We write thousands of log files at a time

Hello comrades!

What happens if I set Spring Boot in an application logging.level.root=TRACE?

All logs will be written to a single file, in which there will be a bunch of small classes.

In the best case, with the help sifting appenderyou can divide a maximum of 1 criterion ("discriminator") - for example, by the name of the stream or by the name of the class. And even so - there will be recordability from the source code, which should support the discriminator values ​​in "MDC".

Stop, stop, stop! We remove logback, add , and with this simple configuration we write in separate log files for each class and level of logging - and even in separate directories:

Bobbin.yml

destinations:
  - name: io.infinite.bobbin.config.FileDestinationConfig
    fileName: ("./LOGS/INFINITE/${className}/${level}/${className}_${level}_${date}.log")

And this is what happens:

Now imagine that you are asking SRE a colleague:

-      org.visa.api.Transfer
-     ?
-  ERROR 
- !  .

Of course, conceptually everything rushes towards ELK (well, either splunk or grafana). But it is not always configured everywhere - and the good old log files are still quite often relevant.

But we believe that we need to go from private to general - being able to do such tricks as above will give odds to any logger in the cloud. After all, everything depends on the flexibility and ease of configuration.

We recorded a video demonstrating this example in full, and showing how it works in dynamics:

Enjoy watching!

: Java 8 11, Gradle

( — !). , 2020 . logback log4j2. API — , Google Flogger.

— Gradle Maven .

— Open Source .
.

!