What openobserve/openobserve shipped

This week: a Firing tab for triaging live alerts across folders, SLO import from JSON, a decluttered dashboard toolbar, and phase 0 of an enterprise Security solution. Fixes for fractional epoch timestamp precision, top-N aggregate error reporting and custom log download ranges.

This week's updates to the repository cover both new surfaces and correctness fixes. Alerts gain a Firing tab for triaging live alerts across all folders, with a triage drawer and history, and SLOs can now be imported from exported JSON. The dashboard view toolbar was regrouped so frequent actions stay one click away. Phase 0 of an enterprise Security solution adds a header switcher between Observability and Security with new Overview and Events pages. On the fix side, fractional epoch timestamps keep their sub-second precision, a deleted stream no longer causes other streams in the same memtable to lose data during persistence, top-N aggregate errors are now reported instead of returning no rows, and custom log downloads honour the chosen range when the query has a LIMIT clause. RUM sessi

This week: a Dependencies view for Kubernetes workloads and hosts from eBPF network flows, Heatmaps under Product Analytics, and drill down presets in Traces. Also many fixes for flaky tests and CI reruns.

This week for OpenObserve: a new Dependencies view under Infra draws which Kubernetes workloads and Linux hosts talk to each other and how many bytes per second flow between them. Product Analytics gained a Heatmaps sub-tab that rebuilds a page from session replay and ranks the most-clicked elements. Traces added drill down presets, new-tab opening from the results list, and honest list and URL state. Logs got a steadier first paint and a status bar that settles once. Much of the work also went into fixing flaky tests and CI rerun behaviour that was blocking unrelated pull requests.

This week: a state-timeline dashboard panel type, fleet management for OpenTelemetry collectors, per-resource and per-span service_name in OTLP ingest, and a spawned KV watcher so nodes drop stale cache entries. Also report history and alert calibration fixes.

This week in OpenObserve: dashboards gained a state-timeline panel type that draws one horizontal band per entity and colours each stretch by the state in effect. Fleet management for OpenTelemetry collectors landed with shared types, storage, HTTP surface and a web UI. OTLP trace ingest now scopes service_name per resource and per span, fixing wrong attribution when one request carries more than one resource, and the KV watcher is now spawned so other nodes drop stale cache entries after an overwrite or delete. Report history fixes correct weekday rendering, drawer timezone and pausing exhausted one-time reports, and alert calibration lists runs again on digest streams created without the removed coverage column.

OpenObserve v1.1.1 is out. This week: scheduled reports get run history, last-run status and run now; webhook alert templates can send the complete row; on-call policies page by SMS and voice. Also fixed: PromQL alert panics on empty cached results, dead cluster peers kept foreve

OpenObserve v1.1.1 is out. This week's changes include run-by-run history, last-run status and run now for scheduled reports; complete-row payloads in webhook alert templates via {...row}; and SMS plus voice call as on-call paging channels on escalation policies. On reliability, PromQL alerts no longer panic when the result cache trims a series to no samples, cluster nodes now drop peers that fail health checks and were never healthy, and alert notification send errors no longer carry destination URLs that act as credentials for webhooks. Access checks also tightened: downtime previews hide resources a user cannot read, and function updates can no longer switch a pipeline's function to JavaScript.

New installs start without setting an ext auth salt. Stream read access is checked when alerts, SLOs and reports are saved. An unsupported meta store is rejected instead of falling back to sqlite. PromQL matches Prometheus on NaN, quantile and rate cases. CI gates OSS merges on o

This week's changes include a fix so a new install generates and stores the ext auth salt on first start instead of refusing to start until it is set, ported to the 1.1.0 branch as well. Stream read access is now checked when alerts, SLOs, reports and similar background objects are saved, enabled or run. An unsupported meta store or cluster coordinator value is rejected instead of falling back to sqlite. PromQL now matches Prometheus on range-function NaN and overflow handling, quantile, histogram_quantile and rate extrapolation. A CI workflow also fails an OSS pull request merge while a same-named enterprise pull request is open. Alert saving now asks which query mode to use when more than one is in play.

This week: RUM product analytics with funnels and paths, 30x faster stream deletes on long retention, PromQL gauge fixes, function import/export, and improved anomaly bands on charts.

OpenObserve shipped significant improvements this week across multiple areas. RUM now includes product analytics with funnels, paths, and retention tracking. Stream deletion performance improved dramatically for clusters with long retention periods. PromQL queries are more correct with proper series handling in aggregations. New features include function import/export for sharing and anomaly value bands with configurable visualization. Short links work correctly with custom base URIs, and session replays load progressively.

V1.0.4 is out. PromQL subqueries, time(), and vector matching now work like Prometheus. Exemplars link to traces. Large integers preserve precision. Better audit trails and compliance controls.

OpenObserve v1.0.4 ships. PromQL evaluation now matches Prometheus semantics—subqueries evaluate correctly, time() returns per-step values, vector matching handles ignoring() properly. Search results preserve large integer precision. Exemplars on PromQL charts link directly to traces. Alert destination deletes guarded against all consumers. Audit trails and retention controls tightened for compliance.

V1.0.3 ships PromQL fixes for absent_over_time, predict_linear, scalar, and range selectors. HTML injection protection for dashboard tooltips. Compactor now streams merged files to disk. OTLP gRPC handlers properly acknowledge failures.

V1.0.3 brings critical PromQL corrections, dashboard security hardening, and compaction efficiency improvements. PromQL now correctly handles absent_over_time, predict_linear, scalar functions and range selectors. Dashboard tooltips are protected against HTML injection from metric labels and query values. The compactor streams merged files to disk instead of buffering, reducing peak memory. OTLP gRPC handlers now properly report export failures.