new hosts get built-in CPU and memory safeguards that alert immediately, no baseline-learning wait. detectors need no history and fire the moment a value crosses a fixed, known-bad line, like a disk at 95% full or a container at 95% of its memory limit. While a brand-new host or pod is still building the history an anomaly detector needs, these threshold nets are already watching, so genuine saturation on a fresh entity is caught immediately instead of waiting for a baseline to form.

tuned anomaly detection so alerts are the ones worth acting on. detectors learn what normal looks like for each entity from its own recent history and fire only when a reading lands well outside that range, with a setting (High, Medium, or Low) to control how tightly that range is drawn. A breach also has to hold for about 30 minutes across several evaluations before it fires, so a brief spike settles on its own instead of paging anyone.

Check out the Smart Alerts guide for the full breakdown.

Check out the RUM Interface guide and Session Detail guide for the full details.

Understand how users move through your product.

Check out the RUM Interface guide for the full breakdown.

See exactly how people experience your apps, on web and mobile.

play back real user sessions with a smoother player, hour-long session support, and "skip inactivity" to reach the moments that matter. Playback runs from 1x up to 8x speed, and Skip inactivity jumps past the quiet stretches that make up most of a long session. Replay is tied to the same timeline as the rest of the session: clicking an error or a frustration click in the event list moves the replay straight to that moment.

new support for native iOS, React Native, and Expo apps, with the same insights you get on the web. Mobile apps now share the SDK's collector and the same RUM interface as your website, so a workspace can hold a website and its mobile apps side by side. Screen views, app start, crashes, and tap spans are captured on iOS and React Native the same way they already are on Android, with a guided setup for each platform.

accurately spot where users get stuck or frustrated. A rage click is three or more clicks on the same element within one second. A dead click is a click that produced no DOM change and no navigation within 500ms. An error click is a click followed by a JavaScript error within one second. All three show up as badges on the sessions list, as markers on the replay timeline, and as rows in the session's event list, so you can jump straight to the click that frustrated someone.

clearer charts and readable tables for how your pages and endpoints really perform. Performance now opens on Core Web Vitals at the 75th percentile, with a distribution view underneath that shows whether a handful of visitors had a bad time or everyone had a mediocre one. Slowest pages by impact ranks pages by total load time contributed rather than by the raw measurement alone, and an endpoints table puts latency and failure rate on the same row.

a step-by-step flow to add a new application in minutes. The Add application wizard walks through picking a platform (web, Android, or iOS) and generates a ready-to-use snippet with your key and current SDK version already filled in.

Check out the Instrumentation Guide for full setup instructions.

Alert rules can now catch an instance that stops reporting entirely, not just one that crosses a threshold. Turn it on per rule, and KloudMate remembers every instance it's seen, then alerts if any of them goes silent.

An instance is a unique series identified by its labels; group a heartbeat by host_name, and you get one instance per host; group by serviceName + pod_name and you get one per pod. Each is tracked individually, with its own state, history, row on the Instances tab, and dismissal. It's off by default and opt-in per rule, so nothing changes for existing rules until you turn it on. It works with any datasource.

How it fires: an instance missing from query results moves through the normal alert lifecycle, Pending, then Firing, with the reason instance stopped reporting; last seen <timestamp>. It recovers automatically the moment it reports again. Roughly, it fires the query window plus the pending duration after the last data point, so the window should be several times your reporting interval; a 60-second heartbeat with a 60-second window will flap on one late data point; a 5-minute window absorbs the jitter.

Check out the Instance Absence Detection documentation for full details, including how it interacts with routing rules, folders, silences, and maintenance windows.

Kubernetes Monitoring now includes a Storage tab showing Persistent Volume Claims across your clusters, giving visibility into storage alongside your existing pod and node metrics.

Each row shows the Persistent Volume name, cluster, status, claim name, storage class, and capacity. Filter by Persistent Volume type, cluster, or status, or search by name.

This should make storage issues visible in the same place you already check pod and node health, instead of needing kubectl to see what's bound, how much capacity is used, or what's stuck pending.

Check out the Kubernetes Monitoring documentation for full details.

Sharing a dashboard now works the same way no matter where you start it; sharing from the dashboard list and sharing from an individual dashboard page both open the same Share dialog with both options available, instead of the list forcing an immediate JSON copy while the dashboard page only offered a public link.

Check out the Dashboard Details documentation for how sharing works.

Correlates related alerts into a single incident instead of flooding the team with separate alarms. It's opt-in per routing rule: switch a rule's Alert grouping setting to , and the engine takes over from there, learning which alerts tend to fire together over time. No grouping keys to set.

When multiple alerts fire close together, the engine works down a signal ladder to decide whether, and why, they belong together, checked in order:

Each group shows its members and a reason. You can thumbs-up or thumbs-down a grouping decision; a single thumbs-down is a one-off override, but a second person flagging the same pairing stops it from grouping that way again. Groups link directly into for investigation.

Check out Alert Grouping: Static or Auto (AI) and How Alert Grouping Works for full details.

When a panel was built using a custom data source variable, reopening it for editing showed the data source correctly selected but left the query field empty, even though a valid query had been saved. This made a working panel look unconfigured and risked someone overwriting it with a blank query on save. Both the data source and the previously saved query now reload correctly in edit mode.

When building an alert query with two or more dimensions (e.g., + ), the preview concatenated the labels into one unreadable string instead of showing them separately, also affecting Dashboards and the Explore page. This made it impossible to verify a multi-dimension query was configured correctly before saving. Labels now display cleanly and separately per dimension.

Clicking Download on any report failed silently, with a blocked CORS request in the browser network log, meaning no reports could be exported from the UI at all (emailing the report was the only workaround). The cross-origin request issue is fixed, and downloads now complete normally.