Record caching simply stops working

[HitmanInWis]

When I first deploy my project and start testing, record caching is working swimmingly for my ID queries. However, I then throw a barrage of requests at the deployment (load testing), and after a short period of time where I see logging of items being put into the cache ("Version Store: renew") and items being fetched from cache ("UniqueCacheIndex") all of a sudden the logging from record-cache stops, and no more records are being cached.

Any idea what might cause it? I dont see anything in the logs (both rails and server) but perhaps I missed something.

Note that the app continues to work after record-caching stops working, its just that all data is queried from the DB.

[HitmanInWis]

We are using Rails 3.2.17.

As such, I am using commit 800ad8c on the master branch for record-cache, in order to get the 3.2 support.

[robinroestenburg]

@HitmanInWis: can you try version 0.1.3? This is tagged on a later commit than the one you mentioned and contains several improvements/fixes.

[HitmanInWis]

Updated, but still have the same issue. Works on deploy, but stops working in the first few seconds of my stress test on the server.

Interesting to note that when record-caching is working I am loading hundreds of records this way and:

• New Relic verifies that I am indeed not hitting the database for those queries, so the gem is working
• My performance does not increase at all. If anything, it seems that performance is slightly worse than without caching (at least once the database has warmed up).