Upgrading and Testing On A Redundant System

I have a redundant system that I need to upgrade. However, I’d like the ability to test some new features before getting all the clients switched over to the new version. So here’s what I’m thinking:

  1. Shut down the master - All clients switch over to the secondary.
  2. Upgrade the master and do something to prevent it from starting up in redundant master mode (not sure what thought)
  3. Start up master and do testing by starting up clients connected to it.
  4. When satisfied, shut down master and put it back in Master redundant mode
  5. Start up the master and let the existing clients reconnect to it.
  6. Shut down the secondary and upgrade it
  7. Start up the secondary.

So is this possible? That is, starting up the master so it’s no longer master and test? How would I go about it? Thanks for any help.

Kurt

Switching the Redundancy Mode to Independent disables redundancy. You could cause the failover like you described in #1), and then switch the backup node to Independent, which would prevent that node from trying to switch back once the Master is online again.

One important note that I should mention is that you might want to disable the database connections of the node that you are testing on. If you have two Gateways on the same network in Independent mode, then both will be trying to insert rows into your tables, creating double entries.

Instead of shutting down the Master, you could take it off the network, let the Backup take over, switch both nodes to Independent, Disable the Master’s database connections, then bring the Master back on to the network.

Additionally, if you have a spare machine that isn’t on the network, you could always take a Gateway Backup from the Master, and restore it on the spare machine. This way you wouldn’t have to worry about double entries in your database, and could test to your hearts content. Though you wouldn’t have your device connections, but you could always use one of the simulators for this.