Skip to main content

Release Notes - Datadios v1.10.0

Release Date: 2026-08-19 Release Type: Minor Released By: DataDios Team

What's New​

🚀 New Features​

Ready-made data quality checks, and honest pass/fail results​

Data quality rules no longer have to be hand-written SQL. Pick a check type and fill in the fields — DataDios builds the query for you, across all eight supported database engines: PostgreSQL, MySQL, Oracle, SQL Server, Azure Synapse, Snowflake, Trino and Databricks.

Just as importantly, rule results are now recorded correctly. Previously every run was stored as a failure regardless of the actual outcome, so the quality dashboards could not be trusted. Every run now records its real pass or fail, and the dashboards report real numbers.

Key Capabilities:

  • Ready-made check types: empty values, duplicates, out-of-range numbers, invalid patterns, stale data, broken references to another table, and unusual spikes or drops. The invalid-pattern check is not offered on SQL Server or Azure Synapse, which have no regular-expression operator; every other check type is available on all eight engines
  • Evaluate AI-generated answers — accuracy, relevance, safety — against a whole batch of example questions in one run, instead of one question at a time
  • Fleet-wide quality health, per-source quality, and drill-down into a single rule's history over time
  • The rule editor now shows the real configuration for every check type — the source and model you picked, the pass/fail grading criteria, and a plain-text box for AI prompts instead of a SQL editor

Schedule Smart Diff comparisons​

Data Source Diff and Query Diff can now run on a schedule instead of only on demand. Set a cadence — hourly, daily, weekly or monthly — and a time of day, directly from the diff editor header. Scheduled runs appear alongside your other scheduled work.

Available in the new UI only.

The catalog keeps up when tables are renamed or removed​

When a table is renamed at the source, everything you curated against it — tags, classifications, ownership — now follows the rename instead of being stranded on an object that appears to have vanished.

This already worked for sources that expose a durable internal id. It now also works for sources that do not, including PostgreSQL, Redshift and Snowflake, by matching the renamed table on its own structure.

Objects that genuinely were deleted are also handled properly: they are held for 90 days and then retired as deletions, rather than sitting in the catalog as "missing" forever.

⚡ Enhancements​

Timestamps follow your date, time and timezone preferences​

Dates and times shown across the product now use the date format, time format and timezone configured for your account, instead of a fixed format. This applies to Smart Diff workflows, Scheduler, Alerts, and other pages that render timestamps.

Available in the new UI only.

🔧 Bug Fixes​

Smart Diff no longer fails when a mapped column was dropped​

A comparison whose saved column mapping named a column that has since been dropped from the source failed outright, reporting nothing at all and showing only the missing column's name as the error. Those columns are now skipped and the remaining columns are compared as normal, and the results record which columns were skipped — previously a dropped column could still appear in the column list as though it had been compared.

Cost Dashboard loads for every tenant​

The Cost Dashboard returned a server error unless you had connected all six cost-capable data source types. It now shows cost for the sources you actually have.

The SQL Editor's selected database is applied on Snowflake and Databricks​

An unqualified query in the SQL Editor failed on Snowflake with "This session does not have a current database", and on Databricks ran silently against the wrong catalog — the database you picked never reached the connection. Your selection is now applied to the whole statement batch, including DML.

Snowflake database names that are not plain identifiers (for example a name containing a hyphen) are now handled correctly everywhere the session database is switched, including data quality rules — previously the switch silently did nothing for such names.

Snowflake tables created with quoted lowercase names are readable​

A Snowflake table created as, for example, "test1" was listed correctly but failed with "does not exist or not authorized" on every attempt to read its metadata, data, row counts or constraints. All of those paths now address the object correctly.

Snowflake objects no longer break the lineage graph​

Opening the lineage graph for a Snowflake table or view could fail outright. Lineage now resolves the object's details on demand and renders.

Browsing a Redis data source no longer fails​

Opening a Redis data source failed while reading attributes for the database row itself. Existing Redis sources recover without any action on your side.

Azure Blob containers are no longer reported as metadata failures​

Reading metadata for an Azure Blob data source counted every container as a failed object, so a source reported one failure per container even though nothing had gone wrong. Containers are now recognised as folders.

