If I query across the First/Second connection limit boundary, i.e. so that both database sources would need to be queried to return all of the results, is this supposed to work?
I have a very simple test where I've generated a single history record each year over 5 years for a tag, then i've duplicated the database and basically made them so that one database has all the data < 2025 and the other has all the data for 2025. When I try to do a tag history query binding to the tag splitter for the tag across the boundary, it only returns the Second connection's data. If I change the start time to be 2025, then it returns the First connection's data. I can't seem to get it to return data from both connections.
**Query with start = 2022 **
Query with start = 2025
First Database
Second Database
There's no merge. The docs clearly say that one connection will be used or the other. Your attempted architecture is bogus.
1 Like
Well then... that's a bit useless for clients 
lucky I'm not actually trying to use it for that! The "second" database is purely for backup. I'm just scripting some data extraction at the moment and was wondering why I was missing this year's data, and that will do it.
Thanks again
I'm not sure if this is the part you're talking about, but I still think the "Limit First Connection Query" description is open to misinterpreting it to mean it would query the first connection for the time that is within the setting, and the second beyond that 
That's how I interpreted it when I first read it. I had to squint my eyes to interpret it as you did, although now it kinda seems obvious that was the intention...
"Queries that go further back execute against the second connection."
Not even a hint of merging. You are engaging in wishful thinking.
If there was a clarifying sentence after it would be unmistakable:
"Queries that go further back execute against the second connection. This also means that queries whose dates span across the limit boundary will execute only against the second connection."
@Paul.Scott (not sure if I can still ping you for user manual changes?)
1 Like
A manual change that adds the word "only" after "execute" in the sentence I quoted would add clarity. (Not necessary, IMO, but wouldn't hurt.)
1 Like
I disagree. This architecture allows the first connection to be closely pruned to maintain high performance, while the second connection keeps everything. (Or much longer.)
1 Like
