tests: Use waitfornewblock for work queue test in interface_rpc

The work queue exceeded test in interface_rpc.py would repeatedly call
an RPC until the error was achieved. However hitting this error is
dependent on the processing speed of the computer and the optimization
level of the binary. Configurations that result in slower processing
would result in the RPC used being processed before the error could be
hit, resulting the test's runtime having a high variance.

Switching the RPC to waitfornewblock allows it to run in a much more
consistent time that is still fairly fast. waitfornewblock forces
the RPC server to allocate a thread and wait, occupying a spot in the
work queue. This is perfect for this test because the slower the RPC,
the more likely we will achieve the race condition necessary to pass the
test. Using a timeout of 500 ms appears to work reliably without causing
the test to take too long.

1 files changed, 1 insertions, 1 deletions

diff --git a/test/functional/interface_rpc.py b/test/functional/interface_rpc.py
index 48082f3a17..3389746635 100755
--- a/test/functional/interface_rpc.py
+++ b/test/functional/interface_rpc.py
@@ -25,7 +25,7 @@ def expect_http_status(expected_http_status, expected_rpc_code,
 def test_work_queue_getblock(node, got_exceeded_error):
     while not got_exceeded_error:
         try:
-            node.cli('getrpcinfo').send_cli()
+            node.cli("waitfornewblock", "500").send_cli()
         except subprocess.CalledProcessError as e:
             assert_equal(e.output, 'error: Server response: Work queue depth exceeded\n')
             got_exceeded_error.append(True)