Using EV USL failover from clustered prod server to clustered DR server

[Fremont]

Hi,

Customer needs to have a clustered EV server in their production site due to high uptime requirement.  They are also looking at setting up a clustered EV server in their DR site as well.

Can USL be used to failover a clustered EV server into another clustered EV server?  My guess is this should work.  Wanted to check if anyone had done this and if any concerns.

Thanks,

[AndrewB]

what solution are you planning to use for clustering EV?

[Fremont]

2 clusters using windows 2012, one in prod site, one in DR site.

Use USL to fail over from prod to DR and back.

[AndrewB · Accepted solution]

thanks for the additional info. this should help you:

How to configure Enterprise Vault building blocks (Update Service Locations) to implement a disaster recovery solution using Microsoft Cluster

• Article: TECH215320
• Updated: March 10, 2014
• Article URL: http://www.symantec.com/docs/TECH215320

[Fremont]

Thanks.

[Fremont]

Tested this today.  Worked fine.  One comment though:  Both clusters i.e. failed and failover need to have their Admin and Directory services running (online), with the rest of EV services offline for this to work, otherwise USL failover does not work. 

This means USL failover is not exactly a DR solution between 2 clusters.  What if Site is down?  Weird.

[GertjanA]

Hi fremont,

If Admin and directory service on 1 EV server do not work, USL will still work. However, you will then notice in EV Console that one of the aliasses is not switched.

This can be corrected by waiting until affected admin and dir service are running, then rerun USL

[Fremont]

Thanks GertjanA.  Tried running USL many times: no visible response - no moving services prompt, no report, services continue to say (hosted).  Visibly everything was still running on the failed cluster and there was no way to tell if anything had failed back.  Did not try starting services though.

Once I started Admin and Directory on the first cluster everything worked fine.

I was guessing maybe that's what they mean in the article:  http://www.symantec.com/docs/TECH215320

"Before initiating USL after failover on a DR site, ensure that:Only the Enterprise Vault Admin and Enterprise Vault Directory service are running on the cluster nodes.
The Admin service needs to be running so that USL can determine the machine name and hence detect if the alias has changed. It also needs this service to check whether a single machine is represented by multiple computer aliases.​"