On a Jenkins 1.584, using Build Failure Analyser 1.12.1, we host more than 2600 jobs, and we have 123 different BFA rules.
Our log files can be fairly big (ranging from 5Mb to 100 Mb in some exceptional cases) and we have started to have serious performance issues, where the BFA scan on a failing build can take up to 15 minutes. This is of course a big issue.
By looking at the code, I've seen that the scan of a log is performed once per rule being defined, and then line by line. This is of course very costly in our case.
Do you have any suggestion about this kind of situation?