I have a base64 field in my IIS log file. There are 3 very important properties within the base64 string that I want to extract at index time. It looks like everything available within splunk will be translated at search time and not added to the index.
What I don't want to have to do is manage a scheduled process (windows) on each server to run a transform script on the log, make sure it ran, process it intelligently to avoid re-processing already translated rows, having splunk monitor the translated log instead, etc. This was largely the purpose of Splunk.
I would even be ok if splunk orchestrated running the transform script if it couldn't directly do the decode at index time. E.g., splunk runs this script before indexing.
I am currently using a search app to do the decoding with python but doing nothing more than calling the following is a 13-15x performance hit. I want to be able to filter based off of these 3 decoded properties and that makes this approach unacceptable.
results = splunk.Intersplunk.getOrganizedResults()
for r in results
// do nothing
Was there ever an answer to this? I'm in the same situation as this:
"Ideally I'd have all the scripting run on the single indexer from the variable number of forwarder-provided-data. Is that possible? E.g. forwarder sends raw log -> indexer -> decode log from forwarder(s) -> index"
I'm using universal forwarders on each of the target machines. This gets sent via TCP to the single indexer instance. I'm not sure if scripted inputs will run on light forwarders. Ideally I'd have all the scripting run on the single indexer from the variable number of forwarder-provided-data. Is that possible? E.g. forwarder sends raw log -> indexer -> decode log from forwarder(s) -> index