Text logs with fastmatch

The fastmatch parser uses simple string patterns that specify the position of fields within a log line. As the name suggests it is very fast and should be the preferred method to parse text logs. It can handle most cases of structured text logs where the order of fields is known. In fact, it is so fast you can specify multiple patterns that will be tested in order, so you can 'solve' cases where there are a few variations in the structure of the log line.

We will be using the following example log line that is using Apache Common Log format:

127.0.0.1 - frank [10/Oct/2000:13:55:36 -0700] "GET /apache_pb.gif HTTP/1.0" 200 2326

To parse this log type using fastmatch we would define the following log schema:

version: 0
parser:
fastmatch:
# Define an array of patterns to match against.
# In this example we only use one pattern because the log format is the same for all lines.
# If we wanted to include the Apache Extended Log format, we could provide an additional pattern.
match:
- '%{remote_ip} %{identity} %{user} [%{timestamp}] "%{method} %{request_uri} %{protocol}" %{status} %{bytes_sent}'
emptyValues: [ '-' ] # specify that `-` string values are considered null
fields:
- name: remote_ip
type: string
indicators:
- ip
- name: identity
type: string
- name: user
type: string
- name: timestamp
type: timestamp
isEventTime: true
timeFormat: '%d/%b/%Y:%H:%M:%S %z'
- name: method
type: string
- name: request_uri
type: string
- name: protocol
type: string
- name: status
type: int
- name: bytes_sent
type: bigint

Understanding fastmatch patterns

The patterns use %{field_name} placeholders to set where in the log line a field is expected. For example to match the text

2020-10-10T14:32:05 [FOO_SERVICE@127.0.0.1] [DEBUG] "" Something when wrong

We can use this pattern (surrounded by single quotes for clarity):

'%{timestamp} [%{service}@%{ip}] [%{log_level}] %{message}'

Delimiters

The text between two consecutive fields defines the 'delimiter' between them.

Delimiters cannot be empty.

In the example above we cannot omit the "@" between service and ip in the pattern

The field preceding a delimiter cannot contain the delimiter text. In the example above:

  • timestamp cannot contain space " "

  • service cannot contain "@"

  • ip cannot contain "] ["

  • log_level cannot contain "] "

Anonymous fields

Field placeholders without names (%{}) are ignored.

Tail capture

If the last field in a pattern does not have any delimiter text after it, it will capture everything until the end of the text. In the example above message will capture "Something when wrong"

Handling quotes

In some cases fields can be quoted within the text:

2020-10-10T14:32:05 "Some quoted text with \"escaped quotes\" inside"

To properly unescape such fields just surround the field placeholder with quotes:

%{timestamp} "%{message}"

This works for both single and double quotes.