Now, the challenge no1 is that the script that generates the csv, can edit on already existing lines.
challenge no2 is that this does not result in one and the same behaviors all the time.
If a simple value is changed ( from 1 to 2, or from ivan to ivag - important is same number of characters) , the change is no-were to be found in the indexed data. However, if the change includes a change in the number of characters(say ivan becomes johnathan) then, the whole file is re-indexed with the new value, causing lots of duplications.
I am sure that this must be documented somewhere....but I can not find it, thus can not really understand it.
Does anyone know what is going on( I managed to find something in the community about splunk checking the first 256 char. of a file to decide, but I have tested changing both before 256 threshold and after it).....?
I think you already played around with "initCrcLength" config in inputs.conf which just tells Splunk how far into the file to compare the hash. It's just going to re-ingest the whole file no matter what you do with CSV files and you end up with duplicated data with the way you wrote your script (modifying specific values in the CSV file). I think a better way is to create a new CSV file whenever there is a change and modify your inputs.conf accordingly to ingest the files.
I might not have the best answer so wait around for more suggestions.