v0.2.2 -- 2026-08-21

v0.2.2 — 2026-08-21

Fixes the two GitLab release jobs that had never actually run: docker now creates its buildx context with the TLS certs named explicitly instead of relying on inheritance, and chart clears alpine/helm's image entrypoint so its commands stop being wrapped in an extra shell. v0.2.1 ships with no image or chart; this tag is the first one that produces both.

v0.2.1 -- 2026-08-19

v0.2.1 — 2026-08-19

Passive check results can now arrive over the ConnectRPC API in batches, not just through the FIFO. Results for objects owned by another engine are forwarded automatically, and contrib/ingest ships a reference recipe for SNMP traps and logs.

v0.2.0 -- 2026-08-19

v0.2.0 — 2026-08-19

Clustering continues (ADR-015 block 3c): a UI process started with -target ui now forwards commands to the whole engine fleet over jaque.work.cmd.ingress, so acknowledging or scheduling downtimes from the dashboard works regardless of which engine owns the object.

v0.1.0 -- 2026-08-19

v0.1.0 — 2026-08-19

Clustering continues (ADR-015 block 4a): perfdata sinks now run only on the engine that owns an object, avoiding duplicate writes, and can run as their own process with -target sink.

v0.0.13 -- 2026-08-19

v0.0.13 — 2026-08-19

Perfdata sinks gained a log archive sink and a remote_write sink (ADR-016), both usable together via multi-sink fan-out, plus a ready-made Grafana dashboard pack. Separately, the release pipeline now reads PR labels to pick the version bump and fails loudly instead of silently defaulting to a patch release.

v0.0.12 -- 2026-08-19

v0.0.12 — 2026-08-19

Clustering continues (ADR-015 block 3b): commands issued for an object are now forwarded to the engine that owns it over jaque.cmd.<engine-id>, so it no longer matters which engine received the request. A new jaque_objects gauge metric reports object counts per engine.

v0.0.11 -- 2026-08-19

v0.0.11 — 2026-08-19

Clustering continues (ADR-015 block 3): a coordination layer adds a KV store with TTLs, cluster membership tracking, and rendezvous hashing to decide which engine owns which object. Mostly internal plumbing that the rest of the cluster work builds on.

v0.0.10 -- 2026-08-18

v0.0.10 — 2026-08-18

Internal refactor release; no user-visible changes.

v0.0.9 -- 2026-08-18

v0.0.9 — 2026-08-18

Clustering continues (ADR-015 block 2): engines follow object ownership through a follower loop, and the dashboard can now run standalone as its own process with -target ui.

v0.0.8 -- 2026-08-18

v0.0.8 — 2026-08-18

First step of clustering (ADR-015 block 1): a work queue and remote runner let a binary run as a dedicated engine or worker via the new -target engine|worker flag, instead of always doing everything in one process.

v0.0.7 -- 2026-08-18

v0.0.7 — 2026-08-18

Notifications can now be delivered to Telegram.

v0.0.6 -- 2026-08-18

v0.0.6 — 2026-08-18

Downtimes can now be created directly from the Downtimes view in the dashboard, without going through the API or config.

v0.0.5 -- 2026-08-18

v0.0.5 — 2026-08-18

Documentation-only release; no runtime changes. ADR-015 was published, laying out the cluster topology design: engine roles, a pull-based work queue, and per-object ownership.

v0.0.4 -- 2026-08-17

v0.0.4 — 2026-08-17

The dashboard now shows the running jaque version alongside the project logo and favicon.

v0.0.3 -- 2026-08-17

v0.0.3 — 2026-08-17

The dashboard now opens on an overview landing page. You can also select multiple objects and apply bulk actions to them, and a new Downtimes view lists active and scheduled downtimes.

v0.0.2 -- 2026-08-12

v0.0.2 — 2026-08-12

The dashboard gained a host/service drilldown view, and labels shown in the UI are now clickable to filter the object list by them.

v0.0.1 -- 2026-08-12

v0.0.1 — 2026-08-12

Internal build fix release; no user-visible changes. The Docker web build stage now copies pnpm-workspace.yaml, fixing a broken container image build.