Improvements
- AI Assistant: Added an Ask Last9 AI action to the rows of the Firing Alerts, Degraded Services, Unhealthy Workloads, and Exceptions blocks on the Overview. It opens the assistant in a new tab with that row as the question, and it is disabled when AI is turned off for the organization.
- Dashboards
- Added a 1h auto-refresh interval. The shortest interval you can pick grows with the time range: 30s up to 1h, 1m up to 4h, 3m up to 1d, 10m up to 2d, and 1h beyond that. Refreshing pauses while the browser tab is hidden.
- Dashboards save their auto-refresh interval, so it applies the next time anyone opens them. An
autoRefreshvalue in the URL overrides it. See Using dashboards.
- Discover
- Exceptions: Added a Location row to the exception details header with the file, line, and function of the latest occurrence, when its traces carry
code.*attributes. See Discover Exceptions. - Hosts: Added a Logs tab to Host Details with the logs for the host’s Kubernetes node. It appears when the host’s metrics report a single node name; otherwise the tab explains why no logs can be matched. See Discover Hosts.
- Databases: Added an Activity tab under Queries in the details of a PostgreSQL database. It lists query samples, explain plans, and waiting and blocking sessions, and Blocking Only is on by default. Select a row to open the waiting and blocking sessions, the plan, and the recorded details. See Discover Databases.
- Exceptions: Added a Location row to the exception details header with the file, line, and function of the latest occurrence, when its traces carry
- Logs: In the Attributes tab of a log line, the value of the primary trace ID column is labeled TraceId instead of
trace_id. - Alerting
- Monitor: For rules created from a source chart, the alert details chart and the condition and worst-window values show the chart’s unit, such as
479.29 ms. - Rules: Added a Unit field to Rule Settings in the alert rule editor for metric rules. A rule created from a source chart inherits the chart’s unit and the field is locked. See Alert rules.
- Monitor: For rules created from a source chart, the alert details chart and the condition and worst-window values show the chart’s unit, such as
- Control Plane: Added a range select to the Usage Report: Last 30 Days (default), Last 90 Days, Last 6 Months, and Last 12 Months. Downloads accept a custom range of up to 366 days, and ranges over 31 days download as daily totals.
- AI Settings: Added a Describe images and charts checkbox when you add a document to Knowledge. It is off by default; when ticked, AI describes the images and charts in the file. See Knowledge.
Fixes
- AI Assistant
- A question such as “what is the p95 latency” was marked Partial result for missing traces, even when metrics fully answered it.
- “Check active alerts” was marked Partial result because the current alert state was checked against a one-hour window.
- Alerting Rules
- A rule with a negative threshold such as
expr < -16was evaluated asexpr < 16, and the Overview Firing Alerts block and the Alert Monitor failed with “Something went wrong” while such a rule was firing. - Validating an alert expression with two or more indicators always failed.
- The Alert Rules table was missing the View firing history button, the group-name links, and sorting by Severity, Type, and Labels.
- A rule with a negative threshold such as
- Discover ECS: The logs link of an ECS service without an OTel service name opened a search that returned no logs.
- Settings
- Users: Inviting users from a profile with no name created the invitations but failed to send the emails.
- Users: Granting a role, or admin access, again to a user whose access was revoked returned success but left the user without access.
- General
- Icon columns in tables cropped their icons, including the status columns of Alert Rules, Exceptions, Kubernetes events, and Synthetic checks.
- Confirmation dialogs animated in from the top-left corner instead of the center.