database field (errors will be reported in the background, missing columns) and field size (a garbage data occurred, and the saved field size exceeds the table Field size) etc.; if the configuration read from the cache is updated, the background needs to be restarted (some data is initialized every time the background is started),
Wait...