Edit

Configure replication snapshot folder with shares

  • Article

Applies to: SQL Server - Linux

The snapshot folder is a directory that you have designated as a share; agents that read from and write to this folder must have enough permissions to access it.

Diagram showing replication and the snapshot share.

Replication snapshot folder share explained

Before the examples, let's walk through how SQL Server uses samba shares in replication. Following is a basic example of how this works.

  1. Samba shares are configured that files written to /local/path1 by the replication agents on publisher can be seen by the subscriber
  2. SQL Server is configured to use share paths when setting up the publisher on the distribution server such that all instances would look at the //share/path
  3. SQL Server finds the local path from the //share/path to know where to look for the files
  4. SQL Server reads/writes to local paths backed by a samba share

Configure a samba share for the snapshot folder

Replication agents will need a shared directory between replication hosts to access snapshot folders on other machines. For example, in transactional pull replication, the distribution agent resides on the subscriber, which requires access to the distributor to get articles. In this section, we'll go through an example of how to configure a samba share on two replication hosts.

Steps

As an example, we will configure a snapshot folder on Host 1 (the distributor) to be shared with Host 2 (the subscriber) using Samba.

Install and start Samba on both machines

sudo yum install samba
sudo service smb start
sudo service smb status

Set up the Samba share on distributor (host1)

  1. Set-up user and password for samba:

    sudo smbpasswd -a mssql

  2. Edit the /etc/samba/smb.conf to include the following entry and fill in the share_name and path fields

    <[share_name]>
path = </local/path/on/host/1>
writable = yes
create mask = 770
directory mask
valid users = mssql

    Example

    [mssql_data]    <- Name of the shared directory
path = /var/opt/mssql/repldata <- location of directory we wish to share
writable = yes  <- determines if the share is writable from other hosts
create mask = 770  <- Linux permissions for files created
directory mask = 770 <- Linux permissions for directories created
valid users = mssql   <- list of users who can login to this share

Mount the Samba share on subscriber (host2)

Edit the command with the correct paths and run the following command on machine2:

sudo mount //<name_of_host_1>/<share_name> </local/path/on/host/2> -o user=mssql,uid=mssql,gid=mssql

Example

mount //host1/mssql_data /var/opt/mssql/repldata_shared -o user=mssql,uid=mssql,gid=mssql

user=mssql <- sets the login name for samba
uid=mssql  <- makes the mssql user as the owner of the mounted directory
gid=mssql  <- sets the mssql group as the owner of the mounted directory

Configure SQL Server on both Linux hosts to use snapshot share

Add the following section to mssql.conf on both machines. Use wherever the samba share for the //share/path. In this example, it would be //host1/mssql_data.

[uncmapping]
//share/path = /local/path/on/hosts/

Example

On host1:

[uncmapping]
//host1/mssql_data = /local/path/on/hosts/1

On host2:

[uncmapping]
//host1/mssql_data = /local/path/on/hosts/2

Configure publisher with shared paths

  • When setting up replication, use the shares path (example //host1/mssql_data
  • Map //host1/mssql_data to a local directory and the mapping added to mssql.conf.

Related content