Splunk Search

How can I find delta between events using transaction command?

limalbert
Path Finder

 

I could retrieve the list of the transactions as a single event below. Transactions start with "Dashboard Load:" and end with "Total Dashboard Load Time"

 

| rex "Account: (?<account>[^\,]*), SiteID: (?<siteID>[^\s]*) - (?<duration_in_ms>[^\s]*) milliseconds on Requestor: (?<requestor>[^\,]*)"
| transaction agent endswith="Total Dashboard Load Time"
| eval duration_in_sec=round(avg(duration_in_ms)/1000,2)
| table agent account siteID flow duration_in_sec

 

 

I could find the delta between each event as multiple events with below

 

| rex "Account: (?<account>[^\,]*), SiteID: (?<siteID>[^\s]*) - (?<duration_in_ms>[^\s]*) milliseconds on Requestor: (?<requestor>[^\,]*)"
| streamstats current=f last(_time) as last_time2 by agent
| rename _time as current_time2
| eval delta= last_time2 - current_time2
| eval last_time =strftime(last_time2, "%m/%d, %H:%M:%S.%6N")
| eval current_time =strftime(current_time2, "%m/%d, %H:%M:%S.%6N")
| table agent account siteID flow duration_in_sec last_time current_time delta

 

 

I'm looking to see if I can find a delta between events in a single transaction. The transaction starts with "Dashboard Load" and ends with "Total Dashboard Load Time"

Expected Result Format:

 

agent, account, siteID, list of flow (all 5 Dashboard Load actions), duration_in_sec, list of event times, list of delta between event time

 

  1. agent - the name of the agent
  2. account - account #
  3. siteID - site#
  4. list of flow - all action flows
  5. durations_in_sec - Total Dashboard Load Time
  6. list of event times - List showing all event time
  7. list of the delta between event time - List showing 4 delta times between each flow

 

Below are a sample of 2 data of 5 events

 

2021-12-16 05:30:43,834 alpha - Dashboard Load: User Clicked on New
2021-12-16 05:30:46,498 alpha - Dashboard Load: User Clicked on AccountSearch
2021-12-16 05:31:05,420 alpha - Dashboard Load: User Clicked on Search with String abcdef
2021-12-16 05:31:08,557 alpha - Dashboard Load: User clicked on Searched Result 123456
2021-12-16 05:31:12,234 alpha - Total Dashboard Load Time for Account: 000111222, SiteID: 123- 3438.0 milliseconds on Requestor: xoxoxo

 

  

Labels (2)
0 Karma
1 Solution

yuanliu
SplunkTrust
SplunkTrust

You can combine transaction with streamstats.  I'll illustrate two slightly different tweaks.

The first is perhaps closest to the described end result.

 

| rex "(?<agent>\w+)\s-\s(?<event>.+)"
| rex field=event "Account: (?<account>[^\,]*), SiteID: (?<siteID>[^\s]*) - (?<duration_in_ms>[^\s]*) milliseconds on Requestor: (?<requestor>[^\,]*)"
| reverse
| streamstats current=f last(_time) as last_time2 by agent
| eval current_time2 = _time
| eval delta= if(isnull(last_time2), "n/a", current_time2 - last_time2)
| eval event = strftime(_time, "%Y-%m-%d %H:%M:%S,%3Q") . " <" . delta . "> " . event
| eval last_time =if(isnull(last_time2), "n/a", strftime(last_time2, "%m/%d, %H:%M:%S.%6N"))
| reverse
| transaction agent endswith="Total Dashboard Load Time" startswith="Clicked on New"
| eval duration_in_sec=round(avg(duration_in_ms)/1000,2)
| rename event as flow
| table _time agent account siteID flow duration duration_in_sec last_time

 

The first note is that the two reverse's will be very expensive for large volumes.  I removed delta as a column because transaction takes away its ordering; instead, "current time" and "delta" are both embedded into flow.  If desired, you can extract them out as columns for display purposes. (Note that delta "n/a" in the first event will be replaced by the delta between that event and the preceding last event.)  Additionally, "duration_in_sec" is actually the dashboard load time; duration of the transaction is provided by the transaction command, which I included for illustration.   Sample output as follows:

_timeagentaccountsiteID
flow
durationduration_in_sec
last_time
2021-12-16 05:30:43.834alpha000111222123
2021-12-16 05:30:43,834 <n/a> Dashboard Load: User Clicked on New
2021-12-16 05:30:46,498 <2.664000> Dashboard Load: User Clicked on AccountSearch
2021-12-16 05:31:05,420 <18.922000> Dashboard Load: User Clicked on Search with String abcdef
2021-12-16 05:31:08,557 <3.137000> Dashboard Load: User clicked on Searched Result 123456
2021-12-16 05:31:12,234 <3.677000> Total Dashboard Load Time for Account: 000111222, SiteID: 123 - 3438.0 milliseconds on Requestor: xoxoxo
28.43.44
12/16, 05:30:43.834000
12/16, 05:30:46.498000
12/16, 05:31:05.420000
12/16, 05:31:08.557000
n/a

 

The second tweak is more economic because there is no reverse, but you have to tolerate some odd artifacts, or correct them in some way after you get the list.

