Concurrent data inspect, Strimzi support, and consumer group improvements

Kpow 95.4 introduces concurrent data inspect queries, Strimzi support, and improved consumer group management and monitoring.

Data inspect improvements

Concurrent workers

Kpow's data inspect engine now supports concurrent access. If multiple team members use data inspect at the same time, you may have noticed occasional "query is queued" messages. This release addresses that.

By default, Kpow runs a single data inspect worker, meaning only one query can execute at a time. Other requests are queued until the worker is free. This is usually fine (queries complete quickly once the result limit is reached) but can become a bottleneck when multiple users are active or long-running queries monopolise the worker. The result limit is the first of SAMPLER_TIMEOUT_MS or 100 consumed records.

You can now increase the number of concurrent workers with the SAMPLER_N_WORKERS environment variable. The right value depends on your team's usage patterns. Each worker maintains its own consumer thread pool (default: 3 threads), so increasing workers will proportionally increase Kpow's resource usage and the number of concurrent consumers reading from your brokers. Worth considering if your cluster is already under significant load.

If the queue buffer is at capacity (default: 1000), requests are rejected and users will see a "Rejected" state for their query.

Data policies and String deserialization

Kpow now allows deserialization of string data even when matching data policies exist, provided the Auto deserializer is in use. The Auto deserializer exhaustively checks all structured message formats (JSON, Avro, Protobuf, etc.) before falling back to String, so the message is genuinely unstructured and not subject to any masking rules.

Base64 deserializer

Kpow now includes a Base64 deserializer in Data Inspect, making it easy for teams working with binary message formats to decode and view raw topic data directly. Like the String deserializer, Base64 is non-redactable if you have data policies configured.

Strimzi support

Kpow now supports connecting to Strimzi Kafka clusters. Strimzi users will need to pull a tagged version of the Docker image:

docker pull factorhouse/kpow:95.4-strimzi


Read more about our Strimzi integration in the docs.

Consumer group improvements

Offset management controls

You can now reset consumer group offsets to start, commit, or end positions with a single click, at either the topic or individual partition level. Previously this required manually entering offset values.

Improved lag calculation for empty consumer groups

We've improved lag calculation for Empty consumer groups. Previously, lag was only calculated for active groups, derived from active member assignments. Since Empty groups have no active members, lag is now calculated from all known assignments.

We've also fixed a bug where offset management actions were not visible for empty assignments in the UI.

Simple consumer groups

Kpow classifies a consumer as "simple" if the group uses manual partition assignment with no group coordination (e.g. legacy consumers). This release brings two changes to simple consumer support:

• Dimensionality: monitor simple consumer groups from Group, Topic, and Broker dimensions
• Offset management: now available for simple consumer groups from the UI and API

Read more about simple consumer management in the docs.