Release 9.2.49.1¶
These are the Rolling Release Notes for Release 9.2.49.1. These release notes are applicable to both Privacera's Self-Managed version and PrivaceraCloud.
Breaking Changes¶
Breaking Changes
This release includes the following breaking changes and upgrade steps.
All Users Can List S3 Buckets in File Explorer¶
The S3 default policy all - bucketname, objectpath now gives the public group mread on all buckets, so every user can list buckets in File Explorer. New S3 repositories get it on creation, and the upgrade adds it to existing repositories.
- Impact: Every user can list S3 bucket names through File Explorer.
- Action required: To restrict listing, edit the S3 default policy after the upgrade.
Behavioral Changes¶
Behavioral Changes
Customers should know about these changes before they upgrade.
Ranger Downtime During the Upgrade¶
This upgrade requires Ranger downtime of approximately 1 hour, on both Self-Managed and PrivaceraCloud, to apply the database schema changes and complete the migration.
- Action required: Plan a maintenance window for the upgrade.
[PrivaceraCloud only] Ranger Access Audit Retention Is Now 7 Days¶
Ranger access audits on the Audits page are now retained for 7 days, down from 30 days, in PrivaceraCloud and Privacera Data Plane deployments. Existing records keep their original retention. To keep Ranger access audits longer, export them to your own object store. See Configure Audit Backup Settings.
Snowflake: Connector Manages Iceberg Tables¶
After the upgrade, the Snowflake connector applies access, masking and row-level filter policies to Iceberg tables. For each database it starts managing after the upgrade, the connector grants its role ALL on current and future Iceberg tables.
- Impact: Access to Iceberg tables can change on the first sync after the upgrade.
-
Action required: For databases the connector managed before the upgrade, run the following as a role that can grant on them:
SQL This step is not needed if
switch.ownership.roleis set to the connector's own role.
Apache Ranger¶
Zone Administrator And Zone Auditor Policy And Dashboard Performance
Zone Administrator And Zone Auditor Policy And Dashboard Performance¶
Fixed timeouts and slow responses on policy list and dashboard APIs for zone administrators and zone auditors in deployments with many security zones and roles. Ranger resolves the caller's group and role membership once per request instead of rebuilding it for every zone check. Policy list and count APIs return the correct zone-scoped policies for zone administrators and zone auditors, including when no security zone filter is specified in the request.
[Self-Managed only] Limit Service Definitions Created At Ranger Admin Startup
[Self-Managed only] Limit Service Definitions Created At Ranger Admin Startup¶
On Self-Managed deployments, Ranger Admin now creates service definitions, repositories, and service users only for the service-definition names you list in RANGER_ENABLED_SERVICES in vars.ranger.admin.yml, plus tag. Leave RANGER_ENABLED_SERVICES empty to keep the previous full startup behavior, except kms and padmin are no longer created automatically. Ranger does not remove existing service definitions, repositories, or service users when you narrow the list. Remove unneeded definitions and users manually in Ranger Admin or through your upgrade process as a one-time activity after you change the setting. PrivaceraCloud does not read RANGER_ENABLED_SERVICES, but the shared default startup path no longer creates kms or padmin, and new installations get the updated S3 default policy. Existing service definitions, repositories, and users are kept.
Metadata Read for All S3 Repositories
Metadata Read for All S3 Repositories¶
New S3 repositories automatically grant the mread permission to the public group, allowing File Explorer to list buckets through DataServer. Existing S3 repositories will also receive this permission through the schema migration during the upgrade.
Apache Zookeeper¶
Updated Apache Zookeeper Dependencies
Updated Apache Zookeeper Dependencies¶
Upgraded Apache Zookeeper dependencies to address known security vulnerabilities identified in the CVE (Common Vulnerabilities and Exposures) report.
DataServer¶
Boto3 Signer - Fixed Cross-Bucket Copy
Boto3 Signer - Fixed Cross-Bucket Copy¶
Fixed an issue where copying an object from one Amazon S3 bucket to another failed with Access Denied even when the user is allowed to read the source object and write to the destination; the source object is now authorized and audited under its own bucket.
Fixed an issue where a copy from a source other than the one last copied into the same destination was denied because the credentials issued for the earlier copy are reused; cached credentials are now held separately for each copy source.
Discovery¶
Omni Classification
Omni Classification¶
Omni Classification classifies the tables, columns and object-storage files of data sources connected through a Runtime Plane, and lets you review the tags it applies.
- Classification Setup: configure classifiers and dictionaries of values or regex patterns, grouped into region packs (Global, United States, United Kingdom, Canada, Brazil, Australia and New Zealand). With an AI API key, AI can validate low-scoring matches.
- Review Tagged Resources: approve or reject applied tags, and see which classifier suggested each one.
- Classifications: follow each classification's progress and its failed resources.
- Scan Object Storage: choose the Amazon S3 and Google Cloud Storage paths to scan, and the paths to skip.
- Real-time classification: classify new tables as soon as a connector syncs them. Turn it on with the Real-time classification switch in the connector's Access Control configuration. Available on Snowflake, Databricks Unity Catalog, Databricks SQL, BigQuery, Redshift, PostgreSQL and Microsoft SQL Server connectors.