 

| rex "(?<agent>\w+)\s-\s(?<event>.+)"
| rex field=event "Account: (?<account>[^\,]*), SiteID: (?<siteID>[^\s]*) - (?<duration_in_ms>[^\s]*) milliseconds on Requestor: (?<requestor>[^\,]*)"
| streamstats current=f last(_time) as last_time2 by agent
| eval current_time2 = _time
| eval delta= if(isnull(last_time2), "n/a", last_time2 - current_time2)
| eval event = strftime(_time, "%Y-%m-%d %H:%M:%S,%3Q") . " <" . delta . "> " . event
| eval last_time =if(isnull(last_time2), "n/a", strftime(last_time2, "%m/%d, %H:%M:%S.%6N"))
| transaction agent endswith="Total Dashboard Load Time" startswith="Clicked on New"
| eval duration_in_sec=round(avg(duration_in_ms)/1000,2)
| rename event as flow
| table _time agent account siteID flow duration duration_in_sec last_time

 

Sample result follows:

_timeagentaccountsiteID
flow
durationduration_in_sec
last_time
2021-12-16 05:30:43.834alpha000111222123
2021-12-16 05:30:43,834 <2.664000> Dashboard Load: User Clicked on New
2021-12-16 05:30:46,498 <18.922000> Dashboard Load: User Clicked on AccountSearch
2021-12-16 05:31:05,420 <3.137000> Dashboard Load: User Clicked on Search with String abcdef
2021-12-16 05:31:08,557 <3.677000> Dashboard Load: User clicked on Searched Result 123456
2021-12-16 05:31:12,234 <n/a> Total Dashboard Load Time for Account: 000111222, SiteID: 123 - 3438.0 milliseconds on Requestor: xoxoxo
28.43.44
12/16, 05:30:46.498000
12/16, 05:31:05.420000
12/16, 05:31:08.557000
12/16, 05:31:12.234000
n/a

You can see that the delta and "last_time" are misaligned by one event.

Hope these help.

View solution in original post

0 Karma

yuanliu
SplunkTrust
SplunkTrust

Based on your sample code and data, I am guessing that values(X) and list(X) are what you are looking for, not transaction, something like this:

| rex "(?<agent>\w+)\s-\s(?<event>.+)" ```I have to guess that "alpha" is agent```
| rex field=event "Account: (?<account>[^\,]*), SiteID: (?<siteID>[^\s]*) - (?<duration_in_ms>[^\s]*) milliseconds on Requestor: (?<requestor>[^\,]*)"
| streamstats current=f last(_time) as last_time2 by agent
| rename _time as current_time2
| eval delta= if(isnull(last_time2), "n/a", last_time2 - current_time2)
| eval last_time =if(isnull(last_time2), "n/a", strftime(last_time2, "%m/%d, %H:%M:%S.%6N"))
| eval current_time =strftime(current_time2, "%m/%d, %H:%M:%S.%6N")
| eval duration_in_sec=round(avg(duration_in_ms)/1000,2)

| stats values(account) as account values(siteID) as siteID values(duration_in_sec) as duration_in_sec list(event) as flow list(current_time) as current_time list(last_time) as last_time list(delta) as delta by agent
| table agent account siteID flow duration_in_sec last_time current_time delta

 

Tags (2)
0 Karma

limalbert
Path Finder

I would need transaction because each transaction starts with the event of "Total Dashboard Load Time" and ends with "Total Dashboard Load Time for Account:". I would need to group these events together and find the delta for each event in a transaction (in this case 5 events)

 

 

0 Karma

yuanliu
SplunkTrust
SplunkTrust

You can combine transaction with streamstats.  I'll illustrate two slightly different tweaks.

The first is perhaps closest to the described end result.

