Hi Mark,
I understand your logic and having some complex error handling is definitely not a good option.
But I also understand the "pain" from Jan. Assuming you set the max entry size to 10000 bytes, but more than 50% of your requests have values of up to 15000 bytes. Then non of these requests will be cached and the usage of the cache is very poor. Ok, you could compare the amount of incoming requests with those, which are routed to the backend. But that's all complicated and additional effort.
I also had a similar discussion and question last year, if there is some kind of cache-monitoring. Means, some special commands to check how the cache is used (how much entries, the Overall size and RAM consumption). This would be very helpful in tuning the cache parameters and to check how caching is impacting overall system performance and resource consumption.
Is there really nothing to further analyze the usage of the cache?
Thank you!
Ciao Stefan