This is aerospike server on centos7 and version 4.6.0.10
the aerospike client is at 4.6.17 C lang.
In the C program if I set the event loop count to 5. 10. 20 its working good.
But with event loop count = 40, I get the mentioned error.
Which tunable should be checked here?
I get the last number different for different configurations, does it represent the max value which is set or the value by which the client is falling short?
aerospike_key_get_async() returned -7 - Max node/event loop BB9945D147AC40C async connections would be exceeded: 2
aerospike_key_get_async() returned -7 - Max node/event loop BB9945D147AC40C async connections would be exceeded: 5
I am trying to execute the get_async function from N number of threads started from the main
as_config.async_max_conns_per_node
increasing from 300 to 500 or 1600 allows me to create more event loops
Q1. Do we have a formula to decide on the as_config.async_max_conns_per_node as a function of number of event loops?
Q2. What is the drawback of keeping a higher value for as_config.async_max_conns_per_node?
Q3. Is there any profiling tool on the client side which can tell us
– used and unused connections
– used and unused event loops
Maximum number of asynchronous (non-pipeline) connections allowed for each node. This limit will be enforced at the node/event loop level. If the value is 100 and 2 event loops are created, then each node/event loop asynchronous (non-pipeline) connection pool will have a limit of 50. Async transactions will be rejected if the limit would be exceeded. This variable is ignored if asynchronous event loops are not created.