April 30, 2009 at 8:35 am
I’m preparing for an upcoming offsite DR test that requires us to recover a SQL 2005 Clustered instance on a non-Clustered server. The Windows Admin will be restoring the necessary drives from one of the clustered servers along with a copy of the shared drive onto the DR server and then they’ll remove the cluster services so it appears like a standalone server.
The only way I can think of recovering this environment is to remove and then reinstall the SQL server software, restore all of the user databases and then copy all of the necessary objects (logins, jobs, operators, link servers, ect.).
Has anyone done something like this for his or her DR test that can offer some advice?
April 30, 2009 at 8:46 am
There are varying levels of DR that you can build and it sounds like you are going for the real basic version which is most likely fine for your company. The items you have listed are certainly critical components. The jobs and operators can be recovered with a restore of msdb. The logins cannot so I certainly recommend having a job that scripts them out to file on a regular basis and having that backed up to tape.
Going through these recovery scenarios can be great learning experiences so, take notes on things you wish you had done or done differently for the next time.
As for the steps you describe they seem pretty valid for a basic restore sequence of recovering from tape. Make sure you have the media (software) to do the SQL Server install. 🙂
As for future, if you are able to leave that server up and running at that site I would certainly recommend looking at log shipping or other more robust scenarios for DR sites. They save a lot of work in the recovery process, can be tested a whole lot easier and really don't cost more than a server and some disk at the other site. Provides solid coverage moving forward.
Enjoy.
David
@SQLTentmaker“He is no fool who gives what he cannot keep to gain that which he cannot lose” - Jim Elliot
April 30, 2009 at 9:26 am
David,
I would prefer the process where by I REBUILD and RESTORE the master as I’ve done at other SQL DR tests, but since I’m restoring a clustered instance onto a non-clustered stand-alone server that doesn’t seem possible.
I‘ve thought about this approach and others, that might allow me to restore the master database, but I’m afraid there might be information with the master that recognizes it as a cluster.
Do you think that’s a possibility or would it be possible to RESTORE the master, msdb, and model, then restore the other user databases?
Thanks,
Joe
April 30, 2009 at 10:28 am
I can't think of what there would be information kept in master about this being a clustered instance as that is done through windows and the binaries. So, the restore of master should work. Worth a try anyway and see what happens. Just make sure you have the logins scripted out and any other non-default stuff that you keep in master for the "just in case" scenario.
Post back how things go. Always nice to learn from others as they go through DR testing.
David
@SQLTentmaker“He is no fool who gives what he cannot keep to gain that which he cannot lose” - Jim Elliot
Viewing 4 posts - 1 through 3 (of 3 total)
You must be logged in to reply to this topic. Login to reply