Alert not creating an incident for every violation

I would like to get a Slack alert every time a specific error is seen. For now, I am just testing this with SELECT count(*) FROM TransactionError , which triggers a few times an hour. However, these usually don’t cause a slack message (or create an incident at all) to be sent despite showing a violation in the graph. I have tried setting the offset window to 15 minutes and higher, and the INCIDENT PREFERENCE
is by condition AND signal. From what I’ve read these two things are the common cause of not seeing the incident being made. I have tried adjusting the other params as well, currently it is query returns a value “above 0 at least once in 1 minutes” with an aggregation window of 1 minutes and a offset of 10 minutes, no fill data for gaps. Any help appreciated![accountId]=1605532&pane=eyJuZXJkbGV0SWQiOiJjb25kaXRpb24tYnVpbGRlci11aS5jb25kaXRpb24tZWRpdCIsIm5hdiI6IlBvbGljaWVzIiwicG9saWN5SWQiOiIxNDUxOTM3IiwiY29uZGl0aW9uSWQiOjIyMDQ2OTg2LCJyZW5kZXJlZEluUGFuZSI6dHJ1ZX0=&sidebars[0]=eyJuZXJkbGV0SWQiOiJucmFpLm5hdmlnYXRpb24tYmFyIiwibmF2IjoiUG9saWNpZXMifQ==

Hey @ablum,

Thanks for reaching out.

Looking that the NRQL behind the alert condition for the past 1 week, I’m not seeing it violate the threshold (above 0 at least once in 5 minutes). You you have a NRQL example of when the condition violated the threshold, but didn’t generated a violation?


I have since changed the condition, it was previously just SELECT count(*) FROM TransactionError as stated in my post. The reason I changed it is because I got it to sort of work by creating an alert that is “by policy” that also auto closes incidents after 5 minutes (minimum amount of time you can select), and this does correctly create incidents but the granularity is very poor and it closes the incidents which is not ideal. I’d like an incident and corresponding slack message every single time a certain error is seen, and this is what I was unable to get to work – it rarely made incidents for the violations.

I have re-created what I want to have work and posted about so you can look at it:[accountId]=1605532&pane=eyJuZXJkbGV0SWQiOiJjb25kaXRpb24tYnVpbGRlci11aS5jb25kaXRpb24tZWRpdCIsIm5hdiI6IlBvbGljaWVzIiwicG9saWN5SWQiOiIxNDYzMzU1IiwiY29uZGl0aW9uSWQiOjIyMTA2ODI0LCJyZW5kZXJlZEluUGFuZSI6dHJ1ZX0=&sidebars[0]=eyJuZXJkbGV0SWQiOiJucmFpLm5hdmlnYXRpb24tYmFyIiwibmF2IjoiUG9saWNpZXMifQ==&state=e3d5e8f0-cddb-9c8f-2849-5c9e24465f13

Hi, @ablum: Yes, that is not how New Relic Alerts is designed to work. One of the main goals of Alerts is to avoid alert fatigue, that is spamming you with too many notifications. If you want a notification each time an error occurs, a better approach might be to write code to get a list of error occurrences, and send the Slack notifications yourself.