What's Changed

• docs: add guide for installing Coroot with ArgoCD by @def in #907
• Track total redis keys count and TTL in redis section by @vishnukumarkvs in #911
• logs: add "not contains" message filter by @def in #916
• logs: improve attribute filter suggestions for high-cardinality keys by @def in #917
• logs: prefetch filter names together with the logs by @def in #919

Full Changelog: v1.22.0...v1.22.1

ArgoCD monitoring

Coroot now monitors ArgoCD alongside FluxCD, showing the state of every application it manages right on the Kubernetes page.

It works out of the box: GitOps state is collected by the coroot-cluster-agent's embedded kube-state-metrics (read-only), with nothing to configure. Just upgrade the Coroot Operator to the latest version so the agent gets the read permissions it needs on the argoproj.io API group.

What's Changed

• feat: add ArgoCD monitoring to the Kubernetes page by @def in #904
• fix: allow editing SLO config when the app is missing from a stale world by @def in #905

Full Changelog: v1.21.1...v1.22.0

What's Changed

• fix wrong trace sample selected for latency SLO violations by @def in #903
• API keys dialog: submit on Enter by @Sudhanva-Nadiger in #902

New Contributors

• @Sudhanva-Nadiger made their first contribution in #902

Full Changelog: v1.21.0...v1.21.1

Inbound request instrumentation

Until now, Coroot relied only on outbound request instrumentation via eBPF. This was enough to automatically build the service map and understand service-to-service communications without any code changes.

Starting with this release, Coroot also instruments inbound requests. This means that traffic from external clients and load balancers such as AWS ALB is now visible in Coroot.

By default, SLIs for every service are now calculated based on inbound requests, providing a more accurate view of the actual user experience from the server side. For services that cannot be instrumented with eBPF, such as external services or managed databases, Coroot automatically falls back to outbound instrumentation on the client side.

What's Changed

• fix: update Slack integration setup docs by @immanuwell in #889
• fix(clickhouse): partition otel_traces_trace_id_ts by date so TTL reclaims disk by @def in #895
• fix: match only full ReplicaSet hash suffixes by @immanuwell in #896
• use inbound L7 metrics for SLI when available by @def in #897
• perf: stop loading RCA when listing incidents by @def in #901

New Contributors

• @immanuwell made their first contribution in #889

Full Changelog: v1.20.1...v1.21.0

What's Changed

• fix(logs): use indexed ServiceName column for k8s events queries by @Alegrowin in #883
• MCP server by @def in #885

New Contributors

• @Alegrowin made their first contribution in #883

Full Changelog: v1.19.8...v1.20.2

What's Changed

• document --instrumentation-delay flag for coroot-node-agent by @def in #880
• document --max-fqdns-per-container flag for coroot-node-agent by @def in #881
• fix profile query failing on clustered ClickHouse with double-distributed IN by @def in #882

Full Changelog: v1.19.7...v1.19.8

What's Changed

• reduce ClickHouse memory usage for profile queries by @def in #876
• resolve k8s Job owners by @def in #877

Full Changelog: v1.19.6...v1.19.7

What's Changed

• increase RCA timeout by @def in #875

Full Changelog: v1.19.5...v1.19.6

What's Changed

• add flamegraph widget by @def in #870
• materialize per-app DNS metrics as recording rules by @def in #871
• improve memory leak detection and add builtin alert by @def in #872
• reuse MemoryGrowthPct for deployment tracking by @def in #873
• reduce rr_connection_l7_requests cardinality by collapsing status to ok/failed by @def in #874

Full Changelog: v1.19.4...v1.19.5

What's Changed

• drop redundant auditor.Audit call in Incidents.Check by @def in #869
• update docs with new --ca-file flag by @vishnukumarkvs in #865

Full Changelog: v1.19.3...v1.19.4