Skip to content

Instantly share code, notes, and snippets.

@clofresh
Last active March 3, 2023 22:24
Show Gist options
  • Star 5 You must be signed in to star a gist
  • Fork 3 You must be signed in to fork a gist
  • Save clofresh/5550975 to your computer and use it in GitHub Desktop.
Save clofresh/5550975 to your computer and use it in GitHub Desktop.
Multiline datadog agent log parser
[Main]
dogstreams: /path/to/log:/path/to/multiline.py:MultilineParser
''' Log parser that can process multiline log messages. Extracts
counts of log lines by severity, tagging error lines by
the exception type.
'''
import re
from datetime import datetime
import calendar
first_line = re.compile(r'^(?P<timestamp>\d{4}-\d{2}-\d{2} \d{2}:\d{2}:\d{2},\d{3}) \| (?P<severity>[^ ]*) \| (?P<logger>[^ ]*) \| (?P<location>[^ ]*) \|.*$')
stack_trace_end = re.compile(r'^(?P<exception>[a-zA-Z]\w*)(:.*)?$')
def parse_date(date):
return calendar.timegm(datetime.strptime(date, '%Y-%m-%d %H:%M:%S,%f').timetuple())
class MultilineParser(object):
''' Possible parser state transitions:
looking_for_start -> looking_for_start:
* Line doesn't start with a timestamp
* Line starts with a timestamp and isn't severity ERROR (emits point)
looking_for_start -> find_stack_trace_start:
* Line starts with a timestamp and is severity ERROR
find_stack_trace_start -> find_stack_trace_start
* Line doesn't start with "Traceback"
* Line starts with a timestamp (resets state data)
find_stack_trace_start -> find_stack_trace_end
* Line starts with "Traceback"
find_stack_trace_end -> find_stack_trace_start
* Line starts with a timestamp (resets state data)
find_stack_trace_end -> looking_for_start
* Line looks like end of the stack trace (emits point, resets state)
'''
def __init__(self, *args, **kwargs):
self.state = self.looking_for_start
self.data = {}
def parse_line(self, line):
'Entry point to the log parser'
return self.state(line)
def looking_for_start(self, line):
''' Initial state of the parser. Will match lines starting with
a timestamp. If it sees that the log message was of ERROR
severity, transition to the `find_stack_trace_start` state,
otherwise, return a metric point for the line.
'''
global first_line
match = first_line.match(line)
if match:
data = match.groupdict()
severity = data.get('severity')
if severity == 'ERROR':
self.data = match.groupdict()
self.state = self.find_stack_trace_start
elif severity is not None:
return (
'logs.%s.%s' % (data.get('logger'), severity.lower()),
parse_date(data.get('timestamp')),
1,
{'metric_type': 'counter'}
)
def find_stack_trace_start(self, line):
''' Find the start of the stack trace. If found, transition to
`find_stack_trace_end`. If a line beginning with a timestamp
is found first, then reset the state machine.
'''
if line.startswith('Traceback'):
self.state = self.find_stack_trace_end
else:
self.looking_for_start(line)
def find_stack_trace_end(self, line):
''' Find the end of the stack trace. If found, return a metric point
tagged with the type of exception and transition to the
`looking_for_start` state again. If a line beginning with a
timestamp is found first, then reset the state machine.
'''
global stack_trace_end
match = stack_trace_end.match(line)
if match:
exception = match.group('exception')
output = (
'logs.%s.%s' % (self.data.get('logger'), self.data.get('severity').lower()),
parse_date(self.data.get('timestamp')),
1,
{'metric_type': 'counter',
'tags': ['exception:%s' % exception]}
)
self.data = {}
self.state = self.looking_for_start
return output
else:
self.looking_for_start(line)
def test():
sample_log = """2013-05-07 10:44:22,920 | ERROR | dd.collector | checks.batman(__init__.py:453) | Check 'batman' instance #0 failed
Traceback (most recent call last):
File "/Users/carlo/Projects/datadog/dd-agent/checks/__init__.py", line 444, in run
self.check(instance)
File "/Users/carlo/Projects/datadog/dd-agent/checks.d/batman.py", line 55, in check
self.gauge('dd.batman.top', y)
UnboundLocalError: local variable 'y' referenced before assignment
2013-05-07 10:44:22,920 | DEBUG | dd.collector | aggregator(aggregator.py:377) | received 0 payloads since last flush
2013-05-07 10:44:22,921 | DEBUG | dd.collector | aggregator(aggregator.py:377) | received 0 payloads since last flush
