I noticed the deadlock (stack trace attached) when doing embedded queries in a single node with 50 threads (10ms interval between the queries, no writes while querying).

It appears I need to configure a thread pool size to workaround this? cc dan.berindei / rpwburns

Regarding the dump:

• "TaskNode" are the threads used to query the cache
• Each query does a "computeIfAbsent" to interact with a query cache
• This operation blocks in the "non-blocking" pool