使用logstash和jdbc更新复杂的嵌套弹性搜索文档

Question

Let's assume that the Oracle Schema has following tables and columns:

Country
        country_id; (Primary Key)
        country_name;

    Department
        department_id; (Primary Key)
        department_name;
        country_id; (Foreign key to Country:country_id)

    Employee
        employee_id; (Primary Key)
        employee_name;
        department_id; (Foreign key to Department:department_id)

And I have my Elasticsearch document where the root element is a Country and it contains all Departments in that Country which in turn contain all Employees in respective Departments.

So the document structure looks like this:

{
      "mappings": {
        "country": {
          "properties": {
            "country_id": { "type": "string"},
            "country_name": { "type": "string"},        
            "department": {
              "type": "nested",
              "properties": {
                "department_id": { "type": "string"},
                "department_name": { "type": "string"},
                "employee": {
                  "type": "nested",
                  "properties": {
                    "employee_id": { "type": "string"},
                    "employee_name": { "type": "string"}
                  }
                }
              }
            }
          }
        }
      }
    }

I want to be able to have separate input jdbc queries running on each table and they should create/update/delete data in the elasticsearch document whenever the data in the base table are added/updated/deleted.

This is an example problem and actual tables and data structure are more complex. So I am not looking for solution limited to this.

Is there a way to achieve this?

Thanks.

Answer 1

For level one, its straight forward using aggregate filter . You need to have a common id between them to reference.

filter {    

  aggregate {
    task_id => "%{id}"

    code => "     
      map['id'] = event.get('id')
      map['department'] ||= []
      map['department'] << event.to_hash.each do |key,value| { key => value } end    
    "
    push_previous_map_as_event => true
    timeout => 150000
    timeout_tags => ['aggregated']    
  } 

   if "aggregated" not in [tags] {
    drop {}
  }
}

Important : The output action should be update

    output {
        elasticsearch {
            action => "update"
             ...
           }
        }

One way to solve level 2 is to query the already indexed document and update it with the nested record . Again using aggregate filter ; there should be a common id for the document so you can lookup and insert into the correct document.

filter {    
    #get the document from elastic based on id and store it in 'emp'
    elasticsearch {
            hosts => ["${ELASTICSEARCH_HOST}/${INDEX_NAME}/${INDEX_TYPE}"]
            query => "id:%{id}" 
            fields => { "employee" => "emp" }
         }



  aggregate {
    task_id => "%{id}"  
    code => "       
                map['id'] = event.get('id')
                map['employee'] = []
                employeeArr = []
                temp_emp = {}   

                event.to_hash.each do |key,value|                       
                    temp_emp[key] = value
                end     

                #push the objects into an array
                employeeArr.push(temp_emp)

                empArr = event.get('emp')                   

                for emp in empArr
                    emp['employee'] = employeeArr                       
                    map['employee'].push(emp)
                end
    "
    push_previous_map_as_event => true
    timeout => 150000
    timeout_tags => ['aggregated']

  } 

   if "aggregated" not in [tags] {
    drop {}
  } 

}

output {

elasticsearch {
        action => "update"    #important
         ...
        }
 }  

Also, in order to debug the ruby code, use the below in the output

output{
    stdout { codec => dots }
}