furnace O2 problem

Where should I start?

You have told us nothing about the measurement technology (for example, are these nernst cell analyzers?), the fuel being burned, the location of the analyzers in the flue, the age of the sensors, and so on.

If the readings are consistent after calibration, but then drift apart and fluctuate, I would suspect a temperature compensation problem, or an old sensor, or a cracked cell (if Nernst).

But, you have to provide a whole lot more information.

And, please tell us about the flow profile across the flue duct at the location of the sensors. Is it laminar?

I agree with TVP45, more information is required;

Do both analyzers monitor the same sample, i.e. is it a redundant set with A and B on the same duct or are they on separate ducts?

Has anything changed regarding the operation of the furnace operation recently? Has there been a change in flow profiles?

What is the technology used?

Have you tried any independent analysis to verify results?

The obvious; "made in China"????