Filter out master-only nodes when sniffing

Fixes #106, thanks jolynch!

elasticsearch/transport.py

@@ -18,11 +18,19 @@ def get_host_info(node_info, host):
 
     Useful for filtering nodes (by proximity for example) or if additional
     information needs to be provided for the :class:`~elasticsearch.Connection`
-    class.
+    class. By default master only nodes are filtered out since they shouldn't
+    typically be used for API operations.
 
     :arg node_info: node information from `/_cluster/nodes`
     :arg host: connection information (host, port) extracted from the node info
     """
+    attrs = node_info.get('attributes', {})
+
+    # ignore master only nodes
+    if (attrs.get('data', 'true') == 'false' and
+        attrs.get('client', 'false') == 'false' and
+        attrs.get('master', 'true') == 'true'):
+        return None
     return host
 
 class Transport(object):

test_elasticsearch/test_transport.py

@@ -2,7 +2,7 @@
 from __future__ import unicode_literals
 import time
 
-from elasticsearch.transport import Transport
+from elasticsearch.transport import Transport, get_host_info
 from elasticsearch.connection import Connection
 from elasticsearch.exceptions import ConnectionError
 
@@ -37,6 +37,20 @@ def perform_request(self, *args, **kwargs):
     }
 }'''
 
+class TestHostsInfoCallback(TestCase):
+    def test_master_only_nodes_are_ignored(self):
+        nodes = [
+            {'attributes': {'data': 'false', 'client': 'true'}},
+            {'attributes': {'data': 'false'}},
+            {'attributes': {'data': 'false', 'master': 'true'}},
+            {'attributes': {'data': 'false', 'master': 'false'}},
+            {'attributes': {}},
+            {}
+        ]
+        chosen = [ i for i, node_info in enumerate(nodes) if get_host_info(node_info, i) is not None]
+        self.assertEquals([0, 3, 4, 5], chosen)
+
+
 class TestTransport(TestCase):
     def test_request_timeout_extracted_from_params_and_passed(self):
         t = Transport([{}], connection_class=DummyConnection)