Replication and using EMC Autostart
Home > IT Answers > Storage > Replication and using EMC Autostart
Ecann
25 pts.
0
Q:
Replication and using EMC Autostart
SQL replication, EMC Autostart, SQL Database, SQL 2000, Windows Server 2003
I am using EMC Autostart to monitor and mirror an application that uses a SQL 2000 DB. All worked well until I incorporated replication to keep certain tables synchronized with 2 other servers. The Autostasrt uses virtual names and IP's for clients to connect no matter which server is acting as Primary. Autostart stop services and freezes drives on secondary while mirroring.
Problem happens when failed over to secondary.
1. Since all the tables are copied to the secondary, the server name is Primary (server1) in the tables. It retains the name of server1 as it's property, so this causes an error even if I try to setup Replication on secondary (server2). I know there is a script I could run, but hope there is a better way.
2. I've read there is a limitation of not using an alias or virtual ip, but is there a workaround?
Replication is transactional.
Application is defining having to use old 2000 version.
Requirements are to have two physically separate locations for server, sharing of a common storage or quorum not possible.
Thanks for any input. Hopefully there is a solution so I can continue to use replication over the DTS packages I had been previously using.

Software/Hardware used:
propriety software, SQL 2000, EMC Autostart, Windows server 2003
ASKED: Aug 18 2009  3:19 PM GMT
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
RELATED QUESTIONS
0
Ecann
25 pts.
0
A:
 RATE THIS ANSWER
0
Click to Vote:
  •   0
  •  0
  • AddThis Social Bookmark Button
Last Answered: Aug 18 2009  3:19 PM GMT by Ecann   25 pts.
  •   
0
0
RSS - Discussions Help
Discuss This Answer:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _



_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

mrdenny   47440 pts.  |   Aug 18 2009  11:43PM GMT

Can you provide more information about your replication topology?

Replication is very specific about how it works and everything has to be setup just write. You may need to run scripts on the publisher, subscriber and distributor after each failover event in order to get the replication back up and running.

 

Ecann   25 pts.  |   Aug 19 2009  12:18PM GMT

From the sounds of your comments, there is no work around for using the same virtual name for replication setup. It looks like we will need to work on those scripts.

Here is my setup:
Publisher is Distributor.
Replication is transactional.
Selected tables are replicated to 2 other servers.
All servers in same domain and LAN.
SQL application only installed to C:
SQL data stored on E:
SQL Log data on F:
SQL TempDB on G:
E:, F:, G: drives are mirrored through Autostart. Frozen while in standby mode.
SQL cannot run on Secondary while running on Primary.

Replication works correctly as setup on primary server.
Replication fails after fail-over to secondary because it is looking for the physical server name of original primary. (This name is in the database that is mirrored.)

The changes would have to be made post fail over, then again once the system goes back to the Primary.

Have attempted to setup replication using the virtual name but it fails.
Has this issue been fixed in newer versions of SQL? We are forced to use 2000 because of application, but may test upgrading anyway if we can resolve this issue. It is a unique application and the replication is important to us. Also, having this work without manual intervention is important.
Are you aware of any setup that could allow the replication settings to use alias / virtual names?

Thanks for your response.

 

mrdenny   47440 pts.  |   Aug 19 2009  6:31PM GMT

No, this hasn’t been changed the newer versions of SQL Server. Replication still requires that the name that SQL Server knows be used for the replication.

You could create a job that runs every time the SQL Server starts and have it go into the databases and change the server name, and then have it start the replication agents (the agents would need to be changed as well).

Could you setup a geographically distributed cluster? This way you’ve got nodes of the cluster to both facilities, with the SAN handleing replication between the two SANs like you have now. But you have the Microsoft Cluster services using a Majority Node Set cluster to decide which node should be running, and it’ll handle the failover automatically just as if they were local to each other. There would also not need to be any replication changes in this configuration as the clustered name would remain the name no matter which facility was running the active database.

 

Ecann   25 pts.  |   Aug 20 2009  4:49PM GMT

I have not used the Microsoft Cluster and not sure if it will apply to the architecture we have here. I do have a dev environment to try out new things though.
I appreciate all of your help.
You may see another post if the scripts give me problems.

Thanks again mrdenny.

 
0