- Mark as New
- Bookmark Message
- Subscribe to Message
- Mute Message
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Using transaction I have grouped together events for same users.
There are two types of event.
1. Send SMS to user.
2. Log in with this user.
It looks like this:
Jan 22 11:52:15 172.30.112.1 PORTAL: SMS sent to 91813504.
Jan 22 11:54:11 172.30.112.1 AAA: 172.30.44.246 logged in with username 91813504
Jan 22 11:52:04 172.30.112.1 PORTAL: SMS sent to 91300083.
Jan 22 11:53:36 172.30.112.1 AAA: 172.30.53.83 logged in with username 91300083
Jan 22 11:51:10 172.30.112.1 PORTAL: SMS sent to 41237212.
Jan 22 11:51:50 172.30.112.1 AAA: 172.30.40.116 logged in with username 41237212
Jan 22 12:01:37 172.30.112.1 AAA: 172.30.40.113 logged in with username 41237212
Some user logs inn several times.
I do like to calculate the time from sent SMS to the first log in.
Result I need is some like this:
91813504 00:01:56
91300083 00:01:32
41237212 00:00:40
Then I can graph the duration over time and see if there are some wrong with the SMS sent to the user.
How to do this?
Maybe I do not need it use Transaction, but it gives me a visual view of events that belongs to same user.
- Mark as New
- Bookmark Message
- Subscribe to Message
- Mute Message
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content


You are correct about using transaction
.. Add this at the end of your search
... | timechart avg(duration)
Duration is a Splunk keyword which will find the difference between the timestamps
- Mark as New
- Bookmark Message
- Subscribe to Message
- Mute Message
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content


You are correct about using transaction
.. Add this at the end of your search
... | timechart avg(duration)
Duration is a Splunk keyword which will find the difference between the timestamps
- Mark as New
- Bookmark Message
- Subscribe to Message
- Mute Message
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Thanks. This was very close to what I need. Problem is where user has logged inn more than once with same username. avg(duration) gives the time between the first and last event. I need the time between first event and second event. Or to be more specific, time between last received SMS and first login after it. A user may request as many SMS he likes, so he can request two SMS and then login using information on last SMS.
- Mark as New
- Bookmark Message
- Subscribe to Message
- Mute Message
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Since majority av the users only have one SMS and one login, and the transaction need to contain both, I just use eventcount=2 in search after the transaction. This way I ignore all other and get a correct view. Thanks.
- Mark as New
- Bookmark Message
- Subscribe to Message
- Mute Message
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi,
I would recommend you to use "transaction" command to calculate duration.
transaction startswith="PORTAL: SMS sent to" endswith="logged in with username" |concurrency duration=duration
Then you can classify the duration by phone numbers and IP address if you need.
Hope it is ok.
Thanks
Gokhan
- Mark as New
- Bookmark Message
- Subscribe to Message
- Mute Message
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
This does not work since one user may log inn between another user sends SMS and logs inn. transaction startswith="PORTAL: SMS sent to" endswith="logged in with username" This mixes user together.
