很多同事认为filebeat采集日志不能做到多行处理，今天这里讨论下filebeat的multiline与include_lines。

 

先来个案例，以下日志，我们只要求采集error的字段，

2017/06/22 11:26:30 [error] 26067#0: *17918 connect() failed (111: Connection refused) while connecting to upstream, client: 192.168.32.17, server: localhost, request: "GET /wss/ HTTP/1.1", upstream: "http://192.168.12.106:8010/", host: "192.168.12.106"
2017/06/22 11:26:30 [info] 26067#0:
2017/06/22 12:05:10 [error] 26067#0: *17922 open() "/data/programs/nginx/html/ws" failed (2: No such file or directory), client: 192.168.32.17, server: localhost, request: "GET /ws HTTP/1.1", host: "192.168.12.106"


filebeat.yml文件配置如下：

filebeat.prospectors:
- input_type: log
  paths:
    - /tmp/test.log
  include_lines: ['error']
output.kafka:
  enabled: true
  hosts: ["192.168.12.105:9092"]
  topic: logstash-errors-log

查看下kafka队列

果然只有“error”关键字的日志被采集了

{"@timestamp":"2017-06-23T08:57:25.227Z","beat":{"name":"192.168.12.106"},"input_type":"log","message":"2017/06/22 12:05:10 [error] 26067#0: *17922 open() /data/programs/nginx/html/ws failed (2: No such file or directory), client: 192.168.32.17, server: localhost, request: GET /ws HTTP/1.1, host: 192.168.12.106","offset":30926,"source":"/tmp/test.log","type":"log"}
{"@timestamp":"2017-06-23T08:57:32.228Z","beat":{"name":"192.168.12.106"},"input_type":"log","message":"2017/06/22 12:05:10 [error] 26067#0: *17922 open() /data/programs/nginx/html/ws failed (2: No such file or directory), client: 192.168.32.17, server: localhost, request: GET /ws HTTP/1.1, host: 192.168.12.106","offset":31342,"source":"/tmp/test.log","type":"log"}


再来多行案例：

[2016-05-25 12:39:04,744][DEBUG][action.bulk              ] [Set] [***][3] failed to execute bulk item (index) index {[***][***][***], source[{***}}
MapperParsingException[Field name [events.created] cannot contain '.']
    at org.elasticsearch.index.mapper.object.ObjectMapper$TypeParser.parseProperties(ObjectMapper.java:273)
    at org.elasticsearch.index.mapper.object.ObjectMapper$TypeParser.parseObjectOrDocumentTypeProperties(ObjectMapper.java:218)
    at org.elasticsearch.index.mapper.object.ObjectMapper$TypeParser.parse(ObjectMapper.java:193)
    at org.elasticsearch.index.mapper.object.ObjectMapper$TypeParser.parseProperties(ObjectMapper.java:305)
    at org.elasticsearch.index.mapper.object.ObjectMapper$TypeParser.parseObjectOrDocumentTypeProperties(ObjectMapper.java:218)
    at org.elasticsearch.index.mapper.object.RootObjectMapper$TypeParser.parse(RootObjectMapper.java:139)
    at org.elasticsearch.index.mapper.DocumentMapperParser.parse(DocumentMapperParser.java:118)
    at org.elasticsearch.index.mapper.DocumentMapperParser.parse(DocumentMapperParser.java:99)
    at org.elasticsearch.index.mapper.MapperService.parse(MapperService.java:498)
    at org.elasticsearch.cluster.metadata.MetaDataMappingService$PutMappingExecutor.applyRequest(MetaDataMappingService.java:257)
    at org.elasticsearch.cluster.metadata.MetaDataMappingService$PutMappingExecutor.execute(MetaDataMappingService.java:230)
    at org.elasticsearch.cluster.service.InternalClusterService.runTasksForExecutor(InternalClusterService.java:468)
    at org.elasticsearch.cluster.service.InternalClusterService$UpdateTask.run(InternalClusterService.java:772)
    at org.elasticsearch.common.util.concurrent.PrioritizedEsThreadPoolExecutor$TieBreakingPrioritizedRunnable.runAndClean(PrioritizedEsThreadPoolExecutor.java:231)
    at org.elasticsearch.common.util.concurrent.PrioritizedEsThreadPoolExecutor$TieBreakingPrioritizedRunnable.run(PrioritizedEsThreadPoolExecutor.java:194)
    at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
    at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
    at java.lang.Thread.run(Thread.java:745)

