How should I use sql_last_value in logstash?

If you have a timestamp column in your table (e.g. last_updated), you should preferably use it instead of the ID one. So that when a record gets updated, you modify that timestamp as well and the jdbc input plugin will pick up the record (i.e. the ID column won't change its value and the updated record won't get picked up)

input {
    jdbc {
        jdbc_connection_string => "jdbc:mysql://hostmachine:3306/db" 
        jdbc_user => "root"
        jdbc_password => "root"
        jdbc_validate_connection => true
        jdbc_driver_library => "/path/mysql_jar/mysql-connector-java-5.1.39-bin.jar"
        jdbc_driver_class => "com.mysql.jdbc.Driver"
        jdbc_paging_enabled => "true"
        jdbc_page_size => "50000"
        schedule => "* * * * *"
        statement => "SELECT * from mytable where last_updated > :sql_last_value"
    }
}

If you decide to stay with the ID column nonetheless, you should delete the $HOME/.logstash_jdbc_last_run file and try again.

There are a few things to take care of:

1. If you have run Logstash earlier without the schedule, then before running Logstash with schedule, delete the file:

   $HOME/.logstash_jdbc_last_run
   

   In Windows, this file is found at:

   C:\Users\<Username>\.logstash_jdbc_last_run
   

2. The "statement =>" in Logstash config should have "order by" the tracking_column.

3. tracking_column should be given correctly.

Here is an example of the Logstash config file:

    input {
jdbc {
    # MySQL DB jdbc connection string to our database, softwaredevelopercentral
    jdbc_connection_string => "jdbc:mysql://localhost:3306/softwaredevelopercentral?autoReconnect=true&useSSL=false"
    # The user we wish to execute our statement as
    jdbc_user => "root"
    # The user password
    jdbc_password => ""
    # The path to our downloaded jdbc driver
    jdbc_driver_library => "D:\Programs\MySQLJava\mysql-connector-java-6.0.6.jar"
    # The name of the driver class for MySQL DB
    jdbc_driver_class => "com.mysql.cj.jdbc.Driver"
    # our query
    schedule => "* * * * *"
    statement => "SELECT * FROM student WHERE studentid > :sql_last_value order by studentid"
    use_column_value => true
    tracking_column => "studentid"
}
}
output {
stdout { codec => json_lines }
elasticsearch { 
   hosts => ["localhost:9200"]
   index => "students"
   document_type => "student"
   document_id => "%{studentid}"
   }

}

To see a working example of the same you can check my blog post: http://softwaredevelopercentral.blogspot.com/2017/10/elasticsearch-logstash-kibana-tutorial.html

In simple words, sql_last_value allows you to persist data from your last sql run as its name sugets.

This value is specially useful when you schedule your query. But why ... ? Because you can create your sql statement condition based on the value stored in sql_last_value and avoid to retrieve rows that were already ingested for your logstash input or updated after last pipeline execution.

Things to keep in mind when using sql_last_value

• By default, this variable stores a timestamp of last run. Useful when you need to ingest data based in columns like creation_date last_update etc..
• You can define the value of sql_last_value by tracking it with a specific table's column value. Useful when you need to ingest auto increment data based. For that, you need to specify use_column_value => true and tracking_column => "column_name_to_track".

The following example will store the last mytable row's id into :sql_last_value to ingest in the next execution the rows that were not ingested previously, it means the rows which its id is greater than the last ingested id.

input {
    jdbc {
        # ...
        schedule => "* * * * *"
        statement => "SELECT * from mytable where id > :sql_last_value"
        use_column_value => true
        tracking_column => id
    }
}

Extremely important !!!

When you use multiple inputs in your pipeline, each input block will overwrite the value of sql_last_value of the last one. For avoiding that behaviour, you can use last_run_metadata_path => "/path/to/sql_last_value/of_your_pipeline.yml" option, which means that each pipepline will stores its own value in a different file.