Configuration & connecting

flinkit connects to clusters through named profiles in a YAML config file. Define a profile per cluster once; select it by default or with an environment variable.

The config file

By default flinkit reads ~/.flinkit/config.yaml. It has two top-level keys: default, naming the profile to use when none is specified, and profiles, a map of profile name to connection settings.

~/.flinkit/config.yaml

default: prod-oss

profiles:
  prod-oss:
    source: oss
    platform: oss
    endpoint: https://flink.internal.example.com

  local-oss:
    source: oss
    platform: oss
    endpoint: http://localhost:8081

Profile fields

Field	Description
`source`	Where job metadata comes from (e.g. `oss` for the Flink REST API).
`platform`	The platform flinkit is talking to — `oss`, `kubernetes`, or `confluent` — which tunes how it reads the cluster.
`endpoint`	The JobManager REST endpoint, e.g. `http://localhost:8081`.

Run flinkit --help for the complete, version-accurate list of config fields and flags for your build.

Pointing at an alternate file — `FLINKIT_CONFIG`

Set the FLINKIT_CONFIG environment variable to make flinkit read a different config file instead of ~/.flinkit/config.yaml. This is handy for per-project configs or for keeping a demo cluster separate from your real one:

shell

# for this shell
$ export FLINKIT_CONFIG=$PWD/flinkit.config.yaml
$ flinkit scan

# or just for one command
$ FLINKIT_CONFIG=$PWD/flinkit.config.yaml flinkit scan

Connecting to each platform

Open-source Apache Flink

Set platform: oss and point endpoint at the JobManager REST API — the same host and port as the Flink web UI (8081 by default).

Kubernetes — Flink Operator

For jobs managed by the Flink Kubernetes Operator, flinkit reaches the JobManager REST API through the Kubernetes API-server proxy, so you don't need a separately exposed ingress. Set platform: kubernetes and point the endpoint at the proxied JobManager service; flinkit uses your current kubecontext for access.

Exact endpoint and access details depend on your cluster. If you're in the beta and want help wiring up Kubernetes, email hello@flinkit.org.

Confluent Platform & Confluent Cloud

Set platform: confluent to diagnose Flink jobs running on Confluent Platform or Confluent Cloud. flinkit speaks to the Confluent Flink endpoint and reports the same findings as it does for open-source Flink.

Confluent connection and credential specifics vary by deployment — reach out at hello@flinkit.org for the current setup steps.