feat(default detector): create default detector for new projects #104208
Conversation
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## master #104208 +/- ##
===========================================
+ Coverage 80.50% 80.61% +0.11%
===========================================
Files 9342 9344 +2
Lines 399240 400731 +1491
Branches 25560 25560
===========================================
+ Hits 321408 323056 +1648
+ Misses 77383 77226 -157
Partials 449 449 |
23d77c0 to
2d38c85
Compare
roggenkemper
left a comment
There was a problem hiding this comment.
is there a test we can add for this?
| detector = _ensure_metric_detector( | ||
| project, owner_team_id=owner_team.id if owner_team else None, enabled=enabled | ||
| ) |
There was a problem hiding this comment.
Is this so that an org on a plan w/o anomaly detection would have the default detector in the case that they do upgrade? FYI I have a task in my team's backlog to address an edge case I think you're likely to hit here - Seer won't keep data on detectors they're not receiving updates for over 90 days for, so when they upgrade the alert might not work. Here is the task: https://getsentry.atlassian.net/browse/ACI-528
| assert detector.enabled is True | ||
|
|
||
| def test_send_new_detector_data_failure_does_not_block_creation(self): | ||
| """Test that detector is still created even if sending data to Seer fails.""" |
There was a problem hiding this comment.
I don't think you'd actually want this to happen - if sending the data to Seer fails then it will not function. However, it might be worth checking with the Seer team since as I mentioned in a previous comment a new project isn't likely to have historical data anyway so maybe it is ok since we wouldn't expect this monitor to function for 28 days - I can't remember if it'd be ok on their end if they suddenly start getting data on a monitor they don't have any previous data for though, I think they might rely on having the previous historical data though.
There was a problem hiding this comment.
good catch - I will have it skip creating a detector if sending data to seer fails for now and follow up with seer team
Co-authored-by: Colleen O'Rourke <colleen@sentry.io>
Co-authored-by: Colleen O'Rourke <colleen@sentry.io>
Co-authored-by: Colleen O'Rourke <colleen@sentry.io>
create a default metric monitor for new projects
with these settings:
event.type is errorFeature Flag
organizations:default-anomaly-detector