[英]Unable to convert Kafka topic data into structured JSON with Confluent Elasticsearch sink connector
我正在使用Kafka构建数据管道。 数据流如下:在mongodb中捕获数据更改,并将其发送给elasticsearch。
MongoDB
卡夫卡
弹性搜索
由于我仍在测试中,与Kafka相关的系统在单个服务器上运行。
启动Zookeepr
$ bin/zookeeper-server-start etc/kafka/zookeeper.properties
启动引导服务器
$ bin/kafka-server-start etc/kafka/server.properties
启动注册表架构
$ bin/schema-registry-start etc/schema-registry/schema-registry.properties
启动mongodb源connetor
$ bin/connect-standalone \\ etc/schema-registry/connect-avro-standalone.properties \\ etc/kafka/connect-mongo-source.properties $ cat etc/kafka/connect-mongo-source.properties >>> name=mongodb-source-connector connector.class=io.debezium.connector.mongodb.MongoDbConnector mongodb.hosts='' initial.sync.max.threads=1 tasks.max=1 mongodb.name=higee $ cat etc/schema-registry/connect-avro-standalone.properties >>> bootstrap.servers=localhost:9092 key.converter=io.confluent.connect.avro.AvroConverter key.converter.schema.registry.url=http://localhost:8081 value.converter=io.confluent.connect.avro.AvroConverter value.converter.schema.registry.url=http://localhost:8081 internal.key.converter=org.apache.kafka.connect.json.JsonConverter internal.value.converter=org.apache.kafka.connect.json.JsonConverter internal.key.converter.schemas.enable=false internal.value.converter.schemas.enable=false rest.port=8083
启动弹性搜索接收器连接器
$ bin/connect-standalone \\ etc/schema-registry/connect-avro-standalone2.properties \\ etc/kafka-connect-elasticsearch/elasticsearch.properties $ cat etc/kafka-connect-elasticsearch/elasticsearch.properties >>> name=elasticsearch-sink connector.class=io.confluent.connect.elasticsearch.ElasticsearchSinkConnector tasks.max=1 topics=higee.higee.higee key.ignore=true connection.url='' type.name=kafka-connect $ cat etc/schema-registry/connect-avro-standalone2.properties >>> bootstrap.servers=localhost:9092 key.converter=io.confluent.connect.avro.AvroConverter key.converter.schema.registry.url=http://localhost:8081 value.converter=io.confluent.connect.avro.AvroConverter value.converter.schema.registry.url=http://localhost:8081 internal.key.converter=org.apache.kafka.connect.json.JsonConverter internal.value.converter=org.apache.kafka.connect.json.\\ JsonConverter internal.key.converter.schemas.enable=false internal.value.converter.schemas.enable=false rest.port=8084
上面的系统一切都很好。 Kafka连接器捕获数据更改(CDC),并通过接收器连接器成功将其发送到elasticsearch。 问题是我无法将字符串类型的消息数据转换为结构化数据类型。 例如,在对mongodb进行一些更改之后,让我们使用主题数据。
$ bin/kafka-avro-console-consumer \
--bootstrap-server localhost:9092 \
--topic higee.higee.higee --from-beginning | jq
然后,我得到以下结果。
"after": null,
"patch": {
"string": "{\"_id\" : {\"$oid\" : \"5ad97f982a0f383bb638ecac\"},\"name\" : \"higee\",\"salary\" : 100,\"origin\" : \"South Korea\"}"
},
"source": {
"version": {
"string": "0.7.5"
},
"name": "higee",
"rs": "172.31.50.13",
"ns": "higee",
"sec": 1524214412,
"ord": 1,
"h": {
"long": -2379508538412995600
},
"initsync": {
"boolean": false
}
},
"op": {
"string": "u"
},
"ts_ms": {
"long": 1524214412159
}
}
然后,如果我去elasticsearch,我得到以下结果。
{
"_index": "higee.higee.higee",
"_type": "kafka-connect",
"_id": "higee.higee.higee+0+3",
"_score": 1,
"_source": {
"after": null,
