list index out of range on multithreaded environment

Hi, first of all, thank you very much for elasticsearch, you are doing a very nice job. This is my first contribution to the community so please be nice with me.

I’m working on a multithreaded enviroment (using multiprocessing.pool.ThreadPool), and sometimes the following error appear:

File "/opt/clustering/wrappers/DAOElasticSearch/DAOElasticSearch.py", line 77, in indexDocument
    self.index(index = self.indexName, doc_type = self.docType, body = document, **kwargs)
File "/usr/local/lib/python2.7/site-packages/elasticsearch/client/utils.py", line 69, in _wrapped
    return func(*args, params=params, **kwargs)
File "/usr/local/lib/python2.7/site-packages/elasticsearch/client/__init__.py", line 261, in index
    _make_path(index, doc_type, id), params=params, body=body)
File "/usr/local/lib/python2.7/site-packages/elasticsearch/transport.py", line 326, in perform_request
    connection = self.get_connection()
File "/usr/local/lib/python2.7/site-packages/elasticsearch/transport.py", line 182, in get_connection
    return self.connection_pool.get_connection()
File "/usr/local/lib/python2.7/site-packages/elasticsearch/connection_pool.py", line 225, in get_connection
    return self.selector.select(self.connections)
File "/usr/local/lib/python2.7/site-packages/elasticsearch/connection_pool.py", line 66, in select
    return connections[self.rr]

Is this a problem with my code or is this a bug of elasticsearch client? Looking over /usr/local/lib/python2.7/site-packages/elasticsearch/connection_pool.py, I think that this can be solved using some kind of Lock() or Semaphore() protecting connections variable.

Thanks again, Victor.

[marcinn]

Maybe storing rr in threading.local() context in RoundRobinSelector will resolve the issue?

[honzakral]

Fixed via #539, thanks, @HusainZafar!