logdag 0.3.3

Overview

This package generates causal DAGs among time-series events in syslog data. This package works on python3. The input log data is loaded with AMULOG (https://github.com/amulog/amulog). The output DAG is recorded in the format of NetworkX DiGraph.

This project was partially forked from repository LogCausalAnalysis. (https://github.com/cpflat/LogCausalAnalysis)

Usage

All features are available from command line. First you should try following help command python -m logdag -h.

short usage:

1. Generate amulog database (and its config) as the input log time-series source

2. Prepare logdag config file by referring logdag/data/config.conf.default

3. Generate time-series db by logdag.source features

4. Generate DAGs by subcommand makedag

5. See results by commands such as show-subgraphs

See the wiki for full documentation (installation, quick start, configuration, causal-discovery methods, and CLI reference), and the tutorial directory for a runnable sample.

Reference

This project is evaluated in some papers CNSM2019 and TNSM2018. If you use this code, please consider citing:

@inproceedings{Kobayashi_CNSM2019,
  author = {Kobayashi, Satoru and Otomo, Kazuki and Fukuda, Kensuke},
  booktitle = {Proceedings of the 15th International Conference on Network and Service Management (CNSM'20)},
  title = {Causal analysis of network logs with layered protocols and topology knowledge},
  pages = {1-9},
  year = {2019}
}


@article{Kobayashi_TNSM2018,
  author = {Kobayashi, Satoru and Otomo, Kazuki and Fukuda, Kensuke and Esaki, Hiroshi},
  journal = {IEEE Transactions on Network and Service Management},
  volume = {15},
  number = {1},
  pages = {53-67},
  title = {Mining causes of network events in log data with causal inference},
  year = {2018}
}

License

3-Clause BSD license

Author

Satoru Kobayashi