"patch": """{"_id" : {"$oid" : "5ad97f982a0f383bb638ecac"},
"name" : "higee",
"salary" : 100,
"origin" : "South Korea"}""",
"source": {
"version": "0.7.5",
"name": "higee",
"rs": "172.31.50.13",
"ns": "higee",
"sec": 1524214412,
"ord": 1,
"h": -2379508538412995600,
"initsync": false
},
"op": "u",
"ts_ms": 1524214412159
}
}
我要实现的目标如下
{
"_index": "higee.higee.higee",
"_type": "kafka-connect",
"_id": "higee.higee.higee+0+3",
"_score": 1,
"_source": {
"oid" : "5ad97f982a0f383bb638ecac",
"name" : "higee",
"salary" : 100,
"origin" : "South Korea"
}"
}
我一直在尝试并仍在考虑的一些选项如下。
Logstash
情况1:不知道如何解析这些字符(/ u0002,/ u0001)
logstash.conf
input { kafka { bootstrap_servers => ["localhost:9092"] topics => ["higee.higee.higee"] auto_offset_reset => "earliest" codec => json { charset => "UTF-8" } } } filter { json { source => "message" } } output { stdout { codec => rubydebug } }
结果
{ "message" => "H\ \{\\"_id\\" : \\ {\\"$oid\\" : \\"5adafc0e2a0f383bb63910a6\\"}, \\ \\"name\\" : \\"higee\\", \\ \\"salary\\" : 101, \\ \\"origin\\" : \\"South Korea\\"} \\ \\\n0.7.5\\nhigee \\ \172.31.50.13\higee.higee2 \\ ح\\v\\ ̗ \\u\ X", "tags" => [[0] "_jsonparsefailure"] }
情况2
logstash.conf
input { kafka { bootstrap_servers => ["localhost:9092"] topics => ["higee.higee.higee"] auto_offset_reset => "earliest" codec => avro { schema_uri => "./test.avsc" } } } filter { json { source => "message" } } output { stdout { codec => rubydebug } }
测试文件
{ "namespace": "example", "type": "record", "name": "Higee", "fields": [ {"name": "_id", "type": "string"}, {"name": "name", "type": "string"}, {"name": "salary", "type": "int"}, {"name": "origin", "type": "string"} ] }
结果
An unexpected error occurred! {:error=>#<NoMethodError: undefined method `type_sym' for nil:NilClass>, :backtrace=> ["/home/ec2-user/logstash- 6.1.0/vendor/bundle/jruby/2.3.0/gems/avro- 1.8.2/lib/avro/io.rb:224:in `match_schemas'", "/home/ec2- user/logstash-6.1.0/vendor/bundle/jruby/2.3.0/gems/avro- 1.8.2/lib/avro/io.rb:280:in `read_data'", "/home/ec2- user/logstash-6.1.0/vendor/bundle/jruby/2.3.0/gems/avro- 1.8.2/lib/avro/io.rb:376:in `read_union'", "/home/ec2- user/logstash-6.1.0/vendor/bundle/jruby/2.3.0/gems/avro- 1.8.2/lib/avro/io.rb:309:in `read_data'", "/home/ec2- user/logstash-6.1.0/vendor/bundle/jruby/2.3.0/gems/avro- 1.8.2/lib/avro/io.rb:384:in `block in read_record'", "org/jruby/RubyArray.java:1734:in `each'", "/home/ec2- user/logstash-6.1.0/vendor/bundle/jruby/2.3.0/gems/avro- 1.8.2/lib/avro/io.rb:382:in `read_record'", "/home/ec2- user/logstash-6.1.0/vendor/bundle/jruby/2.3.0/gems/avro- 1.8.2/lib/avro/io.rb:310:in `read_data'", "/home/ec2- user/logstash-6.1.0/vendor/bundle/jruby/2.3.0/gems/avro- 1.8.2/lib/avro/io.rb:275:in `read'", "/home/ec2- user/logstash-6.1.0/vendor/bundle/jruby/2.3.0/gems/ logstash-codec-avro-3.2.3-java/lib/logstash/codecs/ avro.rb:77:in `decode'", "/home/ec2-user/logstash-6.1.0/ vendor/bundle/jruby/2.3.0/gems/logstash-input-kafka- 8.0.2/lib/ logstash/inputs/kafka.rb:254:in `block in thread_runner'", "/home/ec2-user/logstash- 6.1.0/vendor/bundle/jruby/2.3.0/gems/logstash-input-kafka- 8.0.2/lib/logstash/inputs/kafka.rb:253:in `block in thread_runner'"]}
python客户端
kafka
库:无法解码消息
