Spent a lot of time trying to debug this simple scenario, so here it is for others to use. I did try to find an answer here but everything was partial or not working, besides a ruby plugin implementation which is overkill.
This is set to import and keep all up to date based on "ru" field which stands for "record updated" and holds precise time of update.
Gotchas: in json field you CAN'T have _id field, it is reserved!
FULLY WORKING CONFIG
input {
jdbc {
jdbc_driver_library => "/postgresql-42.2.6.jar"
jdbc_driver_class => "org.postgresql.Driver"
jdbc_connection_string => "jdbc:postgresql://localhost:32776/customdbs"
jdbc_user => "postgres"
jdbc_password => "root"
schedule => "* * * * *"
statement => "SELECT _id as id, ru, rsearch, rjson::text as rjsontext FROM producers WHERE _id = '5b8ab4e0c3088f8de7a3634d' AND ru > :sql_last_value;"
use_column_value => true
tracking_column => "ru"
tracking_column_type => "timestamp"
record_last_run => true
last_run_metadata_path => "record.last"
}
}
# WORKING CODE BUT OVERKILL
# filter {
# ruby {
# code => "
# require 'json'
# rjson = JSON.parse(event.get('rjsontext').to_s)
# event.set('rjson',rjson)
# "
# }
# }
# NATIVE IMPLEMENTATION WORKING
filter{
json{
source => "rjsontext"
target => "rjson"
remove_field => ["rjsontext"]
}
}
output {
elasticsearch {
hosts => "0.0.0.0:50100"
index => "index1"
document_id => "%{id}"
action => "update"
doc_as_upsert => true
}
}
Cheers!