This tool allows to parse log files and send statistics to statsd endpoint

• {string} config - configuration file
• {string} statsd - statsd endpoint
• {string} prefix - statsd global key prefix, e.g. balancer.
• {int} buffer - buffer interval for metrics, 0 default: one metric - one request

Snith has a particular configuration structure in yml:

sources:
- source: name of source
  file: ./test/log/test2.log
  noFollow: false
  mustExists: false
  reOpen: true
  prefix: "balancer.%HOST%"
  delimiter: "\t"
  keys:
    - key: All.$3.$6
      count: true
      timing: $4
      delimiter: " : "
    - key: All.$3.
      timing: $10
      delimiter: " : "

where is:

• {string} source - source name

• {string} file - where the log file is

• {boolean} noFollow - means no follow new lines that are written in the log file, if true

• {boolean} mustExists - log file have to be existed, if true

• {boolean} reOpen - re-open file, if true (e.g. log rotation)

• {string} prefix - statsd key prefix, %HOST% is used as substitution a hostname

• {string} delimiter - column delimiter in log files, it is reasonable to use \t delimiter in log files

• {[]key} keys - list of keys, which would be send to statsd

  • {string} key - statsd metric key, $N - means column position
  • {boolean} count - boolean, means
  • {string} timing - column with time (time should be in seconds, like 0.001 - means 1ms)
  • {string} delimiter - delimiter into the column, e.g. nginx can to write in one column a few values from upstream the request was in

Snitch allows handle a few sources per instance

Docker build uses multistage build

docker build -t xjewer/snitch:{tag} .

docker run --rm -v "/var/log/nginx/balancer/:/var/log/:ro" xjewer/snitch:0.1 -file /var/log/for_script.access.log -statsd graphite.local:8125 -buffer 10

Benchmark line parsers is:

goos: darwin
goarch: amd64
pkg: github.com/xjewer/snitch
Benchmark_parser-4   	 2000000	       946 ns/op	     416 B/op	       8 allocs/op
PASS
ok  	github.com/xjewer/snitch	2.890s