Merge pull request MicrosoftDocs#8801 from MicrosoftDocs/master

1/17 AM Publish

Author: huypub

docs/2014/database-engine/availability-groups/windows/replication-subscribers-and-always-on-availability-groups-sql-server.md

@@ -1,7 +1,7 @@
 ---
 title: "Replication Subscribers and AlwaysOn Availability Groups (SQL Server) | Microsoft Docs"
 ms.custom: ""
-ms.date: "06/13/2017"
+ms.date: "01/16/2019"
 ms.prod: "sql-server-2014"
 ms.reviewer: ""
 ms.technology: high-availability
@@ -16,7 +16,7 @@ ms.author: mathoma
 manager: craigg
 ---
 # Replication Subscribers and AlwaysOn Availability Groups (SQL Server)
-  When an AlwaysOn availability group containing a database that is a replication subscriber fails over, the replication subscription might fail. For transactional subscribers, the distribution agent will continue to replicate automatically if the subscription is using the name of the availability group listener of the subscriber. For merge subscribers, a replication administrator must manually reconfigure the subscriber, by recreating the subscription.  
+  When an AlwaysOn availability group containing a database that is a replication subscriber fails over, the replication subscription might fail. For transactional replication push subscribers, the distribution agent will continue to replicate automatically after a failover if the subscription was created using the AG listener name. For transactional replication pull subscribers, the distribution agent will continue to replicate automatically after a failover, if the subscription was created using the AG listener name and the original subscriber server is up and running. This is because the distribution agent jobs only get created on the original subscriber (primary replica of the AG). For merge subscribers, a replication administrator must manually reconfigure the subscriber, by recreating the subscription.  
 
 ## What is Supported  
  [!INCLUDE[ssNoVersion](../../../includes/ssnoversion-md.md)] replication supports the automatic failover of the publisher, the automatic failover of transactional subscribers, and the manual failover of merge subscribers. The failover of a distributor on an availability database is not supported. AlwaysOn cannot be combined with Websync and [!INCLUDE[ssNoVersion](../../../includes/ssnoversion-md.md)] Compact scenarios.  

docs/advanced-analytics/install/sql-machine-learning-services-windows-install.md

@@ -4,13 +4,14 @@ description: R in SQL Server or Python on SQL Server installation steps for SQL
 ms.prod: sql
 ms.technology: machine-learning
 
-ms.date: 10/01/2018  
+ms.date: 01/17/2019
 ms.topic: conceptual
 author: HeidiSteen
 ms.author: heidist
 manager: cgronlun
 ---
 # Install SQL Server Machine Learning Services on Windows
+
 [!INCLUDE[appliesto-ss-xxxx-xxxx-xxx-md-winonly](../../includes/appliesto-ss-xxxx-xxxx-xxx-md-winonly.md)]
 
 Starting in SQL Server 2017, R and Python support for in-database analytics is provided in **SQL Server Machine Learning Services**, the successor to [SQL Server R Services](../r/sql-server-r-services.md) introduced in SQL Server 2016. Function libraries are available in R and Python and run as external script on a database engine instance. 
@@ -170,8 +171,8 @@ Use the following steps to verify that all components used to launch external sc
     ```sql
     EXEC sp_execute_external_script  @language =N'R',
     @script=N'
-	OutputDataSet <- InputDataSet;
-	',
+    OutputDataSet <- InputDataSet;
+    ',
     @input_data_1 =N'SELECT 1 AS hello'
     WITH RESULT SETS (([hello] int not null));
     GO
@@ -182,8 +183,8 @@ Use the following steps to verify that all components used to launch external sc
     ```sql
     EXEC sp_execute_external_script  @language =N'Python',
     @script=N'
-	OutputDataSet = InputDataSet;
-	',
+    OutputDataSet = InputDataSet;
+    ',
     @input_data_1 =N'SELECT 1 AS hello'
     WITH RESULT SETS (([hello] int not null));
     GO
@@ -198,6 +199,13 @@ Use the following steps to verify that all components used to launch external sc
     | 1|
 
