Immuta v2020.2.0 Errata

Immuta v2020.2.8

Immuta version 2020.2.8 was released October 19, 2020.

Fixes to improve delta history performance overhead with the Databricks integration.
Fixes to run migrations on Unix socket instead of localhost.

R spark-submit jobs in Databricks may prevent a cluster from auto-terminating on an Immuta-enabled cluster.
Delta table commands - OPTIMIZE, GENERATE, and VACUUM - may sometimes fail on an Immuta-enabled cluster.

Immuta version 2020.2.7 was released October 2, 2020.

Fixes to improve performance with the Databricks integration.

R spark-submit jobs in Databricks may prevent a cluster from auto-terminating on an Immuta-enabled cluster.
Delta table commands - OPTIMIZE, GENERATE, and VACUUM - may sometimes fail on an Immuta-enabled cluster.

Immuta version 2020.2.6 was released September 25, 2020.

Users could not access tables in public schema in the Query Engine if upgrading from Immuta v2.8.
Updating views in Snowflake sometimes failed.
Certain Redshift queries using cardinal positioning in group by failed.
Clean-up job removed external attributes listed in the dropdown menu in the policy builder.
Malformed cookies blocked access to the website.
Error creating policies with custom where clauses in DB2 z/OS sources.
Databricks
- In Databricks 5.5 (but not 6.x) certain self-join queries led to column resolution issues on an Immuta-enabled cluster.
- Re-joining relations in Pyspark sometimes resulted in an error.
- DBFS access was blocked for scratch paths.
- After Immuta v2020.2.4, some performance degradation existed when using spark-submit.
- Delta query stopped responding on Azure Databricks with large transaction logs.
- Governance notification was malformed when updating Databricks usernames.

Some performance degradation exists in Databricks with large tables. Upgrade to this patch only if you need a fix from the list above. Otherwise, wait for 2020.2.7 for this performance fix.
R spark-submit jobs in Databricks may prevent a cluster from auto-terminating on an Immuta-enabled cluster.
Delta table commands - OPTIMIZE, GENERATE, and VACUUM - may sometimes fail on an Immuta-enabled cluster.

Immuta version 2020.2.5 was released August 26, 2020.

Databricks
- Queries sometimes failed if a data source created in the immuta schema contained a masking policy and and if the table was joined multiple times in the same query.
- If runSQLOnFiles was enabled, users may have been able to see data that they did not have access to.
- Added configurable logging.

In Databricks 5.5 (but not 6.x) certain self-join queries may lead to column resolution issues on an Immuta-enabled cluster.
R spark-submit jobs in Databricks may prevent a cluster from auto-terminating on an Immuta-enabled cluster.
Delta table commands - OPTIMIZE, GENERATE, and VACUUM - may sometimes fail on an Immuta-enabled cluster.

Immuta version 2020.2.4 was released August 24, 2020.

Creating a data source against an Impala view created via Beeline in CDH 6+ would sometimes fail.
In Databricks, SHOW DATABASES sometimes returned databases present on different instances.
Spark-submit jobs on Databricks were broken.
Matplotlib was not supported on Databricks clusters.
Db2 11 for z/OS did not support LIMIT.
Queries using an impersonation user were audited as sqlQuery records for an impersonation user with no metadata.
There were performance issues during planning in Databricks.
There were performance issues during execution with sub-query custom where policies in Databricks.
Fixed Azure Databricks workspace remote credentials.
Creating a custom where policy with a "where exists" or "where not exists" clause in Databricks did not work.

In Databricks 5.5 (but not 6.x) certain self-join queries may lead to column resolution issues on an Immuta-enabled cluster.
Self-joins between a data source with a masking policy can fail if the data source was created in the immuta schema.
R spark-submit jobs in Databricks may prevent a cluster from auto-terminating on an Immuta-enabled cluster.
Delta table commands - OPTIMIZE, GENERATE, and VACUUM - may sometimes fail on an Immuta-enabled cluster.

Immuta version 2020.2.3 was released August 5, 2020.

Databricks
- There was significant overhead between Immuta and non-Immuta queries.
- The owner of the workspace directory was being set for remote storage, which caused issues for Azure Databricks workspaces.
- The Databricks cluster init script depended on installing packages from the Internet.
Bim/external usernames: updates to usernames didn't take place unless a user's data source subscription changed.
External IAM/external usernames: an external IAM couldn't be saved after adding and removing external username values.
When using encrypted configuration, superuser password configuration was not being used, causing the password to be reset to a default.

Immuta version 2020.2.2 was released August 3, 2020.

Snowflake and Databricks
- User mappings cannot be removed once they are added to an external IAM configuration.
- Usernames cannot be removed once they are added to users.
Associating a Snowflake user to an existing Immuta user does not automatically update the Snowflake views.
Databricks cluster init script depends on installing remote packages from the Internet.
Creating a Databricks workspace against ADL Gen 2 is broken.

Immuta version 2020.2.1 was released July 13, 2020.