Apache NiFi 2.10.0 - Flow Registry Branch Support

Apache NiFi rel/nifi-2.10.0 was published on June 23, 2026 as NiFi 2.10.0, with branch creation support in Registry Clients as the clearest change for teams that manage versioned flow deployments. This is not a prerelease, and the release also groups JSON reader and writer flexibility, Content Repository cleanup, parameter context references, and cluster safety fixes into one operator focused update.

The full release notes and downloads are on the GitHub release page.

Registry Clients now support branch creation. For NiFi environments that use a git based Flow Registry Client, the release also adds support for recording both Author and Committer. That matters when flow changes move through code review, service accounts, or automated promotion jobs, because authorship and commit metadata are not always the same signal.

The release notes also call out UI support for Connectors. The notes do not spell out a new connector data contract, so platform teams should treat this as a surface area change and validate any internal guidance around connector setup.

NiFi 2.10.0 improves the leniency of JSON readers and the flexibility of JSON Writer. The release notes do not enumerate each parser case, but the direction is useful for ingestion pipelines that receive records from third party APIs, SaaS exports, and event streams where field shape or escaping is not perfectly stable.

Less strict reading can keep a flow moving, while writer flexibility helps downstream systems get the structure they expect. The tradeoff is that leniency can hide source quality problems. Teams that rely on schema strictness should make validation explicit in the flow instead of assuming the reader will be the only guard.

Partial defragmentation of the Content Repository via tail claim truncation is the maintenance item to note. It points at storage efficiency and repository housekeeping, which are not exciting but do affect long running clusters that move large FlowFiles all day. This is a good release to watch repository metrics after upgrade, especially content claim counts, disk pressure, and queue backlogs.

Parameter value references can now target parameters from inherited parameter contexts. That is a practical improvement for shared environment settings. Common values such as host names, bucket names, and endpoint paths can live in a parent context while process groups keep their local overrides.

The release also removes Restricted Component Authorization from the framework and removes the Restricted Annotation from components. That is a framework policy change, so security review should focus on how existing component access rules are represented after the upgrade.

Several bug fixes are directly relevant to operators. A cluster coordinator could disconnect a freshly rejoined node. A receive node could prematurely truncate data when a node was offloaded or load balanced. An extension that stored LOCAL state and then modified the Map could corrupt state. StandardProcessSession.create() also produced FlowFiles with an ambiguous Start Lineage Index.

Those are not cosmetic defects. They sit close to delivery, state, and lineage semantics. If a NiFi cluster is used for regulated data movement, incident replay, or audit trails, the Start Lineage Index fix and state corruption fix deserve specific regression coverage.

Apache links a migration guide for this release, so do not treat 2.10.0 as a drop in update just because the notes summarize it as a feature improvement and bug fix release. Read the Migration Guidance for 2.10.0 before changing production clusters.

The safest smoke tests here are not complicated: rejoin a node, offload a node, run a JSON flow with representative bad input, exercise Registry Client branch creation, and confirm parameter inheritance in a lower environment.

• Release page: https://github.com/apache/nifi/releases/tag/rel/nifi-2.10.0
• Project repo: https://github.com/apache/nifi
• Tag: rel/nifi-2.10.0