Technology Tips And Tricks

Part 1 of 5

This is the first post of an entire series of posts that will outline a Sybase ASE 15 Cluster Edition installation and test environment. The goal is to walk you through the necessary steps to create a working cluster on your laptop or desktop.

The 5 parts of this series are:

• Part 1: System Requirements and Test Setup Concept
• Part 2: Setup of the VMWare Server and the Linux Servers
• Part 3: Install of the Sybase Software
• Part 4: Setup and test the Sybase Cluster
• Part 5: Add an additional node to the cluster

System Requirements:

• Laptop or Desktop Windows PC with dual core or quad core CPU
• 4Gb RAM
• 50Gb disk space
• Sybase ASE 15 CE software (Get your trial copy from Sybase)
• SuSe Linux Enterprise (Download the free ISO file from Novell)
• VMWare Server (Get the free software from VMWare)

Test System Used For This Example:

• Quad core CPU 2.67Ghz
• 8Gb RAM
• 170Gb free disk space
• Windows 7 - 64 bit

The basic idea is to setup a test bed for free. There’s no need to spend money to test software. Sybase supports SuSe Linux very well and requires the least parameter settings to the OS. That’s why SuSe was chosen for this setup.

A word of caution: This is a test setup and should not be used to prepare a production environment.

Test Environment

This diagram schematically explains the test setup. VMWare is instrumental in setting up the shared disk storage and managing 3 Linux servers on a single desktop or laptop.

Using VMWare Server on top of a running Windows system allows configuring the necessary virtual networks and the share disk usage through virtual disk sharing. Sharing disk storage with VMWare and Sybase ASE 15 Cluster Edition is extremely simple and does not require additional software. In a production environment the share storage will be handled by a centralized SAN and assigning LUNs to the servers. And again, no additional software is required to access the raw devices from the nodes in the cluster. An NFS server will be created to provide a centralized filesystem for the Sybase software. This NFS share will be mounted by each node in the cluster.

Next post: Learn how to setup the VMWare Server with the Linux systems used for the cluster.

Sometimes it is necessary to resync the warm standby replication. This could be due to an error, or if replication definitions need to be changed, like changing replication behavior for text and image columns.

In this case the steps to resync a warm standby are as follows.

These are the commands used to re-sync a standby database.

isql -Usa -P<sa_password> -S<rep_server>

>suspend connection to <ws_server>.<ws_db>
>go
>drop connection to <ws_server>.<ws_db>
>go

isql -Usa -P<sa_password> -S<priamary_server>

>use <source_db>
>go
>sp_stop_rep_agent "<source_db>"
>go

isql -Usa -P<sa_password> -S<rep_server>

>create connection to <ws_server>.<ws_db>
>set error class to rs_sqlserver_error_class
>set function string class to rs_sqlserver_function_class
>set username to <source_db>_rep
set password to <password>
>with log transfer on
>as standby for <primary_server>.<source_db>
>use dump marker
>go

isql -Usa -P<sa_password> -S<primary_server>

>use <source_db>
>go
>sp_start_rep_agent "<source_db"
>go

isql -Usa -P<sa_password> -S<rep_server>

>admin logical_status
>go

–it should say Suspended/Awaiting for Enable Marker on <ws_server>.<ws_db>. DO NOT PROCEED if this is not there.

isql -Usa -P<sa_password> -S<primary_server>

>dump database <source_db> to "<file-name>"
>go

isql -Usa -P<sa_password> -S<ws_server>

>load database <ws_db> from "<file-name>"
>go
>online database <ws_db>
>go
>sp_dropalias <source_db>_rep
>go
>sp_dropuser <source_db>_rep
>go
>sp_addalias <source_db>_rep.dbo
>go

The previous steps are only necessary to synchronize the master database syslogin entry with the database users. If your primary database and the warm standby database are in synch with the user definitions, this step can be omitted.

isql -Usa -P<sa_password> -S<rep_server>

>resume connection to <ws_server>.<ws_db>
>go

Check the status of your replication with the System Check commands.

The following commands are being used to check the health of the replication system.

isql -Usa -P -S >admin who_is_down >go

Everything is OK if no rows are returned.

isql -Usa -P -S
>admin health
>go

Everything is OK if the word HEALTHY appears.

isql -Usa -P -S
>admin logical_status
>go

Everything is OK if the word Active appears in both connections.