from kafka import KafkaConsumer consumer = KafkaConsumer( topics='higee.higee.higee', auto_offset_reset='earliest' ) for message in consumer: message.value.decode('utf-8') >>> 'utf-8' codec can't decode byte 0xe4 in position 6: invalid continuation byte
confluent_kafka
与python 3不兼容
知道如何在Elasticsearch中对数据进行JSON处理吗? 以下是我搜索的资源。
提前致谢。
一些尝试
1)我已经如下更改了connect-mongo-source.properties文件以测试转换。
$ cat etc/kafka/connect-mongo-source.properties
>>>
name=mongodb-source-connector
connector.class=io.debezium.connector.mongodb.MongoDbConnector
mongodb.hosts=''
initial.sync.max.threads=1
tasks.max=1
mongodb.name=higee
transforms=unwrap
transforms.unwrap.type = io.debezium.connector.mongodbtransforms.UnwrapFromMongoDbEnvelope
以下是我得到的错误日志。 对于Kafka以及更重要的debezium平台还不满意,我无法调试此错误。
ERROR WorkerSourceTask{id=mongodb-source-connector-0} Task threw an uncaught and unrecoverable exception (org.apache.kafka.connect.runtime.WorkerTask:172)
org.bson.json.JsonParseException: JSON reader expected a string but found '0'.
at org.bson.json.JsonReader.visitBinDataExtendedJson(JsonReader.java:904)
at org.bson.json.JsonReader.visitExtendedJSON(JsonReader.java:570)
at org.bson.json.JsonReader.readBsonType(JsonReader.java:145)
at org.bson.codecs.BsonDocumentCodec.decode(BsonDocumentCodec.java:82)
at org.bson.codecs.BsonDocumentCodec.decode(BsonDocumentCodec.java:41)
at org.bson.codecs.BsonDocumentCodec.readValue(BsonDocumentCodec.java:101)
at org.bson.codecs.BsonDocumentCodec.decode(BsonDocumentCodec.java:84)
at org.bson.BsonDocument.parse(BsonDocument.java:62)
at io.debezium.connector.mongodb.transforms.UnwrapFromMongoDbEnvelope.apply(UnwrapFromMongoDbEnvelope.java:45)
at org.apache.kafka.connect.runtime.TransformationChain.apply(TransformationChain.java:38)
at org.apache.kafka.connect.runtime.WorkerSourceTask.sendRecords(WorkerSourceTask.java:218)
at org.apache.kafka.connect.runtime.WorkerSourceTask.execute(WorkerSourceTask.java:194)
at org.apache.kafka.connect.runtime.WorkerTask.doRun(WorkerTask.java:170)
at org.apache.kafka.connect.runtime.WorkerTask.run(WorkerTask.java:214)
at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
at java.util.concurrent.FutureTask.run(FutureTask.java:266)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
at java.lang.Thread.run(Thread.java:748)
2)这一次,我更改了elasticsearch.properties,但未更改connect-mongo-source.properties。
$ cat connect-mongo-source.properties
name=mongodb-source-connector
connector.class=io.debezium.connector.mongodb.MongoDbConnector
mongodb.hosts=''
initial.sync.max.threads=1
tasks.max=1
mongodb.name=higee
$ cat elasticsearch.properties
name=elasticsearch-sink
connector.class = io.confluent.connect.elasticsearch.ElasticsearchSinkConnector
tasks.max=1
topics=higee.higee.higee
key.ignore=true
connection.url=''
type.name=kafka-connect
transforms=unwrap
transforms.unwrap.type = io.debezium.connector.mongodb.transforms.UnwrapFromMongoDbEnvelope
而且我得到以下错误。
ERROR WorkerSinkTask{id=elasticsearch-sink-0} Task threw an uncaught and unrecoverable exception (org.apache.kafka.connect.runtime.WorkerTask:172)
org.bson.BsonInvalidOperationException: Document does not contain key $set
at org.bson.BsonDocument.throwIfKeyAbsent(BsonDocument.java:844)
at org.bson.BsonDocument.getDocument(BsonDocument.java:135)
