In nova, when exception occurs in MulticallWaiter scheduler the scheduler wont put the connection back to the pool, preventing Nova from running properly
set small pool size, rpc_conn_pool_size=1
cause rpc response timeout ("Timed out waiting for RPC response" in log)
then when rpc.call next time, it will wait forever to get connection from pool
** Impact **
In nova, when exception occurs in MulticallWaiter scheduler the scheduler wont put the connection back to the pool, preventing Nova from running properly
** Development Fix **
This has been addressed in https:/ /review. openstack. org/6804 and fixed in quantal.
** Stable Fix **
This has been addressed in https:/ /review. openstack. org/6833
** Test Case **
set small pool size, rpc_conn_ pool_size= 1
cause rpc response timeout ("Timed out waiting for RPC response" in log)
then when rpc.call next time, it will wait forever to get connection from pool
** Regression Potental **
Minimal this is an edge corner case.