What’s New in Gluesync 2.2
Gluesync 2.2 is the second major refresh of the 2.x line and focuses on a rebuilt agent architecture, stronger security controls, and a refreshed user experience. This page mirrors the current highlights already listed in the release notes so you can quickly understand what changes when planning your upgrade.
Plugin-based agent system
Agents now connect to Core Hub through a plugin interface instead of managing individual WebSocket sessions. This shift unlocks:
-
higher throughput under heavy pipeline workloads thanks to pooled connections;
-
simpler horizontal scaling because each agent installs as a plugin package;
-
safer upgrades handled by Conductor, which decommissions legacy 2.1 agents once the plugin is live.
If you are coming from 2.1, Conductor guides you through the migration so you keep your existing deployment workflow while gaining better reliability.
Enhanced security controls
Gluesync 2.2 brings realm-aware user management and OIDC client authentication. You can now:
-
segregate platform administrators, operators, and tenant users with dedicated realms;
-
register external identity providers for SSO, including OIDC client secrets rotation;
-
enforce least-privilege access to pipelines, modules, and monitoring dashboards.
React-based frontend refresh
The UI is entering a multi-stage refresh that starts with the login screen and gradually expands to the full console. The new React foundation delivers faster load times, improved accessibility, and smoother navigation on both desktop and mobile. Expect more pages to adopt the new experience over the coming sprints.
Third-party API targets
We are opening up the targeting layer so you can push data directly into external SaaS APIs without building bespoke connectors. The initial wave focuses on the most requested API-based destinations and shares the same observability and retry semantics as database or object-storage targets.
Expanded monitoring and observability
Core Hub exports richer metrics and dashboards for replication health. The official Grafana pack (dashboard ID 22907) tracks row and byte throughput, snapshot progress, stalled entities, CDC source-reader freshness (checkpoint age and position), Core Hub internal latency, live JDBC connections by purpose, HikariCP pool pressure, host CPU/memory/network, and JVM heap/GC. Smart-alert threshold gauges can be drawn on the same charts as the measurements.
The Prometheus /metrics endpoint is the source of truth. 2.2.11 adds entity-heartbeat configuration, connection totals, JVM runtime, Ktor HTTP request timings, and checkpoint/pool series when those subsystems are active. Pipeline name filters entity delivery panels; pipeline id filters checkpoint lag. See Grafana dashboards and Metrics reference.
Over-the-air updates
The update experience has been rebuilt from the ground up. Instead of vague progress spinners, you now get crystal-clear progress tracking with a step-by-step breakdown per component and a real changelog for every item being updated.
Smart alerting
Anomaly detection ships with configurable thresholds. You can set a latency ceiling, enable stall detection with warmup, and use a source watchdog (hard floor plus heartbeat). As of 2.2.11, Monitored objects is an allowlist (empty means nobody; new objects stay off until you add them), and sensitivity is Low / Normal / High plus cooldown. Nested EWMA, latency-pattern, and seasonal knobs may still appear on the wire for compatibility, but those legacy detectors no longer fire. Pipeline-level overrides sit on top of the global defaults. See Alert thresholds.
Notifications hub
Core Hub now delivers notifications through multiple channels. Email uses a proper SMTP client with burst-proof delivery and rich templating, while webhooks support CloudEvents, severity filtering, request signing, and custom headers. A dedicated notifications hub lets you browse history, snooze alerts, and choose your preferred interruption style.
Technical columns
Every replicated record can now carry rich technical metadata including transaction ID, transaction timestamp, the user who modified the record, and operation type (INSERT, UPDATE, DELETE). This eliminates the need for custom UDFs to reconstruct audit trails and simplifies compliance reporting.
Query Forge
Gluesync now doubles as a federated query gateway. The new JDBC driver (jdbc:gluesync://<host>:<port>) lets BI tools such as DataGrip, DBeaver, and Tableau run cross-agent SQL queries as if all your data sources were a single database. Authentication is Personal API Token only (gsp_…), each token is scoped to specific pipelines and agents, and all queries are read-only with filter and projection pushdown. See Query Forge for details.
Query Studio
A full-featured SQL workbench embedded directly in the Control Plane. Because it runs inside Core Hub — which is already connected to your databases — you can query remote databases that would otherwise be unreachable without VPN tunnels or firewall exceptions. Features include a multi-tab editor with dialect-aware syntax highlighting, an interactive schema browser, explain plan visualization, real-time query progress streaming via WebSocket, client-side statistics, CSV/JSON/SQL/XLSX export, saved queries, and query auditing with 90-day retention. Read-only by default, with a writable mode toggle for authorized admins.
As of 2.2.11, Query Studio also provides Chart results, XLSX export, edit-in-place, CSV/XLSX import, and manual begin / commit / rollback on a dedicated connection. With Dev Mode enabled, its MCP workbench can run prebuilt and custom tools and organize tools into toolsets. Query Studio also runs on mobile, external modules (including Chronos) can call it, Oracle CLOB / NCLOB cells render as text, and Couchbase can be queried via SQL++. See Query Studio for the full tour and operating details.
