python 使用elasticsearch模块获取ELK系统上的索引数据
这次想要开发用户行为分析系统，所以要把网站日志导入到数据库中，而刚好我们部署了EKL日志系统，所以此次写了个脚本，直接写python脚本，利用elasticsearch模块，直接从ELK上获取所需的数据。
以下所示范内容与真实需求有出入，这只是为了方便理解脚本代码的意思
EKL中，索引文件名如是：logstash-nginx-2017-01-20这样的格式，
我们只要在elasticsearch中获取时间，用户IP，用户访问的路由等（其实真实场景还有更多）。
代码如下：
#!/usr/bin/env python
#coding=utf-8
"""
__author__ = '戴儒锋'
使用elasticsearch模块获取昨天某站点访问日志的所有数据
elasticsearch模块中指定使用scroll用来避免深度分页查找数据时的性能消耗
scan（扫描）搜索类型是和scroll（滚屏）API一起使用来从Elasticsearch里高效地取回巨大数量的结果而不需要付出深分页的代价。
size被应用到每一个分片上，所以我们在每个批次里最多或获得size * number_of_primary_shards（size*主分片数）
scroll= "1m" 指定快照时间为1分钟
"""
import datetime
from elasticsearch import Elasticsearch
# 格式为：2016.7.19 的昨日日期
yesterday = (datetime.datetime.now() + datetime.timedelta(days = -1)).strftime("%Y.%m.%d")
# 格式为：2016-7-19 的昨日日期
filter_yesterday = (datetime.datetime.now() + datetime.timedelta(days = -1)).strftime("%Y-%m-%d")
# 格式为：2016.7.18 的前天日期
before_yesterday = (datetime.datetime.now() + datetime.timedelta(days = -2)).strftime("%Y.%m.%d")
# 请求elasticsearch节点的url
url = "http://192.168.1.41:9200/"
# 使用的索引，因日期时区问题，所以要指定昨天和前天的索引名
index_name = "logstash-apache-www.linuxyw.com-{date},logstash-apache-www.linuxyw.com-{b_date}".format(date=yesterday,b_date=before_yesterday)
# 实例化Elasticsearch类，并设置超时间为120秒，默认是10秒的，如果数据量很大，时间设置更长一些
es = Elasticsearch(url,timeout=120)
# DSL查询语法，在下面es.search使用
data = {
"size": 10000000, #指定每个分片最大返回的数据量，可根据日志量进行设置
"query" : {
"bool":{
# 指定要匹配的字符，这里是查找所有数据
"must" : {"match_all":{}},
# 过滤，指定时间范围，这里设置成昨天0点到24点，代码上||-8h，因为ELK用的是UTC时间，跟北京时间误差8小时，所以要减8小时，这就是日志里的北京时间了
"filter" : {
"range" : { "@timestamp" : {
"gt" : "{date}T00:00:00||-8h".format(date=filter_yesterday),
"lt" : "{date}T23:59:59||-8h".format(date=filter_yesterday),
}
}
}
}
}
}
# 设置要过滤返回的字段值，要什么字段，就在这里添加，这样可以节约返回的数据量（带宽，内存等）
return_fields = [
'_scroll_id',
'hits.hits._source.timestamp',
'hits.hits._source.@timestamp',
'hits.hits._source.clientip',
'hits.hits._source.request',
]
def main():
# 指定search_type="scan"模式，并返回_scroll_id给es.scroll获取数据使用
res = es.search(
index=index_name,
body=data,
search_type="scan",
scroll="1m"
)
scrollId=res["_scroll_id"] # 获取scrollID
response= es.scroll(scroll_id=scrollId, scroll= "1m",filter_path=return_fields,)
print len(response['hits']['hits']) # 打印获取到的日志数量
# for hit in response['hits']['hits']:
# print hit['_source']
if __name__ == "__main__":
main()for hit in response['hits']['hits']:
print hit
输出的结果如：
{u'timestamp': u'19/Jul/2016:08:00:34 +0800', u'@timestamp': u'2016-07-19T00:00:35.380Z', u'request': u'/apk/dwuliu.apk', u'clientip': u'60.221.255.44'}
{u'timestamp': u'19/Jul/2016:08:00:34 +0800', u'@timestamp': u'2016-07-19T00:00:36.507Z', u'request': u'/apk/dwuliu.apk', u'clientip': u'112.5.16.13'}
{u'timestamp': u'19/Jul/2016:08:00:36 +0800', u'@timestamp': u'2016-07-19T00:00:36.508Z', u'request': u'/apk/dwuliu.apk', u'clientip': u'115.238.250.237'}
{u'timestamp': u'19/Jul/2016:00:02:34 +0800', u'@timestamp': u'2016-07-18T16:02:34.934Z', u'request': u'/apk/dwuliu.apk', u'clientip': u'115.231.218.18'}
