+1 (binding)

I verified the release candidate on a local Docker-based cluster (ZooKeeper,
Kafka, Nimbus, Supervisor(s) and UI) running the RC distributions unpacked from
the staged tarballs. Actions taken:

# Signatures and checksums
- Verified GPG signatures and SHA-512 checksums for all artifacts in the RC
  directory: -src, lite and full, both .tar.gz and .zip, plus
RELEASE_NOTES.html.
  All showed a good signature from my code-signing key and matching checksums.

# Distributions
- Built and started a cluster from both the lite
(apache-storm-3.1.0-lite.tar.gz)
  and full (apache-storm-3.1.0.tar.gz) binary distributions. The UI reported
  version 3.1.0, with a revision matching the v3.1.0 tag.

# Topologies from storm-examples
- Ran example topologies built from storm-examples (storm-starter and
  storm-kafka-client-examples): WordCountTopology, MultiThreadWordCountTopology,
  LambdaTopology, and KafkaSpoutTopologyMainNamedTopics (Kafka producers plus a
  Kafka spout consumer).
- All reached ACTIVE and processed data, emitting and acking with no
failed tuples.

# Multi-node and inter-worker communication
- Ran with multiple single-slot supervisors so multi-worker topologies were
  forced onto separate containers.
- Confirmed workers spread across distinct hosts (for example, WordCount across
  three nodes) and that changing the worker count at runtime with "storm
  rebalance" works.

# Lite optional jars
- On the lite distribution, used bin/storm-kafka-monitor-fetch to fetch
  storm-kafka-monitor from the staging repository, restarted the UI, and
  confirmed Kafka spout lag is reported.

# Not tested
- Secure/Kerberos features (storm-autocreds).

Reply via email to