2013-05-07 10:44:22,921 | DEBUG | dd.collector | checks.mcache(mcache.py:89) | Connecting to localhost:11211 tags:['instance:localhost_11211']
2013-05-07 10:44:22,922 | ERROR | dd.collector | checks.mcache(__init__.py:453) | Check 'mcache' instance #0 failed
Traceback (most recent call last):
File "/Users/carlo/Projects/datadog/dd-agent/checks/__init__.py", line 444, in run
self.check(instance)
File "/Users/carlo/Projects/datadog/dd-agent/checks.d/mcache.py", line 165, in check
self._get_metrics(server, port, tags, memcache)
File "/Users/carlo/Projects/datadog/dd-agent/checks.d/mcache.py", line 138, in _get_metrics
raise Exception("Unable to retrieve stats from memcache instance: " + server + ":" + str(port) + ". Please check your configuration")
Exception: Unable to retrieve stats from memcache instance: localhost:11211. Please check your configuration
2013-05-07 10:44:22,922 | DEBUG | dd.collector | aggregator(aggregator.py:377) | received 0 payloads since last flush
2013-05-07 10:44:22,923 | DEBUG | dd.collector | aggregator(aggregator.py:377) | received 0 payloads since last flush
2013-05-07 10:44:22,924 | DEBUG | dd.collector | checks.http_check(http_check.py:25) | Connecting to http://localhost:5000
2013-05-07 10:44:22,924 | DEBUG | dd.collector | checks.http_check(http_check.py:25) | Connecting to https://app.datadoghq.com
2013-05-07 10:44:22,925 | ERROR | dd.collector | checks.mysql(__init__.py:453) | Check 'mysql' instance #0 failed
Traceback (most recent call last):
File "/Users/carlo/Projects/datadog/dd-agent/checks/__init__.py", line 444, in run
self.check(instance)
File "/Users/carlo/Projects/datadog/dd-agent/checks.d/mysql.py", line 52, in check
db = self._connect(host, mysql_sock, user, password)
File "/Users/carlo/Projects/datadog/dd-agent/checks.d/mysql.py", line 225, in _connect
passwd=password)
File "/Users/carlo/Projects/datadog/python/lib/python2.6/site-packages/MySQL_python-1.2.4b4-py2.6-macosx-10.6-universal.egg/MySQLdb/__init__.py", line 81, in Connect
return Connection(*args, **kwargs)
File "/Users/carlo/Projects/datadog/python/lib/python2.6/site-packages/MySQL_python-1.2.4b4-py2.6-macosx-10.6-universal.egg/MySQLdb/connections.py", line 187, in __init__
super(Connection, self).__init__(*args, **kwargs2)
OperationalError: (2002, "Can't connect to local MySQL server through socket '/tmp/mysql.sock' (2)")
2013-05-07 10:44:22,926 | DEBUG | dd.collector | aggregator(aggregator.py:377) | received 0 payloads since last flush
2013-05-07 10:44:22,926 | DEBUG | dd.collector | checks.nagios(log_parser.py:35) | Starting parse for file /tmp/service-perfdata
2013-05-07 10:44:22,927 | WARNING | dd.collector | checks.nagios(log_parser.py:41) | Can't tail /tmp/service-perfdata file
2013-05-07 10:44:22,927 | ERROR | dd.collector | checks.nagios(__init__.py:453) | Check 'nagios' instance #0 failed
Traceback (most recent call last):
File "/Users/carlo/Projects/datadog/dd-agent/checks/__init__.py", line 444, in run
self.check(instance)
File "/Users/carlo/Projects/datadog/dd-agent/checks.d/nagios.py", line 77, in check
parser.parse_file()
File "/Users/carlo/Projects/datadog/dd-agent/checks/log_parser.py", line 37, in parse_file
self.gen.next()
StopIteration
"""
import logging
logging.basicConfig()
parser = MultilineParser()
for line in sample_log.split("\n"):
print parser(logging, line)
if __name__ == '__main__':
test()
@fzsoltf
Copy link

fzsoltf commented Aug 31, 2016

Hi Carlo,

This is a nice example I'd like to tailor according to my log parsing needs. However, when I want to test the parser, I get the following error:

C:\...)\Datadog\Datadog Agent\checks.d>python multiline.py -v
Traceback (most recent call last):
  File "multiline.py", line 162, in <module>
    test()
  File "multiline.py", line 158, in test
    print parser(logging, line)
TypeError: 'MultilineParser' object is not callable

What is the right testing method - is the problem connected with the fact that this is a Datadog log parser? Do I need to import anything before the test?

Thank you

@martindemello
Copy link

This should fix the test() function:

    parser = MultilineParser()
    for line in sample_log.split("\n"):
        out = parser.parse_line(line)
        if out:
            print out

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment