Overview
If an interface request reports an error, we can locate the problematic microservice based on the return of the interface. Under normal circumstances, we can locate according to the URL specification. URLs in the Linkis interface follow certain design specifications. That is, the format of /api/rest_j/v1/{applicationName}/.+, the application name can be located through applicationName. Some applications themselves are microservices. At this time, the application name is the same as the microservice name, and part of the application itself is a microservice, at this time, you should find the belonging microservice by the application name, and check the log under the corresponding microservice. The corresponding relationship between the microservice and the application name is given below.
Ⅱ. community issue column search keywords
On the homepage of the github community, the issue column retains some of the problems and solutions encountered by community users, which is very suitable for quickly finding solutions after encountering problems, just search for keywords that report errors in the filter filter.
“Linkis 1.0 FAQ”, this document contains a summary of common problems and solutions during the installation and deployment process.
Ⅳ. Locating system log
Generally, errors can be divided into three stages: an error is reported when installing and executing install.sh, an error is reported when the microservice is started, and an error is reported when the engine is started.
An error occurred when executing install.sh, usually in the following situations
Missing environment variables: For example, the environment of java/python/Hadoop/hive/spark needs to be configured under the standard version, and the corresponding verification operation will be performed when the script is installed. If you encounter this kind of problem, there will be a lot of problems. Clear prompts for missing environment variables, such as exception -bash spark-submit: command not found, etc.
The system version does not match: Linkis currently supports most versions of Linux. The compatibility of the os version is the best, and some system versions may have command incompatibility. For example, the poor compatibility of yum in ubantu may cause yum-related errors in the installation and deployment. In addition, it is also recommended not to use windows as much as possible. Deploying linkis, currently no script is fully compatible with the .bat command.
-
The former contains the environment parameters that linkis needs to load during execution, and the latter is the database information that linkis itself needs to store related tables. Under normal circumstances, if the corresponding configuration is missing, the error message will show an exception related to the Key value. For example, when db.sh does not fill in the relevant database configuration, unknow will appear mysql server host ‘-P’ is abnormal, which is caused by missing host.
Report error during microservice execution period
The situation of error reporting during the execution of microservices is more complicated, and the situations encountered are also different depending on the environment, but the troubleshooting methods are basically the same. Starting from the corresponding microservice error catalog, we can roughly divide it into three situations:
Manually installed and deployed microservices report errors: The logs of this type of microservice are unified under the log/ directory. After locating the microservice, enter the corresponding directory to view it.
Ⅵ. locate the source code by remote debug
Under normal circumstances, remote debugging of source code is the most effective way to locate problems, but compared to document review, users need to have a certain understanding of the source code structure. It is recommended that you check the in the Linkis WIKI before remote debugging.After having a certain degree of familiarity to the the source code structure of the project, after a certain degree of familiarity, you can refer to How to Debug Linkis.