절차

1. Amazon Elasticsearch Service 시작

2. Ingest Pipeline 설정

$ curl -H "Content-Type: application/json" -XPUT 'https://XXXXX-XXXXXXXXXXXXXXXXXXXXXXXXXX.ap-northeast-1.es.amazonaws.com/_ingest/pipeline/cflog' -d '{
  "processors": [{
    "grok": {
      "field": "message",
      "patterns":[ "%{TIMESTAMP_ISO8601:timestamp} %{NOTSPACE:x_edge_location} (?:%{INT:sc_bytes:int}|-) %{IPORHOST:c_ip} %{WORD:cs_method} %{HOSTNAME:cs_host} %{NOTSPACE:cs_uri_stem} %{INT:sc_status:int} %{GREEDYDATA:referrer} %{GREEDYDATA:User_Agent} %{GREEDYDATA:cs_uri_query} %{GREEDYDATA:cookies} %{WORD:x_edge_result_type} %{NOTSPACE:x_edge_request_id} %{HOSTNAME:x_host_header} %{URIPROTO:cs_protocol} (?:%{INT:cs_bytes:int}|-) %{NUMBER:time_taken:float} %{NOTSPACE:x_forwarded_for} %{NOTSPACE:ssl_protocol} %{NOTSPACE:ssl_cipher} %{WORD:x_edge_response_result_type}" ],
      "ignore_missing": true
    }
  },{
    "remove":{
      "field": "message"
    }
  }, {
    "user_agent": {
      "field": "User_Agent",
      "target_field": "user_agent",
      "ignore_failure": true
    }
  }, {
    "remove": {
      "field": "User_Agent"
    }
  }]
}'

3. AWS Lambda 함수 생성

IAM Role 만들기

Lambda 함수 만들기

$ python3 -m venv dev
$ . dev/bin/activate
(dev) $ mkdir cf_log_to_es_s3
(dev) $ cd cf_log_to_es_s3/
(dev) $ pip install requests requests_aws4auth -t ./
（省略）
(dev) $ rm -rf *.dist-info
(dev) $ vi lambda_function.py
（ここで「lambda_function.py」のコードを入力）
(dev) $ zip -r ../cflog.zip *

import boto3
import os
import gzip
import requests
from datetime import datetime
from requests_aws4auth import AWS4Auth

def lambda_handler(event, context):
    print('Started')
    es_index = os.environ['ES_INDEX_PREFIX'] + "-" + datetime.strftime(datetime.now(), "%Y%m%d")
    region = os.environ["AWS_REGION"]
    service = 'es'
    credentials = boto3.Session().get_credentials()
    awsauth = AWS4Auth(credentials.access_key, credentials.secret_key, region, service, session_token=credentials.token)

    bucket = event["Records"][0]["s3"]["bucket"]["name"]
    key = event["Records"][0]["s3"]["object"]["key"]

    s3 = boto3.resource('s3')
    s3.Bucket(bucket).download_file(key, '/tmp/log.gz')

    with gzip.open('/tmp/log.gz', mode='rt') as f:
        data = ""

        for line in f:
            if not line.strip().startswith('#'):
                data += '{"index":{"_index":"%s","_type":"log"}}\n' % es_index
                data += '{"message":"%s"}\n' % line.strip().replace('"', '\\"').replace('\t', ' ').replace(' ', 'T', 1).replace(' ', 'Z ', 1)

                if len(data) > 3000000:
                    _bulk(data, awsauth)
                    data = ""

        if data != "":
            _bulk(data, awsauth)

    return 'Completed'

def _bulk(data, awsauth):    
    es_host = os.environ['ES_HOST']
    pipeline = os.environ['PIPELINE_NAME']
    url = 'https://%s/_bulk?pipeline=%s' % (es_host, pipeline)

    headers = {'Content-Type': 'application/json'}
    response = request(url, awsauth, headers=headers, data=data)

    if response.status_code != requests.codes.ok:
        print(response.text)

def request(url, awsauth, headers=None, data=None):
    return requests.post(url, auth=awsauth, headers=headers, data=data)