A interesting issues.
molujia opened this issue · comments
I've met a interetsing issues. When emploied to parse tons of logs, Drain3 miner may report a "divided by 0" error (not exactly, I forgot to note down the error's name, but it means the same). But if you use serveral miners to deal with the same data, this error will diminish or even disappear, which means this issue was not caused by data.
And this tool (Drain3) is neat indeed, I appreciate its valid and simplicity.
Hi, @molujia thank you for reporting the issue! Are you referring exclusively to the big_file demo? The 0 division problem, if I recall correctly, might get thrown if the demo process completes too fast and the naive timer in the demo module will attempt to do a 0 division.
Other than that I'm not currently aware of such a problem in the actual application, please try to reproduce the error and post a dataset/traceback here if possible so I can look into the fix. (it is likely related to the profiling)