This post refers to the connection to the cluster through JDBC fails because the concurrent queue is full. Let's see below the process of fixing it.
Symptom
Common users' connections to the cluster through JDBC failed but the connections of initial users and gsql are successful.
Cause analysis
The possible cause is that the connection through JDBC is blocked because the concurrent queue is full.
Analysis procedure
1. Connect to the cluster as an initial user and check the capacity of the concurrent queue.
show max_active_statements;
2. Check the usage of the concurrent queue.
select usename,attribute,status,enqueue,resource_pool,count(*) from pg_session_wlmstat group by 1,2,3,4,5;
Check whether the number of active common users has reached the upper limit specified by max_active_statements.
Solution
Increase the value of max_active_statements as user omm.
gs_guc reload -Z coordinator -N all -I all -c "max_active_statements=40"