 
+<!--  The preceding 'hello' table is NOT rendering properly on live Docs.
+Instead, the RAW markdown for the table is being displayed.  Probable bug in this markdown source,
+due to stricter rules imposed by 'markdig' engine (replaced 'DFM').
+I will inform HeidiSteen  [GeneMi, 2019/01/17]
+-->
+
+
 > [!NOTE]
 > Columns or headings used in the Python script are not returned, by design. To add column names for your output, you must specify the schema for the return data set. Do this by using the WITH RESULTS parameter of the stored procedure, naming the columns and specifying the SQL data type.
 > 

docs/big-data-cluster/deploy-big-data-tools.md

@@ -5,7 +5,7 @@ description: Learn how to install tools used with SQL Server 2019 big data clust
 author: rothja 
 ms.author: jroth 
 manager: craigg
-ms.date: 12/13/2018
+ms.date: 01/17/2019
 ms.topic: conceptual
 ms.prod: sql
 ms.technology: big-data-cluster
@@ -33,7 +33,14 @@ The following table lists common big data cluster tools and how to install them:
 | **sqlcmd** | For some scripts | Legacy command-line tool for querying SQL Server ([More info](https://docs.microsoft.com/sql/tools/sqlcmd-utility?view=sql-server-ver15)). | [Windows](https://www.microsoft.com/download/details.aspx?id=36433) \| [Linux](../linux/sql-server-linux-setup-tools.md) |
 | **curl** <sup>3</sup> | For some scripts | Command-line tool for transferring data with URLs. | [Windows](https://curl.haxx.se/windows/) \| Linux: install curl package |
 
-<sup>1</sup> You must use kubectl version 1.10 or later. Also, the version of Kubectl should be plus or minus one minor version of your Kubernetes cluster. If you want to install a specific version on kubectl client, see [Install kubectl binary via curl](https://kubernetes.io/docs/tasks/tools/install-kubectl/#install-kubectl-binary-using-curl) (on Windows 10, use cmd.exe and not Windows PowerShell to run curl).
+<sup>1</sup> You must use kubectl version 1.10 or later. Also, the version of kubectl should be plus or minus one minor version of your Kubernetes cluster. If you want to install a specific version on kubectl client, see [Install kubectl binary via curl](https://kubernetes.io/docs/tasks/tools/install-kubectl/#install-kubectl-binary-using-curl) (on Windows 10, use cmd.exe and not Windows PowerShell to run curl). 
+
+> [!TIP]
+> To use kubectl with a previously deployed cluster on Azure Kubernetes Service (AKS), you must set the cluster context with the following Azure CLI command:
+>
+>    ```azurecli
+>    az aks get-credentials --name <aks_cluster_name> --resource-group <azure_resource_group_name>
+>    ```
 
 <sup>2</sup> You must be using Azure CLI version 2.0.4 or later. Run `az --version` to find the version if needed.
 

docs/database-engine/availability-groups/windows/replication-subscribers-and-always-on-availability-groups-sql-server.md

@@ -1,7 +1,7 @@
 ---
 title: "Replication Subscribers and Always On Availability Groups (SQL Server) | Microsoft Docs"
 ms.custom: ""
-ms.date: "03/08/2018"
+ms.date: "01/16/2019"
 ms.prod: sql
 ms.reviewer: ""
 ms.technology: high-availability
@@ -18,10 +18,10 @@ manager: craigg
 # Replication Subscribers and Always On Availability Groups (SQL Server)
 [!INCLUDE[appliesto-ss-xxxx-xxxx-xxx-md](../../../includes/appliesto-ss-xxxx-xxxx-xxx-md.md)]
 
-  When an Always On availability group containing a database that is a replication subscriber fails over, the replication subscription might fail. For transactional subscribers, the distribution agent will continue to replicate automatically if the subscription is using the name of the availability group listener of the subscriber. For merge subscribers, a replication administrator must manually reconfigure the subscriber, by recreating the subscription.  
+  When an Always On availability group containing a database that is a replication subscriber fails over, the replication subscription might fail. For transactional replication push subscribers, the distribution agent will continue to replicate automatically after a failover if the subscription was created using the AG listener name. For transactional replication pull subscribers, the distribution agent will continue to replicate automatically after a failover, if the subscription was created using the AG listener name and the original subscriber server is up and running. This is because the distribution agent jobs only get created on the original subscriber (primary replica of the AG). For merge subscribers, a replication administrator must manually reconfigure the subscriber, by recreating the subscription.  
 