- Discovery on connectors: turn on Discovery from the connector's Discovery tab. Amazon S3 and Google Cloud Storage are available as Discovery-only connectors in Add New.
Compliance Policies and Alerts
Compliance Policies and Alerts¶
Apply compliance policies to data products and get an alert for each violation. The policy types are Disallowed Tags, De-identification and Workflow. Policies that change data run in Dry run or Enforce mode.
- Workflow policies move files that break the policy to a quarantine location, and can move clean files to a transfer location.
- Alert detail shows what was found and what the policy did. Reviewers can mark an alert as reviewed or as a false positive, and retry a failed action.
- Turning it on: compliance is off by default. Turn it on with Enable compliance policies in the Discovery Classifier's App Configurations in Runtime Manager.
[Self-Managed only] Omni Classification on Self-Managed runtime planes
[Self-Managed only] Omni Classification on Self-Managed runtime planes¶
Self-Managed runtime planes can now run Omni Classification.
- Turning it on: turn on Omni Classification under Advanced Features in the runtime plane create wizard (AWS) or in Runtime Configuration (AWS and GCP). An administrator must first enable the account setting OMNI Classification (Self-Managed Runtime).
- Requirements: Omni must be on, and the plane needs a PostgreSQL database.
- AI settings: choose the AI provider (OpenAI), the secret that holds the API key, and the model.
Upgrade note
Omni Classification is supported on new Self-Managed runtime planes only; to use it on an existing plane, create a new plane.
Plugin¶
EMR Plugin¶
[PrivaceraCloud only] Added OLAC Support for EMR 7.13 and 7.14
[PrivaceraCloud only] Added OLAC Support for EMR 7.13 and 7.14¶
- OLAC support for EMR 7.13 and 7.14, available on Self-Managed since 9.2.48.1, is now also supported on PrivaceraCloud.
- For the supported version matrix, refer to Supported Versions.
Trino Plugin¶
[PrivaceraCloud only] Added Support for Starburst Enterprise Version 482-e LTS
[PrivaceraCloud only] Added Support for Starburst Enterprise Version 482-e LTS¶
- Starburst Enterprise version 482-e LTS, supported on Self-Managed since 9.2.48.1, is now also supported on PrivaceraCloud.
- For the supported version matrix, refer to Supported Runtime Versions.
PolicySync Connector¶
[PolicySync] Lambda + Kafka path — single on-demand event flow for permission_sync retries
[PolicySync] Lambda + Kafka path — single on-demand event flow for permission_sync retries¶
Added PERMISSION_SYNC on-demand events to re-apply permissions for specific resources after a failed grant. Publish one event to the Apache Kafka topic consumed by the PolicySync event Lambda. The Lambda creates a scoped task in the Ops Server, and the connector reconciles only the named resources and re-applies any missing grants, with no manual API call and no waiting for the next scheduled sync. Support for Databricks Unity Catalog, Databricks SQL Analytics and AWS Lake Formation. With policysync.on-demand.audit.enable set to true, each resource reports an ON_DEMAND audit with your requestor_id and a response code. Results are reported per run, not per requestor.
Databricks Unity Catalog: Row-level filters on foreign catalogs no longer blocked by catalog creation
Databricks Unity Catalog: Row-level filters on foreign catalogs no longer blocked by catalog creation¶
For row-level filters on foreign (federated) catalog tables, the Databricks Unity Catalog connector no longer creates the local privacera_security_<catalog> catalog. On metastores without a storage root, that step failed even when the catalog existed, which reported a false DDL_FAILED and blocked the table's access-policy changes. Create the local catalog before applying row filters.
Databricks Unity Catalog: Configurable Catalog for Federated (Foreign) Row-Filter and Masking Functions
Databricks Unity Catalog: Configurable Catalog for Federated (Foreign) Row-Filter and Masking Functions¶
Added a Databricks Unity Catalog connector property foreign.masking.row.filter.catalog.name. When set, row-filter and masking functions for federated (foreign) catalogs are stored in the configured Unity Catalog catalog (source schema preserved) instead of the derived privacera_security_<source_catalog>. Default is blank (unchanged behaviour).
Lake Formation: Access Audits Use the Connector's AWS Region
Lake Formation: Access Audits Use the Connector's AWS Region¶
Fixed an issue where Athena defaulted to us-east-1 when no Athena region was configured. Athena now uses the connector's AWS Region by default. An explicitly configured Athena region takes precedence.
Snowflake: Fixed Iceberg Tables Not Loading and Masking and Row-Level Filter Policies Failing on Them
Snowflake: Fixed Iceberg Tables Not Loading and Masking and Row-Level Filter Policies Failing on Them¶
-
Fixed an issue where the Snowflake connector skipped Iceberg tables during resource loading. Iceberg tables now load alongside native tables, and the connector's role is granted access to current and future Iceberg tables in each managed database.
-
Fixed an issue where masking and row-level filter (RLF) policies failed on Iceberg tables. The connector now uses
ALTER ICEBERG TABLEfor Iceberg tables and keepsALTER TABLEfor native tables.
