Use case: alerting on individual issues per-attribute


Currently, the most granular we can alert on is individual issues every “Action Interval” of 5 mins.
This means that if I have multiple incoming events of the same type (same fingerprint) from multiple different servers within the same 5min window, they will send out 1 alert instead of multiple alerts.

For example:

  • Issue T from server A
  • Issue T from server B
  • Issue T from server C
  • Issue T from server C
  • Issue T from server C
  • Issue T from server D

If all of the above come within a 5-min window, we will notify once.
Instead, I would like to notify 4 times (once per server (A, B, C, D))

I would like my operations team to receive alerts on OpsGenie to fix issues when they happen on individual devices instead of treating the issues only from an aggregate point of view.

My use case is that I have 100s of IOT devices (for which I have remote access) that can fail in similar ways. I need operations to know when something fails on a specific device so that they can remote into it and fix it as soon as it fails on that device. If 2 errors are generated from 2 devices within the same 5-minute window, I might never know about the second one.

One way I found to go around this is to add the server_name to the fingerprint. However, this means I would lose the aggregate view.

I feel this is a non-standard use case for Sentry. It seems to me Sentry is all about tracking error type and less about tracking specific instances. I care about specific instances and would like to treat them as separate incidents which I can address individually.

Am I thinking about this correctly?