json_lines codec is still not working with file input plugin

[shashank-gupta017]

The json_lines codec when used on the file input plugin causes the file input plugin to not process any data.

input {
  file {
    path => ["/usr/local/Cellar/logstash/7.3.0/bin/grant.log"]
    exclude => ["*.gz"]
    sincedb_path => "/usr/local/Cellar/logstash/7.3.0/bin/grant.sincedb"
    start_position => "beginning"
    codec => "json_lines"
  }
}

output {
  stdout {
      codec => rubydebug
    }
}

When I replace the codec json_lines with json, the codec works and I get the parsed json. But with json_lines, no data is printed on stdout. I read the forums that json_lines should work with file input now, but in this case, it is still giving issues.

I also verified that in the input file, each json is de-limited with a \n by running the command cat input.log|od -tx1c

The sample input looks like this

{"group_id":153,"reason_id":357,"insert_time":"2020-03-04 13:58:48.000","fix_flag":0,"std_change":100,"type":"standard","transaction_id":"01538b8d-c580-3974-b41e-b2b0fb3fc856","timestamp":"2020-03-04 13:58:48.000","refkey":"keya1","transaction_status":"NEW","easy_id":12346738,"subkey":"subkeya1","points":100}

[yaauie]

This is, somewhat unfortunately, by design.

The File input chunks its input by the delimiter, and hands the bytes between delimiters to the codec. This means that when it is configured with the json_lines codec, the codec never receives the delimiter and continues to add these chunks to its internal buffer instead of turning each chunk into a single event.

The solution is to use codec => json.