2 Node cluster in disaster recovery scenario

[Mike Smith_33]

We currently have a two node serviceguard cluster that will be in production soon. I am preparing the hardware list for the Disaster Recovery company. Normally, to save costs, management has us run at a reduced capacity during disaster recovery testing/reality. If I just ask for one machine at the disaster recovery site, will I be able to recover and run the production node only at the DR site even though it is in a SG cluster here?

The second part to my question is, can someone direct me to a good source for service guard cluster disaster recovery, I am new to this.

[A. Clay Stephenson]

A single-node cluster is possible but moreover anything that can run as a package can run standalone as well. In fact, if one of your existing nodes crashes, you are essentially equivalent to your DR scenario.

[Bernhard Mueller]

Mike,

if you mean to say your restore an Ignite image onto one server and want to run that server and all the cluster packages as a "single node cluster", then yes you can manually start the cluster specifying your single node and MC/SG will warn you that it is your responsibility that the cluster is *not* running on any other node.

same is true if you already have the DR node in the cluster and all the rest of the cluster is gone (except for the data I hope..)

Regards,
Bernhard

[John Poff]

Hi,

You can use Ignite-UX to recover your production node on your D/R system, and then bring up MC/SG on that single system by forcing the cluster to run. The advantage there is that you can then use your regular package commands to bring up your packages. You can also just Ignite the D/R box, and then mount filesystems and start things manually. I've done it both ways before, and bringing up the cluster on one node seemed like it was a little easier.

JP

[Sivakumar TS · Accepted solution]

Dear Mike,

If the single server in the DR has the capacity ( be terms of CPU, memory etc) to serve the packages runnning in the service guard cluster you may go for a single machine at DR site.

The following link has more info and related links to high availablity solutions using service guard.

http://h71028.www7.hp.com/enterprise/cache/6468-0-0-0-121.aspx

With Regards,

Siva.

[Mike Smith_33]

Thanks for all the replies. In my case, I have two identical servers and only one package. The second server is the development/standby production so all I should need to do is spec out one server at the Disaster recovery site as I know that is sufficient to run the package.


Thanks all for your help.

[Mike Smith_33]

What I want to do will work and I have a link to additional documentation.