Morning Kevin. Let's work backwards from the database first. The Receiver can receive two identical streams for the same tag. The historian follows a 'first-in-wins' methodology. When it receives identical timestamps for the same tag, it writes the first value and ignores the second, regardless of any quality score differences.
If you changed your architecture and rather than 'shifting' from one OPC server to the other, but instead just had simultaneous streams running constantly, the Canary Historian would have no duplication issues. You would just use the exact same logging configuration on each OPC Collector, pointing to the exact same DataSets.
We do not have any configuration that would allow for a 'switching' between two different OPC servers should one not be available.
Hi Jeff Knepper, great to know that. Is this the only / recommended approach in high availability scenarios (where there are redundant OPC UA servers)? Seems like having a way to hot-swap the source OPC UA server in case of failures instead of running simultaneous streams to OPC Collector would save a lot of network bandwidth.