What’s new in Liberator 2.2
Liberator 2.2 focuses on observability and operability. The release adds an in-product System Monitoring page for super-admins, a standardized way to plug external Grafana instances into the underlying metrics, scheduled cache creation so cold-start latency is gone for predictable access patterns, and the foundations of an ETL on-the-fly dataset mapping workflow. It also lands a reliability fix that prevents long-running query workers from getting stranded across platform restarts.Highlights
In-product System Monitoring
A new admin-only page surfacing cluster health, queue depth, dataset access patterns, long-running queries, and license utilization.
Grafana integration
A Bearer-authenticated, read-only metrics endpoint at
/metrics-api-bearer/, with token issuance and revocation built into the Liberator UI.Scheduled cache creation
Pre-generate dataset caches on a schedule so users never hit a cold-cache penalty for predictable access patterns.
ETL mapping UI
Define cross-dataset value correspondences and enrich queries with
translate=true.Reliable query workers
Query workers no longer get stranded on stale connections when the platform restarts behind them.
UI polish
System Monitoring’s Usage tab uses its own license-appropriate time window, and tab state is preserved as users navigate.
System monitoring
A new super-admin-only System Monitoring page consolidates the operational signals previously scattered across logs and ad-hoc queries. Five tabs:| Tab | Time-range selector | What it answers |
|---|---|---|
| Cluster | 6h / 24h | Is the cluster behaving normally right now? |
| Queue | 6h / 24h | Is anyone being blocked, and by whom? |
| Datasets | 6h / 24h | What are users actually querying? |
| Long Queries | 6h / 24h | Which individual queries are slow enough to warrant attention? |
| Usage | 1d / 1w / 1m / 1y | How close are we to our license caps? |
Grafana integration
CloudQuant Data Liberator now exposes a standardized, Bearer-authenticated metrics endpoint athttps://<your-liberator-host>/metrics-api-bearer/. The endpoint speaks the full standard Prometheus HTTP API surface (/api/v1/query, /api/v1/query_range, /api/v1/series, /api/v1/labels, /api/v1/label/<name>/values, /api/v1/status/buildinfo, /federate), so any off-the-shelf Prometheus client — Grafana, Mimir, federated Prometheus, custom scripts — can consume it without modification.
Token lifecycle is fully self-service from the Liberator UI:
- Issue — a super-admin clicks Generate token in the Grafana Integration dialog. The full token value is shown exactly once.
- Use — point Grafana at
/metrics-api-bearerwithAuthorization: Bearer cqm_<token>as a custom HTTP header. - Rotate — issue a new token, switch Grafana to it, verify, then revoke the old one (zero-downtime).
- Revoke — same dialog, list view.
Scheduled cache creation
For datasets with predictable access patterns (daily morning reports, weekly market-open snapshots), Liberator now supports scheduled cache creation. Super-admins set a cron-style schedule per dataset directly in the Liberator UI’s caching settings; the platform pre-generates the cache slice before users query it, so the first read of the day hits a warm cache rather than triggering a generation. See the cache pre-generation guide for the full Super Admin workflow.ETL mapping UI
Liberator 2.2 lays the foundation for the ETL on-the-fly workflow, letting analysts define cross-dataset value correspondences (e.g. “theticker column in Dataset A maps to the symbol column in Dataset B, with these reconciliations”) and have Liberator surface a virtual unified column across both datasets at query time, without a heavyweight pre-materialization step.
What lands in 2.2:
- ETL mapping UI design — UX research and component specs.
- Data Translation Mapping — the underlying admin-API surface for storing and retrieving mapping definitions.
- Dataset Field Mapping UI — the in-Liberator-UI workflow for defining correspondences and previewing the virtual unified column. See Dataset field mapping.
Reliability improvements
A user-facing reliability bug is fixed in 2.2: previously, after a behind-the-scenes platform restart, a small number of long-running queries could hang indefinitely rather than reconnecting. In 2.2 those workers detect the broken connection within a short keepalive window and reconnect automatically, so long-running queries continue cleanly across platform restarts instead of stalling.UI changes
- Usage tab time-range selector. The global
6h / 24hselector is now hidden when the Usage tab is active. Usage has its own1d / 1w / 1m / 1ywindow because license utilization is measured against per-contract caps that don’t make sense on a 6-hour window. - Controlled Tabs. The System Monitoring page’s tab component is now fully controlled, so per-tab state survives switches without remounting, and the token-rotation flow triggers a clean cache reset across all tabs.
Also in 2.2
Smaller items in the same release:- OneDrive URL share-link paste — paste a full OneDrive or SharePoint share URL into a connection form; see SharePoint and OneDrive.
- Email verification — new accounts receive an email verification message when provisioned.
- Connection-form example text — UX hints in connection-source forms so users know what shape of input each field expects (host, bucket name, region, etc.) before they make a mistake.