Skip to content

Latest commit

 

History

History
86 lines (63 loc) · 2.34 KB

README.md

File metadata and controls

86 lines (63 loc) · 2.34 KB
Raw

KBrowse

This is a tool for exploring data on an Apache Kafka cluster.

Screenshot

Why

Have you ever written a loop like this, for debugging purposes?

for record in topic:
    if record.key == "foo":
        print record

KBrowse provides this as a service, along with a web console.

Cached results can be shared by copy/pasting the URL.

Usage

Defaulting to localhost:

# Local data
./run-zookeeper-and-kafka

# Server
./lein run server

# Web Console
open http://localhost:4000

Using environment overrides:

KAFKA_TIMEOUT=30000 KAFKA_BOOTSTRAP_SERVERS='foo=foo.localhost:9092' ./lein run server

Different yaml config:

CONFIG=config/default-with-avro.yml ./lein run server

Uberjar:

# Build
./lein uberjar

# Run
CONFIG=foo.yml java $JVM_OPTS -jar kbrowse-*-SNAPSHOT-standalone.jar server

Docker:

# Build
docker build -t kbrowse .

# Run on default port 4000
docker run -p 4000:4000 kbrowse 

# Run with local custom config (custom.yml)
docker run -p 4000:4000 -v $PWD/config/custom.yml:/app/default.yml kbrowse

Tests

# Requires npm!
./run-tests

Deserialization

KBrowse deserialization options include String, Msgpack, and Avro (with schemas looked up from the KAFKA_SCHEMA_REGISTRY_URLS).

An attempt is made to parse the deserialized results into JSON, before streaming back as part of the response.

Stateful URL

The web console's state is stored as string-serialized JSON in the browser's window location. This is an effort to distinguish it visually from what might traditionally be a web form query. When the web console loads, the query state is initialized from URL state, and an initial query is sent to the server. If there are any cached results, they are displayed immediately.

This design is intended to support copy/paste-able URLs, so folks can share their findings, with the context baked in. This can save duplicated query time, and also allows verification of the query itself.

Some browsers may have a limit on the window location length, which could lead to a corrupt state. This trade-off is accepted, to allow for share-able contexts.

Kafka Version Compatibility

Kafka 0.9.0.1+

Browser Support

The web console uses Mozilla's Fetch API, and ES6 keywords such as let and const.

It is believed that this limits support to newer versions of Chrome, Firefox, and Safari.