hosts number check failed on ambari 2.5

[zhangyyun]

We are trying to get presto managed with ambari 2.5 recently. We find presto_coordinator starting failed with traceback below:

Traceback (most recent call last): File "/var/lib/ambari-agent/cache/stacks/HDP/2.5/services/PRESTO/package/scripts/presto_coordinator.py", line 91, in Coordinator().execute() File "/usr/lib/ambari-agent/lib/resource_management/libraries/script/script.py", line 375, in execute method(env) File "/var/lib/ambari-agent/cache/stacks/HDP/2.5/services/PRESTO/package/scripts/presto_coordinator.py", line 47, in start smoketest_presto(PrestoClient('localhost', 'root', config_properties['http-server.http.port']), all_hosts) File "/var/lib/ambari-agent/cache/stacks/HDP/2.5/services/PRESTO/package/scripts/presto_client.py", line 41, in smoketest_presto ensure_nodes_are_up(client, all_hosts) File "/var/lib/ambari-agent/cache/stacks/HDP/2.5/services/PRESTO/package/scripts/presto_client.py", line 105, in ensure_nodes_are_up nodes_returned_from_presto, all_hosts)) RuntimeError: Number of hosts returned from Presto [u'8fcebf71-1a27-4970-bedb-afe378792c0b'] do not equal the number of hosts specified by user ['hwslave002', 'hwslave002']

After digging into the presto_coordinator.py, the same host exists in both host_info['presto_worker_hosts'] and host_info['presto_coordinator_hosts'], leads to the error print.

It may be caused by switching on node-scheduler.include-coordinator in ambari, but I stopped here.