Adaptive alerts for exceptions

Feature Improvements Fixes

TL;DR

Exceptions now support adaptive alerting that learns normal patterns and catches anomalies without manual threshold tuning.


Discover Exceptions now support adaptive alerting. The system automatically learns normal exception patterns per service and operation, then alerts when counts deviate from the baseline. No manual threshold tuning required – just enable adaptive alerts from the Exceptions page and let the system learn your environment.

Key benefits:

  • Alerts fire on deviation from learned baselines, not static thresholds
  • Scoped per service and operation for precise, low-noise alerting
  • Works alongside existing adaptive alerts already available on Discover Services

Read more in our docs.

Improvements

  • Dashboards: Importing Grafana dashboards into Last9 now supports table, bar, heatmap, and status history panel types in addition to timeseries (line and area)
  • Discover
    • Operations and endpoints pages now display max response time alongside existing latency metrics, making it easier to spot outlier requests
    • Breakdown page now supports column sorting
    • Truncated operation names in the Top 10 Web Operations table now show the full value on hover
  • Logs: Related logs from traces now support downloading results for sharing or archival
  • AI Assistant: Deep links returned by the AI assistant now use clearer, more descriptive labels

Fixes

  • Discover: Exceptions page was intermittently returning incomplete results for the same time range, which also broke alert inspect links
  • Logs: The logs page was becoming unresponsive for certain query pipelines
  • Traces: Trace Details was hiding the root span when opened from Service Catalog if the root belonged to a different service