KafkaInput & esoutput

heka从kalka中读取数据。

配置：

[hekad]
maxprocs = 2

[KafkaInputExample]
type = "KafkaInput"
topic = "test"
addrs = ["localhost:9092"]

[RstEncoder]

[LogOutput]
message_matcher = "TRUE"
encoder = "RstEncoder"

上述配置只有从kalfka中读取数据并显示到console，写到kalfka中数据，

结果

:Timestamp: 2016-07-21 09:39:46.342093657 +0000 UTC
:Type: heka.kafka
:Hostname: master
:Pid: 0
:Uuid: 501b0a0e-63a9-4eee-b9ca-ab572c17d273
:Logger: KafkaInputExample
:Payload: {"msg":"Start Request","event":"artemis.web.ensure-running1","userid":"12","extra":{"workspace-id":"cN907xLngi"},"time":"2015-05-06T    20:40:05.509926234Z","severity":1}
:EnvVersion: 
:Severity: 7
:Fields:
    | name:"Key" type:bytes value:
    | name:"Topic" type:string value:"test"
    | name:"Partition" type:integer value:0
    | name:"Offset" type:integer value:8

 

读取出来的数据放到了payload中，而fileds中存放了读取kalkfa中的一些信息。那么可以使用jsondecoder进行解析。

[hekad]
maxprocs = 2

[KafkaInputExample]
type = "KafkaInput"
topic = "test"
addrs = ["localhost:9092"]
decoder="JsonDecoder"

[JsonDecoder]
type = "SandboxDecoder"
filename = "lua_decoders/json.lua"

        [JsonDecoder.config]
        type = "artemis"
        payload_keep = true
        map_fields = true
        Severity = "severity"

[RstEncoder]

[LogOutput]
message_matcher = "TRUE"
encoder = "RstEncoder"

 

结果如下：

:Timestamp: 2016-07-21 09:42:34 +0000 UTC
:Type: artemis
:Hostname: master
:Pid: 0
:Uuid: 3965285c-70ac-4069-a1a3-a9bcf518d3e8
:Logger: KafkaInputExample
:Payload: {"msg":"Start Request","event":"artemis.web.ensure-running2","userid":"11","extra":{"workspace-id":"cN907xLngi"},"time":"2015-05-06T    20:40:05.509926234Z","severity":1}
:EnvVersion: 
:Severity: 1
:Fields:
    | name:"time" type:string value:"2015-05-06T    20:40:05.509926234Z"
    | name:"msg" type:string value:"Start Request"
    | name:"userid" type:string value:"11"
    | name:"event" type:string value:"artemis.web.ensure-running2"
    | name:"extra.workspace-id" type:string value:"cN907xLngi"

经过decoder解析之后，fileds发生了改变，但是我们可以看到Logger显示的还是KafkaInputExample，说明数据不是decoder产生，而是Input产生，只不过使用了decoder进行了解析，重写改写了fields而已。

接下来，把数据录入都es中吧。
[hekad]
maxprocs = 2

[KafkaInputExample]
type = "KafkaInput"
topic = "test"
addrs = ["localhost:9092"]
decoder="JsonDecoder"

[JsonDecoder]
type = "SandboxDecoder"
filename = "lua_decoders/json.lua"

        [JsonDecoder.config]
        type = "artemis"
        payload_keep = true
        map_fields = true
        Severity = "severity"

[ESJsonEncoder]
index = "%{Type}-%{%Y.%m.%d}"
es_index_from_timestamp = true
type_name = "%{Type}"
    [ESJsonEncoder.field_mappings]
    Timestamp = "@timestamp"
    Severity = "level"

[ElasticSearchOutput]
message_matcher = "TRUE"
encoder = "ESJsonEncoder"
flush_interval = 1

导入到es中，也需要json，所以使用ESJsonEncoder，同时指定索引名字和类型。执行结果如下，

可以看到，除了heka中元数据field之外，还有JsonDecoder生成field啊，其实是截取JsonDecoder中的fields属性中拿出。注意，Payload不解析。

:Fields:
    | name:"time" type:string value:"2015-05-06T    20:40:05.509926234Z"
    | name:"msg" type:string value:"Start Request"
    | name:"userid" type:string value:"11"
    | name:"event" type:string value:"artemis.web.ensure-running2"
    | name:"extra.workspace-id" type:string value:"cN907xLngi"

这些field当然随着数据不同而不同，那么称之为dynamic fileds。

入es的时候，可以指定提取哪些dynamic fields，

fields=["Timestamp","Uuid","Type","Logger","Pid","Hostname","DynamicFields"]
dynamic_fields=["msg","userid"]

只要使用dynamic_fileds，就必须要在fields中指定DynamicFields。

如果没有dynamic_fileds，那么fields只能列举几个固定的属性，参照官方文档即可。

 

完成的列子：

[hekad]
maxprocs = 2

[KafkaInputExample]
type = "KafkaInput"
topic = "test"
addrs = ["localhost:9092"]
decoder="JsonDecoder"

[JsonDecoder]
type = "SandboxDecoder"
[hekad]
maxprocs = 2

[KafkaInputExample]
type = "KafkaInput"
topic = "test"
addrs = ["localhost:9092"]
decoder="JsonDecoder"

[JsonDecoder]
type = "SandboxDecoder"
filename = "lua_decoders/json.lua"

        [JsonDecoder.config]
        type = "artemis"
        payload_keep = true
        map_fields = true
        Severity = "severity"

[ESJsonEncoder]
index = "%{Type}-%{%Y.%m.%d}"
es_index_from_timestamp = true
type_name = "%{Type}"
fields=["Timestamp","Uuid","Type","Logger","Pid","Hostname","DynamicFields"]
dynamic_fields=["msg","userid"]

raw_bytes_fields=["Payload"]
    [ESJsonEncoder.field_mappings]
    Timestamp = "@timestamp"
    Severity = "level"

[ElasticSearchOutput]
message_matcher = "TRUE"
encoder = "ESJsonEncoder"
flush_interval = 1

结果如下，

转载于:https://my.oschina.net/weikan/blog/715775