In examining the HTTP kernel cache on my ARR machines, I have been spotting extra entries for URI's that I would not expect to be cached. Further, the Cache Entry TTL for these cache entries is very large (> 31500000).
Let me provide some background. Due to issues w/ 3rd party controls when we first tested ARR we engineered our application to segregate cacheable content from data. So we have one farm that purely caches content (with a 99% cache hit rate) and a second that serves dynamic data without caching.
I am seeing HTTP cache entries for resources on the non-cached server farm such as "WebResource.axd" with these unusual long TTL's. I am concened that a users might inadvertently received cached data from a different user's session. However we are not seeing that happen in any of testing. Can you explain what is going on with these cache entries?
At the Server Farm level, my cache setting on the non-cached server farm are as follows:
Memory cache duration = 0
Enable Disk Cache = Off
Enable Request Consolidation = Off
Query String Support = No Caching
Thanks for any replies
Jim