KB4557397 - Cumulative Update 21 for SQL Server 2017

The following update is available from the Microsoft Download Center:

• Download the latest cumulative update package for SQL Server 2017 now

If the download page doesn't appear, contact Microsoft Customer Service and Support to obtain the cumulative update package.

The following update is available from the Microsoft Update Catalog:

• Download the cumulative update package for SQL Server 2017 CU21 now

To update SQL Server 2017 on Linux to the latest CU, you must first have the Cumulative Update repository configured. Then, update your SQL Server packages by using the appropriate platform-specific update command.

For installation instructions and direct links to the CU package downloads, see the SQL Server 2017 Release Notes.

You can verify the download by computing the hash of the SQLServer2017-KB4557397-x64.exe file through the following command:

certutil -hashfile SQLServer2017-KB4557397-x64.exe SHA256

The English version of this package has the file attributes (or later file attributes) that are listed in the following table. The dates and times for these files are listed in Coordinated Universal Time (UTC). When you view the file information, it's converted to local time. To find the difference between UTC and local time, use the Time Zone tab in the Date and Time item in Control Panel.

x64-based versions

SQL Server 2017 Analysis Services

SQL Server 2017 Database Services Common Core

SQL Server 2017 Data Quality Client

SQL Server 2017 Data Quality

SQL Server 2017 sql_dreplay_client

SQL Server 2017 sql_dreplay_controller

SQL Server 2017 Database Services Core Instance

SQL Server 2017 Database Services Core Shared

SQL Server 2017 sql_extensibility

SQL Server 2017 Full-Text Engine

SQL Server 2017 sql_inst_mr

SQL Server 2017 Integration Services

SQL Server 2017 sql_polybase_core_inst

SQL Server 2017 sql_shared_mr

SQL Server 2017 sql_tools_extensions

To apply this cumulative update package, you must be running SQL Server 2017.

You might have to restart the computer after you apply this cumulative update package.

To use one of the hotfixes in this package, you don't have to make any changes to the registry.

This article also provides important information about the following situations:

• Pacemaker: A behavioral change is made in distributions that use the latest available version of Pacemaker. Mitigation methods are provided.

• Query Store: You must run this script if you use the Query Store and you have previously installed Microsoft SQL Server 2017 Cumulative Update 2 (CU2).

Analysis Services CU build version

Beginning in Microsoft SQL Server 2017, the Analysis Services build version number and SQL Server Database Engine build version number don't match. For more information, see Verify Analysis Services cumulative update build version.

Cumulative updates (CU)

Cumulative updates (CU) are now available at the Microsoft Download Center.

Only the most recent CU that was released for SQL Server 2017 is available at the Download Center.

CU packages for Linux are available at https://packages.microsoft.com/.

IMPORTANT

All distributions (including RHEL 7.3 and 7.4) that use the latest available Pacemaker package 1.1.18-11.el7 introduce a behavior change for the start-failure-is-fatal cluster setting if its value is false. This change affects the failover workflow. If a primary replica experiences an outage, the cluster is expected to fail over to one of the available secondary replicas. Instead, users will notice that the cluster keeps trying to start the failed primary replica. If that primary never comes online (because of a permanent outage), the cluster never fails over to another available secondary replica.

This issue affects all SQL Server versions, regardless of the cumulative update version that they are on.

To mitigate the issue, use either of the following methods.

Method 1

Follow these steps:

1. Remove the start-failure-is-fatal override from the existing cluster.

   # RHEL, Ubuntu pcs property unset start-failure-is-fatal # or pcs property set start-failure-is-fatal=true # SLES crm configure property start-failure-is-fatal=true

2. Decrease the cluster-recheck-interval value.

   # RHEL, Ubuntu pcs property set cluster-recheck-interval=<Xmin> # SLES crm configure property cluster-recheck-interval=<Xmin>

3. Add the failure-timeout meta property to each AG resource.

   # RHEL, Ubuntu pcs resource update ag1 meta failure-timeout=60s # SLES crm configure edit ag1 # In the text editor, add `meta failure-timeout=60s` after any `param`s and before any `op`s

   Note

   In this code, substitute the value for <Xmin> as appropriate. If a replica goes down, the cluster tries to restart the replica at an interval that is bound by the failure-timeout value and the cluster-recheck-interval value. For example, if failure-timeout is set to 60 seconds and cluster-recheck-interval is set to 120 seconds, the restart is tried at an interval that is greater than 60 seconds but less than 120 seconds. We recommend that you set failure-timeout to 60s and cluster-recheck-interval to a value that is greater than 60 seconds. We recommend that you do not set cluster-recheck-interval to a small value. For more information, refer to the Pacemaker documentation or consult the system provider.

Method 2

Revert to Pacemaker version 1.1.16.