Azure cost extraction works again​

Azure Service cost extraction failed with "Valid TimePeriod with range not exceeding one year must be present" and returned no cost data. Cost extraction now runs normally.

SQL Server queries ending in ORDER BY no longer break other requests​

Sampling a SQL Server query that ended in a plain ORDER BY crashed the server process handling it, which also failed up to five unrelated requests that happened to be running alongside it. Those queries now sample correctly.

Data sources are no longer removed by background cleanup​

A routine background cleanup could quietly delete data source registrations older than a year, while metrics, metadata, workflow results and the catalog all still referenced them. That cleanup no longer runs against data sources. Existing deployments are corrected automatically on upgrade; no existing data is rebuilt or removed.

Deleting a data source that is already gone is no longer reported as a failure​

Deleting a data source that had already been deleted reported an error. It is now treated as a no-op and reports success.

A catalog sync schedule that cannot run now reports as failed​

When the entities catalog consumer could not determine which queue to use, its scheduled run still reported success, so a genuinely broken configuration looked healthy in the Scheduler. Those runs now show as failed with the reason. A tenant that has simply not configured a queue is still reported as a normal skipped cycle, not a failure.

Revoke works in the Resource Center​

The Resource Center's shared-resources tab showed blank rows, and its Revoke button failed with "Failed to revoke" without removing anyone's access. The tab now shows real resource, owner and permission values, and Revoke removes access.

Available in the new UI only.

Create Data Source no longer demands a password it does not show you​

Creating a data source whose authentication mechanism needs no password — for example Hadoop with "No Auth" — was blocked by a "Password is required" error for a field that was not on the form. Hidden fields are no longer treated as required.

Available in the new UI only.

Share dialog closes on success, and shows teammate names​

Sharing a data source left the dialog open after it had already succeeded, and the "Already shared with" panel listed raw user ids instead of names. The dialog now closes on success, and teammates are shown by name.

Available in the new UI only.

Header and signup form fixes​

The breadcrumb trail could overflow into the search box at narrow window widths or under browser zoom; it now truncates. The signup form states that all fields are mandatory, and rejects an organisation name made only of digits, or a first or last name made only of punctuation.

Available in the new UI only.

🔒 Security​

PostgreSQL connection failures no longer echo raw connection details​

When a PostgreSQL data source connection failed unexpectedly, the raw underlying error was returned to the browser in the response message. That text can contain connection strings and, in some configurations, credentials. PostgreSQL connection failures now return a message that explains the problem without reproducing the underlying error, and the full detail is kept in the server log only.

Separately, and for every data source type: a connection failure that produced no message at all surfaced as an unexplained "Request failed with status code 500" with nothing to act on. Those failures now return a message you can act on.

This release covers PostgreSQL only. Other connectors still include the underlying error text in their connection-failure message; they are being addressed separately. In every case, exposure is limited to the response shown to the signed-in user who triggered the failed connection — it is not visible to other users or to anyone unauthenticated.

How to Use​

Schedule a Smart Diff comparison​

  1. Open a Data Source Diff or Query Diff and configure it as you normally would.
  2. In the editor header, open the Schedule control — it reads Unscheduled by default.
  3. Pick a cadence: Hourly, Daily, Weekly or Monthly, and a time of day.
  4. Save. The comparison now runs on that cadence, and each run appears in the Scheduler alongside your other scheduled work.

To stop a scheduled comparison, set the cadence back to Unscheduled.

Upgrade Instructions​

Deployment​

Pull the latest Docker image.

Post-deployment​

  • Refresh the browser after upgrade
  • No action is needed for the data source cleanup fix — existing deployments are corrected automatically on first start after upgrade
  • Data quality rules created before this release continue to work unchanged; the ready-made check types are available on newly created rules

Known Issues​

Snowflake data source database field: setting database on a Snowflake data source does not select that database for your queries. Use the SQL Editor's database selector, or fully qualify object names, until this is fixed.

File Diff scheduling: Data Source Diff and Query Diff can be scheduled; File Diff cannot yet.

Support​

For technical questions or issues, please contact our support team.

Released by Team Datadios | Document version: 1.0