I have a nodejs app that currently uses bunyan logger to dump big amounts of logs in JSON format on the disk (one file a day). These logs have various nested fields, some about service requests and responses, and some about other app's activity.
Now I want to be able to query these logs using Elasticsearch. What is the best approach for me to reach this goal? Should I use logstash to convert files' data to ELK or it is better to pass logs directly?
I guess I should create an index template to tell which fields should be indexed. In this regard how to choose them? Are there any guidelines for this or just check the current logs one by one and tell which fields should be included in indices?