IMPORTANT

You must run this script if you use Query Store and you're updating from SQL Server 2017 Cumulative Update 2 (CU2) directly to SQL Server 2017 Cumulative Update 3 (CU3) or any later cumulative update. You don't have to run this script if you have previously installed SQL Server 2017 Cumulative Update 3 (CU3) or any later SQL Server 2017 cumulative update.

SET NOCOUNT ON;
DROP TABLE IF EXISTS #tmpUserDBs;

SELECT [database_id], 0 AS [IsDone]
INTO #tmpUserDBs
FROM master.sys.databases
WHERE [database_id] > 4
 AND [state] = 0 -- must be ONLINE
 AND is_read_only = 0 -- cannot be READ_ONLY
 AND [database_id] NOT IN (SELECT dr.database_id FROM sys.dm_hadr_database_replica_states dr -- Except all local Always On secondary replicas
  INNER JOIN sys.dm_hadr_availability_replica_states rs ON dr.group_id = rs.group_id
  INNER JOIN sys.databases d ON dr.database_id = d.database_id
  WHERE rs.role = 2 -- Is Secondary
   AND dr.is_local = 1
   AND rs.is_local = 1)

DECLARE @userDB sysname;

WHILE (SELECT COUNT([database_id]) FROM #tmpUserDBs WHERE [IsDone] = 0) > 0
BEGIN
 SELECT TOP 1 @userDB = DB_NAME([database_id]) FROM #tmpUserDBs WHERE [IsDone] = 0

 -- PRINT 'Working on database ' + @userDB

 EXEC ('USE [' + @userDB + '];
DECLARE @clearPlan bigint, @clearQry bigint;
IF EXISTS (SELECT [actual_state] FROM sys.database_query_store_options WHERE [actual_state] IN (1,2))
BEGIN
 IF EXISTS (SELECT plan_id FROM sys.query_store_plan WHERE engine_version = ''14.0.3008.27'')
 BEGIN
  DROP TABLE IF EXISTS #tmpclearPlans;
  SELECT plan_id, query_id, 0 AS [IsDone]
  INTO #tmpclearPlans
  FROM sys.query_store_plan WHERE engine_version = ''14.0.3008.27''
  WHILE (SELECT COUNT(plan_id) FROM #tmpclearPlans WHERE [IsDone] = 0) > 0
  BEGIN
   SELECT TOP 1 @clearPlan = plan_id, @clearQry = query_id FROM #tmpclearPlans WHERE [IsDone] = 0
   EXECUTE sys.sp_query_store_unforce_plan @clearQry, @clearPlan;
   EXECUTE sys.sp_query_store_remove_plan @clearPlan;
   UPDATE #tmpclearPlans
   SET [IsDone] = 1
   WHERE plan_id = @clearPlan AND query_id = @clearQry
  END;
  PRINT ''- Cleared possibly affected plans in database [' + @userDB + ']''
 END
 ELSE
 BEGIN
  PRINT ''- No affected plans in database [' + @userDB + ']''
 END
END
ELSE
BEGIN
 PRINT ''- Query Store not enabled in database [' + @userDB + ']''
END')
  UPDATE #tmpUserDBs
  SET [IsDone] = 1
  WHERE [database_id] = DB_ID(@userDB)
END

When you deploy an update to a hybrid environment (such as Always On, replication, cluster, and mirroring), we recommend that you refer to the following articles before you deploy the update:

• Upgrade a failover cluster instance

• Upgrade and update of availability group servers that use minimal downtime and data loss

• How to apply a hotfix for SQL Server in a transactional replication and database mirroring topology
• How to apply a hotfix for SQL Server in a replication topology
• Upgrading Mirrored Instances
• Overview of SQL Server Servicing Installation

SQL Server CUs are currently multilingual. Therefore, this CU package isn't specific to one language. It applies to all supported languages.

One CU package includes all available updates for all SQL Server 2017 components (features). However, the cumulative update package updates only those components that are currently installed on the SQL Server instance that you select to be serviced. If a SQL Server feature (for example, Analysis Services) is added to the instance after this CU is applied, you must reapply this CU to update the new feature to this CU.

If other issues occur, or if any troubleshooting is required, you might have to create a service request. The usual support costs apply to additional support questions and to issues that don't qualify for this specific cumulative update package. For a complete list of Microsoft Customer Service and Support telephone numbers, or to create a separate service request, go to the Microsoft support website.

1. In Control Panel, open the Programs and Features item, and then select View installed updates.
2. Locate the entry that corresponds to this cumulative update package under SQL Server 2017.
3. Press and hold (or right-click) the entry, and then select Uninstall.

To uninstall this CU on Linux, you must roll back the package to the previous version. For more information about how to roll back the installation, see Rollback SQL Server.