Skip to content

2026.09.1 — 2026-09-05

Added

  • Slow-consumer detection. A consumer that is attached but not draining is now visible two ways. Studio surfaces the broker's own CONSUMER_SLOW notification on the consumers event topic — the only source that can name the individual consumer — and, for brokers where native detection is not configured, a new ackRatePerConsumer alert metric you can threshold yourself. It only evaluates queues that have consumers attached, a non-zero backlog, and are not paused, so it does not page on idle queues. No rule is created for you: the threshold is workload-specific, so the rule form offers a prefilled template instead. Studio's own detection resolves to a queue on a node, never to one consumer, and says so.
  • Broker configuration diff across a pair. A new Config diff view per cluster compares two nodes' effective configuration — broker attributes, address settings, security settings and acceptors. Differences that are correct by design (a broker's name, its HA policy, node-local paths, acceptor host and port) are shown as expected, and runtime counters as unclassified, so a healthy primary/backup pair reports no drift instead of two dozen false positives. Nothing is hidden: every key either node returns appears somewhere. If either node cannot be read, Studio says so and shows no comparison rather than a half-diff in which the unreachable node's keys look deleted. Read-only, at the same permission as the topology view.
  • Payload inspection in the message detail panel. A browsed message's body now reports its detected format, and JSON and XML are pretty-printed and syntax highlighted, with a formatted/raw toggle plus copy and download. Binary bodies are shown as a hex + ASCII dump instead of being decoded into mojibake, and gzip, zip, Java-serialized and Avro payloads are named. A body the broker truncated says so, rather than reporting your payload as malformed. Very large bodies are shown unformatted with a note, so the panel never blocks on a multi-megabyte message.
  • Syntax highlighting now works throughout the app. No highlighter was ever mounted, so every broker.xml snippet Studio shows — including the one next to a truncated message — rendered as plain text. It is loaded lazily, so first paint is unaffected.
  • The capability list gains a slow-consumer detection row. On brokers that do not expose slow-consumer-threshold over management — which is all of them today — it reports unknown rather than guessing, with the broker.xml to enable it.
  • A message's type now reads text or bytes rather than type 3.

Fixed

  • Topology view. The band carrying each pair's shared NodeID was drawn at a fixed position over the canvas rather than attached to the nodes it grouped, so it lined up only by coincidence and slid out of place on the first pan or zoom. Each logical node is now a real group containing its endpoints. Split-brain reads as two boxes above one group's line instead of a colour change on a detached rule.
  • Topology status marks are told apart by shape — filled disc, hollow ring, half-filled, dashed ring — instead of by two shades of grey eight pixels apart.
  • A two-node cluster is no longer magnified to fill the frame at wide viewport sizes.
  • The topology canvas gains zoom controls, a legend, a keyboard focus indicator, a loading placeholder the size of the graph, and an explanation when a cluster has no nodes yet.
  • The "add a management URL" prompt on a discovered-but-unreachable node was a permanently disabled button. It now opens the dialog that adds the URL.
  • The topology graph re-fits after a failover instead of leaving a stale viewport.

Changed

  • queue_snapshot gains a paused column so paused queues can be excluded from slow-consumer detection. Applied automatically on startup; no action needed.

Apache-2.0. Apache ActiveMQ and Apache ActiveMQ Artemis are trademarks of the Apache Software Foundation. Artemis Studio is an independent project, not produced by, endorsed by, or affiliated with the ASF.