Inputs.conf Monitoring 1 file 1 event

[jeffflynn]

I am monitoring the logs for an application that spits out 3 xml files per day that I want indexed to 1 event per entire file. Everything is setup and working properly when the files are initially indexed because the files are already complete.

The problem is when the application is actually building the file.
1. The application is creates the file
2. Splunk indexes what is created
3. the application adds to the file
4. splunk indexes the addition to another event
5. application adds more to the file
6. splunk indexes changes to another event.
7. so on and so forth.

Since Monitor doesn't use Interval to index new files every however many seconds or via cron job I kinda at a loss. I have tried setting the time_before_close=120 but that still indexes the file at creation and then will index the file modifications of the next 2 min in another event. I am wanting 1 event per file. This isn't a high demand application so indexing the files once a day would be acceptable.

Here is my inputs and props which are pretty simple.

[monitor://whatever]
disabled=false
index=data
whitelist=.xml
blacklist=garbage
sourcetype=info
time_before_close=120

[info]
break_only_before=GOBBLEDEEGOOP
max_events=200000
time_prefix=start-time

[jeffflynn · Accepted solution]

I think my best chance at getting what I need is to run a script every morning that will copy the new files to a subfolder. Then index the files from the subfolder using a batch command instead of monitor.

[Runals]

Have you looked into the batch command as opposed to monitor?

[somesoni2]

You may write a small script to read the file and setup a scripted data input to read script's output.
http://docs.splunk.com/Documentation/Splunk/6.1.1/Data/Setupcustominputs

[jeffflynn]

yea. that just lazy typing on my part. i have everything in caps in my props file

[Ayn]

First of all, props.conf settings are case sensitive so "break_only_before" needs to be BREAK_ONLY_BEFORE, and so on.