Field functions
A library of pre-built, high-performance transformations you can apply directly at the field level — no code, no scripts, no UDFs. Covers date and time conversions, numeric casting (decimals, ints, floats, doubles), string trimming, byte array encoding, number-to-string, and technical fields like auto-generated timestamps and fixed constants. All configurable from the Fields Editor with a simple dropdown.
As of 2.2.11 a target entity can carry both a mapping UDF and field functions (the UDF runs first). A scalar function call can ask the target RDBMS to evaluate an expression per written row. See Field functions for the full catalog.
Chronos: chained events, webhooks, and security
The scheduling module has been significantly upgraded:
-
Chained events: execute a sequence of actions one after another, in sync mode (wait for each step’s completion callback before firing the next) or async mode (fire-and-forget). Ideal for "snapshot first, then start CDC" workflows.
-
Webhook triggers: Chronos can now be triggered by webhooks from CoreHub, so external systems can kick off scheduled flows on demand using secret tokens.
-
Platform-bound events: CoreHub platform events — like pipeline state changes — can trigger Chronos flows automatically via persistent webhook registrations with retry and filter support. No more polling.
-
Security: all Chronos API endpoints are now gated behind JWT authentication, validated against CoreHub’s
/auth/meendpoint. Role-based permissions (SUPER ADMIN, MANAGER, MONITOR, VIEWER) ensure users only see and do what they’re allowed to. SSL/HTTPS support with configurable TLS verification is included. -
Query Studio actions: schedules, chained events, platform events, and webhook-triggered events can run a saved query or custom SQL against a selected pipeline connection. See Query Studio actions.
Conductor module
A new module for managing Gluesync services and agents. Conductor authenticates via CoreHub token introspection (cookie and bearer token support), enforces role-based access control with route-level authorization guards, redacts sensitive data in API responses for users without configuration-modification permissions, and provides readonly functionality for the VIEWER role.
Pipeline wizard
The pipeline creation wizard has been rebuilt from the ground up. The new experience is a clean, step-by-step flow that pulls metadata from your backend agents and pre-fills form fields — connection details, schema info, table lists — so you spend less time copy-pasting. Form validation powered by react-hook-form and yup catches configuration errors before they become runtime problems.
Scheduler and Modules Manager UI
Two management screens received the full UI treatment:
-
Scheduler: a brand-new scheduling interface with weekly and CRON schedule types, a visual chained-events editor, timezone settings, and run-now capabilities — all gated by user role permissions.
-
Modules Manager: a complete view of every agent, module, and third-party service orchestrated by CoreHub. See status at a glance (healthy, error, updating, update available), filter by family, and perform lifecycle actions (start, restart, stop, remove).
Both screens respect the same role hierarchy used across Gluesync: SUPER ADMIN, MANAGER, and VIEWER.
Control Plane UI rebuild
The entire Control Plane received a ground-up rebuild: an all-new pipelines list, a modern entities list that respects screen real estate, refined statuses across the board, and a new entity detail page with smooth status animations. A serious performance boost, especially in loading times, accompanies the visual refresh.
Gluesync Mobile
The first release of the native mobile companion app for the Control Plane. Monitor pipeline statuses and entity health at a glance, with quick access to entity details including current status and last sync information. Available for both iOS and Android with secure authentication using your existing Gluesync credentials.
New agents
2.2 introduces several new agents:
-
Universal File Store agent: writes data in Parquet (default), JSON, or CSV to FTP, FTPS, SFTP, WebDAV, WebDAVS, SMB/CIFS, and NFS. Parquet output is optimized for analytics workloads and BigQuery/Snowflake ingestion. See Universal file store agent.
-
IBM Db2 LUW CDC agent: reads and decodes Db2 transaction logs directly via a custom C native library, capturing transaction IDs, timestamps, operation types, and session user names. No complex transaction log configuration required. See IBM Db2 LUW CDC agent.
-
SAP HANA agent: available as both source (triggers-based) and target, bringing real-time data integration to and from SAP HANA databases.
-
RedShift agent (target): supports bulk operations via AWS S3 as staging area and batch operations for transactional writes.
Pipeline maintenance mode and reset
Pipelines can now be put into maintenance mode to temporarily pause data synchronization for routine maintenance, updates, or troubleshooting without losing connection settings or historical data. Maintenance windows can also be scheduled via Chronos. Additionally, a pipeline reset action clears checkpoint data and lets you start synchronization from scratch — useful when you need to reprocess data or correct synchronization issues.
Performance improvements
Several performance enhancements shipped throughout the 2.2 cycle:
-
Multi-pipeline optimization: a new scheduling algorithm reuses already-initialized connections and resources across pipelines pointing to the same database, significantly reducing overhead (currently applies to Oracle, MS SQL Server, and Informix).
