I got a couple of questions regarding redundancy in Ignition 8.
Where can I make sure that when the master is down a client still can be launched using the vision client launcher?
Because the only client that can be added to the Vision client launcher is the master project.
Do I need to make some configurations or do some ports to be open?
Or does the system automatically needs to open the slave client?
We configured redundancy.
Our project uses a database user source.
The database is installed on the master.
When the master goes down the slave makes sure Ignition is running but the client stops working.
When I log out en back in using the failover user source then the client works fine.
But why do I need to log out and in when the user source isn’t available?
I’m already logged in to the system?
The strange part is that it’s only working this way with the slave, when the master is online and the database connection is lost there isn’t a problem.
So what’s the difference when the slave hasn’t a database connection?
That makes redundancy pretty useless. Install your database in a separate server (or for redundant DBs, a separate cluster).
I know and it will be using a separate database server when it’s installed at the customer. But I find it a little bit strange that I need to login to use the HMI part of the application. The queries are going to SF, so reporting will be available after the connection is restored.
I don’t get this at all. Client security has always been woven into Ignition’s infrastructure. If you choose a database user source, of course that database will have to be available to log in.
I completely agree. But when I’m already logged into the application and I bring the database server down. Then the HMI part of the application still works fine.
But when I’m logged into the application and the master (including the database) is going down then the slave will take over but when this happens I do need to login to the application.
So what makes the difference between a master with no database connection and a slave without a database connection?
When the master goes down, each client must connect to the backup (not slave). Making that connection is where security is (must be) enforced. If you were using an internal user source, you’d be fine. Gateway internals are replicated from master to backup. Databases are supposed to be separate, though in non-redundant situations many get away with putting the DB on the same server as Ignition.
You are using a database user source, so it must be separate, and remain available when failing over from master to backup.
Thank you that makes sense.
Do you also know the answer about launching a new client when the master is down. Does Ignition automatically need to redirect to the backup? Or should you do some settings/ports?
When you create a client launcher shortcut for a particular project in a redundant environment, the shortcut has both master and backup addresses. If your servers have multiple IP addresses, you might need to manual adjust the launcher config.