How do I remove extra information from field names when parsing JSON?

[djain]

Hi

I am trying to ingest JSON data from JSON files. However, instead of just taking the field name, it is also taking header and variable type. See screenshot. I want to remove device.headwaters.watermark.core from the beginning and .long or .string from the end. How can I do that?

[mstjohn_splunk]

hi @djain

were you able to try out the answer below? If it worked, would you mind approving it?

[niketn · Accepted solution]

@djain try the following:

<yourCurrentSearch>
| rename "device.headwaters.watermark.core.*" as "*"

[djain]

Thank you for the answer but I am looking for a more automated process to do this while the data is getting indexed by changing props file or settings at sourcetype

[niketn]

The reason why you are seeing additional name is because of the way your JSON is structured and default parsing will put all node names to make the traversed tree (field name) unique (unless it is a multi-valued field).

Option 1: You will have to get rid of either INDEXED_EXTRACTIONS = json OR KV_MODE=json (whichever is present) toKV_MODE=none without INDEXED_EXTRACTIONS and define your own field extraction. Refer to this answer by @somesoni2 https://answers.splunk.com/answers/484752/how-to-parse-json-which-makes-up-part-of-the-event.html

Option 2: Alternatively, if you are sure there are not events under any of the elements of the tree device.headwaters.watermark.core., you can create a transformation to index only the child JSON element with meaningful data i.e. everything inside code node. You should check out Splunk Answers for Routing and Filtering Data to index only specific part of JSON Payload data. Refer to @woodcock 's answer on similar lines: https://answers.splunk.com/answers/519236/split-indexing-from-json.html