1、配置服务器格式

博主这里用的是tomcat的combined默认格式，格式如下

127.0.0.1 - - [28/Mar/2017:09:23:10 +0800] "GET /manager/html HTTP/1.1" 401 2536 "-" "Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/56.0.2924.87 Safari/537.36"

2、进入hive shell来创建表

使用正则法则来匹配日志格式

CREATE TABLE td_log_analyze(  
  host STRING,  
  identity STRING,  
  usr STRING,  
  time STRING,  
  request STRING,  
  status STRING,  
  size STRING,  
  referer STRING,  
  agent STRING)  
partitioned by (dt string)  
ROW FORMAT SERDE 'org.apache.hadoop.hive.contrib.serde2.RegexSerDe'  
WITH SERDEPROPERTIES (  
  "input.regex" = "([^ ]*) ([^ ]*) ([^ ]*) (-|\\[[^\\]]*\\]) ([^ \"]*|\"[^\"]*\") (-|[0-9]*) (-|[0-9]*)(?: ([^ \"]*|\"[^\"]*\") ([^ \"]*|\"[^\"]*\"))?",  
  "output.format.string" = "%1$s %2$s %3$s %4$s %5$s %6$s %7$s %8$s %9$s"  
)  
STORED AS TEXTFILE;  

3、配置flume-agent

#定义agent名， source、channel、sink的名称  
logAnalyzeAG.sources = s1  
logAnalyzeAG.channels = c1  
logAnalyzeAG.sinks = k1  
#具体定义source  
logAnalyzeAG.sources.s1.type = spooldir  
logAnalyzeAG.sources.s1.spoolDir = /home/data/tomcat/tomcat-8081/logs/access
#设置缓存提交行数  
logAnalyzeAG.sources.s1.deserializer.maxLineLength =1048576  
logAnalyzeAG.sources.s5.fileSuffix =.FINISH
logAnalyzeAG.sources.s5.ignorePattern=^localhost_access_log\.txt$
logAnalyzeAG.sources.s1.consumeOrder = oldest  
logAnalyzeAG.sources.s1.deserializer = org.apache.flume.sink.solr.morphline.BlobDeserializer$Builder  
logAnalyzeAG.sources.s1.batchsize = 5  
#定义拦截器，为消息添加时间戳  
#logAnalyzeAG.sources.r1.interceptors = i1  
#logAnalyzeAG.sources.r1.interceptors.i1.type = org.apache.flume.interceptor.TimestampInterceptor$Builder  
#具体定义channel  
logAnalyzeAG.channels.c1.type = memory  
logAnalyzeAG.channels.c1.capacity = 10000  
logAnalyzeAG.channels.c1.transactionCapacity = 100  
#具体定义sink  
logAnalyzeAG.sinks.k1.type = hdfs  
#%y-%m-%d/%H%M/%S  
#这里对应就是hive 表的目录 此处如果是外部表，则直接对应你的localtion地址，如果普通则对应到你的hive表目录即可  
logAnalyzeAG.sinks.k1.hdfs.path = hdfs://172.16.38.159:8020/apps/hive/warehouse/log_data.db/td_log_analyze/%Y-%m-%d  
logAnalyzeAG.sinks.k1.hdfs.filePrefix = log-%Y-%m-%d  
logAnalyzeAG.sinks.k1.hdfs.fileSuffix = .log  
logAnalyzeAG.sinks.k1.hdfs.fileType = DataStream  
#不按照条数生成文件  
logAnalyzeAG.sinks.k1.hdfs.rollCount = 0  
#HDFS上的文件达到128M时生成一个文件  
logAnalyzeAG.sinks.k1.hdfs.rollSize = 2914560  
#HDFS上的文件达到60秒生成一个文件  
#logAnalyzeAG.sinks.k1.hdfs.rollInterval = 60  
logAnalyzeAG.sinks.k1.hdfs.useLocalTimeStamp = true  
#组装source、channel、sink  
logAnalyzeAG.sources.s1.channels = c1  
logAnalyzeAG.sinks.k1.channel = c1  

PS：采用spoolDir来采集是文件级别的，即扫描新增文件。实时增量可以采用tailDir。上面的配置文件会过滤掉当天打印的文件localhost_access_log.txt。

4、因为对日志进行的分区存放，所以要想将hive表对应指定分区需要创建hive 分区

ALTER TABLE td_log_analyze ADD IF NOT EXISTS PARTITION (dt='2017-03-29') LOCATION '/apps/hive/warehouse/log_data.db/td_log_analyze/2017-03-29/';  

 注意，需要先创建分区，再采集数据到hdfs，hive才能识别到数据

5、编写shell脚本，每天创建第二天的hive分区

#!/bin/bash
#获取明天的日期
day=$(date --date='1 days' +%Y-%m-%d)
cd /home/
touch log.sql
#为明天预添加hive分区
echo "ALTER TABLE td_log_analyze ADD IF NOT EXISTS PARTITION (dt='${day}') LOCATION '/apps/hive/warehouse/log_data.db/td_log_analyze/${day}/';" > /home/log.sql
beeline -u 'jdbc:hive2://node0.hdp:2181,node2.hdp:2181,node1.hdp:2181/log_data;serviceDiscoveryMode=zooKeeper;zooKeeperNamespace=hiveserver2' -f '/home/log.sql'

6、设置定时任务（centos）

crontab -e

 添加任务信息，每天凌晨两点执行

0 2 * * * /home/shell-test/hive-partitions.sh

重新载入配置

service crond reload

更多定时任务功能参考：http://julyme.com/20170329/75.html

PS：附上tomcat日志格式的配置方法

转载请注明出处: http://www.julyme.com/20170327/74.html