Last9
Book demo
Sep 1st, 2026

WebView pages as their own views, 14 days of alert history, and a shared environment in Discover

Improvements Fixes

TL;DR

Dashboard panels take descriptions and threshold bands, and wide tables in Logs, Traces and Metrics stop hiding the end of a value.

WebView pages as their own views, 14 days of alert history, and a shared environment in Discover

Improvements

  • Dashboards
    • Panel descriptions, written in the panel editor and shown in the panel info tooltip. An imported Grafana description is kept.
    • Threshold bands on timeseries and bar panels.
    • Gauge thresholds are configurable, instead of a fixed 85 and 95 percent band.
  • Discover
    • Each WebView page is now its own view, named by its URL and carrying its own duration, error count, and resource count, so the path a user took inside a WebView is visible. The new view.type=webview view replaces the webview_page_load span event, so move any dashboard or alert that queries it. Needs the Android or iOS RUM SDK at v1.6.8 or later.
    • The environment you select stays selected across every Discover screen, and an Env column appears when the selection is All, so you can see which environment a row belongs to.
  • Logs
    • Long cell values end in an ellipsis, and group-by and label columns scroll sideways. The Traces and Metrics Explorer tables behave the same way.
    • A recent log line is cut to fit its column, with the full text on hover.
  • Metrics: The aggregated-query table supports sideways scrolling, column resizing, and column sorting, with the full series name on hover.
  • Alerting
    • Click View firing history in the Rules table for a 14-day, hour-by-hour view of when a rule fired. Open any episode’s chart from there, and share the view by link.
    • A click on the Alert Group name filters the rules list to that group, instead of opening its settings dialog.
    • Every alert notification link opens that alert, ready to acknowledge or mute.

Fixes

  • AI Assistant
    • A question about logs over a wide time range stopped part-way and answered with internal wording such as “Log Query Budget Exhausted for This Turn”.
    • The assistant said it could not calculate a p99 from logs, and offered trace data for a service with no traces.
    • A p99 for a week or a month averaged the p99 of each interval, so it disagreed with the dashboard.
    • A question that asked for a root cause lost that intent, and came back with only latency numbers.
    • A question about service performance over a month or longer returned empty data or ended the session.
  • Dashboards: A short panel had no readable plot, and some queries lost the labels they group by.
  • Discover
    • View Logs did nothing for an ECS service with no OpenTelemetry service name.
    • The iOS Applications (RUM) SDK emptied streamed POST bodies during network instrumentation, which broke requests that other SDKs own, including Google sign-in.
    • An iOS event lost attributes when it fired many times in quick succession.
    • An Android release build with minification on failed to compile, with missing OpenTelemetry classes.
  • Alerting: Alerts sent to Zenduty were rejected with “summary cannot be empty”.
  • Grafana: An imported Grafana dashboard showed no data.
  • Synthetic Monitoring: A Cloudflare-hosted check dropped the HTTP authentication you configured and failed with a 401.