Splunk Search

Remove specific substrings from specific values

techspec
Explorer

I have a query that looks for certain error messages and displays a list sorted by most common occurrence. My problem is that some of these error messages contain unique IDs, causing them to show as separate results.

The first 3 messages below end with unique values - how do I update this query to output those error messages without the unique values after the * mark, or even replace with another string like "redacted"?

index=mint event_name=APIServiceError appEnvironment=Release appVersionName>=3.0 
  (
    extraData.message="Incorrect email or password for*"  OR
    extraData.message="Street address and/or city is too long*" OR
    extraData.message="The following address could not be found:*" OR
    extraData.message="We failed to authorize your payment card. Please verify your payment card is valid." OR
    extraData.message="The network connection was lost."
  )  
| stats count by extraData.message | sort -count

 

Labels (1)
0 Karma
1 Solution

Pathik
Path Finder

Try adding below to replace message with the one you are looking for.

index=mint event_name=APIServiceError appEnvironment=Release appVersionName>=3.0
(
extraData.message="Incorrect email or password for*" OR
extraData.message="Street address and/or city is too long*" OR
extraData.message="The following address could not be found:*" OR
extraData.message="We failed to authorize your payment card. Please verify your payment card is valid." OR
extraData.message="The network connection was lost."
)
| eval extraData.message=if(like(extraData.message, "Street address and/or city is too long*"), "Street address and/or city is too long",extraData.message) | stats count by extraData.message | sort -count 

View solution in original post

0 Karma

Pathik
Path Finder

Try adding below to replace message with the one you are looking for.

index=mint event_name=APIServiceError appEnvironment=Release appVersionName>=3.0
(
extraData.message="Incorrect email or password for*" OR
extraData.message="Street address and/or city is too long*" OR
extraData.message="The following address could not be found:*" OR
extraData.message="We failed to authorize your payment card. Please verify your payment card is valid." OR
extraData.message="The network connection was lost."
)
| eval extraData.message=if(like(extraData.message, "Street address and/or city is too long*"), "Street address and/or city is too long",extraData.message) | stats count by extraData.message | sort -count 

0 Karma

techspec
Explorer

Thanks - what you suggested makes sense, but I used it exactly as you showed and get "no results found": 

Screen Shot 2020-10-09 at 11.32.57 AM.png

0 Karma

isoutamo
SplunkTrust
SplunkTrust

techspec
Explorer

Thanks - still no luck. Even tried on an error message with no wildcards, just a simple match and replace:

Screen Shot 2020-10-09 at 12.34.49 PM.png

0 Karma

isoutamo
SplunkTrust
SplunkTrust
Try to put ‘ surround your field names.

techspec
Explorer

That did it - thanks!

0 Karma

isoutamo
SplunkTrust
SplunkTrust
Hi
Other option is use rex mode=sed
r. Ismo
0 Karma
Get Updates on the Splunk Community!

Updated Data Type Articles, Anniversary Celebrations, and More on Splunk Lantern

Splunk Lantern is a Splunk customer success center that provides advice from Splunk experts on valuable data ...

A Prelude to .conf25: Your Guide to Splunk University

Heading to Boston this September for .conf25? Get a jumpstart by arriving a few days early for Splunk ...

4 Ways the Splunk Community Helps You Prepare for .conf25

.conf25 is right around the corner, and whether you’re a first-time attendee or a seasoned Splunker, the ...