sentry
sentry copied to clipboard
ref(backpressure): Track reasons for unhealthy consumers and services
This refactors backpressure management so that when a service exceeds its memory threshold or there's an exception when checking, that information is attached to the Sentry error for that service. Likewise, when a consumer is unhealthy, the error contains the list of unhealthy services.
This doesn't change the logic of backpressure management in any way, only the presentation. Unless I've introduced a bug somewhere, which I would obviously never do.
Codecov Report
Attention: 11 lines in your changes are missing coverage. Please review.
Comparison is base (
70898e1) 81.37% compared to head (f7ffda9) 81.39%. Report is 50 commits behind head on master.
Additional details and impacted files
@@ Coverage Diff @@
## master #64998 +/- ##
==========================================
+ Coverage 81.37% 81.39% +0.01%
==========================================
Files 5242 5251 +9
Lines 231348 232102 +754
Branches 45384 45573 +189
==========================================
+ Hits 188266 188909 +643
- Misses 37230 37303 +73
- Partials 5852 5890 +38
| Files | Coverage Δ | |
|---|---|---|
| src/sentry/processing/backpressure/memory.py | 75.67% <87.50%> (+1.38%) |
:arrow_up: |
| src/sentry/processing/backpressure/monitor.py | 61.11% <76.92%> (+0.54%) |
:arrow_up: |
| src/sentry/processing/backpressure/health.py | 84.05% <65.00%> (-1.91%) |
:arrow_down: |
Suspect Issues
This pull request was deployed and Sentry observed the following issues:
- ‼️ KeyError: 'total_system_memory'
backpressure.monitoringView Issue
Did you find this useful? React with a 👍 or 👎
PR reverted: 956db53336cc15a019ea84aad6948a4d390673ce
I think reverting this might have caused new monitor failures: https://sentry.sentry.io/issues/?project=1&query=is%3Aunresolved&referrer=issue-list&sort=date&statsPeriod=1h