2013-09-20 16:53:04,723 INFO[Thread-3]EndTime=20/09/2013 16:53:04 TransactionID=A, Event=completed, Result=sent
2013-09-20 16:03:04,723 INFO[Thread-3]EndTime=20/09/2013 16:53:04 TransactionID=b, Event=completed, Result=sent
2013-09-20 13:52:04,723 INFO[Thread-3]StartTime=20/09/2013 16:52:04 TransactionID=Ab, Event=start_process
2013-09-20 09:52:04,723 INFO[Thread-3]StartTime=20/09/2013 16:52:04 TransactionID=Ab, Event=start_process
If i need to get the # of transaction every hour as belows, what would be the search statement?
Date Transaction count
2013-09-20 09 1
2013-09-20 13 1
2013-09-20 16 2
Another method (which is more accurate then bucket) is as below:-
sourcetype=TranData | eval Date=strftime(_time,"%Y-%m-%d %H") | stats count as "Transaction Count" by Date
Problem with bucket is that if the query is executed in first 30 min, the base time to calculate may become
Given your sample data, what you are talking about are not transactions
in splunkese, but rather simple events
. The following will present a table that is quite near what you are looking for.
... | bucket _time span=1h | stats count AS "Transaction Count" by _time
/Kristian