Log4j - how to tokenize logs like Elastic Search and group by values, filter and visualize them?
When you know the logging format/pattern, is there a way to visualize/classify/filter logs with certain thread name/client id/... without using Elastic Search? I want to show for example, "this thread has these lines, ordered by timestamp, with these keywords". I don't want to manually search by thread names anymore.
I can manually grep, of course, but I want to visualize them in a better way. I can search them if I know the thread names, but I want to show all of the thread names without knowing them previously.
Also, ES marks session boundary, but grep does not. You can see that a session/request is different from another, because the session id has changed. A thread can undertake several sessions in a period of time.
I have only seen that in ES for now. I think it can be called as a log tokenizer and classifier tool, but towards static log files which haven't been fed into ES.
Now, I am trying to do this in Excel: adding ; between tags, and parse them to columns, and classify/filter. In the screenshot, thread names is filtered to be one. But I found it very hard to do if logs are too many because separating by space is not good, need to manually add ; or other unique separator.
Log4j2 has a companion project to visualize logs: Chainsaw. It does not have a lot of features, but it might be enough for your needs.