How to manage the logs of each node in a unified manner after clustering? For example, if a certain function reports an error, you need to go to each node to check the log. This is too troublesome. How is it usually managed?
三叔2017-06-12 09:26:37
There is one log file every day, and the analysis is done through merging. It is not recommended to unify all logs on one server.
A function reported an error
You should check the error log, and there are not many error logs. Do not keep error logs and normal logs together.
为情所困2017-06-12 09:26:37
Distributed open source log analysis system under Google, you can refer to other people's implementation.
The overall idea should be: a distributed queue, log collection + log search engine + graphical interface display
