Databricks Runtime 6.5 (EoS)

Article
09/30/2024

Note

Support for this Databricks Runtime version has ended. For the end-of-support date, see End-of-support history. For all supported Databricks Runtime versions, see Databricks Runtime release notes versions and compatibility.

Databricks released this version in April 2020.

The following release notes provide information about Databricks Runtime 6.5, powered by Apache Spark.

New features

Operation metrics for all writes, updates, and deletes on a Delta table now shown in table history

You can now see operation metrics (for example, number of files and rows changed) for all writes, updates, and deletes on a Delta table in the table's history. The generation of these metrics is enabled by default. You can see the history by running:

DESCRIBE HISTORY deltaTableName

Operation metrics

Rate-limit the data processed in Delta Lake streaming micro-batches

When you process data from Delta Lake using Structured Streaming, you can now set the option maxBytesPerTrigger to rate-limit how much data gets processed in each micro-batch. This sets a "soft max," meaning that a batch processes approximately this amount of data and may process more than the limit.

If you use Trigger.Once for your streaming, this option is ignored. In addition, if you use this option in conjunction with maxFilesPerTrigger, the micro-batch processes data until either the maxFilesPerTrigger or maxBytesPerTrigger limit is reached.

Improvements

Snowflake connector updated to 2.5.9

The updated connector promotes the Snowflake JDBC driver to 3.12.0.

Java upgraded to 1.8.0_242

This version introduces new checks on trust anchor certificates to ensure that trust anchors are CA certificates and contain proper extensions. Trust anchor certificates must include a Basic Constraints extension with the ca field set to true. In addition, if they include a Key Usage extension, the keyCertSign bit must be set.

If you use self-signed trust anchor certificates, you may hit the following exceptions introduced by the new checks:

TrustAnchor with subject "..." is not a CA certificate

TrustAnchor with subject "..." does not have keyCertSign bit set in KeyUsage extension

To keep using self-signed trust anchor certificates, you have two options to address the above exceptions:

Set Java system property jdk.security.allowNonCaAnchor introduced with Java 1.9.0_242 to true to restore the previous behavior. You can use spark.driver.extraJavaOptions and spark.executor.extraJavaOptions to set this system property to Spark driver and executors, respectively.
Include a Basic Constraints extension with the ca field set to true to your trust anchor certificates. If your trust anchor certificates include a Key Usage extension, set keyCertSign bit.

If you use Java Keytool to generate public/private key pairs, set -ext BasicConstraints=ca:true when generating trust anchor certificates. If your trust anchor certificates include a Key Usage extension, set keyCertSign bit -ext KeyUsage=...,keyCertSign when generating trust anchor certificates.

Upgraded R libraries

Upgraded several installed R libraries. See Installed R Libraries.

Deprecations

The boto package is deprecated and will be removed in an upcoming major release of Databricks Runtime. Azure Databricks recommends that you use boto3 instead.
The following R packages will be removed in an upcoming major release of Databricks Runtime:
- abind
- bitops
- car
- carData
- doMC
- gbm
- h2o
- littler
- lme4
- mapproj
- maps
- maptools
- MatrixModels
- minqa
- mvtnorm
- nloptr
- openxlsx
- pbkrtest
- pkgKitten
- quantreg
- R.methodsS3
- R.oo
- R.utils
- RcppEigen
- RCurl
- rio
- sp
- SparseM
- statmod
- zip

Azure Databricks recommends using the following code to install and load R packages that are not installed by Databricks Runtime:

if (!require("package_name")) {
  install.packages("package_name")
  library("package_name")
}

Apache Spark

Databricks Runtime 6.5 includes Apache Spark 2.4.5. This release includes all Spark fixes and improvements included in Databricks Runtime 6.4 (EoS), as well as the following additional bug fixes and improvements made to Spark:

[SPARK-27099] [SQL] Add 'xxhash64' for hashing arbitrary columns to Long
[SPARK-26151] [SQL] Return partial results for bad CSV records
[SPARK-30927] [SS] StreamingQueryManager should avoid keeping reference to terminated StreamingQuery
[SPARK-22590] Copy sparkContext.localproperties to child thread in BroadcastExchangeExec.executionContext
[SPARK-30671] emptyDataFrame should use a LocalRelation
[SPARK-30556] [SQL] Reset the status changed in SQLExecution withThreadLocalCaptured
[SPARK-30763] [SQL] Fix java.lang.IndexOutOfBoundsException No group 1 for regexp_extract
[SPARK-30806] [SQL] Evaluate once per group in UnboundedWindowFunctionFrame
[SPARK-30731] Update deprecated Mkdocs option
[SPARK-30857] [SQL] Fix truncations of timestamps before the epoch to hours and days
[SPARK-30798] Scope Session.active in QueryExecution
[SPARK-30793] [SQL] Fix truncations of timestamps before the epoch to minutes and seconds
[SPARK-30826] [SQL] Respect reference case in StringStartsWith pushed down to parquet
[SPARK-30797] [SQL] Set tradition user/group/other permission to ACL entries when setting up ACLs in truncate table
[SPARK-30065] [SQL] DataFrameNaFunctions.drop should handle duplicate columns
[SPARK-29890] [SQL] DataFrameNaFunctions.fill should handle duplicate columns
[SPARK-30310] [CORE] Resolve missing match case in SparkUncaughtExceptionHandler and added tests
[SPARK-30512] Added a dedicated boss event loop group

Maintenance updates

See Databricks Runtime 6.5 maintenance updates.

System environment

Operating System: Ubuntu 16.04.6 LTS
Java: 1.8.0_252
Scala: 2.11.12
Python: 3.7.3
R: R version 3.6.3 (2020-02-29)
Delta Lake: 0.5.0