The logparser toolkit is implemented with Python and requires a number of dependency requirements installed. Users are encouraged to set up the local environment for logparser with Anaconda. However, for ease of reproducing our benchmark results, we have built docker images for the running evironments. Docker is a popular container technology used in production. If you have docker installed, you can easily pull and run docker containers as follows:

$ mkdir logparser
$ docker run --name logparser_py2 -it -v logparser:/logparser logpai/logparser:py2 bash

Note that if you are going to try MoLFI, which requires Python 3, please run the following container:

$ mkdir logparser
$ docker run --name logparser_py3 -it -v logparser:/logparser logpai/logparser:py3 bash

After starting the docker containers, you can run the demos of logparser on the HDFS sample log:

$ git clone /logparser/
$ cd /logparser/demo/
$ python

The logparser demo/benchmark scripts will produce both event templates and structured logs in the result directory:

  • HDFS_2k.log_templates.csv
  • HDFS_2k.log_structured.csv