Dedicated cluster management is now available via API. Three new endpoints let you rebalance traffic weights, add replicas, and remove replicas, all scriptable.

• PUT /v0/organizations/{org_id}/clusters/{cluster_id}/weights to rebalance traffic.
• POST /v0/organizations/{org_id}/clusters/{cluster_id}/replicas to add a replica.
• DELETE /v0/organizations/{org_id}/clusters/{cluster_id}/replicas/{replica_name} to remove a replica.

Wire cluster scaling into your runbooks, automate from your own control plane, or standardize changes across organizations.

Available for dedicated infrastructure where self-serve replica management is enabled. Read the blog post or the cluster management docs.

Deployments can now apply schema changes to Data Sources using ALTER instead of rewriting and backfilling the entire table. This means faster deployments with no data movement for the following operations:

• Add a column to a Data Source.
• Drop a column from a Data Source.
• Make a column Nullable.
• Modify TTLs.
• Add or drop an index.

Update your datafile, deploy, and Tinybird applies the change almost instantly when promoting the deployment without any data movement. Read more about ALTER operations in Forward deployments.

Branch limit increased to 6¶

Workspaces now support up to 6 branches, doubled from the previous limit of 3. Useful for teams running multiple preview environments or feature branches in parallel.

UI improvements¶

• Resources table performance. Row virtualization and loading optimizations make the Resources page faster, especially for Workspaces with many resources.
• Polling stops when the tab is hidden. Pipe detail pages were firing queries even when the browser tab wasn't visible. Polling now pauses automatically and refreshes when you return.
• Playground loading. Nodes load lazily with better loading state visuals. Default query limit is now 50 rows.
• Error visibility. Time series charts group errors, the results bar shows on error, and the side panel shows an error indicator when a resource fails.
• Resource actions moved to the top bar for easier access in split-screen view.
• No changelog popup for new users. New users no longer see the changelog on first visit.

CLI updates¶

CLI 4.3.0: tb deploy --auto/--no-auto keeps the same behavior (default --auto) while using server-side deployment auto_promote under the hood.

CLI 4.2.2: Improved the max-depth warning message to clarify when the warning can be safely ignored.

Also fixed: tb deploy now prints the deployment URL consistently, warns when you Ctrl-C after a deployment has been sent, and correctly handles cancellation.

Bug fixes and improvements¶

• [All]: Fixed double Nullable wrapping for LowCardinality(Nullable()) columns that caused schema issues.
• [All]: Fixed EXPLAIN JSON parsing crash on invalid UTF-8 from unhex() in FixedString values.
• [All]: Fixed CSV column name mapping where headers with spaces or special characters didn't match normalized Data Source column names.
• [All]: POST request bodies are now encoded as UTF-8 instead of Latin-1, fixing potential corruption of non-ASCII characters.
• [All]: Sending an empty file body for a replace operation now returns a proper 400 error instead of silently creating issues.
• [All]: Improved S3 connector error message when using deprecated settings.
• [All]: Increased CSV importer queue wait timeout to 240 seconds.

From the blog¶

Cluster management is now scriptable: Replica management in dedicated clusters is now available via API. Rebalance weights, add replicas, remove replicas.