aerospike-client-c does not resize thread pool size if needed about aerospike-client-c HOT 2 CLOSED

The thread pool is not designed to automatically detect load and adjust thread pool size accordingly. That would be a much more complex implementation.

Synchronous batch/scan/query are the only commands that use this thread pool. These commands subdivide into subcommands for each server node so they can be run in parallel. The thread pool is used to run these subcommands.

The thread pool size is configured per each aerospike client instance. Different client processes can have different thread pool sizes at instance creation.

It's important for the user to size the pool according to how many concurrent batch/scan/query commands need to run concurrently on that instance. If the cluster size is 8 and at least 4 batch/scan/queries need to run in parallel, then the thread pool size should be 32 (8 * 4).

The thread pool implementation is shared with server. as_thread_pool_resize() is only called by the server when batch thread size is modified by the user via an info command on a running server.

from aerospike-client-c.

Thanks for the answer. It is clearer.

We circumvented the hanging issue by defining a different thread pool size per aerospike client instance. A small thread pool size (1 thread) for the first context mentioned and a larger size (default value so 16 threads) for solving the hanging issue in the second context.

from aerospike-client-c.