Sending too many IDs in elasticsearch-py mtermvectors

(Patrick Lam) #1

Cross-posted on SO:

I'm trying to use mtermvectors() in the python elasticsearch-py client to retrieve term vectors for a large number of documents. I'm running into an error when trying to request too many documents. For example, if I request termvectors for 200 documents:

>> es.mtermvectors(index = index_name, doc_type = index_type, ids = docs.keys()[0:200])

Traceback (most recent call last):
  File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/elasticsearch/connection/", line 74, in perform_request
    response = self.pool.urlopen(method, url, body, retries=False, headers=self.headers, **kw)
  File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/urllib3/", line 607, in urlopen
  File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/urllib3/util/", line 222, in increment
    raise six.reraise(type(error), error, _stacktrace)
  File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/urllib3/", line 557, in urlopen
    body=body, headers=headers)
  File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/urllib3/", line 374, in _make_request
    httplib_response = conn.getresponse(buffering=True)
  File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/", line 1132, in getresponse
  File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/", line 453, in begin
    version, status, reason = self._read_status()
  File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/", line 417, in _read_status
    raise BadStatusLine(line)
ProtocolError: ('Connection aborted.', BadStatusLine("''",))
ConnectionError                           Traceback (most recent call last)
<ipython-input-172-74371d038a71> in <module>()
      1 st = time.time()
----> 2 tv = es.mtermvectors(index = index_name, doc_type = index_type, ids = docs.keys()[0:200])
      3 time.time()-st

/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/elasticsearch/client/utils.pyc in _wrapped(*args, **kwargs)
     67                 if p in kwargs:
     68                     params[p] = kwargs.pop(p)
---> 69             return func(*args, params=params, **kwargs)
     70         return _wrapped
     71     return _wrapper

/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/elasticsearch/client/__init__.pyc in mtermvectors(self, index, doc_type, body, params)
   1100         """
   1101         _, data = self.transport.perform_request('GET', _make_path(index,
-> 1102             doc_type, '_mtermvectors'), params=params, body=body)
   1103         return data

/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/elasticsearch/transport.pyc in perform_request(self, method, url, params, body)
    306             try:
--> 307                 status, headers, data = connection.perform_request(method, url, params, body, ignore=ignore, timeout=timeout)
    309             except TransportError as e:

/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/elasticsearch/connection/http_urllib3.pyc in perform_request(self, method, url, params, body, timeout, ignore)
     83         except Exception as e:
     84             self.log_request_fail(method, full_url, body, time.time() - start, exception=e)
---> 85             raise ConnectionError('N/A', str(e), e)
     87         if not (200 <= response.status < 300) and response.status not in ignore:

ConnectionError: ConnectionError(('Connection aborted.', BadStatusLine("''",))) caused by: ProtocolError(('Connection aborted.', BadStatusLine("''",)))

I did some testing and it seems to be that the problem is that I'm sending too many document IDs in the request and somehow there is a limit. For example, in the above example, it works when I request 193 documents, but breaks down when I request >= 194 documents. The IDs here are 18 characters long. In another test, I used shorter IDs and was able to request a larger number of documents before it broke down.

Is there a way around this problem or am I limited in the number of document termvectors I can request at once?

(Honza Král) #2

It looks like you are hitting the maximum allowed URL size. Unfortunately that is the limitation of the protocol and many libraries. The work-around here is to pass in the ids in the body instead:

es.mtermvectors(index=index_name, doc_type=index_type, body={'ids': docs.keys()})

Hope this helps!

(system) #3