Is there any option in Splunk to run a search in a loop?

[sunnyb147]

Hi All, Good morning,
Is there any option in Splunk to run a search in a loop?

Basically what I want to say is I have a search which is producing some result in a tabular format and further I am piping that in a CSV file, a single iteration is working fine.

But I want to run that search lets say for 7 days based on date_mday, so I was wondering if there is a way like for loop kind of a thing so that every time when the search is executed it appends the output to a csv file.

Sample search:

index=test1 country=india 
| dedup txn-id| stats count(txn-id) as unique_txns by date_mday, date_month 
| table date_mday, date_month, unique_txns
| outputlookup append=true sunny_test.csv

Any help/guidance would be really appreciated.

Thanks,
Sunny

[sunnyb147 · Accepted solution]

Found the solution without using loop, and its working fine.

Instead of doing, dedup then counting the unique transactions.. counted distinct count of transactions.

[amitm05]

Hi @sunnyb147

Can't exactly say why are you looking for a loop for the scenario you described. It more looks like you want to run it on a cron schedule.
In case there is a gap in understanding your logic, below is a sample of how you could use a loop -

| makeresults | eval  Feature.Flags.1 = "True", Feature.Flags.2 = "abc", Feature.Flags.3 = "" | eval HostFlags="" | foreach "Feature.Flags"* [eval HostFlags='<<FIELD>>'] | where HostFlags!="" | table Feature.Flags*

Also, take a look at - https://docs.splunk.com/Documentation/Splunk/7.3.0/SearchReference/Foreach
Hope this helps. Let me know.

[sunnyb147]

Hi @amitm05 ,

Thanks for your feedback, let me try explaining the scenario, basically what I am up-to:

1. If I run above query , it will give me a unique count of all transaction-ids for a specific day which I select in time-picker.
2. And if I add this thing in the cron schedule, it will append the file on daily basis.
3. What I want is, to have a while loop or for loop in which if I pass this query then it should take the date via date_mday field and extract the results and further append that in the csv file.

The main issue here is if I use 7 days in time-picker and dedup the transaction-id it doesn't give me correct count reason being some transaction-ids are being duplicated over multiple dates.

Example: If I search for a unique count on 30th of June 2019 the count is 515 but if I select the tenure of 7 days it gives me a count of 483 for 30th of June 2019.

I tried using MAP and FOREACH but didn't got what I was looking for.

Thanks,
Sunny

[amitm05]

Let me know if this answers your query. Or if there is more to it yet. Please accept if you are ok with the answer. Thaks

[skalliger]

Where is the problem exactly? Create a scheduled search that runs every seven days with earliest=-7d@d AND latest=@d for example and that's it. Your outputlookup already uses the option append=true so the files should get appended.

Skalli

[sunnyb147]

Hi @skalliger,

Thanks for your feedback, The main issue here is if I use 7 days in time-picker or by earliest/latest and dedup the transaction-id it doesn't give me correct count reason being some transaction-ids are being duplicated over multiple dates.

Example: If I search for a unique count on 30th of June 2019 the count is 515 but if I select the tenure of 7 days it gives me a count of 483 for 30th of June 2019.

What I am looking for: Is to have a while loop or for loop in which if I pass this query then it should take the date via date_mday field and extract the results and further append that in the csv file.

Thanks,
Sunny