Flink Manager 0.0.1 is out

Today we are releasing Flink Manager 0.0.1 — the first public version of our management console for Apache Flink clusters. Flink is a phenomenal stream processing engine, but running it day to day means juggling the JobManager REST API, the stock dashboard, JAR uploads from a terminal, and SQL Client sessions that live in a shell window. Flink Manager puts all of that behind one console — for the admins who keep the cluster healthy and for the data teams who ship streaming jobs on it.

For Flink admins: operate the cluster with confidence

If you are the person on call for the Flink cluster, 0.0.1 gives you a live operational picture without touching a REST client:

Cluster and TaskManager visibility. The cluster overview shows running jobs, slots, and resource usage at a glance. Every TaskManager is listed with its slots, CPU cores, task heap and memory — and its log files can be opened directly in the browser. No SSH hop just to read a log.

Job lifecycle, end to end. Run, stop and restart jobs with savepoint support — the savepoint directory, drain and format defaults are configurable. Before you even submit a JAR, the job plan preview shows how Flink will execute it: the topology, operators, and shuffles, drawn as a graph. After submission, the job detail view follows the job through its whole lifecycle with per-operator metrics.

JAR management in the console. Upload job JARs, list and delete them, and inspect the detected entry point — all from the browser instead of the upload script.

Operations is more than watching dashboards: it is knowing who did what, when, and being notified when it matters.

Governance and audit. Role-based access control ships with four seeded roles — admin, operator, developer, viewer — and users can authenticate with local accounts, LDAP, or OIDC (provider secrets are encrypted at rest). Every job operation, JAR upload, SQL statement, and configuration change lands in an append-only audit trail, with dedicated job, SQL, and JAR history views. Inbound HTTPS with optional mutual TLS is built in.

Schedulers and notifications. Background schedulers collect metrics and purge expired data on configurable intervals, and persistent notifications surface job and system events — so the console tells you what happened, even after you closed the tab.

For the data team: SQL that actually runs on Flink

Writing Flink SQL usually means the SQL Client in a terminal — no autocomplete, no catalog browsing, no history. The Flink Manager SQL workbench, backed by the Flink SQL Gateway, moves that workflow into the browser:

Open a named session, browse catalogs and tables in the explorer, and write your statements in the editor. DDL runs directly: a CREATE TABLE returns a result panel with the outcome, the schema, and the raw JSON, with the execution time next to it.

The moment that matters most: an INSERT INTO is not a simulation — it submits a real Flink job. The result is the submitted job id, and the job appears in the jobs list with a SQL source badge and the originating statement. From there, admins and developers look at the same execution graph, in the same console. Saved queries, session and operation history, and query history keep the team's work traceable.

Metrics that mean something

Raw metrics are easy to collect; useful metrics are curated. Flink Manager keeps a metrics history with time-window navigation, pulls reporter metrics from JMX or Prometheus endpoints, and maintains a metric dictionary with warning and critical thresholds — those thresholds drive the cluster health widgets, so green/yellow/red reflects rules your team actually tuned.

And more

0.0.1 also includes a connectors view that detects what your DataStream pipelines and SQL tables use, a visual flow designer for assembling pipelines from sources, transformations, and sinks, and dataset management — the designer and datasets ship as preview features in this release.

Availability

Flink Manager 0.0.1 is available as a Docker image and as a release bundle. The full details — including the complete 0.0.1 feature list — are in the changelog, and the documentation covers everything from the first login to production configuration.

Want to try it on your own cluster? Request a free trial license — it is issued instantly and kept renewed by Vinteros Labs, so you can evaluate the latest release without any paperwork.