{u'timestamp': u'19/Jul/2016:00:02:34 +0800', u'@timestamp': u'2016-07-18T16:02:34.934Z', u'request': u'/apk/dwuliu.apk', u'clientip': u'123.134.186.178'}
{u'timestamp': u'19/Jul/2016:00:02:34 +0800', u'@timestamp': u'2016-07-18T16:02:34.934Z', u'request': u'/apk/dwuliu.apk', u'clientip': u'218.61.8.10'}
{u'timestamp': u'19/Jul/2016:00:02:34 +0800', u'@timestamp': u'2016-07-18T16:02:34.934Z', u'request': u'/apk/dwuliu.apk', u'clientip': u'118.212.147.71'}
{u'timestamp': u'19/Jul/2016:00:02:34 +0800', u'@timestamp': u'2016-07-18T16:02:34.935Z', u'request': u'/apk/dwuliu.apk', u'clientip': u'210.32.125.68'}
{u'timestamp': u'19/Jul/2016:00:02:34 +0800', u'@timestamp': u'2016-07-18T16:02:34.935Z', u'request': u'/apk/dwuliu.apk', u'clientip': u'222.163.80.18'}
{u'timestamp': u'19/Jul/2016:00:02:34 +0800', u'@timestamp': u'2016-07-18T16:02:34.935Z', u'request': u'/apk/dwuliu.apk', u'clientip': u'222.186.33.170'}
{u'timestamp': u'19/Jul/2016:00:02:34 +0800', u'@timestamp': u'2016-07-18T16:02:34.935Z', u'request': u'/apk/dwuliu.apk', u'clientip': u'218.199.110.39'}
{u'timestamp': u'19/Jul/2016:00:02:34 +0800', u'@timestamp': u'2016-07-18T16:02:34.935Z', u'request': u'/apk/dwuliu.apk', u'clientip': u'114.80.201.18'}
{u'timestamp': u'19/Jul/2016:00:02:34 +0800', u'@timestamp': u'2016-07-18T16:02:34.935Z', u'request': u'/apk/dwuliu.apk', u'clientip': u'115.238.250.237'}
{u'timestamp': u'19/Jul/2016:08:04:35 +0800', u'@timestamp': u'2016-07-19T00:04:36.209Z', u'request': u'/apk/dwuliu.apk', u'clientip': u'58.215.186.208'}
{u'timestamp': u'19/Jul/2016:08:04:35 +0800', u'@timestamp': u'2016-07-19T00:04:36.209Z', u'request': u'/apk/dwuliu.apk', u'clientip': u'122.227.164.103'}
{u'timestamp': u'19/Jul/2016:08:04:35 +0800', u'@timestamp': u'2016-07-19T00:04:36.209Z', u'request': u'/apk/dwuliu.apk', u'clientip': u'222.186.33.170'}
{u'timestamp': u'19/Jul/2016:08:04:35 +0800', u'@timestamp': u'2016-07-19T00:04:36.208Z', u'request': u'/apk/dwuliu.apk', u'clientip': u'223.100.7.69'}
{u'timestamp': u'19/Jul/2016:08:04:35 +0800', u'@timestamp': u'2016-07-19T00:04:36.209Z', u'request': u'/apk/dwuliu.apk', u'clientip': u'218.6.9.4'}
{u'timestamp': u'19/Jul/2016:08:04:35 +0800', u'@timestamp': u'2016-07-19T00:04:36.115Z', u'request': u'/apk/dwuliu.apk', u'clientip': u'183.136.232.136'}
{u'timestamp': u'19/Jul/2016:08:04:35 +0800', u'@timestamp': u'2016-07-19T00:04:36.114Z', u'request': u'/apk/dwuliu.apk', u'clientip': u'221.203.236.213'}
{u'timestamp': u'19/Jul/2016:08:04:35 +0800', u'@timestamp': u'2016-07-19T00:04:36.115Z', u'request': u'/apk/dwuliu.apk', u'clientip': u'124.119.87.204'}
{u'timestamp': u'19/Jul/2016:08:04:35 +0800', u'@timestamp': u'2016-07-19T00:04:36.115Z', u'request': u'/apk/dwuliu.apk', u'clientip': u'43.254.144.227'}
{u'timestamp': u'19/Jul/2016:08:04:35 +0800', u'@timestamp': u'2016-07-19T00:04:36.115Z', u'request': u'/apk/dwuliu.apk', u'clientip': u'59.173.18.243'}最后推荐些Elasticsearch教程资料：
Elasticsearch 权威指南（中文版）：http://es.xiaoleilu.com/index.html
Python Elasticsearch Client：http://elasticsearch-py.readthedocs.io/en/master/