I'm looking for some documentation/help on how to implement failover for the primary hub, specifically what should be installed on the secondary hub and how should it be configured.
Have you checked out the HA Probe? For Hub to hub failover, that's one option... Other options worth reviewing are:
Secondary hub should have all components of the primary hub plus the HA probe itself. HA probe will be actively monitoring the primary hub and then will take action if the primary fails to respond (heartbeat lost).
HA probe configuration specifies which probes/queues to activate on the secondary should the primary become unavailable. Conversely, it will deactivate the probes/queues on the secondary when the primary comes back.
Careful if you are actually doing any sort of polling from the primary hub as not only will those probes need to be installed on the secondary, but you will also have to replicate the probe configs/profiles.
Documentation in the Nimsoft Probes - GA Versions is actually not too bad. Take a look.
I use the attached powerpoint every time I talk to a prospect or customer about the HA probe. It highlights what Brad mentions about HA in NMS.
Has anyone implemented HA on the Primary HUB where WASP and UMP was deployed?
I Tried this around a year ago as part of a POC and never managed to get it working correctly.
Thanks everyone for the input and especially the document Sachin, it works a treat although I'd agree you probably don't want to be performing any remote monitoring from the primary if you do this.
Since Adrian asked, I thought I'd add wasp/UMP to the failover and it works fine, although again you probably don't want to do this unless you really need to because of the issue of synchronizing the content and managing the change of URL.
Charlie and I tested out HA with UMP on our POC and it works out for the most part. We had a primary hub with UMP installed. We installed the HA probe on a secondary hub, which worked great for everything but UMP. Then we installed UMP on the secondary hub. Once the primary went down, we set up the wasp probe to automatically start (I guess you could leave it activated too) and it did. We had to point our browser to the URL of the secondary hub, like John mentions. We also have to make sure the secondary wasp probe is in synch with the primary wasp probe, just like the rest of the probes that HA controls. But since everything is written to the database, and the data_engine on the secondary hub points to the same database as the primary, we didn't lose any of the customization done in UMP with tabs and pages.
Another option that I just became aware of today was that in UMP 2.0, load balancing was introduced, so you can have UMP on multiple servers for scalability and availability : http://d1guc7pjy56zs9.cloudfront.net/ump2.0-ga/Nimsoft-UMP-MultiserverInstallationGuide-2.0.pdf. I didn't get a chance to play with this, but it looks very interesting.
Retrieving data ...