spellpy 0.0.9

Auto parser for system raw log without human guidance

Spell

Use python implement the paper Spell: Streaming Parsing of System Event Logs from Min Du, Feifei Li @University of Utah.

This implement is refactored and enhancement version of logpai's logparser.

Install

pip install spellpy

Getting Started

python example.py

After executing the line above, the result folder will be created and you will see two files: structured.csv and templates.csv.

 

• *_main_structured.csv

...	Level	Component	Content	EventId	EventTemplate	ParameterList
...	INFO	dfs.DataNode$DataXceiver	Receiving block blk_-1608999687919862906 src: /10.250.19.102:54106 dest: /10.250.19.102:50010	f57d69cf	Receiving block blk_-1608999687919862906 src <*> <*> dest <*> 50010	['/10.250.19.102:54106', '/10.250.19.102']
...	INFO	dfs.DataNode$PacketResponder	PacketResponder 1 for block blk_-1608999687919862906 terminating	7b619377	PacketResponder <*> for block blk_-1608999687919862906 terminating	['1']
...	INFO	dfs.DataNode$DataXceiver	Receiving block blk_-1608999687919862906 src: /10.250.10.6:40524 dest: /10.250.10.6:50010	f57d69cf	Receiving block blk_-1608999687919862906 src <*> <*> dest <*> 50010	['/10.250.10.6:40524', '/10.250.10.6']

 

• *_main_templates.csv

EventId	EventTemplate	Occurrences
6af214fd	Receiving block <*> src <*> <*> dest <*> 50010	5
26ae4ce0	BLOCK* NameSystem.allocateBlock <*>	2
dc2c74b7	PacketResponder <*> for block <*> terminating	4

 

As you see, there have three test log files. Use for loop to simulate (nearly) streaming situation.

In the result folder, there are _main_*.csv files and *.log_*.csv files. The _main_*.csv files will keep appending the new coming log when it has been parse.

Plot

We can use graphviz to visualize the tree-structured of the parser.

python plot_tree.py

 

Tests

sh test.sh

Name	Stmts	Miss	Cover
spell/init.py	0	0	100%
spellpy/spell.py	319	188	41%
test/test_spellpu.py	65	1	98%
TOTAL	387	189	51%

TODO

• Refactor spellpy.py
• Test coverage