Splunk Enterprise

How to not include a pattern when line merging?

ricotries
Communicator

I have a log from an application that isn't structured in any standard format and I am struggling with dropping certain lines at index time due to the line merging configuration.

This is a pseudo sample of the data:

----- <application version> -----
(<timestamp>) <data> 
(<timestamp>) <data>

----- <application version> -----
(<timestamp>) <data> 
(<timestamp>) <data>

----- <application version> -----
(<timestamp>) <data> 
(<timestamp>) <data>
<data>
<data>
<data>
(<timestamp>) <data>

As you can see, for some events the message is broken down into multiple lines, so the best way to break events would be by the timestamp, so this is the props.conf I wrote for this source type:

[my_new_sourcetype]
SHOULD_LINEMERGE = true
BREAK_ONLY_BEFORE_DATE = true
TRANSFORMS-drop_header = new_sourcetype_drop_header

 And the associated transforms.conf:

[new_sourcetype_drop_header]
REGEX = ^-{5}.+-{5}$
DEST_KEY = queue
FORMAT = nullQueue

 

The issue becomes that when the data is indexed, any event that would have been the <application version> header by itself is dropped, but then there are events with a linecount of 2 that look like:

(<timestamp>) <data>
----- <application version> -----

 

How do I force it so that the <application version> header is always made into its own event so that it can be dropped by the transforms configuration?

Labels (1)
0 Karma
1 Solution

richgalloway
SplunkTrust
SplunkTrust

Rather than breaking only before a date, try breaking before a date OR a header.

[my_new_sourcetype]
SHOULD_LINEMERGE = false
LINE_BREAKER = ([\r\n]+)(<<timestamp regex>>|-{5})
TRANSFORMS-drop_header = new_sourcetype_drop_header
TIME_FORMAT = <<timestamp regex>>
TIME_PREFIX = \(

Be sure to replace <<timestamp regex>> with a regular expression that matches your timestamp strings.

---
If this reply helps you, Karma would be appreciated.

View solution in original post

richgalloway
SplunkTrust
SplunkTrust

Rather than breaking only before a date, try breaking before a date OR a header.

[my_new_sourcetype]
SHOULD_LINEMERGE = false
LINE_BREAKER = ([\r\n]+)(<<timestamp regex>>|-{5})
TRANSFORMS-drop_header = new_sourcetype_drop_header
TIME_FORMAT = <<timestamp regex>>
TIME_PREFIX = \(

Be sure to replace <<timestamp regex>> with a regular expression that matches your timestamp strings.

---
If this reply helps you, Karma would be appreciated.
Get Updates on the Splunk Community!

Splunk Smartness with Brandon Sternfield | Episode 3

Hello and welcome to another episode of "Splunk Smartness," the interview series where we explore the power of ...

Monitoring Postgres with OpenTelemetry

Behind every business-critical application, you’ll find databases. These behind-the-scenes stores power ...

Mastering Synthetic Browser Testing: Pro Tips to Keep Your Web App Running Smoothly

To start, if you're new to synthetic monitoring, I recommend exploring this synthetic monitoring overview. In ...