filebeat.yml文件配置如下：

filebeat.prospectors:
- input_type: log
  paths:
    - /tmp/test.log
   multiline:
        pattern: '^\['
        negate:  true
        match:   after
  fields:
    beat.name: 192.168.12.106
  fields_under_root: true
output.kafka:
  enabled: true
  hosts: ["192.168.12.105:9092"]
  topic: logstash-errors-log
kafka队列如下：

{"@timestamp":"2017-06-23T09:09:02.887Z","beat":{"name":"192.168.12.106"},"input_type":"log",
"message":"[2016-05-25 12:39:04,744][DEBUG][action.bulk              ] [Set] [***][3] failed to execute bulk item (index) index {[***][***][***], source[{***}}\n
MapperParsingException[Field name [events.created] cannot contain '.']\n    at org.elasticsearch.index.mapper.object.ObjectMapper$TypeParser.parseProperties(ObjectMapper.java:273)\n    
at org.elasticsearch.index.mapper.object.ObjectMapper$TypeParser.parseObjectOrDocumentTypeProperties(ObjectMapper.java:218)\n    
at org.elasticsearch.index.mapper.object.ObjectMapper$TypeParser.parse(ObjectMapper.java:193)\n    
at org.elasticsearch.index.mapper.object.ObjectMapper$TypeParser.parseProperties(ObjectMapper.java:305)\n    
at org.elasticsearch.index.mapper.object.ObjectMapper$TypeParser.parseObjectOrDocumentTypeProperties(ObjectMapper.java:218)\n    
at org.elasticsearch.index.mapper.object.RootObjectMapper$TypeParser.parse(RootObjectMapper.java:139)\n    
at org.elasticsearch.index.mapper.DocumentMapperParser.parse(DocumentMapperParser.java:118)\n    
at org.elasticsearch.index.mapper.DocumentMapperParser.parse(DocumentMapperParser.java:99)\n    
at org.elasticsearch.index.mapper.MapperService.parse(MapperService.java:498)\n    
at org.elasticsearch.cluster.metadata.MetaDataMappingService$PutMappingExecutor.applyRequest(MetaDataMappingService.java:257)\n    
at org.elasticsearch.cluster.metadata.MetaDataMappingService$PutMappingExecutor.execute(MetaDataMappingService.java:230)\n   
at org.elasticsearch.cluster.service.InternalClusterService.runTasksForExecutor(InternalClusterService.java:468)\n    
at org.elasticsearch.cluster.service.InternalClusterService$UpdateTask.run(InternalClusterService.java:772)\n    
at org.elasticsearch.common.util.concurrent.PrioritizedEsThreadPoolExecutor$TieBreakingPrioritizedRunnable.runAndClean(PrioritizedEsThreadPoolExecutor.java:231)\n    
at org.elasticsearch.common.util.concurrent.PrioritizedEsThreadPoolExecutor$TieBreakingPrioritizedRunnable.run(PrioritizedEsThreadPoolExecutor.java:194)\n   
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)\n    
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)\n    
at java.lang.Thread.run(Thread.java:745)\n\n\n\n","offset":35737,"source":"/tmp/test.log","type":"log"}

可以看出multiline将多行日志汇总。


multiline与include_lines，结合使用。

filebeat.yml文件配置如下：

filebeat.prospectors:
- input_type: log
  paths:
    - /tmp/test.log
  include_lines: ['error']
  multiline:
        pattern: '^\['
        negate:  true
        match:   after
output.kafka:
  enabled: true
  hosts: ["192.168.12.105:9092"]
  topic: logstash-errors-log

即日志中如果有"error"关键字的日志，进行多行合并，发送至kafka.

经验证，在日志不断输入的情况，会把不含"error"的行也进行合并，日志有间隔的情况输入，过滤效果比较好，具体结合业务情况实用吧。

总之一句话，filebeat可以多行合并和进行关键字日志采集。