 

| rex "(?<agent>\w+)\s-\s(?<event>.+)"
| rex field=event "Account: (?<account>[^\,]*), SiteID: (?<siteID>[^\s]*) - (?<duration_in_ms>[^\s]*) milliseconds on Requestor: (?<requestor>[^\,]*)"
| reverse
| streamstats current=f last(_time) as last_time2 by agent
| eval current_time2 = _time
| eval delta= if(isnull(last_time2), "n/a", current_time2 - last_time2)
| eval event = strftime(_time, "%Y-%m-%d %H:%M:%S,%3Q") . " <" . delta . "> " . event
| eval last_time =if(isnull(last_time2), "n/a", strftime(last_time2, "%m/%d, %H:%M:%S.%6N"))
| reverse
| transaction agent endswith="Total Dashboard Load Time" startswith="Clicked on New"
| eval duration_in_sec=round(avg(duration_in_ms)/1000,2)
| rename event as flow
| table _time agent account siteID flow duration duration_in_sec last_time

 

The first note is that the two reverse's will be very expensive for large volumes.  I removed delta as a column because transaction takes away its ordering; instead, "current time" and "delta" are both embedded into flow.  If desired, you can extract them out as columns for display purposes. (Note that delta "n/a" in the first event will be replaced by the delta between that event and the preceding last event.)  Additionally, "duration_in_sec" is actually the dashboard load time; duration of the transaction is provided by the transaction command, which I included for illustration.   Sample output as follows:

_timeagentaccountsiteID
flow
durationduration_in_sec
last_time
2021-12-16 05:30:43.834alpha000111222123
2021-12-16 05:30:43,834 <n/a> Dashboard Load: User Clicked on New
2021-12-16 05:30:46,498 <2.664000> Dashboard Load: User Clicked on AccountSearch
2021-12-16 05:31:05,420 <18.922000> Dashboard Load: User Clicked on Search with String abcdef
2021-12-16 05:31:08,557 <3.137000> Dashboard Load: User clicked on Searched Result 123456
2021-12-16 05:31:12,234 <3.677000> Total Dashboard Load Time for Account: 000111222, SiteID: 123 - 3438.0 milliseconds on Requestor: xoxoxo
28.43.44
12/16, 05:30:43.834000
12/16, 05:30:46.498000
12/16, 05:31:05.420000
12/16, 05:31:08.557000
n/a

 

The second tweak is more economic because there is no reverse, but you have to tolerate some odd artifacts, or correct them in some way after you get the list.

 

| rex "(?<agent>\w+)\s-\s(?<event>.+)"
| rex field=event "Account: (?<account>[^\,]*), SiteID: (?<siteID>[^\s]*) - (?<duration_in_ms>[^\s]*) milliseconds on Requestor: (?<requestor>[^\,]*)"
| streamstats current=f last(_time) as last_time2 by agent
| eval current_time2 = _time
| eval delta= if(isnull(last_time2), "n/a", last_time2 - current_time2)
| eval event = strftime(_time, "%Y-%m-%d %H:%M:%S,%3Q") . " <" . delta . "> " . event
| eval last_time =if(isnull(last_time2), "n/a", strftime(last_time2, "%m/%d, %H:%M:%S.%6N"))
| transaction agent endswith="Total Dashboard Load Time" startswith="Clicked on New"
| eval duration_in_sec=round(avg(duration_in_ms)/1000,2)
| rename event as flow
| table _time agent account siteID flow duration duration_in_sec last_time

 

Sample result follows:

_timeagentaccountsiteID
flow
durationduration_in_sec
last_time
2021-12-16 05:30:43.834alpha000111222123
2021-12-16 05:30:43,834 <2.664000> Dashboard Load: User Clicked on New
2021-12-16 05:30:46,498 <18.922000> Dashboard Load: User Clicked on AccountSearch
2021-12-16 05:31:05,420 <3.137000> Dashboard Load: User Clicked on Search with String abcdef
2021-12-16 05:31:08,557 <3.677000> Dashboard Load: User clicked on Searched Result 123456
2021-12-16 05:31:12,234 <n/a> Total Dashboard Load Time for Account: 000111222, SiteID: 123 - 3438.0 milliseconds on Requestor: xoxoxo
28.43.44
12/16, 05:30:46.498000
12/16, 05:31:05.420000
12/16, 05:31:08.557000
12/16, 05:31:12.234000
n/a

You can see that the delta and "last_time" are misaligned by one event.

Hope these help.

0 Karma
Get Updates on the Splunk Community!

Routing logs with Splunk OTel Collector for Kubernetes

The Splunk Distribution of the OpenTelemetry (OTel) Collector is a product that provides a way to ingest ...

Welcome to the Splunk Community!

(view in My Videos) We're so glad you're here! The Splunk Community is place to connect, learn, give back, and ...

Tech Talk | Elevating Digital Service Excellence: The Synergy of Splunk RUM & APM

Elevating Digital Service Excellence: The Synergy of Real User Monitoring and Application Performance ...