Hello,
I am struggling a bit with regex and field extractions. I need to write my own sourcetype because I haven't found anything pre-made for dnstap. Maybe I was blind and you have something ready to hand.
I have the following raw event text:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 24094 ;; flags: qr aa rd ra ; QUESTION: 1, ANSWER: 1, AUTHORITY: 1, ADDITIONAL: 1 ;; OPT PSEUDOSECTION: ; EDNS: version: 0, flags:; udp: 4096 ; COOKIE: 67816834b9432822c5a508fd59b65054fb5bbab0c5fe14f8 ;; QUESTION SECTION: ;www.test.aa. IN A ;; ANSWER SECTION: www.test.aa. 60 IN CNAME testserver.domain www.test.aa. 60 IN A 192.168.1.20 ;; AUTHORITY SECTION: test.aa. 60 IN NS localhost.
I want to extract the "ANSWER SECTION", but my regex fails:
;;\sANSWER\sSECTION:\v(?<response_query>\S+)\s+(?<response_ttl>\S+)\s+(?<response_class>\S+)\s+(?<reponse_type>\S+)\s+(?<response>\S+)
The problem is that only the first line of the section is captured, but I need to capture every line because I need all the values. The "ANSWER SECTION" can consist of one line or several lines.
I'm using regex101.com with the regex flags "multi line" and "single line" as described in props.conf -> EXTRACT-<class>.
Solution found with reddit community
So that's exactly as we said - two-step approach. Firstly you parse the whole section, then you parse separate entries from it.
Still my warning about multivalued fields holds.
One thing is - as @isoutamo already pointed out - you should first "split" the event into sections, then parse "sets" of fields from each set.
But there is an additional problem - if you parse the answer section into multivalued fields, you will have separate mvfields with no relation between them. Splunk doesn't handle multi-level structures very well.
Hi
one option is do this with two steps like
| makeresults
| eval _raw = ";; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 24094
;; flags: qr aa rd ra ; QUESTION: 1, ANSWER: 1, AUTHORITY: 1, ADDITIONAL: 1
;; OPT PSEUDOSECTION:
; EDNS: version: 0, flags:; udp: 4096
; COOKIE: 67816834b9432822c5a508fd59b65054fb5bbab0c5fe14f8
;; QUESTION SECTION:
;www.test.aa. IN A
;; ANSWER SECTION:
www.test.aa. 60 IN CNAME testserver.domain
www.test.aa. 60 IN A 192.168.1.20
;; AUTHORITY SECTION:
test.aa. 60 IN NS localhost."
| rex max_match=0 ";;\sANSWER\sSECTION:\v(?<line>[^;]+)"
| rex max_match=0 field=line "(?<response_query>\S+)\s+(?<response_ttl>\S+)\s+(?<response_class>\S+)\s+(?<reponse_type>\S+)\s+(?<response>\S+)"
|fields - _raw _time
If you do it on props.conf, you need to ensure that extraction names are correct for getting line extracted firsts.
r. Ismo