Lenses® for Apache Kafka
This is the official Docker for Lenses that enables you to:
- View and Query Kafka Topic Data (Avro & JSon) - Browse and query with Lenses SQL
- View data topologies and monitor every aspect of your Kafka cluster
- View and manage your Data Schemas
- Build and monitor ETL pipelines with open source Kafka Connectors
- Execute KStreams processors instrumented in SQL in Kubernetes or Connector workers
- Set up alerting and external notifications on liveniness of streaming systems
- Data governance: Auditing on all actions, data lineage & multi-tenancy
- Fine-grained security. Role based access with LDAP support
- Manage Quotas, ACLs
As a state-less application lenses fits naturally in containers and run on Kubernetes or Openshift. It integrates and helps you instrument and view your streaming data pipelines; as well as operate them with confidence !
The documentation is always available at https://docs.lenses.io and data engineering talk is on Discourse and Slack
The Docker Image
Please check out the docker image documentation at docs.lenses.io for the most recent docs and the complete set of features, settings and tweak knobs.
This image is aimed for our enterprise clients, though anyone with a free
developer license may use it. Visit
our download page to get a free developer
license or an enterprise trial. Only Lenses is included in this docker. Our
development environment image, which additionally includes Kafka, Connect,
Schema Registry and our open-source Stream Reactor collection of connectors can
be found as lensesio/box
.
This image has to be run alongside a Kafka cluster.
How to run
In the current iteration lensesio/lenses
uses environment variables as the
primary means for configuration and alternatively configuration files.
Setup with environment variables
For any lenses configuration option, set an environment variable by converting
the option name to uppercase and with dots replaced by underscores. As an
example lenses.port
should be converted to LENSES_PORT
. Optionally settings
may be mount as volumes under /mnt/settings
or /mnt/secrets
. As an example
you could set —file— volume /mnt/settings/LENSES_PORT
with the port number as
the content of the file.
A brief example of a docker-compose file to setup Lenses, would be:
version: '2'
services:
lenses:
image: lensesio/lenses
environment:
LENSES_PORT: 9991
# # Users are managed within Lenses. Here you can change the superuser username:
# LENSES_SECURITY_USER: admin
# # Users are managed within Lenses. Here you can change the superuser password:
# LENSES_SECURITY_PASSWORD: admin
ports:
- 9991:9991
- 9102:9102
network_mode: host
The docker image has two volumes where data are saved: /data/log
for logs and
/data/kafka-streams-state
for storing the state of Lenses SQL
processors. Depending on your queries and the topics volume, the latter can
become pretty large. You should monitor space and plan for adequate
capacity. Maintaining the streams state directory between Lenses restarts is
mandatory for the SQL processors to be able to continue from where they left.
The container starts with root privileges and drops to nobody:nogroup
(65534:65534
) before running Lenses. If you start the image as a custom
user:group
, it falls under your responsibility to make sure that the two
volumes are writeable by the custom user:group
.
Setup with configuration files
Lenses software configuration is driven by two files: lenses.conf
and
security.conf
. In the docker image we create them automatically from
environment variables but it is possible to set directly these files instead.
Create your configuration files according to
the
documentation and
mount them under /mnt/settings
and /mnt/secrets
respectively —i.e
/mnt/settings/lenses.conf
and /mnt/secrets/security.conf
. You can set either
one or both together. Please for lenses.conf
omit the settings
lenses.secret.file
. If by any chance you set them,
you have to make sure lenses can find the files described in these settings.
How to build
If you want to build the image yourself, you can just run:
docker build -t lensesiolocal/lenses .
If you are on an older version of Docker which does not support multi-arch builds, you can emulate a multi-arch build via args:
docker build \
--build-arg TARGETOS=linux --build-arg TARGETARCH=amd64 \
-t lensesiolocal/lenses .
For more information, please visit our documentation. Enterprise customers may use the support channels made available to them. Developer Edition users are encouraged to visit our slack community. We are always happy to help and hear from you.
With respect,
The Lenses Team.
Copyright 2017-2023, Lenses.io Ltd