at io.debezium.connector.mongodb.transforms.UnwrapFromMongoDbEnvelope.apply(UnwrapFromMongoDbEnvelope.java:53)
at org.apache.kafka.connect.runtime.TransformationChain.apply(TransformationChain.java:38)
at org.apache.kafka.connect.runtime.WorkerSinkTask.convertMessages(WorkerSinkTask.java:480)
at org.apache.kafka.connect.runtime.WorkerSinkTask.poll(WorkerSinkTask.java:301)
at org.apache.kafka.connect.runtime.WorkerSinkTask.iteration(WorkerSinkTask.java:205)
at org.apache.kafka.connect.runtime.WorkerSinkTask.execute(WorkerSinkTask.java:173)
at org.apache.kafka.connect.runtime.WorkerTask.doRun(WorkerTask.java:170)
at org.apache.kafka.connect.runtime.WorkerTask.run(WorkerTask.java:214)
at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
at java.util.concurrent.FutureTask.run(FutureTask.java:266)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
at java.lang.Thread.run(Thread.java:748)
3)更改了test.avsc并运行了logstash。 我没有收到任何错误消息,但即使在给定非null值的情况下,结果也不是我期望的那样,因为origin
, salary
和name
字段都是空的。 我什至能够通过控制台用户正确读取数据。
$ cat test.avsc
>>>
{
"type" : "record",
"name" : "MongoEvent",
"namespace" : "higee.higee",
"fields" : [ {
"name" : "_id",
"type" : {
"type" : "record",
"name" : "HigeeEvent",
"fields" : [ {
"name" : "$oid",
"type" : "string"
}, {
"name" : "salary",
"type" : "long"
}, {
"name" : "origin",
"type" : "string"
}, {
"name" : "name",
"type" : "string"
} ]
}
} ]
}
$ cat logstash3.conf
>>>
input {
kafka {
bootstrap_servers => ["localhost:9092"]
topics => ["higee.higee.higee"]
auto_offset_reset => "earliest"
codec => avro {
schema_uri => "./test.avsc"
}
}
}
output {
stdout {
codec => rubydebug
}
}
$ bin/logstash -f logstash3.conf
>>>
{
"@version" => "1",
"_id" => {
"salary" => 0,
"origin" => "",
"$oid" => "",
"name" => ""
},
"@timestamp" => 2018-04-25T09:39:07.962Z
}
您必须使用Avro Consumer,否则您将获得'utf-8' codec can't decode byte
即使这个例子中是行不通的 ,因为你仍然需要架构注册表中查找的模式。
Confluent的Python客户端的前提条件要求它可与Python 3.x一起使用
没有什么可以阻止您使用其他客户端,因此不确定为什么只尝试使用Python就将其保留了。
$oid
来代替_id
您的AVSC实际上应该是这样的
{
"type" : "record",
"name" : "MongoEvent",
"namespace" : "higee.higee",
"fields" : [ {
"name" : "_id",
"type" : {
"type" : "record",
"name" : "HigeeEvent",
"fields" : [ {
"name" : "$oid",
"type" : "string"
}, {
"name" : "salary",
"type" : "long"
}, {
"name" : "origin",
"type" : "string"
}, {
"name" : "name",
"type" : "string"
} ]
}
} ]
}
但是, Avro不允许以[A-Za-z_]
的正则表达式开头的名称开头 ,因此$oid
将是一个问题。
尽管我不推荐这样做(也没有实际尝试过),但从Avro控制台使用者将JSON编码的Avro数据导入Logstash的一种可能方法是使用Pipe输入插件
input {
pipe {
codec => json
command => "/path/to/confluent/bin/kafka-avro-console-consumer --bootstrap-server localhost:9092 --topic higee.higee.higee --from-beginning"
}
}
请注意,
after
值始终是一个字符串,并且按照惯例,它将包含文档的JSON表示形式
http://debezium.io/docs/connectors/mongodb/
我认为这也适用于patch
值,但我真的不了解Debezium。
如果不使用简单消息转换(SMT),Kafka将不会在运行中解析JSON。 阅读链接到的文档,您可能应该将它们添加到Connect Source属性中
transforms=unwrap
transforms.unwrap.type=io.debezium.connector.mongodb.transforms.UnwrapFromMongoDbEnvelope
值得一提的是,路线图上也将进行场平坦化-DBZ-561
如果不使用Logstash或其JSON Processor之类的东西,Elasticsearch不会解析和处理编码的JSON字符串对象。 相反,它仅将它们索引为整个字符串主体。
如果我没记错的话,Connect只会将Elasticsearch映射应用于顶级Avro字段,而不应用于嵌套字段。
换句话说,生成的映射遵循此模式,
"patch": {
"string": "...some JSON object string here..."