Ops Server: Faster On-Demand Task Polling and Task Cleanup
Ops Server: Faster On-Demand Task Polling and Task Cleanup¶
- Improved performance at scale: On-demand task polling and duplicate-task checks are now significantly faster through indexed database lookups.
- More efficient task cleanup: Ops Server now purges failed and cancelled tasks as well as completed ones once they are older than
ops.server.purge.interval.days(30 days by default), using batched cleanup to minimize database impact. The first purge after the upgrade removes the existing backlog. To keep failed and cancelled task history longer, raiseops.server.purge.interval.daysbefore the upgrade.
Portal¶
Multiple DataServers in File Explorer
Multiple DataServers in File Explorer¶
File Explorer now supports multiple DataServers. You can switch between them using the dropdown next to the page title. Applications not connected to a DataServer will appear under “Other applications”. If there is only one DataServer, File Explorer will appear as usual, without the dropdown.
Portal File Explorer In D2P
Portal File Explorer In D2P¶
File Explorer now works with D2P runtimes: pick a runtime and dataserver to browse S3, with files moving directly between S3 and the browser through signed URLs. Only AWS dataservers are supported for now, and File Explorer in PrivaceraCloud deployments is unchanged.
SAML SSO rejects invalid IdP metadata files at upload
SAML SSO rejects invalid IdP metadata files at upload¶
Fixed SAML single sign-on so that invalid IdP metadata files are rejected at upload with an error, instead of replacing the current file and breaking SSO login. A rejected upload no longer changes the IdP metadata file name shown on the Single Sign On form, and an accepted IdP metadata file is used at the next SSO login on every portal instance.
[PrivaceraCloud only] Server-Side Authorization for UserSync APIs
[PrivaceraCloud only] Server-Side Authorization for UserSync APIs¶
APIs now check permissions on the server, so only users with data source access can use them. Admins see no change.
[PrivaceraCloud only] Access Control Dashboard
[PrivaceraCloud only] Access Control Dashboard¶
Policies by Service now counts all policy types, so every service with policies appears.
[Self-Managed only] About Dialog Uses Trust3 AI Branding
[Self-Managed only] About Dialog Uses Trust3 AI Branding¶
The portal’s About dialog now uses Trust3 AI branding instead of Privacera.
Updated Portal Dependencies and Base Image
Updated Portal Dependencies and Base Image¶
Upgraded Portal dependencies and updated the base image to a newer version to address known security vulnerabilities identified in the CVE (Common Vulnerabilities and Exposures) report.
Privacera Manager¶
Privacera Manager Base Image Upgraded
Privacera Manager Base Image Upgraded¶
Updated the Privacera Manager base image to a newer Debian version to address known security vulnerabilities identified in the CVE (Common Vulnerabilities and Exposures) report.
Updated Post Install Job Dependencies
Updated Post Install Job Dependencies¶
Upgraded Post Install Job dependencies to address known security vulnerabilities identified in the CVE (Common Vulnerabilities and Exposures) report.
Privacera Monitoring¶
Updated Grafana Dependencies
Updated Grafana Dependencies¶
Upgraded Grafana dependencies to address known security vulnerabilities identified in the CVE (Common Vulnerabilities and Exposures) report.
Updated Prometheus Dependencies
Updated Prometheus Dependencies¶
Upgraded Prometheus dependencies to address known security vulnerabilities identified in the CVE (Common Vulnerabilities and Exposures) report.
Privacera UserSync¶
Security Fixes: Jackson, amqp-client, Guava and Base Image Upgrades
Security Fixes: Jackson, amqp-client, Guava and Base Image Upgrades¶
Upgraded the following UserSync dependencies to address known security vulnerabilities identified in the CVE (Common Vulnerabilities and Exposures) report:
- Jackson: Jackson 2 is now 2.21.8 and Jackson 3 is now 3.1.8. This fixes CVE-2026-68497, CVE-2026-91776, CVE-2026-91777, CVE-2026-89407 and CVE-2026-89425.
- amqp-client: Upgraded from 5.33.1 to 5.37.0. This fixes CVE-2026-75516.
- Guava: Upgraded from 32.0.1-jre to 33.7.2-jre. This fixes CVE-2026-102554.
- Base image: Updated the UserSync base image to the latest version to pick up operating system package fixes.
Trust3 AI Cloud¶
Runtime Planes¶
[Self-Managed only] Runtime planes on Azure Kubernetes Service (AKS)
[Self-Managed only] Runtime planes on Azure Kubernetes Service (AKS)¶
Self-Managed runtime planes can now be deployed on Azure Kubernetes Service (AKS).
Applications no longer get stuck after a failed Helm operation
Applications no longer get stuck after a failed Helm operation¶
The runtime agent now recovers when an application's Helm release is left mid-upgrade or was changed by hand, instead of blocking every later deployment. A failed uninstall is now reported as failed and can be retried.
Connector and UserSync settings are checked as you type
Connector and UserSync settings are checked as you type¶
Connector and UserSync configuration fields are now validated as you type, not only when you save.
- Prev topic: Releases