-
RDBMS snapshot performance: expanded query pre-fetch size (10,000 by default), configurable query parallelism (4 by default), and reduced round-trip operations across all RDBMS sources.
-
IBM i journal reading: transitioned from the classic
DISPLAY JOURNALcommand to native IBM i APIs for more efficient journal reading with less overhead. -
Columns length, precision, and scale: refined discovery of column metadata for all source and target databases, enabling finer-grained control during table creation, data type mapping, and schema synchronization.
-
UDF performance: 2–3x faster in typical workloads, with improved stability.
Public roadmap
A public roadmap is now available at roadmap.molo17.com. You can vote on upcoming features, submit your own feature requests, and see what’s in progress, what’s planned, and what’s already shipped. A direct link is also available from the Gluesync Control Plane main menu.
2.2.11 additions
Shipped September 9, 2026. Full detail lives in the release notes.
Core Hub MCP server
An MCP server runs inside Core Hub (same JVM, same RBAC). Point Claude, Cursor, or any MCP client at /mcp or /mcp/http for 55 built-in tools covering pipelines, entities, metrics, Query Studio, and the connection catalogue (list_connections, get_connection, get_connection_usages), plus per-agent SQL tools discovered from the caller’s token. No extra process and no extra privileges. Tool calls still use the caller’s bearer token (loopback HTTP by default). Remote clients can connect by IP; discovery works against a self-signed HTTPS Core Hub. Optional GLUESYNC_MCP_IN_PROCESS (defaults false, dev-only) switches the client from self-HTTP loopback to in-process.
Core Hub-managed CDC checkpoints
Cache-based CDC readers persist position in Core Hub instead of a local checkpoint file, so an agent move or restart no longer resets the reader. Read errors retry instead of jumping the checkpoint. AS/400, MySQL, MariaDB, Informix, and Oracle LogMiner ship in this release, with REST/WS exposure and Prometheus scrape series when a reader has persisted a position.
Live connection registry
Core Hub tracks every live SQL connection and what it is for (pipeline task, Query Studio, Query Forge). Admins can see personal access tokens and last-used time. The same snapshot is scraped as Prometheus gauges (gluesync_db_connections_active / _total). See Connection pool management.
Connections catalog (internal)
Database credentials are extracted from the per-pipeline encrypted blob into a reusable global catalog. Custom-property secrets such as logminerPassword are no longer visible to view-only users. There is no Control Plane UI for picking a shared connection in this release; that follows.
Insert conflict strategy
When an insert hits a row that already exists on the target, each entity can choose UPSERT, SKIP, or FAIL. Unsupported targets hide or caveat the setting. Settable over REST in this release.
Other 2.2.11 notes
-
Snapshot-only entities via a dedicated flag.
-
Pipelines in maintenance or warm-up can be deleted; shared connections keep working when some of the pipelines that use them are in maintenance.
-
Trigger, CDC/shadow, and staging objects are dropped when an entity is deleted on trigger-based and CDC/shadow source agents.
-
Entity status is WARNING when a smart-alert threshold fires; real exceptions stay ERROR.
2.2.11.2 ArenaCache and Couchbase DCP
Shipped September 14, 2026. Source agents that used Chronicle Queue as their local cache now use ArenaCache, Gluesync’s own disk-backed cache, first proven on Oracle LogMiner. The role is unchanged — decouple capture from Core Hub, survive short outages, expire rows by "Source change retention in hours" — but files are per entity, cheaper on disk, and recover more cleanly after a crash. Upgrade when lag is low: Chronicle Queue files are not read by the new agent. See Source agent caching layer.
This patch also ships the Couchbase DCP source agent (couchbase-dcp): source-only CDC from Couchbase’s native DCP stream, without the Eventing service, on Community Edition and Enterprise Edition. See Change data capture (DCP).
2.2.11.1 Query Studio Hub API and UI
Shipped September 11, 2026 (patch notes in the release notes). The 2.2.11.1 Hub API adds distinct-value retrieval (scopeSql, column, optional parameters, sampleLimit, limit, includeNull; response values, sampleLimit, limit, truncated; Hub clamps sampleLimit to 5,000 and limit to 200), explicit cell-update modes (VALUE, NULL, and DEFAULT), cancellation by a client-provided queryId, and agent-dialect identifier quoting. The matching Control Plane chrome (UI !538) adds one-click browse, browse-filter chips that use that Hub request with a current-page fallback, a value viewer, record view, a session statement feed, Copy as Markdown/HTML/XML, and Set NULL / Set DEFAULT cell actions. These are 2.2.11.1 features, not part of 2.2.11. See Query Studio.
What’s next
2.2 is being delivered iteratively (following our Agile development process). As additional features roll out (for example, more UI pages on React or new API targets), this page and the main release notes will be updated. Subscribe to the release feed if you want an email whenever the list grows.