How to configure timestamp to recognize multiline timestamp?

[iljubicic]

I am a new Splunk user and I am having difficulties resolving this problem. I have an xml log file as an input structured like this:

<LOG>
 <DATE>DDMMYYYY</DATE>
 <TIME>HHMMSS</TIME>
 <CC>\d{16}</CC>
 <AMOUNT>\d+\.\d{2}</AMOUNT>
</LOG>

where logs are separated with LOG tag. What do I need to do to make Splunk to correctly read the timestamp that is separated in the two lines (DATE and TIME)?

[maciep · Accepted solution]

You'll need to configure props.conf at the parsing layer (probably on your indexer) appropriately. That includes creating a new stanza for your sourcetype and telling Splunk how to deal with the format.

http://docs.splunk.com/Documentation/Splunk/latest/admin/Propsconf

http://pubs.opengroup.org/onlinepubs/009695399/functions/strptime.html

Something like this should work. Tell splunk to merge multiple lines into one event and where to break events. Then tell it where the timestamp is and what the format looks like.

[your sourcetype]
SHOULD_LINEMERGE = true
BREAK_ONLY_BEFORE = <LOG>
TIME_PREFIX = <DATE>
TIME_FORMAT = %d%m%Y</DATE>%n<TIME>%H%M%S