What coroot/coroot shipped

Coroot 1.29.2 is out. GOMEMLIMIT is now set automatically from the container memory limit, alerting rules accept 0s for for and keepFiringFor, and fixes cover Prometheus headers and series errors, chunk v4 encoding, and JSON encoding. New code is now checked by golangci-lint in C

Coroot 1.29.2 is released. GOMEMLIMIT is now set automatically from the container memory limit, so memory behavior follows the limit you already configure. Alerting rules now accept 0s for for and keepFiringFor in config. The release also includes fixes for custom Prometheus headers coming from the environment, the actual error returned by the Prometheus series endpoint, bounded label pair count and key index in chunk v4 encoding, and checked errors from JSON encoding. Vulnerable transitive dependencies were bumped to patched versions, and a golangci-lint config with a CI check was added so all new code is checked by the same rules.

Coroot 1.29.0 adds eBPF disk I/O profiles and groups pods by DisaggregatedSet and LeaderWorkerSet for llm-d and multi-node vLLM. 1.29.1 follows, attributing database metrics to instances by container_id.

Coroot 1.29.0 is out with eBPF disk I/O profiles and pod grouping by DisaggregatedSet and LeaderWorkerSet, covering llm-d and multi-node vLLM. Coroot 1.29.1 follows and attributes database metrics to instances by container_id, so database activity lines up with the instances running it. See the changelog for details.

Coroot 1.28.1 skips corrupted metrics cache chunks instead of failing the whole query. 1.28.0 groups KServe InferenceService pods and pods under non-controller owners into single applications, and disables SLO tracking for job-like apps.

Coroot 1.28.1 is out: corrupted metrics cache chunks are now skipped instead of failing the whole query. Coroot 1.28.0 disables SLO tracking and alerting for job-like apps, groups KServe InferenceService pods into a single application, and groups pods by a non-controller owner. A database fix closes check config rows on errors, so repeated failed reads no longer keep pooled connections open. Alerting also now counts events within KeepFiringFor and regroups unresolved Kubernetes events.

Coroot 1.27.3 adds support for Amazon ECS on EC2. 1.27.4 lets dashboards and alerting set a scope on PromQL queries, defaulting to global, and adds a MySQL hint for enabling performance_schema. Also fixed: service account creation with autofilled email fields.

Two releases this week. 1.27.3 adds support for Amazon ECS on EC2. 1.27.4 lets dashboards and alerting set a scope on PromQL queries, defaulting to global, and adds a hint for enabling performance_schema in MySQL monitoring. It also loads raw SLIs only for the last alert window and fixes service account creation failing when the browser autofills the email field.

Coroot 1.27.2 is out. Testing Slack, MS Teams, and webhook integrations now sends a test alert. The query state SQLite database opens in WAL mode. Trace tools fall back to eBPF spans when there are no OpenTelemetry traces. Docs drop read-only and do-not-check-slo flags.

Coroot 1.27.2 is out. Testing Slack, MS Teams, and webhook integrations now sends a test alert, so notification setup can be verified. The query state SQLite database opens in WAL mode with synchronous=NORMAL. Trace tools fall back to eBPF spans when there are no OpenTelemetry traces. Multi-cluster resolves external services to services on hosts and cloud instances of member projects. The docs drop the read-only and do-not-check-slo flags that are no longer in the code. 1.27.1 also updated RHEL 7 and CentOS 7 support notes, stated the minimum supported Postgres version on the Postgres page, and fixed time range selection on the profiling chart.

Coroot 1.27.0 is out. The Service Map now has a Topology mode that visualizes applications and dependencies as an interactive graph. Hover over services to highlight connections and see metrics.

Coroot 1.27.0 is now available. We've added a new Topology mode to the Service Map that renders applications and their dependencies as an interactive graph visualization. Hover over a service to highlight its connections and view key metrics alongside. This makes exploring complex service architectures much more intuitive. We also fixed time range selection in profiling charts and optimized log filtering with primary keys.

Coroot 1.26.8 is out with service accounts and API keys for programmatic access, plus expanded performance benchmarks for Postgres, MongoDB, and OpenTelemetry SDKs.

Coroot 1.26.8 brings service accounts and user API keys for programmatic access to the MCP endpoint and API, enabling better automation workflows. We've also significantly expanded our performance documentation with benchmarks for MySQL, Postgres, MongoDB instrumentation, and OpenTelemetry SDK overhead across multiple languages including C++. These additions help teams make informed decisions about monitoring impact.

Five releases this week: OCI and GCP integrations, RDS/ElastiCache support, remote Coroot TLS fixes, and WebSocket ClickHouse tunneling. Cloud observability expanding across major platforms.

Shipped five releases of Coroot this week spanning major cloud observability improvements. Added first-class support for OCI (MySQL HeatWave, PostgreSQL, Cache), GCP (Cloud SQL, Memorystore), and AWS (RDS, ElastiCache) with managed database details on node pages. Fixed critical TLS routing through remote Coroot instances and switched to WebSocket tunneling for ClickHouse. Also improved Kubernetes alert filtering and PostgreSQL vacuum horizon tracking.