},
您实际需要的位置-也许手动定义ES索引
"patch": {
"properties": {
"_id": {
"properties" {
"$oid" : { "type": "text" },
"name" : { "type": "text" },
"salary": { "type": "int" },
"origin": { "type": "text" }
},
同样,不确定是否允许使用美元符号。
如果以上都不起作用,则可以尝试使用其他连接器
我能够使用python kafka客户端解决此问题。 以下是我的管道的新架构。
即使Confluent文档说支持python3,我也使用了python 2。 主要原因是有一些python2语法代码。 例如...(不完全是下面一行,但语法相似)
except NameError, err:
为了与Python3配合使用,我需要将上述行转换为:
except NameError as err:
话虽如此,以下是我的python代码。 请注意,此代码仅用于原型设计,尚不用于生产。
码
from confluent_kafka.avro import AvroConsumer c = AvroConsumer({ 'bootstrap.servers': '', 'group.id': 'groupid', 'schema.registry.url': '' }) c.subscribe(['higee.higee.higee']) x = True while x: msg = c.poll(100) if msg: message = msg.value() print(message) x = False c.close()
(在mongodb中更新文档之后)让我们检查message
变量
{u'after': None, u'op': u'u', u'patch': u'{ "_id" : {"$oid" : "5adafc0e2a0f383bb63910a6"}, "name" : "higee", "salary" : 100, "origin" : "S Korea"}', u'source': { u'h': 5734791721791032689L, u'initsync': False, u'name': u'higee', u'ns': u'higee.higee', u'ord': 1, u'rs': u'', u'sec': 1524362971, u'version': u'0.7.5'}, u'ts_ms': 1524362971148 }
码
patch = message['patch'] patch_dict = eval(patch) patch_dict.pop('_id')
检查patch_dict
{'name': 'higee', 'origin': 'S Korea', 'salary': 100}
from confluent_kafka import avro
from confluent_kafka.avro import AvroProducer
value_schema_str = """
{
"namespace": "higee.higee",
"name": "MongoEvent",
"type": "record",
"fields" : [
{
"name" : "name",
"type" : "string"
},
{
"name" : "origin",
"type" : "string"
},
{
"name" : "salary",
"type" : "int32"
}
]
}
"""
AvroProducerConf = {
'bootstrap.servers': '',
'schema.registry.url': ''
}
value_schema = avro.load('./user.avsc')
avroProducer = AvroProducer(
AvroProducerConf,
default_value_schema=value_schema
)
avroProducer.produce(topic='python', value=patch_dict)
avroProducer.flush()
剩下的唯一事情是通过以以下格式设置配置,使elasticsearch sink连接器响应新主题“ python”。 除topics
外,其他所有内容均保持不变。
name=elasticsearch-sink
connector.class= io.confluent.connect \
elasticsearch.ElasticsearchSinkConnector
tasks.max=1
topics=python
key.ignore=true
connection.url=''
type.name=kafka-connect
然后运行elasticsearch接收器连接器,并在elasticsearch上对其进行检查。
{
"_index": "zzzz",
"_type": "kafka-connect",
"_id": "zzzz+0+3",
"_score": 1,
"_source": {
"name": "higee",
"origin": "S Korea",
"salary": 100
}
}
+1 @ cricket_007的建议-使用io.debezium.connector.mongodb.transforms.UnwrapFromMongoDbEnvelope
单个消息转换。 您可以在此处阅读有关SMT及其优势的更多信息。
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.