hi all, I have created an experimental tool for mining patterns from logfiles. I have been using it for a month now and it has worked quite well in my environment, so I decided to put it on the web. The tool uses clustering algorithm to divide the logfile into clusters, so that each cluster corresponds to a certain pattern that occurs frequently enough. Lines that do not belong to any detected clusters (so called outliers) are written to a separate file. This allows one to detect rare lines that are possibly anomalous. Also, one can run the analysis process iteratively by reapplying the tool for the outliers file for several times. The tool is available at http://kodu.neti.ee/~risto/slct/, and any feedback is welcome. br, risto _______________________________________________ LogAnalysis mailing list LogAnalysisat_private http://lists.shmoo.com/mailman/listinfo/loganalysis
This archive was generated by hypermail 2b30 : Tue Apr 15 2003 - 11:25:03 PDT