Regex data from its position in string?

[pshankland]

Hi,

I have just installed Splunk as want to get some reports out of a Barracuda Spam firewall we have installed that sends all data to a syslog server. I have installed Splunk on the same machine as syslog so getting the file imported was easy.

The problem I am having is with trying to extract fields from the data as I can't seem to 'teach' the system the correct regex.

The following is an example line from syslog (anonomiyzed obviously):

Sep 24 15:34:59 192.168.0.1 inbound/pass1[26165]: 114-38-48-47.dynamic.hinet.net[114.38.48.47] 1285338898-663591fe0001-1ljPNx 1285338898 1285338899 RECV sender@domain.com receiver@domain.com 2 62 114.38.48.47

The data can be totally different which is why Splunk seems to have a problem with it, but the data is always in the same order:

Timestamp:  Sep 24 15:34:59
Host:  192.168.0.1
Process:  inbound/pass1[26165]:
Sender:  114-38-48-47.dynamic.hinet.net[114.38.48.47]
MessageID:  1285338898-663591fe0001-1ljPNx
StartTime:  1285338898
EndTime:  1285338899
Service:  RECV
From:  sender@domain.com
To:  receiver@domain.com
ActionCode:  2
ReasonCode:  62
SenderIP:  114.38.48.47

It was all going well until I got to the ActionCode!

There is always whitespace between the 'parts' so am sure it is just a matter of getting the regex correct but am struggling.

Would appreciate some help.

Thanks.

Pete.

[pshankland]

Thanks for the replies.

Just to explain what I am trying...

I am in the Search window and then click next to one of the entries and select "Extract Fields". I have then highlighted the bit I want to extract and dragged it into the Example Values box. Finally, I have then gone through deleting the sample extractions that were wrong.

A pattern never correctly generates as the figures are so small, this is why I thought RegEx would be the correct way to look at the issue.

Twinspop - could you let me know where I should be using the regex as assumed it would have been when extracting but that just failed 🙂

Thanks.

[southeringtonp]

Does it always log a single To address, or can you have multiple entries?

[twinspop]

\w+ \d+ \d{2}:\d{2}:\d{2} (?<host>[0-9.]+)\s+(?<process>\S+)\s+(?<sender>\S+)\s+(?<msgid>\S+)\s+(?<starttime>\d+)\s+(?<endtime>\d+)\s+(?<service>\S+)\s+(?<from>\S+)\s+(?<to>\S+)\s+(?<actioncode>\S+)\s+(?<reasoncode>\S+)\s+(?<senderip>[0-9.]+)

[christopherutz]

Can you post the regex you tried?