I'm having problems with Terracotta 3.7.2 (opensource) due to memory issues. On client's memory dumps we can see one of our Query cache taking up more than 1gb. This is something we have to deal with, but this query cache is referring to the distributed kind of object taking up the biggest size on Terracotta Server (according to developer console information).
Is it possible local query cache references are keeping too much objects alive on terracotta distributed cache? Could it be the reason of our memory leak? We don't have distributed cache queries, but only distributed entity caches.
We guess the problem is keeping alive on client memory a kind of distributed element cached on Terracotta server. But we also think this scenario wouldn't survive restarts, and it does.
- Terracotta 3.7.2 server array.
- 8 instances connected sharing distributed hibernate second-level caches.
Any help on what we are doing wrong?
Thank you very much in advance.