 ## What is Supported  
- [!INCLUDE[ssNoVersion](../../../includes/ssnoversion-md.md)] replication supports the automatic failover of the publisher and the automatic failover of transactional subscribers. The failover of a distributor on an availability database is not supported. Merge subscribers can be part of an availability group, however manual actions are required to configure the new susbcriber after a failover. Availability Groups cannot be combined with Websync and SQL Server Compact scenarios.  
+ [!INCLUDE[ssNoVersion](../../../includes/ssnoversion-md.md)] replication supports the automatic failover of the publisher and the automatic failover of transactional subscribers. Merge subscribers can be part of an availability group, however manual actions are required to configure the new susbcriber after a failover. Availability Groups cannot be combined with Websync and SQL Server Compact scenarios.  
 
 ## How to Create Transactional Subscription in an Always On Environment  
  For transactional replication, use the following steps to configure and failover a subscriber availability group:  

docs/relational-databases/replication/configure-distribution-availability-group.md

@@ -1,7 +1,7 @@
 ---
 title: "Configure SQL Server distribution database in availability group | Microsoft Docs"
 ms.custom: ""
-ms.date: "11/13/2018"
+ms.date: "01/16/2019"
 ms.prod: sql
 ms.reviewer: ""
 ms.technology: replication
@@ -28,7 +28,9 @@ This article explains how to set up a SQL Server replication distribution databa
 SQL Server 2017 CU6 and SQL Server 2016 SP2-CU3 introduces support for replication distribution database in an AG through the following mechanisms:
 
 - The distribution database AG needs to have a listener. When the publisher adds the distributor, it uses the listener name as the distributor name.
-- The replication jobs are created with the listener name as the distributor name.
+- The replication jobs are created with the listener name as the distributor name. Replication snapshot, log reader and distribution agent (push subscription) jobs created on the distribution server gets created on all secondary replicas of the AG for Distribution DB.
+ >[!NOTE]
+ >Distribution agent jobs for pull susbcriptions are created on the subscriber server and not on the distribution server. 
 - A new job monitors the state (primary or secondary in AG) of the distribution databases and disables or enables the replication jobs based on the distribution databases state.
 
 After a distribution database in the AG is configured based on the steps described below, replication configuration and run time jobs can run properly before and after distribution database AG failover.

docs/t-sql/functions/approx-count-distinct-transact-sql.md

@@ -1,7 +1,7 @@
 ---
 title: "APPROX_COUNT_DISTINCT (Transact-SQL) | Microsoft Docs"
 ms.custom: ""
-ms.date: "07/23/2018"
+ms.date: "01/17/2019"
 ms.prod: sql
 ms.prod_service: "database-engine, sql-database, sql-data-warehouse, pdw"
 ms.reviewer: ""
@@ -50,7 +50,7 @@ An [expression](../../t-sql/language-elements/expressions-transact-sql.md) of an
 
 The function implementation guarantees up to a 2% error rate within a 97% probability. 
 
-`APPROX_COUNT_DISTINCT` requires less memory than an exhaustive COUNT DISTINCT operation.  Given the smaller memory footprint, `APPROX_COUNT_DISTINCT` is less likely to spill memory to disk compared to a precise COUNT DISTINCT operation. 
+`APPROX_COUNT_DISTINCT` requires less memory than an exhaustive COUNT DISTINCT operation.  Given the smaller memory footprint, `APPROX_COUNT_DISTINCT` is less likely to spill memory to disk compared to a precise COUNT DISTINCT operation. To learn more about the algorithm used to achieve this, see [HyperLogLog](https://en.wikipedia.org/wiki/HyperLogLog).
 
 > [!NOTE]
 > With collation sensitive strings, the Public Preview version of APPROX_COUNT_DISTINCT uses a binary match and provides results that would have been generated in the presence of BIN collations and not BIN2.